Ulrich Oron, Utrecht

Transcription

1

2 Cover illustration: Cover design and lay-out inside work: Photography: Printed by: Ulrich Oron, Utrecht Francis te Nijenhuis, zonnezijn creaties, s Hertogenbosch Lisa Terry photograpy, Utrecht Optima Grafisch Communicatie, Rotterdam The work presented in this thesis was performed at the Department of Clinical Pharmacy of the University Medical Center Utrecht in affiliation with the Patient Safety Center of the University Medical Center Utrecht and the Division of Pharmacoepidemiology & Clinical Pharmacology of the Utrecht Institute for Pharmaceutical Sciences (UIPS), Faculty of Science, Utrecht University. Parts of this thesis were performed in collaboration with the following institutes: Nederlandse Vereniging van Ziekenhuisapothekers (NVZA), Den Haag De Orde van Medisch Specialisten (OMS), Utrecht Koninklijke Nederlandse Maatschappij ter Bevordering der Pharmacie (KNMP), Den Haag Kring-apotheek B.V., s-hertogenbosch SIR Institute for Pharmacy Practice and Policy, Leiden The PHARM-study presented in this thesis was part of a larger study program, entitled Implementation of interventions for preventing adverse drug events in high risk patient populations in primary care and care institutions, by a team of doctors and hospital pharmacists. This study program was financially supported by the Patient Safety Program of the Netherlands Organisation for Health research and Development (ZonMw) (file number ). CIP-gegevens Koninklijke Bibliotheek, Den Haag: Leendertse, A.J. Hospital Admissions Related to Medication: prevalence, provocation and prevention Thesis Utrecht University with ref. with summary in Dutch ISBN/EAN: Anne J. Leendertse For articles published or accepted for publication, the copyright has been transferred to the respective publisher. No part of thesis thesis may be reproduces, stored in a retrieval system, or transmitted in any form or by any means without the permission of the author, or when appropriate, the publisher of the manuscript.

3 Hospital Admissions Related to Medication: Prevalence, Provocation and Prevention Geneesmiddel gerelateerde ziekenhuisopnames: prevalentie, provocatie en preventie (met een samenvatting in het Nederlands) Proefschrift ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof.dr. J.C. Stoof, ingevolge het besluit van het college voor promoties in het openbaar te verdedigen op donderdag 30 september des middags te 2.30 uur door Anne Josselien Leendertse geboren op 17 mei 1971 te Amsterdam

4 P rom otor e n: Prof.dr. A.C.G. Egberts Prof.dr. J.J. de Gier C o-promotoren Dr. P.M.L.A. van den Bemt Dr. G.H.P. de Koning Dr. A.N. Goudswaard

5 Voor mijn ouders Jet en Matthijs, Voor Ulrich en Lennart Gelukkig is de mens die wijsheid vindt, de mens die inzicht verkrijgt, want men kan beter inzicht verwerven dan zilver, beter wijsheid winnen dan goud. Spreuken 3:13-14

6

7 Chapter 1 Introduction 9 Chapter 2 Hospital Admissions Related to Medication The relationship between study characteristics and the prevalence of medication-related hospitalizations: a literature review and novel analysis 2.2 Frequency of and risk factors for preventable medication-related hospital admissions in the Netherlands 2.3 The contribution of renal impairment to preventable medication-related hospital admissions 2.4 Preventable Hospital Admissions Related to Medication (HARM): Cost analysis of the HARM-study Chapter 3 Preventing Hospital Admissions by Reviewing Medication Preventing Hospital Admissions by Reviewing medication (PHARM) in primary care: design of the cluster randomised, controlled, multicentre PHARM-study 3.2 The effect of a pharmaceutical care process intervention on medication related hospital admissions in the elderly in an integrated primary care setting: results of the PHARM study Chapter 4 General discussion

8 Chapter 5 From Summary to About the author Summary Samenvatting Dankwoord List of co-authors List of publications About the Author 211

9

10 Introduction Medication is one of the most commonly applied medical interventions in health care. It has shown to improve health and quality of life and to increase life expectancy. 1 Many people are prescribed multiple and long-term medication. In the Netherlands 11.3 million patients, 68% of all inhabitants, were prescribed and dispensed at least one drug in the second half of The vast majority (76%) of all prescriptions are repeat prescriptions, showing that most medication is prescribed for long-term use. In the elderly medication consumption is even higher: three times higher than average in people over 65 and five times higher in people over Although on a population level the benefit-risk balance is positive for the approved indication, population and setting, for the individual patient it may be that the adverse effects outweigh the benefits, even at normal doses and at normal use. These risks can be considerable: iatrogenesis has been estimated as one of the leading causes of death. 3,4 Adverse effects of medication can be severe and may lead to hospital admissions or death. 5 From several studies outside the Netherlands we know that this burden is considerable. For example in the United Kingdom 6.5% of all admissions are found to be related to an adverse drug event. The complexity and toxicity of modern pharmacotherapy is blamed for these high numbers of hospital admissions, although the majority of these admissions related to medication are caused by well-established medicines, and crop up with depressing regularity in studies: medication such as aspirin, warfarin, non-steroidal anti-inflammatory drugs and diuretics. 5 By combining and extrapolating the data from foreign studies the number of these hospital admissions would approximately be yearly in the Netherlands. 6 The extrapolation was based on computing a mean prevalence of all studies combined (irrespective of the definition of events, methodology, patient population and setting) which might not be a valid estimate of the true number of medication-related hospital admissions in the Netherlands. Adverse effects of medication may be caused by either the product itself, an adverse drug reaction, or by a human factor: a medication error. 7,8 The notion of this human factor has grown in recent years, especially since the publication of To Err Is Human by the Institute of Medicine. 9 This report showed that many human errors, including medication errors, occur within health care systems and that these contribute substantially to the occurrence of adverse events in patients. Medication errors may occur during prescribing, dispensing, administration and during evaluation of drug therapy 10 and can therefore be divided in prescribing, 10

11 Introduc tion Chapter 1 transcribing, dispensing, administration and monitoring errors. 11 In many studies medication errors have been assessed but the information is limited because these often focus on specific errors, such as prescribing, 12,13 dispensing 14 or administration errors, 15 rather than on the whole process from prescribing till use and evaluation of pharmacotherapy. Consequently, solutions to medication errors have usually concentrated on just one part of the process, 16 or on a specific error, such as prescribing, 17 dispensing and administration errors 18 or one specific administration error: i.e. non-adherence. 19 Some patients are more prone to adverse drug events (including those due to medication errors) than others. Especially elderly patients are at risk because of the changed pharmacokinetics and pharmacodynamics of many drugs, frailty syndrome, impaired cognition, vision and hearing, and their high risk of depression. 6,20 23 Elderly patients also more frequently suffer from multi morbidities, use more medication, and are treated by a larger number of health care professionals. 23 Unfortunately there is limited knowledge as to the pathophysiology of this group of patients, as well as the effectiveness of drug therapy when it concerns multi morbidity and old age. 24 These factors hamper the provision of effective and safe drug therapy to this group of patients and therefore need special attention. 25 Given the incomplete knowledge about the outcomes of drug therapy on the level of the individual patient as well as on a population level, studies to improve the use and effects of drug therapy in daily clinical practice are warranted. Several methods and instruments have been developed to improve patient outcomes, prevent adverse drug events and medication errors, like quality management systems. These have been developed to identify the underlying healthcare system defects in which an error could occur and analyses possible improvements of the system to prevent future errors. 26 Also clinical guidelines, protocols and formularies have been developed to help clinicians select the safest and most effective medication. Clinical decision support systems support the prescribing and dispensing practices, according to these clinical guidelines, protocols and formularies. This can for example avoid the use of medication that might be contra-indicated or that interacts with other medication. 27,28 Finally dose packaging and drug distribution systems have been implemented to reduce the risk of wrong medication and wrong dose in patients. 29 These methods and instruments contain knowledge on diseases and medication which are developed to improve outcomes and safety of drug therapy for defined groups of patients, while in practice medication is used in a different context of co- 11

12 morbidity and polypharmacy. Since patients are individuals and patients responses are unpredictable, each patient is the subject of an experiment (n=1). Therefore the medication of the patient needs to be validated against agreed expectations. Medication use is altered to individualized treatment, as necessary to optimize safety and effectiveness. The decision options at first are those that are within a normal protocol locally agreed best practice, according to the clinical guidelines, within the known clinical evidence base. But thereafter in some patients the decision-making may be taken outside what might be accepted as the normal protocol. Therefore unusual drug choices, combinations and doses may be necessary. This patient-centered approach to improve outcomes and safety of drug therapy, can be found in the concept of pharmaceutical care, 30 which emerged from the concept of clinical pharmacy. The process of pharmaceutical care can be seen as a clinical risk management cycle 30 for an individual patient. This pharmaceutical care process is continuous and comparable to models used in quality management, like the Deming cycle (see Figure 1). 31 Within the pharmaceutical care process information about the patient is gathered from the patient in a pharmaceutical anamnesis, from the medical and the medication history. With this information the patient s needs, drug therapy related problems and potential drug therapy problems are identified by a pharmacotherapy review. The management of these problems is implemented by the pharmaceutical care plan involving the patient and the clinicians. The care plan is followed up, Figure 1 Pharmaceutical care process model 12

13 Introduc tion Chapter 1 monitored and evaluated by the patient outcomes. 32 The care plan takes the form of a quality management cycle. In medication use patients needs are met from the choice of possible treatment options. Setting treatment expectations precedes the design of the plan. The follow up assessment of potential good or harm achieved requires documented clinical monitoring. Early detection of treatment success or failure is important. Patient monitoring allows the plan to be adjusted (individualized dose or switch of common combinations) or modified in some additional way. At some point the treatment will be judged to have been optimized and evidence sought to confirm the medication to be doing good or responsible for harm. The overall medical assessment will be informed by the pharmaceutical care process and help the decision of whether the medication be continued or abandoned. Abandonment of the use of the medication leads the plan to need a complete review of the patients condition and a revision of the expectations from medication use. Continuation leads to continued monitoring to manage the situation in order to maintain success or respond to change. The pharmaceutical care process is performed in daily practice in which the doctor takes ultimate care of patients and the pharmacist is responsible for managing the patient s drug therapy. 33 In doing so pharmacists co-operate with doctors, patients and carers in designing, implementing and monitoring the pharmaceutical care plan. The pharmaceutical care process has the potential to reduce medication errors, drug therapy related problems, and to improve patients quality of life. 32 Several studies have been conducted studying the effect of parts of the pharmaceutical care process, especially on the pharmacotherapy review part. However, most published randomised controlled trials on pharmacotherapy reviews showed no or little effect on morbidity and mortality. 34,35 These studies differed largely with respect to the nature and extensiveness of the review techniques, the outcomes studied, setting and follow-up time and results are therefore difficult to compare. One study reported an increase in emergency readmissions, 36 while in the other studies there is no suggestion that patients were harmed by the interventions, 37,38 and even some consistency in suggesting that falls 39 and hospital admissions 40,41 might be reduced. Potentially relevant elements in these studies were a review of a full available medical and medication history, structured pharmaceutical care plan approach, combined effort of pharmacist and GP and involvement and commitment of the patient. Evidence on the effect of the whole pharmaceutical care process on patient outcomes and medication safety is limited. Therefore we need more information on the effect of the pharmaceutical care process as an ongoing iterative process, which can be performed in daily practice. 13

14 Objective Patients, clinicians and health policy makers are requesting more information on the burden of medication errors. Moreover, there is an urge for an effective method to prevent these errors and to improve patient outcomes. The focus in this thesis will be on medication errors, which may lead to hospital admissions. An effective method to reduce these admissions needs to be developed, based on information showing, which patients are at risk, which medication is hazardous and which medication errors cause hospital admissions. The objective of this thesis is therefore to gain more information on medication-related hospital admissions and on how to prevent these admissions and to improve patient outcomes. Ou tline The thesis consists of two parts. In the first part studies will be presented about the prevalence and the burden of medication-related hospital admissions (the HARM study); which patients are at risk, what kind of medication is hazardous and which types of medication errors are associated with these admissions. In Chapter 2.1 a review of literature is presented that shows the influence of various study characteristics on the prevalence of medication-related hospital admissions. In Chapter 2.2 the prevalence of medication-related hospital admissions in the Netherlands is studied. Also details are presented on the type of medication, type of medication error and risk factors associated with medication-related hospital admissions. In Chapter 2.3 more information is provided on the costs associated with the medication-related hospital admissions. Chapter 2.4 focuses on renal impairment as a cause of medication-related hospital admissions. More information is presented about medication errors in renal impairment that are associated with these admissions. Based upon results from the first part of the thesis an intervention was designed to prevent medication-related hospital admissions: the pharmaceutical care process (the PHARM study). This process is patient centered and performed periodically in daily practice. Chapter 3.1 describes the method, which was used to prevent medication-related hospital admissions in an integrated primary care setting. This chapter also provides information on the magnitude of the population at risk in primary care. 14

15 Introduc tion Chapter 1 In Chapter 3.2 the results of the study on the effect of the pharmaceutical care process as described in the previous chapter is presented. The effect was studied on 364 patients in the intervention group and 310 patients in the control group in 42 different integrated primary care settings. In the Appendix to Chapter 3.2 the discrepancy between the intended and actual number of included patients is discussed. Finally, the results of these studies are summarized and put into a broader perspective in the discussion presented in Chapter 4. This perspective also embodies a description of implications and recommendations of our studies for medical and pharmacy practice in primary care. References Schnittker J, Karandinos G. Methuselah s medicine: Pharmaceutical innovation and mortality in the United States, Soc Sci Med. 2010;70: Stichting Farmaceutische Kengetallen (SFK). Data en Feiten Den Haag; August Available from: http// [Accessed Dec 2009] Starfield B. Is US health really the best in the world? JAMA. 2000;284: Steel K, Gertman PM, Crescenzi C, Anderson J. Iatrogenic illness on a general medical service at a university hospital. N Engl J Med. 1981;304: Pirmohamed M, James S, Meakin S, et al. Adverse drug reactions as cause of admission to hospital: prospective analysis of patients. BMJ. 2004;329: Beijer JHM, De Blaey CJ. Hospitalisations caused by adverse drug reactions (ADR): a metaanalysis of observational studies. Pharm World Sci. 2002;24: Edwards IR, Aronson JK. Adverse drug reactions: definitions, diagnosis, and management. Lancet. 2000;356: Nebeker JR, Barach P, Samore MH. Clarifying adverse drug events: a clinician s guide to terminology, documentation, and reporting. Ann Intern Med. 2004;140: Kohn L, Corrigan J, Donaldson M, Committee on Quality of Health Care in America, Institute of Medicine. To err is human: building a safer health system. Washington, DC: National Academy Press; Garfield S, Barber N, Walley P, Willson A, Eliasson L. Quality of medication use in primary care--mapping the problem, working to a solution: a systematic review of the literature. BMC Med. 2009;7: Van den Bemt PMLA, Egberts ACG. Geneesmiddelgerelateerde problemen gedefinieerd en geclassificeerd. Bijwerkingen en medicatiefouten systematisch ingedeeld. Pharm Weekbl. 2002;137: Buurma H, De Smet PA, Egberts AC. Clinical risk management in Dutch community pharmacies: the case of drug-drug interactions. Drug Saf. 2006;29:

16 13. Bates DW, Cullen DJ, Laird N, et al. Incidence of adverse drug events and potential adverse drug events: Implications for prevention. JAMA.1995;274: Dean Franklin B, O Grady K. Dispensing errors in community pharmacy: frequency, clinical significance and potential impact of authentication at the point of dispensing. Int J Pharmacy Practice. 2007;15: Beardon PHG, McGilchrist MM, McKendrick AD, et al. Primary non-compliance with prescribed medication in primary care. BMJ. 1993;307: Karapinar-Carkit F, Borgsteede SD, Zoer J, et al. Effect of medication reconciliation with and without patient counseling on the number of pharmaceutical interventions among patients discharged from the hospital. Ann Pharmacother. 2009;43: Van Doormaal JE, van den Bemt PM, Zaal RJ, et al.the influence that electronic prescribing has on medication errors and preventable adverse drug events: an interrupted time-series study. J Am Med Inform Assoc. 2009;16: Idzinga JC, de Jong AL, van den Bemt PM. The effect of an intervention aimed at reducing errors when administering medication through enteral feeding tubes in an institution for individuals with intellectual disability. J Intellect Disabil Res. 2009;53: World Health Organization. Adherence to long-term therapies: evidence for action. Geneva, Switzerland: World Health Organization; Mangoni AA, Jackson SH. Age-related changes in pharmacokinetics and pharmacodynamics: basic principles and practical applications. Br J Clin Pharmacol. 2004;57: Frazier SC. Health outcomes and polypharmacy in elderly individuals: an integrated literature review. J Gerontol Nurs. 2005;31: Beers MH, Ouslander JG. Risk factors in geriatric drug prescribing. A practical guide to avoiding problems. Drugs. 1989;37: Higashi T, Shekelle PG, Solomon DH, et al. The quality of pharmacologic care for vulnerable older patients. Ann Intern Med 2004;140: Boyd CM, Darer J, Boult C, Fried LP, Boult L, Wu AW. Clinical practice guidelines and quality of care for older patients with multiple comorbid diseases: implications for pay for performance. JAMA. 2005;294: ZonMw. Programma priority-medicines-voor-ouderen [Internet]. Den Haag: ZonMw; July Available from: [Accessed 24 May 2010]. 26. Nolan TW. System changes to improve patient safety. BMJ. 2000;320: Grimshaw JM, Russell IT. Effect of clinical guidelines on medical practice: a systematic review of rigorous evaluations. Lancet. 1993;342: Buurma H. Clinical risk management in community pharmacy [dissertation]. Utrecht: Utrecht University; Barker KN, Pearson RE, Hepler CD, Smith WE, Pappas CA. Effect of an automated bedside dispensing machine on medication errors. Am J Hosp Pharm. 1984;41: Hepler CD. Pharmaceutical care. Pharm World Sci. 1996;18: Edwards Deming W. Out of the Crisis. Cambridge, Mass. : Massachusetts Institute of Technology, Center for Advanced Engineering Study;

17 Introduc tion Chapter Cipolle R, Strand LM, Morley PC. Pharmaceutical Care Practice. New York: McGraw Hill; Hepler CD, Strand LM. Opportunities and responsibilities in pharmaceutical care. Am J Hosp Pharm. 1990;47: Holland R, Desborough J, Goodyer L, Hall S, Wright D, Loke YK. Does pharmacist-led medication review help to reduce hospital admissions and deaths in older people? A systematic review and meta-analysis. Br J Clin Pharmacol. 2008;65: Zermansky AG, Silcock J. Is medication review by primary-care pharmacists for older people cost effective?: a narrative review of the literature, focusing on costs and benefits. Pharmacoeconomics. 2009;27: Holland R, Lenaghan E, Harvey I, et al. Does home based medication review keep older people out of hospital? The HOMER randomised controlled trial. BMJ. 2005; 330: Richmond S, MortonV, Cross B, et al.; RESPECT trial team. Effectiveness of shared pharmaceutical care for older patients: RESPECT trial findings. Br J Gen Pract. 2010;59: Krska J, Cromarty JA, Arris F, et al. Pharmacist-led medication review in patients over 65: a randomized, controlled trial in primary care. Age Ageing. 2001;30: Zermansky AG, Alldred DP, Petty DR, et al. Clinical medication review by a pharmacist of elderly people living in care homes: randomised controlled trial. Age Ageing. 2006;35: Schnipper JL, Kirwin JL, Cotugno MC, et al. Role of pharmacist counseling in preventing adverse drug events after hospitalization. Arch Intern Med. 2006;166: Stewart S, Pearson S, Luke CG, Horowitz JD. Effects of home-based intervention on unplanned readmissions and out-of-hospital deaths. J Am Geriatr Soc. 1998;46:

18

19

20

21

22 Abstract Background Studies on medication-related hospitalizations differ in study setting, studied population, outcome, and method of data collection. Thus, extrapolations based on a meta-analysis of unselected studies may be biased. Objective To explore the influence of study characteristics on the prevalence of medicationrelated hospitalizations. Methods After a structured literature search, the retrieved studies were categorized based on the following aspects: (1) study setting (e.g. all hospital admissions vs. only acute hospital admissions); (2) study population (e.g. an entire hospital, study ward(s), selected population and/or age group); (3) outcome of medication-related problem (e.g. adverse drug reaction [ADR] vs. adverse drug event [ADE]); (4) method of data collection (e.g. medical chart review, spontaneous reporting or database research); and (5) continent in which the study took place (only for studies looking at all acute admissions). We then examined the relationship between these factors and reported prevalence of medication-related hospital admissions. Resu lts Ninety-five studies were analysed, with a range of reported prevalence of medication-related hospitalizations from 0.1% to 54%. Higher prevalences were found in the studies examining all hospital admissions than in the studies examining only acute hospital admissions. In addition, higher prevalences were found in the elderly population than in children. As would be expected, higher prevalences were also found in studies examining ADEs than in studies examining only ADRs. With respect to the method of data collection, medical chart screening resulted in higher prevalences of medication-related hospitalizations than database methods or spontaneous reporting. Combined studies in Europe show lower prevalences of medication-related hospital admissions than in other continents included in the study. Discussion The reported prevalence of medication-related hospital admissions varies as a function of the setting (all admissions or only acute admissions), studied population (entire hospital, specific wards, selected population and age group), outcome 22

23 Study charac teristics and medication- related hospitalizations Chapter 2.1 (ADR/ADE), the method of data collection and the continent in which the study is performed. C onclusion Extrapolation using national hospital admission data and the prevalence identified by pooling international studies should be carried out with great caution. 23

24 Background Medication safety, which is part of the broader context of patient safety, is considered an essential element of high-quality healthcare systems. This notion has gained recognition in recent years, especially since the publication of the report To err is human by the Institute of Medicine in the USA. 1 This report fuelled social awareness and pressure to improve patient safety worldwide. Data on patient and medication safety mainly originate from English speaking countries, which are often extrapolated to other countries and settings to estimate the burden of medication-related harm. An example of such an extrapolation is the meta-analysis of Beijer and de Blaey 2 regarding the prevalence of medicationrelated hospitalizations. Based upon the findings of previously published studies, they estimated that 4.8% of all hospital admissions were medication related, of which 29% were preventable. These findings were extrapolated to the Netherlands, taking age into account, and resulted in an estimated number of medicationrelated hospitalizations annually, of which almost admissions in the elderly were considered to be preventable. These figures have often been cited in the Dutch lay press and have resulted in political attention. However, after publication of the review by Beijer and de Blaey, 2 three Dutch studies were published that specifically investigated the prevalence of medication-related hospital admissions in the Netherlands. 3 5 The reported prevalences (1.8%, 3.5% and 2.4% in the three studies, respectively) of all medication-related hospital admissions were lower than the point estimate reported in the meta-analysis of Beijer and de Blaey. 2 One explanation for the differences seen may be due to actual differences in healthcare systems in the USA, the UK and the Netherlands. Another reason may be that the published studies on this topic differ in certain key aspects that were examined, such as setting (e.g. entire hospital vs. specific wards), type of admissions (e.g. all hospitalizations or only acute admissions), study population (e.g. specific wards), outcome (e.g. focus on adverse drug reaction [ADR] vs. adverse drug event [ADE]), data collection method (e.g. clinical coding database, medical record screening or spontaneous reporting) and continent in which the study was performed. Therefore, the objective of this study is to explore the relationship between certain study factors and the prevalence estimate of medication-related hospital admissions. 24

25 Study charac teristics and medication- related hospitalizations Chapter 2.1 Methods Data S ources and Study S election Data for this review came from two sources. First, all studies included in the metaanalysis of Beijer and de Blaey 2 were included (these comprised studies performed up to 2001 and were counted as studies identified by screening reference lists). For retrieving studies performed since 2001, a PubMed search was performed using the search terms hospitalization or hospital admission or admission to hospital and medication, with the exclusion of search terms drug abuse and alcohol abuse. Other limits in this search were English language and publication date from April 2001 until March Additionally, the reference lists of all retrieved publications (both from the meta-analysis and from the PubMed search) were screened for additional eligible studies. Studies were further selected from the above sources if they provided data that estimated the prevalence of medication-related hospitalizations of outpatients and if they were written in the English language. Studies were excluded if they addressed only specific types of ADEs (e.g. blood dyscrasias or cardiovascular events), ADEs in patients with a particular disease (e.g. heart failure patients or depressive patients), emergency visits without hospitalization, hospitalization after the use of one specific drug (e.g. digoxin or chemotherapy), ADEs without an evaluation of the relationship to admission or ADEs during admission. Data Extraction and Definitions For each included study the following data were extracted: bibliographic characteristics (year of publication, first author), characteristics of the study setting and population (country, hospital, study period, included admissions and patient age group), the outcome focus (ADE or ADR), methods of data collection (medical chart review, database method or spontaneous reporting method) and continent in which the study was performed. Every paper was studied for these characteristics and classified according to the definitions as stated below. The study setting was defined as the location where the study participants were included either admission to the entire hospital or specific categories of the hospitalized population. The first category of admission to the entire hospital was further subdivided into all admissions or only acute admissions. An acute admission was defined as any unscheduled admission to a hospital. All admissions were defined as a combination of planned admissions and acute admissions. The second specific category of the hospitalized population was subdivided into admission to one or more specific wards (internal medicine, cardiology, gastroenterology, respiratory medicine, medical wards, paediatrics, paediatric 25

26 oncology, neonatology, psychiatry, care of the elderly), a specific patient population (selection of hospital admissions of patients with polypharmacy or medicationrelated problems) or age group of patients (all ages, adults, children, elderly [defined by the authors or a cut-off of 65 years or older]). An ADE was defined as medication-related harm, caused by either an ADR or a medication error. ADEs caused by medication errors are potentially preventable. 6 An ADR is defined by the WHO as a response to a drug which is noxious and unintended and which occurs at doses normally used in man for the prophylaxis, diagnosis or therapy of disease, or for the modification of physiological functions. 7 These definitions were used to identify the exact outcome of the study. When any mention was made on preventability in the study then the outcome was defined as ADE (even when the authors themselves defined the outcome as ADR). Intentional and unintentional overdoses were classified as medication errors and thus the harm they caused was classified as an ADE. A database method was defined as a method of data collection in which the medication-related admissions are identified from a database of hospital discharge diagnoses. A spontaneous reporting method was defined as a method of data collection in which the medication-related hospitalizations are identified from doctor and/or nurse reports of adverse events. Medical chart review was defined as the third method of data collection in which the medication-related hospitalizations were identified from the patients medical record or case notes of the admission. This method could be retrospective or prospective. Finally, the following six geographical areas ( continents ) were defined: Australasia (Australia and New Zealand), North America (Canada and the US), Europe, Middle and South America, Asia and Africa. To better compare studies from these continents we only included the studies that examined the outcome (ADRs or ADEs) for all acute admissions and that used medical chart review as the method. Data Analysis From the extracted data, the prevalence of medication-related hospital admissions was calculated by dividing the number of patients admitted to the hospital because of a medication-related event (the numerator) by the number of patients admitted to the hospital within the study period and within the study setting (the denominator). The included studies were stratified with respect to setting (included admissions), study population (subdivided into study ward, selected population and age group), outcome focus and method of data collection. For all strata, the weighted mean prevalences were calculated using the inverse variance method 8 in 26

27 Study charac teristics and medication- related hospitalizations Chapter 2.1 Figure 1 Study design and numbers of included and excluded studies a) Cross references from other studies retrieved from the electronic search and from the meta-analysis of Beijer and de Blaey. 2 which the studies are weighted by their standard error (SE), which is related to the population size. The SE is calculated using the formula ((p (1 p))/n), where p is the proportion of medication-related admissions of the studied hospital admissions and n is the total number of studied hospital admissions. The confidence interval is calculated using the formula prevalence ± (1.96 SE). 27

28 Table 1 No. of studies presented by setting, population, method of data collection and definition of outcome Studies ADR (n=36) ADE (n=59) Total (n=95) Setting Included admissions all hospital admissions all acute hospital admissions Population Studied ward or other population internal medicine cardiology gastroenterology respiratory medicine medical wards paediatrics all hospital admissions paediatrics acute hospital admissions paediatric oncology neonatology psychiatry care of the elderly intensive care unit others (polypharmacy, previous medicationrelated admission) Studied age group all ages or adults children elderly Continent of study Australasia North America Middle and South America Europe Asia Africa Method of data collection Medical chart review Database/computer screening Spontaneous reporting ADR = adverse drug reaction; ADE = adverse drug event 28

29 Study charac teristics and medication- related hospitalizations Chapter 2.1 Result s L iterature S earch Results The electronic search resulted in 17 articles; the study of the references of these articles and of several literature studies and the meta-analysis of Beijer and de Blaey 2 resulted in the identification of an additional 72 articles, resulting in 89 relevant articles (Figure 1). 3,4,5,9 94 Of these 89 articles, four 10,33,74,88 concerned a combination of two studies and one 31 concerned a combination of three studies. Thus, 95 studies were analysed (see Appendix I). Table 1 shows the number of studies subdivided by study setting, study population (further subdivided into study ward, selected population and age group), outcome focus, data collection method and continent. The prevalence estimate for medication-related hospital admissions varied in the identified studies from 0.1% to 54%. Also, a large variation was found in the number of included admissions in the denominator, from 41 admissions studied by medical chart review 86 to a clinical coding database study of almost 89 million admissions. 44 Pre valence Estimates All studies combined (irrespective of study setting, study population, outcome focus and data collection method) resulted in a very low weighted mean prevalence of medication-related hospitalizations of 0.46% (95%CI ). Study S etting The weighted mean prevalence of medication-related hospitalizations was calculated for studies looking at all hospital admissions and only acute hospital admissions as a denominator. Higher prevalences were found in the studies examining all hospital admissions (3.03% [95%CI ] for admissions related to ADRs, 9 12 and 5.35% [95%CI ] for admissions related to ADEs 4,40 42 ) than in studies only examining acute hospital admissions (1.14% [95%CI ] for admissions related to ADRs, 15 and 3.68% [95%CI ] for admissions related to ADEs 5,47 63 [Table 2]). Study Population Prevalences were also calculated for the different study populations found on different study wards. Studies examining ADEs on wards for care of the elderly (10.18% [95%CI ]) and psychiatric wards (23.05% [95%CI ] 85,86 including intentional overdoses) showed high prevalences, whereas relatively low prevalences were found in studies in paediatric wards (3.94% [95%CI ] including all admissions related to ADEs, 74,83 and 4.33% [95%CI

30 Table 2 Prevalence of drug-related hospital admissions with adverse drug reaction (ADR)/adverse drug event (ADE) as definition of outcome and various settings and studied populations (all chart review methods) a ADR ADE weighted mean prevalence no. of studies weighted mean prevalence no. of studies % (95% CI) % (95%CI) Setting Included admissions all admissions 3.03 ( ) 4 b 5.35 ( ) 4 b all acute admissions 1.14 ( ) 1 b 3.68 ( ) 18 Population Studied ward or other population internal medicine 2.67 ( ) ( ) 16 cardiology ( ) 1 0 gastroenterology ( ) 1 respiratory medicine ( ) 1 medical wards c ( ) 1 paediatrics all admissions 0.19 ( ) ( ) 2 paediatrics acute admissions 0.93 (CI not available) ( ) 1 paediatric oncology 21.7 ( ) 1 0 neonatology 0.20 ( ) 1 0 psychiatry 7.48 ( ) ( ) 2 care of the elderly ( ) ( ) 6 intensive care unit ( ) 1 polypharmacy (> 4) ( ) 1 previous admission for drug-related problem 0 b ( ) 1 30

31 Study charac teristics and medication- related hospitalizations Chapter 2.1 (Table 2 continued) ADR ADE weighted mean prevalence no. of studies weighted mean prevalence no. of studies % (95% CI) % (95%CI) Studied age group all ages or adults 1.79 ( ) 19 b 4.31 ( ) 43 b children 0.19 ( ) ( ) 3 elderly 3.60 ( ) 5 b ( ) 9 Continent of study setting: all acute admissions Australasia 0 d 8.09 ( ) 3 d North America 0 d ( ) 3 d Europe 1.10 ( ) 1 d 3.30 ( ) 11 d Asia 0 d 6.90 ( ) 1 d Africa 0 d 0 a) b) c) d) Only studies with similar study method, same denominator and same numerator compared. Studies with different study methods, such as spontaneous reporting or database method, were excluded. Admissions to geriatric, internal medicine, cardiology, respiratory and gastroenterology wards. Studies with different study methods, such as spontaneous reporting or database method, and studies with different settings, other than all acute admissions, were excluded. 31

32 5.07] including only acute admissions related to ADEs 84 ) [see Table 2]. This pattern of results was also seen when specific age groups were analysed: 3.60% (95%CI ) related to ADRs in the elderly 12,24,36 39 versus 0.19% (95%CI ) in children with ADR as the studied event, and 12.30% (95%CI ) related to ADEs in the elderly 48,49,54,87 91 versus 4.16% (95%CI ) in children with ADEs as the studied event 74,83,84 (Table 2). High prevalences were found in patients who had been admitted previously because of a medication-related problem (19.46% [95%CI ][93] and 17.71% [95%CI ][39]) or in patients with polypharmacy (five or more drugs; 6.63% [95%CI ][94]). Of the 95 studies, 59 examined ADEs and 36 examined ADRs. As would be expected, the weighted mean prevalence of studies that use ADE as an outcome was higher (see Study Setting section and Table 2. Method of Data C ollection All medical chart review studies combined (irrespective of study setting, study population and outcome) resulted in a low weighted mean prevalence of medicationrelated hospitalizations of 0.85% (95%CI ). 4,5,9 12,15 38,40 42,47 94 A number of studies use methods other than medical chart review to find cases of medicationrelated hospitalization. These kinds of studies also show low prevalences of %, 3,13,14,39,43 46 with the exception of a study that included only patients who had been previously admitted because of an ADR (17.71% [95%CI ]) 39 (Table 3). C ontinent Where the Study Took Place Studies conducted in the US, Europe or Australasia report similar prevalences of admission to hospital that are related to medication. The calculated weighted mean prevalence of studies in North America on ADEs and all acute admissions was 14.73% (95%CI ), 40,47,48 combined studies in Australasia was 8.09% (95%CI ) 49,50,54 and for combined studies in Europe was 3.30% (95%CI ). 5,51 62 Studies conducted in Europe therefore seem to have lower prevalences than studies in both Australasia and North America (see also Table 2). For Asia and Africa, too few studies were available. 32

33 Study charac teristics and medication- related hospitalizations Chapter 2.1 Table 3 Prevalence of drug-related hospital admissions with adverse drug reaction (ADR)/adverse drug event (ADE) as definition of outcome, and database methods of data collection and spontaneous reporting method of data collection a Data collection method ADR ADE weighted mean prevalence no. of studies weighted mean prevalence % (95%CI) % (95%CI) no. of studies Database Denominator: all admissions 0.14 ( ) ( ) 2 acute admissions 1.83 ( ) 1 previous admission for ADR ( ) 1 Spontaneous reporting Denominator: all admissions 0.70 ( ) ( ) 2 a) Only studies with a similar study method, the same denominator and the same numerator were compared. Discussion This literature review shows that in studies of medication-related hospitalizations a wide range of prevalences is reported. This variation can be explained by the differences in study setting, study population (and thus choice of denominator for prevalence calculation), outcome (ADRs or ADEs as the numerator for prevalence calculation) and data collection method. Furthermore, the variation may arise from differences in healthcare systems between countries. Therefore, it is not possible to combine all these different studies and present a representative prevalence of medication-related hospital admissions. Outcome, S ettings, Study Populations, Methods of Data C ollection and C ontinent In studies looking at ADEs, higher prevalences are found than in studies on ADRs. This higher percentage is to be expected because ADEs include not only ADR events but also harm caused by medication errors (i.e. the potentially preventable harm). Therefore, when studying medication-related hospitalizations, the broader focus of ADE as an outcome is to be recommended as the focus of the study. 33

34 An unexpected finding was the difference found in the prevalence of medicationrelated hospital admission when examined for all admissions versus only acute admissions. It is to be expected that ADEs and ADRs are acute problems and hospitalization due to these problems cannot be a scheduled admission. Therefore, all admissions should result in lower percentages of medication-related hospital admissions. Nevertheless, a higher percentage was found in this analysis. This can be partly explained by an outlier study on all admissions in a hospital in Iran by Zargarzadeh et al. 42 They found a large percentage of medication-related hospital admissions in the population (76%) due to literacy problems. This reason for medication-related problems is not widely recognized in other studies and therefore the study of Zargarzadeh et al. 42 in Iran might not be representative for developed countries. Also, Pirmohamed et al. 41 found a higher percentage than expected for all admissions. This could be explained by the fact that most patients in this study were admitted through either the accident and emergency department or the acute medical and surgical assessment units; in fact the admissions included were largely unplanned, which might have resulted in a higher proportion of medication-related hospital admissions. 41 This unexpected difference in mean prevalence disappears when the study of Zargarzadeh et al. 42 is excluded and the study of Pirmohamed et al. 41 is reclassified as acute admissions due to ADEs. The prevalence of all admissions due to ADEs decreases from 5.35% to 3.34% and the mean prevalence of only acute admissions due to ADEs increases from 3.68% to 4.31%. Thus, in this re-analysis, the mean prevalence of only acute admissions is higher than the mean prevalence of all admissions, which is in accordance with expectations. A higher prevalence was also found for the selection of specific study wards and for selection in the study population. Although the number of studies included was small, the identified prevalences can be regarded as a reflection of the different types of patients in the selected wards and their risk of an ADR or ADE. For example, elderly patients are regarded as high-risk patients because of their multiple drug use, which results in high prevalences in studies performed in an elderly population. 95,96 The high prevalences found on psychiatric wards 85,86 can be explained by the multiple drug use in psychiatric patients, together with their many drug-related problems, 97,98 and the inclusion of admissions due to drug abuse and intentional overdose in the included studies. Children, on the other hand, use few or no drugs, which results in low prevalences in studies performed in paediatric admissions or in children. 99,100 Yet, as stated before, these results need to be assessed with caution because of the relatively low number of studies per subcategory of patients. With respect to the method of data collection, (retrospective or prospective) medical chart screening results in higher prevalences than database methods or spontaneous 34

35 Study charac teristics and medication- related hospitalizations Chapter 2.1 reporting. It can be concluded that ADEs and ADRs are under-reported when using spontaneous reporting or database methods of data collection. Finally, when looking at the effect of the continent in which the study was performed, European studies seemed to show smaller prevalences than North American and Australasian studies. This may be the result of different healthcare systems in the different countries, but conclusions on this outcome need to be drawn with care because of the relative limited number of studies on ADEs using a medical chart review method investigating all acute admissions performed in Australasia and North America. C omparing Results The meta-analysis of Beijer and de Blaey 2 does make a correction for sample size of the study and also divides the studies into elderly versus younger study populations, but the choices of numerator (ADRs or ADEs) and denominator (all admissions, only acute admissions, specific wards) were not taken into account. Disregarding the influence of the denominator will lead to an overestimation of the prevalence of medication-related hospitalizations. Combining studies, regardless of whether they look at ADRs or ADEs as a numerator, could result in an underestimation of the prevalence. The results of the published, prospective, multicentre, HARM (Hospital Admissions Related to Medication) study 5 into medication-related hospitalizations in the Netherlands confirms that the meta-analysis of Beijer and de Blaey 2 overestimates the number of medication-related hospitalizations. Extrapolation of the results to the Dutch situation results in potentially preventable hospital admissions each year according to the HARM study, while Beijer and de Blaey 2 estimated this number to be as high as admissions. Lazarou et al. 101 studied only ADRs and state that the heterogeneity in their results is due to variation in the examined population. However, they did not study ADRs exclusively, since a few included articles also studied medication errors. Furthermore, despite their statement on heterogeneity, the estimated number of hospital admissions was not adjusted for ward type when extrapolated. By overrepresenting medical wards in the analysis and including admissions due to medication errors, the incidence and number of admissions related to ADRs are likely to be overestimated. Wiffen et al. 102 also studied ADRs and showed heterogeneity in the results. They identified a prevalence of 2.6% from predominantly North American studies, which was shown to be about one-half the prevalence in Europe and the UK. Therefore, extrapolation on the basis of North American studies can result in an 35

36 underestimation of the number of medication-related hospital admissions in Europe and the UK. Finally, Kongkaew et al. 103 analysed studies on hospital admissions associated with ADRs only. This systematic review suggests that approximately 5.3% of hospital admissions are associated with ADRs. An overall prevalence was calculated, while heterogeneity between the studies, especially higher rates for elderly patients, were observed. Therefore, applying this prevalence of 5.3% to total numbers of admissions (including non-elderly patients) can result in an overestimation of the number of medication-related hospital admissions. Lazarou et al., 101 Wiffen et al. 102 and Kongkaew et al. 103 studied only ADRs as a numerator, while in our analysis, studies with ADEs as a numerator were also included. Because of the broader focus of ADEs than ADRs, more medicationrelated admissions in the numerator are expected. Therefore, the extrapolation based on ADRs only can give an underestimation of the number of medicationrelated hospital admissions. Over- or underestimation may still be the case when applying the prevalences identified in our review. Even though we tried to group similar studies and separate those that are different, the definitions we used may still be too broad or the settings may not be comparable. Another limitation is the inclusion of studies limited to the English language. Notwithstanding our rigorous search method, we still may have missed some studies. However, to our knowledge this review is the first that systematically analyses the influence of choice of outcome, settings, study populations and methods of data collection on the results reported in studies on medication-related hospitalizations. This reveals important information for the interpretation as well as the design of future studies on this subject. Future studies should use clear definitions, should preferably study ADEs instead of ADRs, should include entire hospital populations instead of subpopulations, and should use the more reliable method of chart review. Furthermore, studies into the differences between various countries and healthcare systems are needed. C onclusion In conclusion, this study shows that the prevalences of medication-related hospitalizations reported in studies looking at ADEs and/or specific wards such as those for care of the elderly and psychiatric patients are higher than in studies investigating ADRs as the numerator and/or all admissions as the denominator. 36

37 Study charac teristics and medication- related hospitalizations Chapter 2.1 Furthermore, the prevalences of medication-related hospitalizations are lower in studies using other methods than medical chart review for the prevalence calculation. As prevalences of hospital admissions related to medication depend on setting and focus of the outcome, extrapolation of these prevalences using local hospitalization data should be carried out with great caution. Acknowledgement - The authors would like to thank Pamela M. Kato, PhD EdM, University Medical Centre Utrecht, for her comments on the manuscript. References Kohn L, Corrigan J, Donaldson M. To err is human: building a safer health system. Committee on Quality of Health Care in America, Institute of Medicine. Washington, DC: National Academy Press; Beijer JHM, de Blaey CJ. Hospitalisations caused by adverse drug reactions (ADR): a metaanalysis of observational studies. Pharm World Sci. 2002;24: Van der Hooft CS, Sturkenboom MCJM, Van Grootheest K, et al. Adverse drug reactionrelated hospitalisations: a nationwide study in the Netherlands. Drug Saf. 2006;29: Van der Hooft CS, Dieleman JP, Siemes C, et al. Adverse drug reaction-related hospitalisations: a population-based cohort study. Pharmacoepidemiol Drug Saf. 2008;17: Leendertse AJ, Egberts AC, Stoker LJ, et al. Frequency of and risk factors for preventable medication-related hospital admissions in the Netherlands. Arch Intern Med. 2008;168: Ferner RE, Aronson JK. Clarification of terminology in medication errors: definitions and classification. Drug Saf. 2006;29: Edwards IR, Aronson JK. Adverse drug reactions: definitions, diagnosis, and management. Lancet. 2000;356: Deeks JJ, Higgins JPT, Altman DG, editors. Analysing and presenting results. In: Higgins JPT, Green S, editors. Cochrane handbook for systematic reviews of interventions [updated May 2005]. Chichester: John Wiley & Sons Ltd; Miller RR. Hospital admissions due to adverse drug reactions: a report from the Boston Collaborative Drug Surveillance Program. Arch Intern Med. 1974;134: Levy M, Kewitz H, Altwein W, et al. Hospital admissions due to adverse drug reactions: a comparative study from Jerusalem and Berlin. Eur J Clin Pharmacol. 1980;17: Larmour I, Dolphin RG, Baxter H, et al. A prospective study of hospital admissions due to drug reactions. Aust J Hosp Pharm. 1991;2:90-5. Onder G, Pedone C, Landi F, et al. Adverse drug reactions as cause of hospital admissions: results from the Italian Group of Pharmacoepidemiology in the Elderly (GIFA). J Am Geriatr Soc. 2002;50:

38 Classen DC, Pestotnik SL, Scott Evans R, et al. Computerized surveillance of adverse drug events in hospital patients. JAMA. 1991;266: Ramesh M, Pandit J, Parthasarathi G. Adverse drug reactions in a south Indian hospital: their severity and cost involved. Pharmacoepidemiol Drug Saf. 2003;12: Ibanez L, Laporte JR, Carne X. Adverse drug reactions leading to hospital admission. Drug Saf. 1991;6: Smith JW, Seidl LG, Cluff LE. Studies on the epidemiology of adverse drug reactions: V. Clinical factors influencing susceptibility. Ann Intern Med. 1966;65: Sidel VW, Koch-Weser J, Barnett GO, et al. Drug utilization and adverse reactions in a general hospital. Hospitals. 1967;41:80-8. Caranasos GJ, Steart RB, Cluff LE. Drug-induced illness leading to hospitalisation. JAMA. 1974;228: Cooke DI, van der Merwe W, Pudifin DJ. Hospital admissions for adverse reactions to drugs and deliberate self-poisoning. S Afr Med J. 1985;67: Lin SH, Lin MS. A survey on drug-related hospitalisation in a community teaching hospital. Int J Clin Pharmacol Ther Toxicol. 1993;31:66-9. Huic M, Mucolic V, Vrhovac B, et al. Adverse drug reactions resulting in hospital admission. Int J Clin Pharmacol Ther. 1994; 2: Moore N, Lecointre D, Noblet C, et al. Frequency and cost of serious adverse drug reactions in a department of general medicine. Br J Clin Pharmacol. 1998;45: Pouyanne P, Haramburu F, Imbs JL, et al. Admissions to hospital caused by adverse drug reactions: cross sectional incidence study. BMJ. 2000;320:1036. Mannesse CK, Derkx FH, de Ridder MA, et al. Contribution of adverse drug reactions to hospital admission of older patients. Age Ageing. 2000;29:35-9. Fattinger K, Roos M, Vergères P, et al. Epidemiology of drug exposure and adverse drug reactions in two Swiss departments of internal medicine. Br J Clin Pharmacol. 2000;49: Mjörndal T, Boman MD, Hägg S, et al. Adverse drug reactions as a cause for admissions to a department of internal medicine. Pharmacoepidemiol Drug Saf. 2002;11: Schneeweiss S, Hasford J, Göttler M, et al. Admissions caused by adverse drug events to internal medicine and emergency departments in hospitals: a longitudinal population-based study. Eur J Clin Pharmacol. 2002;58: Von Euler M, Eliasson E, Öhlén G, et al. Adverse drug reactions causing hospitalisation can be monitored from computerized medical records and thereby indicate the quality of drug utilization. Pharmacoepidemiol Drug Saf. 2006;15: Davidsen F, Haghfelt T, Gram LF, et al. Adverse drug reactions and drug non-compliance as primary causes of admission to a cardiology department. Eur J Clin Pharmacol. 1988;34:83 6. McKenzie MW, Stewart RB, Weiss CF, et al. A pharmacist based study of the epidemiology of adverse drug reactions in pediatric medicine patients. Am J Hosp Pharm. 1973;30: Mitchel AA, Lacouture PG, Sheehan JE, et al. Adverse drug reactions in children leading to hospital admission. Pediatrics. 1988;82:

39 Study charac teristics and medication- related hospitalizations Chapter Martinez-Mir I, Garcia-Lopez M, Palop V, et al. A prospective study of adverse drug reactions as a cause of admission to a paediatric hospital. Br J Clin Pharmacol. 1996;42: Jonville-Béra AP, Giradeau B, Blanc P, et al. Frequency of adverse drug reactions in children: a prospective study. Br J Clin Pharmacol. 2002;53: Lamabadusuriya SP, Athiada G. Adverse drug reactions in children requiring hospital admission. Ceylon Med J. 2003;48:86-7. Hermesh H, Shalev A, Munitz H. Contribution of adverse drug reaction to admission rates in an acute psychiatric ward. Acta Psychiatr Scand. 1985;72: Williamson J, Chopin JM. Adverse reactions to prescribed drugs in the elderly: a multicentre investigation. Age Aging. 1980;9: Popplewell PY, Henschke PJ. Acute admissions to a geriatric assessment unit. Med J Aust. 1982;1: Smucker WD, Kontak JR. Adverse drug reactions causing hospital admission in an elderly population: experience with a decision algorithm. J Am Board Fam Pract. 1990;3: Zhang M, Holman CD, Price SD, et al. Comorbidity and repeat admission to hospital for adverse drug reactions in older adults: retrospective cohort study. BMJ. 2009;338: Senst BL, Achusim LE, Genest RP, et al. Practical approach to determining costs and frequency of adverse drug events in a health care network. Am J Health Syst Pharm. 2001;58: Pirmohamed M, James S, Meakin S, et al. Adverse drug reactions as cause of admission to hospital: prospective analysis of patients. BMJ. 2004;329:15-9. Zargarzadeh AH, Emami MH, Hosseini F. Drug-related hospital admissions in a generic pharmaceutical system. Clin Exp Pharmacol Physiol. 2007;34: Waller P, Shaw M, Ho D, et al. Hospital admissions for drug-induced disorders in England: a study using the Hospital Episodes Statistics (HES) database. Br J Clin Pharmacol. 2005;59: Patel H, Bell D, Molokhia M, et al. Trends in hospital admissions for adverse drug reactions in England: analysis of national hospital episode statistics BMC Clin Pharmacol. 2007;7:9. Pearson TF, Pittman DG, Longley JM, et al. Factors associated with preventable adverse drug reactions. Am J Hosp Pharm ;51: McDonnell PJ, Jacobs MR. Hospital admissions resulting from preventable adverse drug reactions. Ann Pharmacother. 2002;36: Bigby J, Dunn J, Goldman L, et al. Assessing the preventability of emergency hospital admissions: a method for evaluating the quality of medical care in a primary care faculty. Am J Med. 1987;83: Grymonpre RE, Mitenko PA, Sitar DS, et al. Drug-associated hospital admissions in older medical patients. J Am Geriatr Soc. 1988;36: Atkin PA, Finnegan TP, Ogle SJ, et al. Functional ability of patients to manage medication packaging: a survey of geriatric inpatients. Age Ageing. 1994;23: Dartnell JGA, Anderson RP, Chohan V, et al. Hospitalisation for adverse events related to drug therapy: incidence, avoidability and costs. Med J Aust. 1996;1634:

40 Raschetti R, Morgutti M, Menniti-Ippolito F, et al. Suspected adverse drug events requiring emergency department visits or hospital admissions. Eur J Clin Pharmacol. 1999;54: Green CF, Mottram DR, Rowe PH, et al. Adverse drug reactions as a cause of admission to an acute medical assessment unit: a pilot study. J Clin Pharm Ther. 2000;25: Wasserfallen J, Livio F, Buclin T, et al. Rate, type and cost of adverse drug reactions in emergency department admissions. Eur J Intern Med. 2001;12: Chan M, Nicklason F, Vial JH. Adverse drug events as a cause of hospital admission in the elderly. Intern Med J. 2001;31: Olivier P, Boublés O, Tubery M, et al. Assessing the feasibility of using an adverse drug reaction preventability scale in clinical practice: a study in a French emergency department. Drug Saf. 2002;25: Howard RL, Avery AJ, Howard PD, et al. Investigation into the reasons for preventable drugrelated admissions to a medical admissions unit: observational study. Qual Saf Health Care. 2003;12: Capuano A, Motola G, Russo F, et al. Adverse drug events in two emergency departments in Naples, Italy: an observational study. Pharmacol Res. 2004;50: Juntti-Patinen L, Kuitunen T, Pere P, et al. Drug-related visits to a district hospital emergency room. Basic Clin Pharmacol Toxicol. 2006;98: Patel KJ, Kedia MS, Bajpai D, et al. Evaluation of the prevalence and economic burden of adverse drug reactions presenting to the medical emergency department of a tertiary referral centre: a prospective study. BMC Clin Pharmacol. 2007;7:8. Queneau P, Bannwarth B, Carpentier F, et al. Emergency department visits caused by adverse drug events. Drug Saf. 2007;30:81-8. Alexopoulou A, Dourakis SP, Mantzoukis D, et al. Adverse drug reactions as a cause of hospital admissions: a 6-month experience in a single center in Greece. Eur J Intern Med. 2008;19: Hopf Y, Watson M, Williams D. Adverse-drug-reaction related admissions to a hospital in Scotland. Pharm World Sci. 2008;30: Zed PJ, Abu-Laban RB, Balen RM, et al. Incidence, severity and preventability of medicationrelated visits to the emergency department: a prospective study. CMAJ. 2008;178: McKenney JM, Harrison WI. Drug-related hospital admissions. Am J Hosp Pharm. 1976;33: Ghose K. Hospital bed occupancy due to drug-related problems. J Royal Soc Med. 1980;73: Bergman U, Wiholm BE. Drug-related problems causing admission to a medical clinic. Eur J Clin Pharmacol. 1981; 20: Ramsay LE, Freestone S, Silas JH. Drug-related acute medical admissions. Human Toxicol. 1982; 1: Lakshmanan MC, Hershey CO, Breslau D. Hospital admissions caused by iatrogenic disease. Arch Intern Med. 1986; 146: Ives TJ, Bentz EJ, Gwyther RE. Drug-related admissions to a family medicine inpatient service. Arch Intern Med. 1987;147:

41 Study charac teristics and medication- related hospitalizations Chapter Hallas J, Harvald B, Gram LF, et al. Drug-related hospital admissions: the role of definitions and intensity of data collection, and the possibility of prevention. J Intern Med. 1990;228: Stanton LA, Peterson GM, Rumble RH, et al. Drug-related admissions to an Australian hospital. J Clin Pharm Ther. 1994;19: Lin Wu FL, Yang CC, Shen LJ, et al. Adverse drug reactions in a medical ward. J Formos Med Assoc. 1996;95: Nelson KM, Talbert RL. Drug-related hospital admissions. Pharmacotherapy. 1996;16: Major S, Badr S, Bahlawan L, et al. Drug-related hospitalisation at a tertiary teaching centre in Lebanon: incidence, associations, and relation to self-medicating behavior. Clin Pharmacol Ther. 1998;64: Gholami K, Shalviri G. Factors associated with preventability, predictability, and severity of adverse drug reactions. Ann Pharmacother. 1999;33: Jha AK, Kuperman GJ, Rittenberg E, et al. Identifying hospital admissions due to adverse drug events using a computer-based monitor. Pharmacoepidemiol Drug Saf. 2001;10: Peyriere H, Cassan S, Floutard E, et al. Adverse drug events associated with hospital admission. Ann Pharmacother. 2003;37:5-11. Dormann H, Criege-Rieck M, Neubert A, et al. Lack of awareness of community-acquired adverse drug reactions upon hospital admission: dimensions and consequences of a dilemma. Drug Saf. 2003;26: Hardmeier B, Braunschweig S, Cavallaro M, et al. Adverse drug events caused by medication errors in medical inpatients. Swiss Med Wkly. 2004;134: Hallas J, Jensen KB, Grodum E, et al. Drug related admissions to a department of medical gastroenterology: the role of self-medicated and prescribed drugs. Scand J Gastroenterol. 1991;26: Hallas J, Davidsen O, Grodum E, et al. Drug related illness as a cause of admission to a department of respiratory medicine. Respiration. 1992;59:30-4. Hallas J, Gram LE, Grodum E, et al. Drug related admissions to medical wards: a population based survey. Br J Clin Pharmacol. 1992;33:61-8. Easton KL, Parsons BJ, Starr M, et al. The incidence of drug-related problems as a cause of hospital admissions in children. Med J Aust. 1998;169: Easton KL, Chapman CB, Brien J. Frequency and characteristics of hospital admissions associated with drug-related problems in paediatrics. Br J Clin Pharmacol. 2004;57: Stewart RB, Sprinker PK, Adams JE. Drug-related admissions to an inpatient psychiatric unit. Am J Psychiatry. 1980;137: Salem RB, Keane TM, Williams JG. Drug-related admissions to a Veterans Administration psychiatric unit. Drug Intell Clin Pharm. 1984;18:74-6. Col N, Fanale JE, Kronholm P. The role of medication noncompliance and adverse drug reactions in hospitalisations of the elderly. Arch Intern Med. 1990;150: Wong ME, Ioannides-Demos LL, Li SC, et al. Drug-related hospital admissions of geriatric patients [abstract]. Aust J Hosp Pharm. 1993;23:75. Courtman BJ, Stallings SB. Characterization of drug-related problems in elderly patients on admission to a medical ward. Can J Hosp Pharm. 1995;48:

42 Cunningham G, Dodd TR, Grant DJ, et al. Drug-related problems in elderly patients admitted to Tayside hospitals, methods for prevention and subsequent reassessment. Age Ageing. 1997;26: Malhotra S, Karan RS, Pandhi P, et al. Drug-related medical emergencies in the elderly: role of adverse drug reactions and non-compliance. Postgrad Med J. 2001;77: Rivkin A. Admissions to a medical intensive care unit related to adverse drug reactions. Am J Health Syst Pharm. 2007;64: Frisk PA, Cooper JW, Campbell NA. Community-hospital pharmacist detection of drugrelated problems upon patient admission to small hospitals. Am J Hosp Pharm. 1977;34: Koh Y, Kutty FB, Li SC. Therapy related hospital admission in patients on polypharmacy in Singapore: a pilot study. Pharm World Sci. 2003;25: Kaufman DW, Kelly JP, Rosenberg L, et al. Recent patterns of medication use in the ambulatory adult population of the United States: the Slone Survey. JAMA. 2002;287: Walker J, Wynne H. Review: the frequency and severity of adverse drug reactions in elderly people. Age Ageing. 1994;23: Tranulis C, Skalli L, Lalonde P, et al. Benefits and risks of antipsychotic polypharmacy: an evidence-based review of the literature. Drug Saf. 2008;31:7-20. Rittmannsberger H, Meise U, Schauflinger K, et al. Polypharmacy in psychiatric treatment: patterns of psychotropic drug use in Austrian psychiatric clinics. Eur Psychiatry. 1999;14: Bonati M. Epidemiological evaluation of drug use in children. J Clin Pharmacol. 1994;34: Sturkenboom MC, Verhamme KM, Nicolosi A, et al. Drug use in children: cohort study in three European countries. BMJ. 2008;337: Lazarou J, Pomeranz BH, Corey PN. Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies. JAMA. 1998;279: Wiffen P, Gill M, Edwards J, et al. Adverse drug reactions in hospital patients. Bandolier Extra 2002: 1 15 [online]. Available from URL: Extraforbando/ADRPM.pdf [Accessed 2009 May 7] Kongkaew C, Noyce PR, Ashcroft DM. Hospital admissions associated with adverse drug reactions: a systematic review of prospective observational studies. Ann Pharmacother. 2008;42:

43 Study charac teristics and medication- related hospitalizations Chapter 2.1 Appendix I Included studies in the literature review with author, year of publication, study setting and population (country, hospital, study period, included admissions, patient age group), the definitions used (ADE or ADR), and methods of data collection (medical chart review, database method or spontaneous reporting method) and the reported prevalence with 95% confidence interval and reference in reference list. Table sorted by nominator, denominator, method of data collection and year of publication a First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection Studied admissions (n) b Cases (n) Prev (95%CI) Miller et al. USA, Canada and Israel 5 hospital in USA, 1 in Canada and 1 in Israel Apr 1969 Jan 1972 all ages ADR all admissions medical chart review % ( ) Levy et al. Israel university hospital all ages ADR all admissions medical chart review % ( ) Larmour et al Australia teaching hospital Melbourne 6 months; Jun Nov 1987 all ages ADR all admissions medical chart review % ( ) Onder et al. Italy regional and university hospitals 14 months elderly ADR all admissions medical chart review % ( ) Classen et al. USA 1 hospital Salt Lake City May 1989 Oct 1990 all ages ADR all admissions database % ( ) Ramesh et al. India hospital in Mysore 7 months; Dec 2001 Jul 2002 all ages ADR all admissions spontaneous reporting % ( ) Ibanez et al. Spain university hospital Sept 1986 Jun 1989 all ages ADR acute admissions medical chart review % ( ) Hooft et al. NL database (discharge diagnoses) 2001 all ages ADR acute admissions database % ( ) 43

44 (Appendix I continued) First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection Smith et al. USA hospital Jan Dec all ages ADR admissions to internal medicine ward medical chart review Sidel et al. USA Boston Oct Jan 1967 all ages ADR admissions to internal medicine ward medical chart review Caranasos USA university hospital Florida Aug 1969 Jul 1972 all ages ADR admissions to internal medicine ward medical chart review Levy et al. Israel and Germany university hospital Berlin all ages ADR admissions to internal medicine ward medical chart review Cooke et al. South-Africa Addington hospital Durban Aug Oct 1982 all ages ADR admissions to internal medicine wards medical chart review Lin et al. Taiwan general hospital 10 months; Jan ct 1990 > 17 ADR admissions to internal medicine ward medical chart review Huic et al. Croatia university hospital Zagreb 14 months; Jan 1992 Apr 1993 > 15 ADR admissions to internal medicine ward medical chart review Moore et al. France general hospital Jacques Monod in le Havre 6 months; May Oct 1993 > 16 ADR admissions to internal medicine ward medical chart review Studied admissions (n) b Cases (n) Prev (95%CI) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) 44

45 Study charac teristics and medication- related hospitalizations Chapter 2.1 (Appendix I continued) First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection Pouyanne et al France 33 hospitals Mar Apr 1998 all ages ADR admissions to internal medicine wards medical chart review Mannesse et al NL university hospital Rotterdam 4 months; Feb May 1994 > 69 ADR admissions to internal medicine wards medical chart review Fattinger et al Switzerland university and general hospital Jan 1996 Dec 1998 all ages ADR admissions to internal medicine wards medical chart review Mjörndal et al Sweden university hospital Umea 36 weeks (1 week per ward) 1997/1998 all ages ADR admissions to internal medicine and cardiology ward medical chart review Schneeweiss et al Germany hospitals of the Jena and Rostock area 2.5 years; Oct 1997 Mar 2000 all ages ADR admissions to internal medicine ward (except patients on chemotherapy) medical chart review Von Euler et al Sweden university hospital Stockholm spring 2002 > 18 ADR admissions to internal medicine ward medical chart review Studied admissions (n) b Cases (n) Prev (95%CI) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) 45

46 (Appendix I continued) First author, publication year Country Hospital Study period Davidsen et al Denmark Odense university hospital 2 months; Apr June 1986 Mckenzie et al USA teaching hospital Gainesville 8 months; Feb Sep 1971 Mitchell et al USA several hospitals around Boston 11 years; Sept 1974 Dec 1985 Martínez-Mir et al Spain university paediatric hospital in Valencia Jun Oct 1992 and Jan Apr 1993 Jonville-Bèra et al France paediatric hospital, University of Tours 1 week; May June 1998 Lamabadusuriya et al Sri-Lanka hospital Colombo 11 months; Feb 2002 Dec 2002 Jonville-Bèra et al France Paediatric hospital, University of Tours 1 week; May June 1998 Age group Nominator Denominator Method of data collection all ages ADR admissions to cardiology ward medical chart review < 26 ADR paediatric admissions medical chart review < 16 ADR paediatric admissions medical chart review 1 24 months ADR paediatric admissions medical chart review children (mean age 6 years) ADR paediatric admissions medical chart review children (mean age 5 years) ADR paediatric admissions medical chart review children (mean age 7 years) ADR acute paediatric admissions medical chart review Studied admissions (n) b Cases (n) Prev (95%CI) % ( ) % ( ) % ( ) % ( ) % c % ( ) % c 46

47 Study charac teristics and medication- related hospitalizations Chapter 2.1 (Appendix I continued) First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection Studied admissions (n) b Cases (n) Prev (95%CI) Mitchell et al. USA several hospitals around Boston 11 years; Sept 1974 Dec 1985 children with cancer ADR admission of children with cancer medical chart review % ( ) Mitchell et al. USA several hospitals around Boston 11 years; Sept 1974 Dec 1985 new-borns ADR admissions of newborns medical chart review % ( ) Hermesh et al Denmark 1 hospital 17 months all ages ADR acute psychiatric admissions medical chart review % ( ) Williamson et al UK care of the elderly wards of 42 hospitals start Nov 1975 elderly ADR admissions of elderly patients medical chart review % ( ) Popplewell et al Australia hospital 1981 > 74 ADR admissions of elderly patients medical chart review % ( ) Smucker et al USA community hospital 1990 > 64 ADR admissions of elderly patients medical chart review % ( ) Zhang et al. Australia all hospitals in Western Australia February 2005 > 59 ADR admission for an ADR during database % ( ) 47

48 (Appendix I continued) First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection Studied admissions (n) b Cases (n) Prev (95%CI) Senst et al. USA 4 hospitals, of which 2 paediatric hospitals and 1 psychiatric hospital 53 days in 1998 all ages (except newborns) ADE all admissions medical chart review % ( ) Pirmohamed et al UK 2 NHS hospitals, 1 teaching hospital, 2 regional hospitals 6 months; Nov 2001 Apr 2002 > 15 ADE all admissions medical chart review % ( ) Zargarzadeh et al Iran 3 hospitals 6 months; 2002 (summer winter) all ages ADE all admissions medical chart review % ( ) Hooft et al. NL electronic patient records of 150 GPs 2003 all ages ADE all admissions medical chart review % ( ) Waller et al. UK database of NHS hospitals all ages ADE all admissions database % ( ) Patel et al. UK database of NHS hospitals all ages ADE all admissions database % ( ) Pearson et al. USA general hospital Atlanta 7 months; Jul 1992 Jan 1993 all ages ADE all admissions spontaneous reporting % ( ) 48

49 Study charac teristics and medication- related hospitalizations Chapter 2.1 (Appendix I continued) First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection McDonnell et al USA university hospital Philadelphia 11 months: Jul all ages ADE all admissions spontaneous reporting Bigby et al. USA Brigham and Womens hospital all ages ADE acute admissions medical chart review Grymonpre et al Canada Hospital Winnipeg May Aug 1983 > 50 ADE acute admissions medical chart review Atkin et al. Australia teaching hospital 1991 elderly ADE acute admissions medical chart review Dartnell et al. Australia hospital Melbourne Nov Dec 1994 all ages ADE acute admissions medical chart review Raschetti et al Italy emergency department of 1 hospital Oct 1994 Sept 1995 (first week of every month) all ages ADE acute admissions medical chart review Green et al. UK university hospital Liverpool 1996 > 18 ADE acute admissions medical chart review Wasserfallen et al Switzerland university hospital Lausanne Jan Jul 1994 > 16 ADE acute admissions medical chart review Chan et al. Australia hospital in Hobart Aug Sept 1998 > 75 ADE acute admissions medical chart review Studied admissions (n) b Cases (n) Prev (95%CI) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) 49

50 (Appendix I continued) First author, publication year Country Hospital Study period Olivier et al. France university hospital of Toulouse 4 weeks in May, July, Aug, Oct 1998 Howard et al. UK hospital in Nottingham 6 months 2001 Capuano et al Italy 2 hospitals March and June 2000 Juntti et al. Finland general hospital Nov 2001 Apr 2002 Patel et al. India 1 hospital 1 May Jun 2005 Queneau et al France 5 university hospitals and 5 general hospitals 1 week in June and December 1999 Alexopoulou et al Greece university hospital Jan Jun 2005 Hopf et al. UK 1 hospital 22 May Jun 2006 Leendertse et al NL 21 hospitals Sept 2005 June 2006 Age group Nominator Denominator Method of data collection > 14 ADE acute admissions medical chart review all ages ADE acute admissions medical chart review all ages ADE acute admissions medical chart review all ages ADE acute admissions medical chart review > 17 ADE acute admissions medical chart review all ages ADE acute admissions medical chart review all ages ADE acute admissions medical chart review > 15 ADE acute admissions medical chart review > 17 ADE acute admissions medical chart review Studied admissions (n) b Cases (n) Prev (95%CI) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) 50

51 Study charac teristics and medication- related hospitalizations Chapter 2.1 (Appendix I continued) First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection Zed et al. Canada emergency department of 1 hospital 12 weeks, 13 Mar 4 Jun 2006 all ages ADE acute admissions medical chart review McKenny et al USA teaching hospital Virginia 2 months; Feb Mar 1974 all ages ADE admissions to internal medicine ward medical chart review Ghose UK Carlisle hospital Oct Dec 1979 all ages ADE admissions to internal medicine ward medical chart review Bergman et al Zweden university hospital 3.5 months > 15 ADE admissions to internal medicine ward medical chart review Ramsay et al. UK teaching hospital Sheffield 1978 all ages ADE admissions to internal medicine ward medical chart review Lakshmanan et al USA teaching hospital Cleveland 2 months; Jul Aug 1984 all ages ADE admissions to internal medicine ward medical chart review Ives et al. USA hospital North Carolina 12 months all ages (except newborns) ADE admissions to internal medicine ward medical chart review Hallas et al. Denmark Odense University Hospital Mar May 1988 all ages ADE admissions to internal medicine ward medical chart review Stanton et al. Australia teaching hospital Tasmania 10 weeks; Mar Aug 1993 > 16 ADE admissions to internal medicine ward medical chart review Studied admissions (n) b Cases (n) Prev (95%CI) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) 51

52 (Appendix I continued) First author, publication year Country Hospital Study period Lin Wu et al. Taiwan university hospital Taiwan Apr 1992 Dec 1993 Nelson et al. USA university hospital Bexar County Jul Aug 1993 Major et al. Lebanon American University hospital Beirut 6 months; Oct 1996 May 1997 (except Dec) Gholami et al. Iran university hospital Teheran Mar Dec 1996 Jha et al. USA tertiary care hospital Boston Oct 1994 May 1995 Peyriere et al. France university hospital Montpellier 48 days (6 8 days over period of 1 year) Dormann et al Germany Erlangen- Nuremberg University hospital 13 months; Dec 1997 Jan 1999 Age group Nominator Denominator Method of data collection > 17 ADE admissions to internal medicine ward medical chart review all ages ADE admissions to internal medicine ward medical chart review > 17 ADE admissions to internal medicine ward medical chart review all ages ADE admissions to internal medicine wards medical chart review all ages ADE admissions to internal medicine ward medical chart review all ages ADE admissions to internal medicine ward medical chart review > 17 ADE admissions to two internal medicine wards medical chart review Studied admissions (n) b Cases (n) Prev (95%CI) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) 52

53 Study charac teristics and medication- related hospitalizations Chapter 2.1 (Appendix I continued) First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection Hardmeier et al Switzerland university hospital and regional hospital Zurich all ages ADE admissions to internal medicine ward medical chart review Hallas et al. Denmark Odense University hospital Aug Nov 1988 > 17 ADE admissions to gastroenterology ward medical chart review Hallas et al. Denmark Odense University hospital Aug Nov 1988 > 14 ADE admissions to respiratory ward medical chart review Hallas et al. Denmark Odense University hospital Mar 1988 May 1989 all ages ADE admissions to care of the elderly, internal medicine, cardiology, respiratory and gastroenterology wards medical chart review Easton et al. Australia paediatric hospital Jun Aug 1996 < 19 ADE paediatric admissions medical chart review Major et al. Lebanon American University hospital Beirut 6 months; Oct 1996 May 1997 (except Dec) < 18 ADE paediatric admissions medical chart review Studied admissions (n) b Cases (n) Prev (95%CI) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) 53

54 (Appendix I continued) First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection Easton et al. Australia paediatric and regional hospital 1998/1999 < 18 ADE acute paediatric admissions medical chart review Stewart et al. USA hospital Florida 6 months all ages ADE admissions to psychiatric ward medical chart review Salem et al. USA Veteran s Administration hospital 4 months all ages ADE admissions to psychiatric ward medical chart review Col et al. USA teaching hospital Jun Aug 1987 > 64 ADE admissions of elerly medical chart review Wong et al. Australia hospital 8 weeks; > ADE admissions of elderly patients medical chart review Wong et al. Australia hospital 4 weeks; > 64 ADE admissions of elderly patients medical chart review Courtman et al Canada tertiary care hospital in Toronto Sep 1992 Febr 1993 > 64 ADE admissions of elderly patients medical chart review Cunningham et al UK hospitals in Tayside region of Scotland Mar Dec 1992 > 65 ADE admissions of elderly patients medical chart review Malhotra et al India tertiary care hospital in Northern India 7 months; Jan Jul 2000 > 64 ADE admissions of elderly patients medical chart review Studied admissions (n) b Cases (n) Prev (95%CI) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) % ( ) 54

55 Study charac teristics and medication- related hospitalizations Chapter 2.1 (Appendix I continued) First author, publication year Country Hospital Study period Age group Nominator Denominator Method of data collection Studied admissions (n) b Cases (n) Prev (95%CI) Rivkin USA hospital New York Dec 2004 May 2005 all ages ADE admissions to intensive care unit medical chart review % ( ) Frisk et al. USA 2 hospitals Dec Jun 1974 (6 months per hospital) all ages ADE selection of drug-related hospitalisations medical chart review % ( ) Koh et al. Singapore emergency hospital 4 days; Nov Dec 2000 > 15 ADE admissions of patients using more than 5 medications medical chart review % ( ) prev = prevalence; 95%CI = 95% confidence interval; ADR = adverse drug reaction; ADE = adverse drug event; USA = United States of America; NL = the Netherlands; UK = United Kingdom a) Available from (table, Supplemental Digital Content 1). b) Number of studied hospital admissions, the denominator. c) Numbers to small to calculate 95%CI (n prevalence < 5). 55

56

57

58 ABSTRACT Background Medication-related problems that lead to hospitalization have been the subject of many studies, many of which were limited to one hospital or lacked patient followup. Furthermore, little information exists on potential risk factors associated with preventable medication-related hospitalizations. Methods A prospective multicenter study was conducted to determine the frequency and patient outcomes of medication-related hospital admissions. A case-control design was used to determine risk factors for potentially preventable admissions. All unplanned admissions in 21 hospitals were assessed during 40 days. Controls were patients admitted for elective surgery. Cases and controls were followed up until hospital discharge. The frequency of medication-related hospital admissions, potential preventability, and outcomes were assessed. For potentially preventable medication-related admissions, risk factors were identified in the case-control study. Resu lts Almost unplanned admissions were screened, of which 714 (5.6%) were medication related. Almost half (46.5%) of these admissions were potentially preventable, resulting in 332 case patients matched with 332 controls. Outcomes were favorable in most patients. The main determinants of preventable medicationrelated hospital admissions were impaired cognition (odds ratio [OR] 11.9; 95% confidence interval [95%CI] ), 4 or more comorbidities (OR 8.1; 95%CI ), dependent living situation (OR 3.0; 95%CI ), impaired renal function (OR 2.6; 95%CI ), nonadherence to medication regimen (OR 2.3; 95%CI ), and polypharmacy (OR 2.7; 95%CI ). C onclusi ons Adverse drug events are an important cause of hospitalizations, and almost half are potentially preventable. The identified risk factors provide a starting point for preventing medication-related hospital admissions. 58

59 Hospital admissions related to medication- study Chapter 2.2 INTRODUCTION Patient safety is considered an essential element of high-quality health care systems. This notion has grown in recent years, especially since publication of To Err Is Human by the Institute of Medicine. 1 Medications can be an important source of unintended patient harm, which may be caused by either nonpreventable adverse effects of medication use or by medication errors (potentially preventable). Medication-related problems can cause serious adverse drug effects (ADEs) that may lead to hospitalization of the patient. These have been the subject of many published studies, which have been summarized in 2 meta-analyses 2 3 and in some more recent studies. 4 5 Many of these studies were limited to one hospital or to one specific type of ward, 6 9 had a retrospective design, 10,11 or provided no information on preventability. 2 Furthermore, little information exists on the risk factors associated with preventable medication-related hospital admissions in a general population. Given these limitations and the need for information on potential risk factors, we conducted a multicenter study aimed at identifying the frequency and preventability of medication-related hospitalizations in the Netherlands and risk factors for the preventable hospitalizations. METHODS Design The multicenter, observational Hospital Admissions Related to Medication (HARM) Study, with a prospective follow-up, was conducted between 1 September 2005 and 30 June Medication-related hospitalizations were defined as hospitalizations due to ADEs: harm due to adverse effects of medication use (as defined by the World Health Organization) 12 or due to medication errors (i.e., preventable medicationrelated hospitalizations). A medication error was defined as any error in the process of prescribing, dispensing, or administering the medication. 13 A case-control design was used to identify potential risk factors for the subset of preventable admissions. Conceptually, control patients should be a representative sample from those at risk for an ADE that would necessitate hospital admission. This would imply sampling control patients from the community. The disadvantage of such sampling would be the risk of information bias; for example, nonhospitalized patients have fewer renal function tests available and, therefore, decreased renal function as a risk factor may be oversampled in the case group, leading to inflated risk estimates. This can be dealt with by prospectively performing additional diagnostic tests in control patients, but this was considered unfeasible. 59

60 The problem of information bias is largely overcome by using hospital-based controls. However, the potential for selection bias is the main problem in using hospital-based controls. We considered selecting only unplanned admissions not related to an ADE, but we believed that we could never be sure that the admission was definitely unrelated to an ADE (and the same problem arises when selecting controls from all admissions). Therefore, we decided to select controls from the planned surgery population (reason for admission definitely unrelated to ADEs) and to match on age and sex (thereby increasing the comparability of cases and controls regarding the use of medications). The study protocol was approved by a medical ethics committee (Medisch-Ethische Toetsing Onderzoek Patiënten en Proefpersonen, Tillburg, the Netherlands). S etting and par ticipants The data were collected in 21 of the 104 Dutch hospitals. 14 The hospitals (university, teaching, and general hospitals) were selected from all regions of the Netherlands to obtain a representative sample of hospitalizations. In each hospital, a specially trained researcher screened all unplanned admissions for a potentially medication-related cause of hospitalization during 40 days in two consecutive months. The exclusion criteria were age younger than 18 years and admission for obstetric indications, to a psychiatric ward, or for self-poisoning. For all remaining admissions, the documented reason for admission and medication use before admission were assessed by means of a trigger list. This trigger list consisted of 537 combinations of symptoms and medicines, that have been mentioned in the literature15 as possible causes of ADEs that may lead to medication-related hospitalizations. Examples of symptom/medication combinations on the list are asthma exacerbation and nonsteroidal anti-inflammatory drugs (NSAIDs)/aspirin, thrombocytopenia and antiepileptic medication, and hyponatremia and selective serotonin reuptake inhibitors (asthma exacerbation, thrombocytopenia, and hyponatremia were, of course, related to a variety of other medicines as well on the trigger list). In addition, some symptoms on the trigger list referred to no particular medication, for example, trauma as a symptom referred to check whether any medicines used have sedating potential. Any admission that matched this list was discussed with the hospital physician of the patient. If a relation with medication use was deemed possible, the patient was included as a case and was followed up during admission. Hospital physicians were also asked to report potential cases not identified by the trigger list. For each case, one control was selected in the same hospital matched on age (by 5-year age group) 60

61 Hospital admissions related to medication- study Chapter 2.2 and sex. Cases and controls provided written informed consent to use their medical information for research purposes. Data collection For included cases and controls, relevant information from the medical record (medical history, diagnostic procedures, and outcomes) was collected. The medical record abstracters were not blinded to the nature of the study or to whether the patient was a case. All clinical laboratory data from one year before the present admission were recorded. On the basis of serum creatinine values nearest to the hospitalization, renal function before admission was calculated using the Cockcroft- Gault formula. 16 Medications dispensed for one year before admission were obtained from the patient s community pharmacy records. From this medication history, adherence to the regimen (compliance) was estimated for all oral medicines by calculating the refill rate for one year before hospitalization. The refill rate is defined as the number of daily doses dispensed divided by the total number of days between the first and last prescription in this period. Only medicines indicated for long-term use and dispensed at least three times during the year were considered. Patients were classified as adherent to the medication regimen if the refill rates of all these medicines were between 0.8 and Information about the living situation (independent vs dependent [i.e., in a nursing home, in a care home, or at home with nursing care]), sex, and age was obtained from the patient s medical record. Cognitive function before admission was obtained from the medical record or was discussed with the physician and assessed as normal or impaired (this is the way cognitive function is assessed in everyday practice in the Netherlands; formal tests, such as the Mini-Mental State Examination, are not routinely used). Information about the number of previous admissions and the number of physicians was obtained from the hospital information system. Previous admissions are mentioned in the medical history of the patient in the hospital information system, even when the admissions took place in other hospitals. The duration of the hospital admission and the outcome were recorded for each case. Assessment of causality and pre ventability Two clinical pharmacists (A.J.L. and P.M.L.A.v.d.B) independently assessed all the patients initially included as cases with respect to the causal relationship between the suspected medicine and the reason for hospitalization, according to an adjusted version of the algorithm by Kramer et al. 18 In this version, three questions need to be answered (in contrast to six questions in the original algorithm): whether the reason for admission is known to be an adverse event of the suspected medicine, whether 61

62 alternative causes can explain the relationship between the suspected medicine and the adverse event, and whether a plausible time relationship exists between the adverse event and the start of medication administration (or the occurrence of the medication error). On the basis of the answers, causality is classified as possible, probable, or unlikely. Cases with an assessment of unlikely were excluded. The same pharmacists also assessed the preventability of the admissions, according to a modified version of the algorithm by Schumock and Thornton. 19 In this algorithm, an admission was assessed as preventable when a medication error was made with the medication that caused the hospital admission. The original Schumock-Thornton algorithm assesses prescribing errors, which can be defined as dosing errors or therapeutic errors, such as medication not indicated (based on patient history), medication contraindicated, recorded medication allergies, drug-drug interaction (included only if the interaction is inadequately monitored or if the medication involved in the interaction may never be combined [absolute contraindication for the combination]), inadequate monitoring of therapy, therapeutic duplication medication, and underprescribing (defined as an essential medicine not being prescribed). 20 The algorithm was expanded to include dispensing errors (errors at the dispensing stage in the pharmacy) and administration errors (errors when administering medication to the patient either by caretakers or by the patient, e.g., nonadherence to the medication regimen). If the assessments of the pharmacists disagreed, they met to reach consensus (2.5% of the cases for the causality assessment and 26% of the cases for the preventability assessment). Main outcome measures The frequency of medication-related hospitalizations was the main outcome measure, defined as the number of medication-related hospitalizations divided by the number of all unplanned admissions of persons older than 18 years (excluding obstetric admissions, self-poisonings, and psychiatric admissions). In addition, the percentage of potentially preventable medication-related admissions and patient outcomes was determined. The following determinants were assessed as potential risk factors in the casecontrol part of the study: medication regimen adherence (determined by calculation of the refill rate), 17 patient living situation (independent or dependent), cognition, renal function before admission, number of diseases in the medical history, number of previous admissions (in the year before the present admission), polypharmacy (defined as five medicines in long-term use at the time of admission), 21 and number of prescribing physicians. 62

63 Hospital admissions related to medication- study Chapter 2.2 Data analysis Data were entered into local databases, which were merged. Before data analysis, all electronic case report forms were verified centrally on missing values, extreme values, and coding of the medical history. For validation purposes, a random sample of cases was assessed completely on correct input into the database. Data were analyzed using statistical software (SPSS, version 11; SPSS Inc, Chicago, Illinois). The mean age of the potentially preventable cases was compared with that of all patients admitted in the same period in the same hospitals using the Mann-Whitney test. The sex of these cases was compared with that of all admissions using a χ 2 test. Cases were compared with all hospitalized patients for these patient characteristics because cases and controls were matched on age and sex and, therefore, could not be compared with each other. Duration of hospitalization was tested against the national mean after logarithmic transformation of the length of hospitalization of the potentially preventable medication-related admissions. In this case, national data were used because duration of hospitalization of all hospitalized patients in the 21 hospitals was not available (in contrast to age and sex). Cases were not compared with controls regarding length of hospitalization because only cases were prospectively followed up. For the analysis of potential risk factors for preventable admissions, a univariate conditional logistic regression analysis was performed, with stratification on matching variables. Determinants identified in the univariate analysis as being statistically significantly associated with the outcome (p < 0.05) were considered confounding variables that were included in a stepwise multivariate conditional logistic regression analysis. RESULT S Frequenc y During the study, patients were admitted to the 21 participating hospitals, of which were unplanned admissions. Seventy-two patients refused participation, mostly because they did not understand the questions asked or did not speak the Dutch language. Initially, 743 admissions were considered to be medication related but, after central causality assessment, 714 admissions were included as possibly or probably medication related, representing a frequency of 5.6% of unplanned admissions. A total of 332 of these cases (46.5%) were assessed as potentially preventable (Figure 1). 63

64 Figure 1 Study design with the main outcomes a) 815 admissions were assessed by the physician as drug related of which 72 patients refused to participate in the study and 29 admissions were assessed centrally as unlikely to be drug related. The median length of hospital stay of the 332 potentially preventable medicationrelated cases was eight days, and 24 (7.2%) of these cases were admitted to an intensive care unit. Of the 332 potentially preventable medication-related admissions, 233 patients (70.2%) recovered completely, but 21 (6.3%) died and 31 (9.3%) experienced a disability after discharge; for 47 cases (14.2%), the outcome was uncertain at the time of discharge. Whether patients died of the actual ADE leading to hospitalization or of other causes (e.g., hospital-acquired infection and comorbidities) was not assessed. 64

65 Hospital admissions related to medication- study Chapter 2.2 The most common reasons for hospitalization of the potentially preventable cases were gastrointestinal tract problems: 14.5% (48 of 332) of these cases were admitted for gastrointestinal bleeding and 6.6% (22 of 332) for other gastrointestinal tract symptoms, such as constipation and diarrhea. Other common problems were cardiovascular symptoms (10.5% [35 of 332]), respiratory symptoms (7.8% [26 of 332]), and poor glycemic control (6.0% [20 of 332]) (Table 1). Table 1 Reasons for potentially preventable medication-related hospital admissions and the associated drugs Reason for Admission Preventable admissions Associated drugs (no. of admissions) a Gastro-intestinal (GI) system n=332 (100%) GI tract bleeding 48 (14.5%) Antiplatelets (34), NSAIDs (14), anticoagulants (12), oral corticosteroids (4) GI tract symptoms (e.g., diarrhea, constipation) Circulatory system 22 ( 6.6%) Oral antidiabetics (4), laxatives (4), diuretics (4), opiates (3), loperamide (3), statins (3), antibacterial drugs (3) Cardiovascular symptoms (e.g., dysrhythmias, heart failure) 35 (10.5%) β-blockers (15), drugs affecting the RAAS (9), calcium antagonist (9), diuretics (9), anticoagulants (7) Respiratory system Respiratory symptoms (e.g., dyspnea) 26 ( 7.8%) Diuretics (12), respiratory drugs (6), β-blockers (6), NSAIDs (5) Endocrine system Hypoglycemia or hyperglycemia 20 ( 6.0%) Insulin (18), oral antidiabetics (12), corticosteroids (3), diuretics (3) GI = gastrointestinal; NSAIDs = nonsteroidal anti-inflammatory drugs; RAAS = renin angiotensin aldosterone system a) An admission can be associated with more than one drug and is then mentioned more than once in the list. Medicines associated most often with potentially preventable medication-related hospital admissions were those that affect blood coagulation, such as antiplatelet drugs (8.7% [29 of 332 patients]), oral anticoagulants (6.3% [21 of 332]), NSAIDs (5.1% [17 of 332]), and a combination of these medicines (10.5% [35 of 332]). 65

66 Antidiabetic drugs were related to the reason for admission in 41 of 332 cases (12.3%). Medications that act on the central nervous system (5.1% [17 of 332 patients]) were most often related to a trauma (Table 1). A total of 509 medication errors were identified in the 332 potentially preventable medication-related hospitalizations. Lack of a clear indication for the medication (n = 84), nonadherence to the medication regimen (n = 78), inadequate monitoring (n = 71), and drug-drug interactions (n = 70) were the most common errors found. Underprescribing of gastroprotective drugs in the case of NSAID or aspirin use (only in high-risk patients) and drug-drug interactions were the most common errors found in patients admitted with gastrointestinal tract bleeding (Table 2). Table 2 Medication errors associated with potentially preventable medicationrelated hospital admissions Medication Error Prescribing error a All Admissions (n=332) Reasons for potentially preventable medication-related hospital admissions GI tract bleeding (n=48) CVD (n=35) Respiratory symptoms (n=26) Drug not indicated 84 (16.5%) 13 (16.7%) 4 ( 9.3%) 6 (16.2%) Inadequate monitoring 71 (13.9%) 5 ( 6.4%) 8 (18.6%) 3 ( 8.1%) Drug-drug interaction 70 (13.8%) 20 (25.6%) 6 (14.0%) 6 (16.2%) Underprescribing 57 (11.2%) 23 (29.5%) 3 ( 7.0%) 3 ( 8.1%) Contraindication 45 ( 8.8%) 8 (10.3%) 1 ( 2.3%) 4 (10.8%) Dose too high 29 ( 5.7%) 6 ( 7.7%) 1 ( 2.3%) 0 Administration error a Nonadherence to medication regimen 78 (15.3%) 1 ( 1.3%) 17 (39.5%) 12 (32.4%) Incorrect use 36 ( 7.1%) 1 ( 1.3%) 1 ( 2.3%) 1 ( 2.7%) Other 39 ( 7.7%) 1 ( 1.3%) 2 ( 4.7%) 2 ( 5.4%) Total b 509 (100.0%) 78 (100.0%) 43 (100.0%) 37 (100.0%) GI = gastrointestinal; CVD = cardiovascular disease a) Presented as number (% of total). b) Because of rounding, percentages may not total

67 Hospital admissions related to medication- study Chapter 2.2 Potential risk factors The mean age of patients admitted with a potentially preventable medication-related cause was 68 years, which differed significantly from the mean age of all unplanned admissions: 60 years (Mann-Whitney test, p < 0.001). The robust mean of the duration of hospitalization (the Hampel M-estimator: days; 95% confidence interval [95%CI] of length of stay in the hospital, days) also differed from the national mean hospital stay (5.6 days). No significant difference was found for sex between cases and the entire patient population admitted during the study period in the participating hospitals (χ 2 test, p = 0.73) (Figure 1). The most important patient-related, statistically significant potential risk factors identified were impaired cognition (odds ratio [OR] 13.0; 95%CI ), four or more diseases in the patient s medical history (OR 11.3; 95%CI ), dependent living situation (OR 4.5; 95%CI ), impaired renal function before hospital admission (OR 2.6; 95%CI ), and nonadherence to the medication regimen (OR 2.6; 95%CI ). After adjustment for several confounders, the effect of these risk factors remained statistically significant (Table 3). Polypharmacy was a medication-related determinant that was associated with medication-related hospital admissions. For the use of five or more medicines at the time of admission (polypharmacy), a statistically significant effect was found that remained so in the multivariate model (OR 2.7; 95%CI ). The determinant previous admissions was not statistically significantly associated with risk of preventable medication-related hospitalization (OR 1.3; 95%CI ), and neither was the number of prescribers after correction for the confounding factor polypharmacy (four prescribers: OR 1.4; 95%CI ) (Table 3). C omment The HARM Study was the first multicenter study of medication-related hospitalizations in the Netherlands. The results of this study show that a considerable proportion (5.6%) of all unplanned admissions is medication related. Almost half (46.5%) of these admissions are potentially preventable. The frequency of 5.6% is comparable to the frequencies of 4.9% and 5.2% reported in a meta-analysis of studies on medication-related hospital admissions 2 and in a large study in the United Kingdom, respectively. 4 As was shown in a recent systematic review, 23 the HARM Study identified anticoagulant and antiplatelet drugs as major causes of medication-related hospital admissions. Other groups identified in the HARM Study (antidiabetic drugs, NSAIDs, and medications that act on the central nervous system) are also well known from the literature as medications with increased risks

68 Table 3 Determinants associated with potentially preventable medication related hospital admissions before and after adjustment for confounders Frequency a Univariate Multivariate Analysis Variable Cases Controls OR (95%CI) OR (95%CI) Adjusted for n=332 (100%) n=332 (100%) Patient related Elderly (> 65 yr) 226 (68.1%) 221 (66.7%) NA b NA b NA b Female sex 164 (49.4%) 164 (49.4%) NA b NA b NA b Living situation (dependent) 90 (29.8%) 37 (12.3%) 4.5 ( ) 3.0 ( ) Polypharmacy, cognition, medication regimen adherence, and renal function Impaired cognition 51 (21.6%) 8 ( 3.3%) 13.0 ( ) 11.9 ( ) Polypharmacy and living situation Impaired renal function 124 (40.9%) 39 (20.5%) 2.6 ( ) 2.6 ( ) Nonadherence to medicationregimen 157 (65.1%) 102 (45.9%) 2.6 ( ) 2.3 ( ) Polypharmacy and cognition No. of diseases 0 16 ( 4.8%) 56 (16.9%) 1 (reference) 1 (reference) (39.2%) 144 (43.4%) 7.0 ( ) 5.9 ( ) (55.7%) 132 (39.8%) 11.3 ( ) 8.1 ( ) Polypharmacy and medication regimen adherence No. of previous admissions (51.8%) 192 (57.8%) 1 (reference) 1 (reference) (48.2%) 140 (42.2%) 1.3 ( ) 1.3 ( ) 68

69 Hospital admissions related to medication- study Chapter 2.2 Table 3 continued Frequency a Univariate Multivariate Analysis Variable Cases Controls OR (95%CI) OR (95%CI) Adjusted for n=332 (100%) n=332 (100%) Physician related No. of prescribers (incl. general practitioner) (47.9%) 178 (53.6%) 1 (reference) 1 (reference) 2 or (44.3%) 142 (42.8%) 1.2 ( ) 1.0 ( ) Polypharmacy 4 26 ( 7.8%) 12 ( 3.6%) 2.5 ( ) 1.4 ( ) Medication related Polypharmacy ( 5 drugs,chronically used) 180 (54.2%) 96 (28.9%) 3.0 ( ) 2.7 ( ) Medication regimen adherence OR = odds ratio; 95%CI = 95% confidence interval; NA = not applicable a) Frequency is calculated without missing values. b) Matching variable. 69

70 Impaired cognition, number of comorbidities, impaired renal function, dependent living situation, and nonadherence to the medication regimen were identified as the most important patient-related determinants, whereas polypharmacy was the most important medication-related potential risk factor identified. Many of these risk factors have been mentioned in the literature The HARM Study has several limitations. First, the frequency of medication-related hospitalizations may be underestimated because of the conservative assessment of cases using a 3-step approach (trigger list, confirmation by a physician, and central assessment). On the other hand, this approach is likely to result in high specificity, adding to the reliability of the results. A second limitation is the exclusion of children and psychiatric patients, which may limit the generalizability of the results. Yet, by excluding these very specific populations, we intended to generate results that are more representative of the general population. Third, although the study used a central assessment of preventability, the proportion of preventable cases may not reflect reality. In real life, medical decisions depend on many circumstances that cannot be extracted from medical records. That is the reason we preferably use the term potential preventability. A fourth limitation concerns the use of 21 different researchers, which could have led to variability in the inclusion of cases. By using strict protocols and thorough training, this variability was reduced as much as possible. Finally, the control patients may not be ideal. In the Methods section, we considered the reasons for selecting this group. Despite these considerations, one may still argue that the control patients may be less ill than the unplanned admissions because, to be able to undergo surgery, they need to be reasonably well. However, Table 3 shows that the number of previous admissions is relatively comparable between cases and controls, suggesting that both groups are at equal risk for hospitalization. By matching on age and sex, controls are likely to be equally exposed to medications as are cases. Even when the risks are overestimated in the case group, the ORs are of such a magnitude that taking the overestimation into account would probably still result in identification of the same risk factors. Notwithstanding these limitations, the HARM Study differs from many other studies of medication-related hospitalizations. Major differences concern its prospective design, the number of hospitals included, and the focus on preventability and risk factors. In this study, patients were prospectively included on admission and were followed up until discharge. Other prospective studies 4,7,26 27 identified frequencies of fatal medication-related hospitalizations of 0.15% to 9%, which are comparable to the present results. One other study28 reported that this frequency was 19%, but it was conducted among intensive care admissions only and, therefore, reflects the 70

71 Hospital admissions related to medication- study Chapter 2.2 most serious ADEs. The length of stay mentioned in these studies 4,7,27 ranged from five to ten days, which is also comparable to the present results. Furthermore, the HARM Study was performed in a large representative sample of Dutch hospitals, screening a large number of admissions from all patient groups and wards, thus providing more generalizable outcomes. This study is one of the few multicenter studies on this subject. Other multicenter studies 5,7,29 31 that used comparable methods identified frequencies of medication-related hospital admissions of 2.4% to 17%. However, only one of these studies 5 also investigated the aspect of preventability. Thus, the HARM Study confirms findings from previous studies, thereby strengthening the evidence base of medication-related hospital admissions, and also adds new evidence regarding the frequency, outcome, and risk factors of preventable medication-related admissions. The meta-analysis by Beijer and de Blaey 2 concluded from the available evidence at that time that large-scale studies on preventable medication-related hospital admissions were needed, and this is what the HARM Study provides. Based on findings from the present study, several recommendations can be made. First, the medication use of high-risk patients (e.g., elderly patients with polypharmacy) should be reviewed regularly for potential medication-related problems, such as underprescription and overprescription, interactions, and user convenience. Such an evaluation should also include a cognitive assessment and identification of barriers for medication regimen adherence and should provide tools to facilitate the proper use of medication. The patient should be actively involved in this process and should be given his or her own responsibility in achieving treatment goals. Second, we recommend that physicians and pharmacists exchange more information relevant to adequate medication surveillance, such as comorbidities and clinical laboratory data (e.g., renal function). Third, policy makers should facilitate the development of information technology designed to provide all relevant health care professionals the information necessary for medication assessment and evaluation and to document changes in medication and reasons for these changes. Finally, when analyzing the ADEs most frequently involved in preventable medication-related hospitalizations, the following medication-specific actions should be taken whenever possible: provide gastroprotection for NSAID and low-dose aspirin users at risk for gastrointestinal events, limit the duration of benzodiazepine use, avoid combinations of psychotropic medications, educate users of diuretics and antidiabetics how to act in periods of low food and fluid intake, 71

72 and monitor blood glucose levels in patients in whom corticosteroid therapy is initiated. Further study is needed into the effectiveness of the aforementioned recommendations in reducing the risks of medication-related hospitalizations. Also, confirmation of the risk factors identified in this study and in other large casecontrol studies with different control selections should be undertaken. In addition, other potential risk factors for preventable medication-related hospitalizations should be studied, such as the dosage of medication taken (e.g., in relation to body surface). The HARM Study Group - Bovenij Ziekenhuis, Amsterdam: Marjo Janssen, PhD, and Dana Appelo, PharmD; Bronovo ziekenhuis, Den Haag: Paul van der Linden, PhD, Mireille Toet, BSc, and Zina Brkic, BSc; Diakonessen Huis, Utrecht: Lennart Stoker, PharmD, and Francine Prak, BSc; Erasmus Medisch Centrum, Rotterdam: Pieter Knoester, PhD, and Liselotte Soeting, BSc; HAGA ziekenhuis, Den Haag: Paul le Brun, PhD and Christine Evertse, MSc; Isala klinieken, Zwolle: Hans Harting, PharmD, Frank Jansman, PhD, and Djoek Vogel, PharmD; Langeland ziekenhuis, Zoetermeer: Rob Moss, PharmD, and Martijn Brummer, BSc; Leids Universitair Medisch Centrum, Leiden: Irene Twiss, PhD, Nielka van Erp, PharmD, and Jeroen Diepstraten, PharmD; Maxima Medisch Centrum, Veldhoven: Sjoukje Troost, PharmD and Ebby Ruiz, PharmD; Medisch Centrum Alkmaar, Alkmaar: Paul Kloeg, PharmD, Ingrid van Haelst, PharmD, and Jeroen Doodeman, BSc; Medisch Centrum Haaglanden, Den Haag: Hans Overdiek, PhD, Christien Schmitz, BSc, and Gert de Marie, BSc; Medisch Centrum Leeuwarden, Leeuwarden: Eric van Roon, PhD, Jan Maarten Langbroek, PharmD, and Danielle de Keizer, BSc; Onze Lieve Vrouwe Gasthuis, Amsterdam: Erik Franssen, PhD and Milly Attema, PhD; Reinier de Graaf Groep, Delft: Erik Meijer, PharmD, and Vincent Tan, PharmD; St Antonius ziekenhuis, Nieuwegein: Ed Wiltink, PharmD, and Bram van Arkel, BSc; St Elisabeth Ziekenhuis, Tilburg: Patricia van den Bemt, PhD and Rixt Nynke Eggink, PharmD; Universitair Medisch Centrum Groningen, Groningen: Marian Laseur, PharmD, and Koen Oosterhuis, PharmD; Universitair Medisch Centrum Utrecht, Utrecht: Piet Nauta, PharmD, Moniek Boekweit, BSc, and Marrit van Buuren, BSc; VieCuri Medisch Centrum voor Noord-Limburg, Venlo: Liesbeth van Dijk, PharmD, and Maartje Wijnhoven, PharmD; Wilhelmina Ziekenhuis, Assen: Elsbeth Helfrich, PharmD, Marjan Bouma, PhD, Karen van Loon, PharmD, and Mariet Heins, PharmD; and Zaans Medisch Centrum, Zaandam: Martin Schuitenmaker, PharmD, and Yvonne Scheffer, BSc. Acknowledgements - The authors would like to thank Martin G. Schuitenmaker, PharmD, for making it possible to conduct the study; Svetlana V. Belitser, MSc, for 72

73 Hospital admissions related to medication- study Chapter 2.2 providing statistical support; Patrick C. Souverein, PhD, for performing database handling and programming; Youssef Chahid, PharmD, for performing database validation; and Stephen A. Hudson, PharmD, for his comments on the manuscript. REFERENCES Kohn L, Corrigan J, Donaldson M, Committee on Quality of Health Care in America, Institute of Medicine. To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press; Beijer HJ, de Blaey CJ. Hospitalisations caused by adverse drug reactions (ADR): a metaanalysis of observational studies. Pharm World Sci. 2002;24: Lazarou J, Pomeranz BH, Corey PN. Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies. JAMA. 1998;279: Pirmohamed M, James S, Meakin S, et al. Adverse drug reactions as cause of admission to hospital: prospective analysis of patients. BMJ. 2004;329:15-9. Queneau P, Bannwarth B, Carpentier F; et al, Association Pédagogique Nationale pour l Enseignement de la Thérapeutique (APNET). Emergency department visits caused by adverse drug events. Drug Saf. 2007;30:81-8. Howard RL, Avery AJ, Howard PD, Patridge M. Investigation into the reasons for preventable drug related admissions to a medical admissions unit: observational study. Qual Saf Health Care. 2003;12: Pouyanne P, Haramburu F, Imbs JL, Bégaud B, for the French Pharmacovigilance Centres. Admissions to hospital caused by adverse drug reactions: cross sectional incidence study. BMJ. 2000;320:1036. Von Euler M, Eliasson E, Öhlén G, Bergman U. Adverse drug reactions causing hospitalization can be monitored from computerized medical records and thereby indicate the quality of drug utilization. Pharmacoepidemiol Drug Saf. 2006;15: Stewart RB, Sprinker PK, Adams JE. Drug-related admissions to an inpatient psychiatric unit. Am J Psychiatry. 1980;137: Van der Hooft CS, Sturkenboom MCJM, Van Grootheest K, Kingma HJ, Stricker BHC. Adverse drug reaction-related hospitalizations: a nationwide study in the Netherlands. Drug Saf. 2006;29: Waller P, Shaw M, Ho D, Shakir S, Ebrahim S. Hospital admissions for drug-induced disorders in England: a study using the Hospital Episodes Statistics (HES) database. Br J Clin Pharmacol. 2005;59: Edwards IR, Aronson JK. Adverse drug reactions: definitions, diagnosis, and management. Lancet. 2000;356: Van den Bemt PM, Egberts ACG, de Jong-van den Berg LT, Brouwers JR. Drug-related problems in hospitalised patients. Drug Saf. 2000;22:

74 Ministerie van Volksgezondheid, Welzijn en Sport. Brancherapport ziekenhuiszorg. Available from: [Accessed June 25, 2007]. Aronson JK. Meyler s side effects of drugs: The international encyclopedia of Adverse drug reactions and interactions. 14th ed. Amsterdam, the Netherlands: Elsevier; Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16: Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol. 1997;50: Kramer MS, Leventhal JM, Hutchinson TA, Feinstein AR. An algorithm for the operational assessment of adverse drug reactions, I: background, description, and instructions for use. JAMA. 1979;242: Schumock GT, Thornton JP. Focusing on the preventability of adverse drug reactions. Hosp Pharm. 1992;27:538. Kuijpers MA, van Marum RJ, Egberts AC, Jansen PA. Relationship between polypharmacy and underprescribing. Br J Clin Pharmacol. 2008;65: Rollason V, Vogt N. Reduction of polypharmacy in the elderly: a systematic review of the role of the pharmacist. Drugs Aging. 2003;20: Hampel FR, Ronchetti EM, Rousseeuw PJ, Stahel WA. Robust Statistics: The Approach Based on Influence Functions. New York, NY: John Wiley & Sons; Howard RL, Avery AJ, Slavenburg S, et al. Which drugs cause preventable admissions to hospital? a systematic review. Br J Clin Pharmacol. 2007;63: Hanlon JT, Pieper CF, Hajjar ER, et al. Incidence and predictors of all and preventable adverse drug reactions in frail elderly persons after hospital stay. J Gerontol A Biol Sci Med Sci. 2006;61: Col N, Fanale JE, Kronholm P. The role of medication noncompliance and adverse drug reactions in hospitalizations of the elderly. Arch Intern Med. 1990;150: Zargarzadeh AH, Emami MH, Hosseini F. Drug-related hospital admissions in a generic pharmaceutical system. Clin Exp Pharmacol Physiol. 2007;34: Patel KJ, Kedia MS, Bajpai D, Mehta SS, Kshirsagar NA, Gogtay NJ. Evaluation of the prevalence and economic burden of adverse drug reactions presenting to the medical emergency department of a tertiary referral centre: a prospective study. BMC Clin Pharmacol. 2007;7:8. Rivkin A. Admissions to a medical intensive care unit related to adverse drug reactions. Am J Health Syst Pharm. 2007;64: Onder G, Pedone C, Landi F, et al. Adverse drug reactions as cause of hospital admissions: results from the Italian Group of Pharmacoepidemiology in the Elderly (GIFA). J Am Geriatr Soc. 2002;50: Williamson J, Chopin J. Adverse reactions to prescribed drugs in the elderly: a multicentre investigation. Age Ageing. 1980;9: Schneeweiss S, Hasford J, Göttler M, Hoffmann A, Riethling AK, Avorn J. Admissions caused by adverse drug events to internal medicine and emergency departments in hospitals: a longitudinal population-based study. Eur J Clin Pharmacol. 2002;58:

75

76 Summary Aim Medication errors related to impaired renal function can lead to severe adverse drug events, which may lead to hospital admission. The aim of this study was to determine if medication errors and renal impairment contributes to hospital admission and to gain more information on these errors, which can help to prevent these hospital admissions in the future. Methods We analyzed the 714 medication-related hospital admissions from the HARM (Hospital Admissions Related to Medication)-study. The patients were divided into three groups based on the availability of creatinine levels: Group A, the homemonitored group (n=227), group B, the in-hospital-monitored group (n=420) and group C, the non-monitored group (n=67). Resu lts After assessment 70 admissions (10%) were related to a medication error and renal impairment (A: 29, B: 41, C: none). A dosing error was found in 46 patients (A: 14, B: 32), a drug-drug interaction in 22 patients (A: 13, B: 9) and a drug-disease interaction in 17 patients (A: 10, B: 7). C onclusi ons From these results we conclude that renal impairment and medication may lead to medication-related hospital admissions. From the admissions in the hospitalmonitored group B we conclude that monitoring renal function is relevant and may help to prevent admissions. Although renal function was monitored in group A, relevant medication errors still occurred in this group. Therefore we conclude that adjusting pharmacotherapy according to renal function is relevant and may help to prevent hospital admissions. 76

77 Renal impairment and medication causing hospital admission Chapter 2.3 Introduction Renal impairment is an important health problem. Estimates indicate that 8% of the European population has some stage of chronic renal impairment. 1 This figure has been steadily increasing due to the ageing population and an increase in diabetes and cardiovascular diseases. 2,3 If the present trend continues, the number of people with a chronic kidney disease will double over the next decade. 4 For renally eliminated drugs dosage adjustments are necessary to prevent overdosing or concentration dependent adverse drug reactions. In addition, some medicines can worsen renal function and should therefore be avoided or carefully considered if a patient has renal impairment. Several consensus-based guidelines have been developed to tailor a patient s pharmacotherapy to his or her renal function. 5 9 Medication errors related to impaired renal function; such as lack of monitoring of renal function, ignoring renal impairment or non-compliance with dosing guidelines, frequently occur in the inpatient setting For example, studies have shown that medication doses were not adjusted to renal impairment in 19-67% of hospitalized patients This can lead to serious adverse events 20 such as a major bleeding. 21 Whether clinically relevant errors occur as often within the outpatient setting is less well known. 22 Dosing errors were found in 7 (32%) of the 22 elderly patients admitted to hospital due to renal impairment. 22 Studies that evaluate medication-related hospital admissions show that impaired renal function is a potential risk factor for these admissions, but more detailed information for the outpatient setting from large studies is often lacking. Therefore, in this study we analyzed the medication-related hospital admissions from the multicenter HARM (Hospital Admissions Related to Medication)-study 23 to determine to which extent renal impairment contributes to preventable hospital admissions due to medication errors. Methods S etting 714 cases included in the HARM-study were analyzed in more detail. 23 These 714 cases represent 5.6% of medication-related hospital admissions from a sample of unplanned admissions to a hospital. 332 (46%) of these mediation-related hospital admissions were considered potentially preventable. Relevant information from the medical records of all 714 patients (medical history, diagnostic procedures and outcomes) was recorded. All clinical laboratory data during the year before the present admission and at the time of admission were 77

78 obtained from the admitting hospital laboratory and from the patient s general practitioner (GP) record. A list of the medications dispensed, during one year before admission, was obtained from the patient s community pharmacy records, including detailed information on dispensed drug, amount, date and dosage. The medication related to the reason for admission was termed the HARM-medication. Renal function For all included patients serum creatinine levels were collected and creatinine clearance was estimated by using the Jelliffe-II formula. 26 These patients were divided into three groups based on the availability of creatinine levels. For the first group (group A, the home-monitored group) of patients, at least one creatinine level was available up to 12 months before admission. For this group the most recent creatinine level before hospitalization was used for creatinine clearance calculation. For the second group (group B, the hospital-monitored group) information on creatinine levels was only available on the first day of the admission. For the last group (group C, the non-monitored group), no creatinine levels were available before nor during the first day of hospitalization. Causality assessment All HARM-cases were assessed for a causal relationship between renal function and the medication-related hospital admission. This was done by two independent clinical pharmacists (authors: AL and LD), based on the Kramer algorithm. 27 They assessed if the HARM-medication was prescribed according to the guideline on medication and renal impairment. 5,6 They compared the prescription of the HARM-medication with the recommendations in the guideline. If the drug was not prescribed according to the guideline this was assessed as a medication error such as a drug-drug interaction, inappropriate dosing or a drug-disease interaction related to renal function. The admission was then assessed as a hospital admission related to medication and renal function if all of the following criteria were met: first, the medication error was related to the reason for the hospital admission because the reason for admission was consistent with the outcome of the error as described in the guideline; second, no other more likely explanations were present for the reason for admission, besides the impaired renal function and the medication error; and third: the time relation between the medication error and the reason for admission was as expected. If no information was available on renal function (group C) the prescription of the HARM-medication was also compared with the guidelines. 5,6 If the HARMmedication appeared in the guideline with monitoring recommendations, this was 78

79 Renal impairment and medication causing hospital admission Chapter 2.3 assessed as a monitoring error. Then the reason for admission was compared with the outcome as described in the guideline. Further causality assessment was done as described above. The assessment of the different medication errors is described in more detail below. Drug-drug interaction For all patients the HARM-medication was compared with a list of medications that can cause a drug-drug interaction of which the outcome can be influenced by impaired renal function, e.g. the interaction between an ACE-inhibitor and potassium-sparing diuretic which leads to hyperkalaemia in impaired renal function. If there was a drug-drug interaction in which the renal function (groups A + B) could have played an important role, the reason for admission was compared with the described outcome in the guideline 5,6 and when these two matched, the case was marked as a drug-drug interaction. For group C, with no information on renal function available, all HARM-drugs were compared with the list of medications that can cause a drug-drug interaction of which the outcome can be influenced by impaired renal function. If there was a drug-drug interaction detected from this list, the reason for admission was compared with the described outcome in the guideline. 5,6 When these two matched, the case was marked as a monitoring-error, because monitoring of renal function would be required for the prescribed combination of these drugs. Dosage For all patients with information on the renal function (groups A + B) the dosage regimen of the HARM-medication was assessed in relation to the patient s renal function. The prescribed dosage regimen before admission and the renal function were compared with the recommended dosage regimen in the guideline on medication and renal impairment. 5,6 If the dosage regimen of the HARM-medication was not according to the guidelines the described outcome was compared with the reason for admission and marked as a wrong dose, when these two matched. For group C, with no information on renal function available, all HARM-drugs were compared with the guideline and checked if monitoring of renal function was required for the prescribed dose of the HARM-drug. If so, the described outcome was compared with the reason for admission and marked as a monitoring-error. Drug-disease interaction For all patients the HARM-medication was compared with a list of medications, which can cause a drug-disease interaction with impaired renal function. 5,6 For 79

80 all patients with information on renal functioning (groups A + B) the HARMmedication and renal functioning were compared with the recommendations of the guidelines on medication and renal impairment. 5,6 If the HARM-medication was prescribed despite the guidelines, the described outcome was compared with the reason for admission and marked as a drug-disease interaction if these two matched. For group C, with no information on renal function available, all HARM-drugs were compared with the guidelines and checked if monitoring of renal function was required for prescribing the HARM-drug to prevent a drug-disease interaction. If so the described outcome was compared with the reason for admission and marked as a relevant monitoring-error if these two matched. Main outcome measures The frequency of medication-related hospital admissions due to impaired renal functioning was the main outcome measure. This is defined as the number of medication-related hospital admissions in which the prescribed medication combined with the impaired renal function, or possibly impaired renal function, was associated with the reason for admission, divided by the number of all acute admissions as included in the HARM-study. In addition, the number of the different medication errors was determined in the different subgroups with different availability of information on renal function (group A, B or C: see above under Renal function ). Data analysis All relevant patient data were entered into local databases of the participating hospitals, which were merged. For the comparison of data with the guideline, Microsoft Access 2000 was used. Microsoft Excel was used for calculations. The differences in groups A, B and C were tested with appropriate statistical tests (Student t-test, Chi-square test or Kruskal-Wallis test) using SPSStatistics Result s Renal function Serum-creatinine levels before admission to the hospital were available from 30.8% (227) of the 714 patients included in the HARM-study (group A, the home-monitored group). Creatinine levels were available only on admission to hospital (group B, the hospital-monitored group) for 58.8% (420) of the patients, 80

81 Renal impairment and medication causing hospital admission Chapter 2.3 Figure 1 Hospital admissions related to medication (HARM) and medication errors based on renal impairment a) Number of medication errors. The admission of one patient can be related to more than one medication error. 81

82 information on renal functioning was not available, neither before nor during the first day of hospital stay (group C, the non-monitored group) for 9.4% (67) of the patients (see Figure, box 1). The average estimated creatinine clearance (CrCl) of the patients in group A and group B were similar (mean CrCl in group A: 48.5 ml/min and mean CrCl in group B: 50.6 ml/min; t-test p = 0.359). In group A, 133 patients (59%) had a CrCl below 50 ml/min, 56 patients (25%) had a CrCl below 30 ml/min and 8 patients (4%) had a CrCl below 10 ml/min. In group B, 228 patients (54%) had a CrCl below Table 1 Patient characteristics of the patient groups Group A CrCl available before admission Group B CrCl available on admission Group C CrCl not available p Total number of patients Calculated CrCl in ml/min mean ± SD 48.5 ± ± a Age in years mean ± SD 68.2 ± ± ± b median Gender (male) 50.7% 49.8% 52.2% c No. of medications mean ± SD 11.9 ± ± ± c median No. of diseases mean ± SD 4.76 ± ± ± c,d median Non-adherence e 47.1% 48.7% 44.8% c No. of prescribing physicians mean ± SD 2.04 ± ± ± c median CrCl = creatinine clearance; SD = standard deviation a) t-test. b) Kruskal-wallis test. c) Pearson s Chi-square test. d) Test on difference between group A and group B (Chi-square p = 0.093) shows no difference in the number of diseases. e) A patient is classified as non-adherent if one or more chronic prescribed medications has a calculated refillrate of < 0.8 or > 1.2. The refill rate was calculated by summing the dispensed daily doses, divided by the number of days between the first and the last prescription dispensed

83 Renal impairment and medication causing hospital admission Chapter ml/min, 105 patients (25%) had a CrCl below 30 ml/min and 16 patients (4%) had a CrCl below 10 ml/min. There were no differences in age, gender, number of prescribed medications, adherence to prescribed medication or number of prescribing physicians. In group C fewer diseases were documented (mean: 3 diseases) than in group A and group B (mean for both: 4 diseases) (see Table 1). Causality assessment After causality assessment (Figure 1:2) 29 (13%) of the medication-related hospital admissions in group A (Figure 1:3) and 41 (10%) of the medication-related hospital admissions in group B (Figure 1:4) were assessed as probably due to impaired renal function. None of the admissions in group C could be considered as medication and renal impairment related (Figure 1:5). Therefore a total of 70 or 10% of all cases included in the HARM-study were assessed as medication and renal impairment related. In all these cases at least one medication error was considered to have contributed to the reason for admission to hospital. Many patients had several medication errors related to renal function and admission, as can be seen in the Figure 1. Within the 29 patients in group A (Figure 1:3) a total of 85 medication errors (Figure 1:6-8) were associated with the hospital admission (on average, 2.9 medication errors per patient). Twenty-three patients had medication errors in one class (drug-drug interaction, dosage or drug-disease interaction), four patients had medication errors in two classes and two patients had medication errors in three classes. Table 2 shows the number of patients per class and therefore the totals of these classes are larger than the total of patients in group A (because four patients are in two classes and two patients are in three classes). Within the 41 patients in group B (figure 1:4) a total of 66 medication errors (Figure 1:9-11) were associated with the hospital admission (on average, 1.6 medication errors per patient). Thirty-four patients had medication errors in one class and seven patients had medication errors in two classes (see Table 2 for the numbers of patients per class). Of all studied admissions ( admissions) 0.55% (n=70) was considered as a hospital admission due to medication and impaired renal function. Drug-drug interaction Thirteen patients (5.7%) in group A (Figure 1:6) and nine patients (2.1%) in group B (Figure 1:9) had one drug-drug interaction. The outcome of the drug-drug interactions all contributed to the reason for hospital admission. In group C (Figure 1:5), no such drug-drug interactions were found (see also Table 2). 83

84 Table 2 Number of patients with medication errors related to renal impairment which is also associated with the reason for admission to hospital Group A CrCl available before admission Group B CrCl available on admission Group C CrCl not available Total Total number of patients 227 (31.8%) 420 (58.8%) 67 (9.4%) 714 (100%) Number of patients with a drug-drug interaction (possibly) related to impaired renal function Number of patients with a dosing error dose not according to guideline renal function and reason for hospital admission (possibly) related to wrong dose Number of patients with a drug-disease interaction contra-indicated medication due to impaired renal function and described clinical outcome of drug-disease interaction (possibly) related to reason for admission Total number of patients a hospital admission (possibly) related to medication and impaired renal function CrCl = creatinine clearance a) Number of patients in separate classes do not add up to total number of patients per group (A, B or C), because a patient can have errors in two or three classes. Dosage In group A the dosage of 58 medicines (Figure 1:7) was not adjusted according to the guideline. This was assessed as related to the reason for admission in 14 (6.2% of the patients in group A) patients. The most common medications with a prescribed dosage regimen that was not according to the guideline dosing in renal impairment were ACE-inhibitors (8 patients), diuretics (6 patients) and metformin (5 patients). In group B the dosage of 38 medicines (Figure 1:10) was not adjusted according to the guideline. In 32 patients (7.6% of the patients in group B) this was assessed as a clinically relevant dose-adjustment and this could be related to the reason for admission. Diuretics (8 patients), digoxin (4 patients) and ACE-inhibitors 84

85 Renal impairment and medication causing hospital admission Chapter 2.3 (4 patients) were the most common medicines found with a relevant dosing error in renal impairment. In group C one monitoring error was detected based on the dose of the HARMmedication which should be adjusted if the patient had an impaired renal function. This error was assessed as unlikely to be related to the reason for admission and renal impairment (Figure 1:5). Drug-disease interaction In group A, 14 medicines (Figure 1:8) were found with a possible drug-disease interaction. In 10 (4.4%) patients this was related to the reason for admission. In group B, 19 medicines (Figure 1:11) were found with a possible drug-disease interaction based on the CrCl at the time of admission. In seven patients (1.7%) the outcome as described in the guideline was related to the reason for admission. In group C, ten patients used a medicine that was unsuitable in renal impairment. In these cases the reason for admission was not related to the potential outcome of the drug-disease interaction. Therefore these admissions were not classified as hospital admissions related to medication and renal impairment (Figure 1:5). Discussion Of the 714 medication-related hospital admissions from the HARM-study, 70 admissions (10%) were related to medication and impaired renal function. In these cases one or more medication errors (lack of monitoring, dosing errors, drug-drug interactions and drug-disease interactions), were deemed to contribute to the reason for admission. As errors might have been prevented, we therefore regard all these admissions as potentially preventable. This percentage seems lower than found in inpatients settings but the measured outcome in these latter studies was medication errors instead of medication errors related to a serious adverse event such as an admission. Also the setting was different, namely inpatients instead of outpatients admitted to hospital with a medication-related cause. Our percentage also seems lower than found in the elderly (32%), 22 which is expected as renal function decreases with age. Therefore, these results are not comparable. Nevertheless our study confirms that medication errors can lead to severe renal ADEs in outpatients as other studies have found in different patient populations Another important finding in this study, is the size of the hospital-monitored group B (420 of the 714 patients) and therefore non-compliance with the guideline on 85

86 monitoring which also has been recognized by others. 28 No data were available on renal function before admission in this group, while a considerable number of the medication-related hospital admissions were probably related to impaired renal function and one or more medication errors. In 41 of the 70 cases (59%) no information on renal function was available before the event, which would have been essential for the medication safety in these patients. Therefore an important implication of the results is that improving monitoring of renal function is essential to avoid medication-related events by adjusting the dosage regimen and avoiding critical medication in patients with impaired renal function. Another remarkable finding is the difference in errors in the home-monitored group A and hospital-monitored group B. Although the renal function in group A was monitored in an outpatient setting, more medication errors occur in this group (Figure 1:6-8). Even though the number of medicines used and the number of co-morbidities did not differ between the groups, the patients in this homemonitored group might have had more severe diseases. This may increase their risk for a medication-related hospital admission despite the proper monitoring of renal function. Another explanation may be that after monitoring of the renal function, the proper actions were not taken which may have been due to lack of knowledge on these actions. Several explanations may exist for the identified medication errors. These include prescribers underestimation of the consequences of mild renal impairment and of the slow deteriorating of renal function with age, in terms of subsequent iatrogenic risk. 29 Another explanation might be the poor knowledge of the medication requiring adjustment in renal impairment. A third explanation would be that prescribers use serum creatinine levels as reference, resulting in an overestimation of renal function, instead of calculated creatinine clearances. This number is only a rough guide and renal impairment can remain undetected, especially in the frail and elderly. A final explanation might be the lack of communication and the assumption that another health care provider would monitor renal function and adjust medication according to the patient s renal function. A number of limitations need to be mentioned. The results of this study show that medication in impaired renal function can cause severe adverse drug events that lead to hospitalization but provide no information about the size of the problem in the general population that do not require hospitalization. Also the monitoring practices in non-hospitalized patients remain largely unknown. A second limitation is that our frequency estimation of medication-related- and impaired renal function related hospital admissions may have been too conservative. The frequency of medication-related hospitalizations may be underestimated 86

87 Renal impairment and medication causing hospital admission Chapter 2.3 because of the conservative assessment of admissions using a 3-step approach (trigger list, confirmation by a physician, and central assessment) in the HARMstudy and adding an extra assessment on the relationship with renal function. On the other hand, this approach is likely to result in high specificity, adding to the reliability of the results. Additionally, admissions to psychiatric and pediatric wards and pregnancy related admissions were excluded. In particular, the frequency of medication-related admissions to psychiatric wards can be considerable (10%, 30 23% 31 ) and antipsychotics, hypnotics and antidepressants can cause severe adverse drug events in impaired renal function. This suggests that the number of cases related to medication and renal function in this study is underestimated. A third limitation is the estimation of the renal function. We roughly estimated renal function by calculating creatinine clearance with the Jelliffe-II equation, based on serum creatinine values and age but not on weight, which was often missing in the HARM-database. The results should be interpreted with some caution because it is an estimation of the Glomerular Filtration Rate (GFR) and not directly measured GFR. However, the Jelliffe-II formula has been proven to predict the GFR with a high degree of accuracy also in the elderly, at the low GFR range and in patients with a normal weight Since creatinine clearance based on serum creatinine levels overestimates renal function in the lower region and correlation with renal function decreases with increasing age, the percentage of patients with creatinine clearance less than or equal to 50 ml/min might even be higher and this could contribute also to an underestimation of the results. A fourth limitation is the missing information of laboratory data. Information was collected of the laboratory of the admitting hospital, of the primary care laboratory and from the patients GP. It is possible that we missed information from other laboratories. But in these cases this information would not have been readily available to the patients GP as well, who often prescribed the HARMmedication and therefore could not have adjusted to medication according to the renal function. In addition we only used a single serum creatinine determination prior to admission as an estimation of the GFR. The use of two or more values of outpatient serum creatinine concentration, would likely reduce misclassification and provide a more accurate causality assessment. But this information would also not have been readily available to the patients GP, who made the decision including the medication error based on this information. A final limitation in this study is the central assessment of medication errors and thereby preventability. The proportion of preventable admissions may not reflect reality because in real life, medical decisions depend on many circumstances which cannot be extracted from medical records. 35 Although medication may cause a 87

88 drug-disease interaction in a patient because of its renal function, an alternative drug might not have been possible and the risk of not taking a drug might have been worse. That is the reason we preferably use the term potential preventability. On the other hand, central assessment with strict guidelines ensures equal assessment of every case, independent of the individual hospital and therefore improves the reliability of the outcomes. Despite these limitations this is the first large study providing detailed information on monitoring practices with respect to renal function and medication errors related to impaired renal function in outpatient care leading to hospital admissions. Improvements in monitoring and subsequent prevention of medication errors are necessary, as our study strongly suggests. Computerized clinical decision support systems and pharmacy services may improve the quality of prescribing for patients with renal impairment Programs designed to alert physicians and pharmacists may help to prevent potential adverse drug events if they are sensitive and specific to the prescribed drugs and the patients renal function. 40 These support systems and pharmacy services, which combine laboratory values and medication data, are also needed outside the hospital to prevent renal medication errors in outpatient care by prescribers and pharmacists. But further studies into such systems, services and implementation of both, need to be conducted to provide additional knowledge on the most optimal methods to prevent renal impairment related medication errors in outpatients. C onclusions From this study we conclude that renal impairment and medication may lead to medication-related hospital admissions. A considerable part of these admissions can probably be prevented by monitoring renal function and adjusting pharmacotherapy accordingly (selection of medication and dosage regimen) for every individual patient. Acknowledgement - The authors would like to thank Pamela M. Kato, PhD EdM, UMC Utrecht, for her comments on the manuscript and the HARM-study group for collecting the data. 88

89 Renal impairment and medication causing hospital admission Chapter 2.3 References The European Kidney Health Alliance. Factsheet The kidney in health and disease. Available from: [Accessed November 2009]. Coresh J, Astor BC, Greene T, et al. Prevalence of chronic kidney disease and decreased kidney function in the adult US population: Third National Health and Nutrition Examination Survey. Am J Kidney Dis. 2003;41:1 12. Coresh J, Selvin E, Stevens LA, et al. Prevalence of chronic kidney disease in the United States. JAMA. 2007;298: Zhang QL, Rothenbacher D. Prevalence of chronic kidney disease in population-based studies: systematic review. BMC Public Health. 2008;11:117. Verminderde nierfunctie, doseringsadviezen voor geneesmiddelen. Den Haag: Koninklijke Nederlandse Maatschappij ter bevordering van de Pharmacie; G-standaard. Den Haag: Koninklijke Nederlandse Maatschappij ter bevordering van de Pharmacie; Aronoff GR, Berns JS, Brier ME, et al. 5th ed. Drug prescribing in renal failure: dosing guidelines for adults. Philadelphia: American College of Physicians; Joint Formulary Committee. British national formulary. 58th ed. London: British Medical Association and Royal Pharmaceutical Society of Great Britain; Available from: McEvoy GK, Miller J, Snow EK, et al. American Hospital (AHFS) drug information Bethesda: American Society of Health-System Pharmacists; Bates DW, Cullen DJ, Laird N, et al; ADE Prevention Study Group. Incidence of adverse drug events and potential adverse drug events: implications for prevention. JAMA. 1995;274: Classen DC, Pestotnik SL, Evans RS, et al. Adverse drug events in hospitalized patients. Excess length of stay, extra costs, and attributable mortality. JAMA 1997;277: Lesar TS, Briceland L, Stein DS. Factors Related to Errors in Medication Prescribing. JAMA. 1997;277:312-7 Cantu TG, Ellerbeck EF, Yun SW, et al. Drug prescribing for patients with changing renal function. Am J Hosp Pharm. 1992;49: Van Dijk EA, Drabbe NR, Kruijtbosch M, et al. Drug dosage adjustments according to renal function at hospital discharge. Ann Pharmacother. 2006;40: Falconnier AD, Haefeli WE, Schoenenberger RA, et al. Drug dosage in patients with renal failure optimized by immediate concurrent feedback. J Gen Intern Med. 2001;16: Hu KT, Matayoshi A, Stevenson FT. Calculation of the estimated creatinine clearance in avoiding drug dosing errors in the older patient. Am J Med Sci. 2001;322: Papaioannou A, Clarke JA, Campbell G, et al. Assessment of adherence to renal dosing guidelines in long-term care facilities. J Am Geriatr Soc. 2000;48: Salomon L, Deray G, Jaudon MC, et al. Medication misuse in hospitalized patients with renal impairment. Int J Qual Health Care. 2003;15: Wong NA, Jones HW. An analysis of discharge drug prescribing amongst elderly patients with renal impairment. Postgrad Med J. 1998;74:

90 Hug BL, Witkowski DJ, Sox CM, et al. Occurrence of adverse, often preventable, events in community hospitals involving nephrotoxic drugs or those excreted by the kidney. Kidney Int. 2009;76: Tsai TT, Maddox TM, Roe MT, et al.; National Cardiovascular Data Registry. Contraindicated medication use in dialysis patients undergoing percutaneous coronary intervention. JAMA. 2009;302: Helldén A, Bergman U, von Euler M, et al. Adverse drug reactions and impaired renal function in elderly patients admitted to the emergency department: a retrospective study. Drugs Aging. 2009;26: Leendertse AJ, Egberts AC, Stoker LJ, et al.: HARM Study Group. Frequency of and risk factors for preventable medication-related hospital admissions in the Netherlands. Arch Intern Med. 2008;168: Helldén A, Bergman U, von Euler M, et al. Adverse drug reactions and impaired renal function in elderly patients admitted to the emergency department: a retrospective study. Drugs Aging. 2009;26: McDonnell PJ, Jacobs MR. Hospital admissions resulting from preventable adverse drug reactions. Ann Pharmacother. 2002;36: Jelliffe RW. Creatinine Clearance: Bedside Estimate. Ann Int Med. 1973;79: Kramer MS, Leventhal JM, Hutchinson TA, et al. An algorithm for the operational assessment of adverse drug reactions. I. Background, description, and instructions for use. JAMA. 1979;242: Ten Berg MJ, van den Bemt PM, Huisman A, et al. Compliance with platelet count monitoring recommendations and management of possible heparin-induced thrombocytopenia in hospitalized patients receiving low-molecular-weight heparin. Ann Pharmacother. 2009;43: Chadban SJ, Briganti EM, Kerr PG, et al. Prevalence of kidney damage in Australian adults: The Aus Diab kidney study. J Am Soc Nephrol. 2003;14(7 Suppl 2):S Salem RB, Keane TM, Williams JG. Drug-related admissions to a Veterans Administration psychiatric unit. Drug Intell Clin Pharm. 1984;18:74-6. Stewart RB, Sprinker PK, Adams JE. Drug-related admissions to an inpatient psychiatric unit. Am J Psychiatry. 1980;137: Duncan L, Heathcote J, Djurdjev O, et al. Screening for renal disease using serum creatinine: who are we missing? Nephrol Dial Transplant. 2001;16: Hahn T, Yao S, Dunford LM, et al. A comparison of measured creatinine clearance versus calculated glomerular filtration rate for assessment of renal function before autologous and allogeneic BMT. Biol Blood Marrow Transplant. 2009;15: Xun L, Cheng W, Hua T, et al. Assessing glomerular filtration rate (GFR) in elderly Chinese patients with chronic kidney disease (CKD): A comparison of various predictive equations. Arch Gerontol Geriatr. 2010;51: Van Doormaal JE, Mol PG, van den Bemt PM, et al. Reliability of the assessment of preventable adverse drug events in daily clinical practice. Pharmacoepidemiol Drug Saf. 2008;17:

91 Renal impairment and medication causing hospital admission Chapter Hassan Y, Al-Ramahi YR, Aziz NA, et al. Impact of a renal drug dosing service on dose adjustment in hospitalized patients with chronic kidney disease. Ann Pharmacother. 2009;43: Field TS, Rochon P, Lee M, et al. Computerized clinical decision support during medication ordering for long-term care residents with renal insufficiency. J Am Med Inform Assoc. 2009;16: Chertow GM, Kuperman GJ, Burdick E, et al. Guided medication dosing for inpatients with renal insufficiency. JAMA. 2001;286: Nash IS, Rojas M, Hebert P, et al. Reducing excessive medication administration in hospitalized adults with renal dysfunction. Am J Med Qual. 2005;20:64-9. Indermitte J, Beutler M, Bruppacher R, et al. Management of drug-interaction alerts in community pharmacies. J Clin Pharm Ther. 2007;32: Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol. 1997;50:

92

93

94 Abstract Introduction Adverse drug events (ADEs) can cause serious harm to patients and can lead to hospitalization or even death. ADEs are not only a burden to patients and their relatives but also to society, potentially involving high costs. To provide more information on the economic burden of preventable adverse drug events of outpatients, we performed a cost study on the data collected in the HARM (Hospital Admissions Related to Medication)-study, which looked at the frequency, preventability and risk factors for hospital admissions related to medication. Methods The average costs for a preventable medication-related hospital admission were calculated by summing the direct medical costs and the production losses of all the preventable admissions, taking into account the different types of hospitals (academic and general) and the age of the admitted patients. Resu lts The average medical costs for one preventable medication-related hospital admission were The average production loss costs for one admission were 1712 for a person below 65 years of age. Combining the medical costs and the costs of production losses resulted in average costs of 6009 for one, potentially preventable, medication-related hospital admission for all ages. C onclusi ons The costs of potentially preventable hospital admissions related to medication are considerable. Patient safety interventions to prevent ADEs and hospital admissions may therefore be cost-effective or even cost saving. 94

95 Cost analysis of the HARM- study Chapter 2.4 Introduction Adverse drug events (ADEs) can cause serious harm to patients and can lead to hospitalization or even death. 1,2 Adverse drug events are not only a burden to patients and their relatives but also to society, potentially involving high costs. 3 5 On the one hand, improvement of medication safety and patient safety is a major concern to health care workers and policymakers and has the potential to reduce health care costs. But the increasing budgetary constraints often hamper investments in patient safety improvements. Thus, more insight into the costs of preventable hospital admissions may help to prioritize areas to improve patient safety from an economic perspective in addition to the patient and health care perspective. Some information is already available on costs associated with adverse events and preventable adverse events that occur inside hospitals. A study in the United States of America (US) estimated the costs attributable to an ADE at $ 2595 for all ADEs and $ 4685 for preventable ADEs in Based on these costs and data about the incidence of ADEs, the authors extrapolated that the annual costs attributable to all ADEs and to the preventable ADEs for a 700-bed teaching hospital would be $ 5.6 million and $ 2.8 million, respectively. 3 The direct medical costs in Dutch hospitals 4 (total number of beds in the Netherlands: ) 6 were estimated at a total of 355 million for all adverse events (not just events caused by drugs) and 161 million for preventable adverse events in 2004, which is 1.1% of the expenses of the Dutch health care budget. 7 Information on costs of outpatient adverse drug events leading to hospital admissions is still lacking in The Netherlands, but some information is available from studies performed in the US and the United Kingdom (UK). Estimates of the costs of one medication-related hospital admission vary from US$ 1507 to US$ ,9 Exchanging UK into US$, a large study in the UK estimated these costs at the lower range of this interval. They also suggested that these admissions cost the NHS up to 466 (US$ 786; 542) million annually, 1 which is 0.59% of the British health care budget. 10 Unfortunately only direct medical costs were reported 11 and many of the published studies were either limited to only one 12 or two hospitals, individual units or patient groups 8,13 or reported no information on preventable costs. 14,15 Given the wide range of costs mentioned in literature and the need for information on the economic burden of preventable adverse drug events of outpatients we performed a cost analysis on the data we have previously collected in the HARM (Hospital Admissions Related to Medications)-study. 2 The previous HARM-study was a prospective, multicenter, case-control study where we collected data on approximately unplanned admissions in 21 hospitals in the Netherlands. 95

96 Results revealed that 5.6% (n=714) of hospital admissions were thought to be medication-related. One-half of these (n=332) were considered to be potentially preventable. In the current study, we present the total short-term costs associated with preventable medication-related hospital admissions. In addition, we report costs of different subgroups of admissions based on type of hospital, age, preventability and reason of admission, in order to gain further insight in the potential sizes and areas for cost savings due to possible strategies to prevent ADEs. Method S etting and study population Data were collected from the prospective, multicenter, case-control, HARMstudy on medication-related hospital admissions, which has been described in more detail in a previous publication. 2 Briefly, in this study unplanned (acute) admissions from four university and 17 general hospitals from all regions in the Netherlands were screened for a potential medication-related cause of hospitalization. An unplanned admission was defined as an admission that was not scheduled by the hospital 24 hours before the actual admission. A case-control design was used to determine risk factors for potentially preventable admissions. Controls were patients admitted for elective surgery. The exclusion criteria were age younger than 18 years and admission for obstetric indications, to a psychiatric ward, or for self-poisoning. The causality assessment of admissions was done by using a 3-step approach (trigger list, confirmation by a physician, and central assessment). The central causality assessment was performed by two independent clinical pharmacists according to an adjusted version of the algorithm by Kramer et al. 2,16 In the adjusted version of the algorithm by Kramer et al, three questions need to be answered (in contrast to six questions in the original algorithm): whether the reason for admission is known to be an adverse event of the suspected medicine, whether alternative causes can explain the relationship between the suspected medicine and the adverse event and whether a plausible time relationship exists between the adverse event and the start of medication administration (or the occurrence of the medication error). On the basis of the answers, causality is classified as possible, probable, or unlikely. Cases with an assessment of unlikely were excluded. Preventability was also assessed centrally according to a modified version of the algorithm by Schumock and Thornton. 2,17 In this algorithm, an admission was assessed as potentially preventable when a medication error was made with the medication that caused the hospital admission. If the assessments 96

97 Cost analysis of the HARM- study Chapter 2.4 of the pharmacists disagreed, they met and discussed to reach consensus. This resulted in 714 (5.6%) medication-related hospital admissions of which 332 (46%) were considered potentially preventable. The median length of hospital stay of the 332 potentially preventable medication-related cases was eight days and 24 (7.2%) of these cases were admitted to an intensive care unit. Lack of a clear indication for the medication, nonadherence to the medication regimen, inadequate monitoring and drug-drug interactions were the most common medication errors found. Most of the included admitted patients had many comorbidities: 56% had four or more diseases in their medical history. Besides the number of comorbidities, other risk factors to medication related hospital admissions were identified: impaired cognition, impaired renal function, dependent living situation, non-adherence to the medication regimen and polypharmacy. 2 For inclusion in the cost analysis, the HARM-admissions had to comply with the following inclusion criteria: potentially preventable and availability of information on type of admitting hospital (university or general hospital), length of stay in hospital, length of stay on an intensive care unit (ICU) during the admission, reason for admission and age of the patient. Data collection Of the 332 potentially preventable medication-related hospital admissions, one admission was excluded due to lack of information on length of stay. For all 331 remaining potentially preventable admissions, data were collected on visits to the emergency department, length of hospital stay and length of stay in an ICU. Other admissions, such as controls, non-medication related and non-preventable admissions, were not included in this costing study. Based on the three items mentioned a cost estimate was performed using separate prices for university and general hospitals. 18 For a subset of 153 of the included HARM-admissions (one university hospital and three general hospitals) it was possible to retrieve more detailed information on diagnostic tests, treatment during hospitalization (including medication), specialist consultations and transportation by ambulance by medical chart review. This information was used to determine a more precise cost estimate for this subset. Medical costs For the subset of 153 HARM-admissions, all costs to the health care system were identified during the hospital admission, both related and unrelated to the adverse drug event. For every included admission all costs were valued according to the Dutch Manual for Costing in economic evaluations. 18,19 Application of this manual 97

98 is recommended according to the Dutch guidelines for pharmacoeconomic research. 20 All the identified costs were summed for every admission and deflated up to the year 2006, the year in which the data were collected. Production loss Lost productivity of patients during admission in hospital was also valued for the 331 included admissions. Productivity costs included cost estimates for time off work and reduced productivity on the job. Based on the friction-costing method, standardized costs per day were derived from the costing manual according to the sex and age of the admitted patient, up to the age of The costs for all 331 admissions were then calculated by multiplying the number of days admitted to hospital times the costs per day. As this figure will overestimate the productivity costs because colleagues often undertake the absentees work during normal working hours and after short-term absence productivity is also compensated by the patient during normal working hours, absence from work may not lead to a productivity loss corresponding to 100% of the absence. 22 This compensating mechanism is taken into account within the friction-costing method by applying an elasticity factor. This elasticity factor reflects the change in production compared to the change in labour time. Costs of absence from work shorter than the friction period were calculated as being 80% of the production value during the period of absence. (assuming a heterogeneous labour market and labour time elasticity of production = 0.823) The friction period was not explicitly taken into account, as production losses were only counted during hospitalizations, which were all within the assumed friction period (123 days). 18,23 Note that this is a conservative method of estimating productivity loss. It is limited to the production loss during the admission only while it might be expected that days of absence from work extend beyond the actual days of admission. Extrapolation A medical costs multiplier was calculated to estimate all the direct medical costs of all preventable medication-related hospital admissions. This multiplier was based on the detailed data from the subset of 153 HARM-admissions from four hospitals, separately for the different type of hospitals, and was subsequently applied to the other hospitals lacking this detailed information. For the subset of 153 HARMadmissions, firstly the standardized costs (A) 18,19 of the emergency room (ER) visits, time spent on an ICU and standard costs of the total number of bed days were summed. Secondly, all medical costs related to, for example, diagnostic tests, treatment during hospitalization (including medication), specialist consultation 98

99 Cost analysis of the HARM- study Chapter 2.4 and transportation by ambulance were retrieved by medical chart review and using hospital billings. The sum of the standardized and other medical costs were then calculated (A+B) and divided by the standardized costs (A) to drive the multiplier used to inflate the costs for ER, ICU and other bed days to totals for those hospitals lacking the detailed information. Ergo, the total medical costs of the preventable admissions were calculated by summing the standard costs of a day in the specific type of hospital times the number of bed days, the costs of time spent on an ICU and visits to the accident and emergency rooms, and subsequently multiplying this by the multiplier according to the type of hospital. The average costs for a preventable medication-related hospital admission were calculated by summing the direct medical costs of all the preventable admissions together with the production losses of all the preventable admissions, taking into account the different types of hospitals and different age groups, divided by the total number of included preventable admissions. These average costs per preventable admission were extrapolated to the Dutch situation using national admissions data, regarding the type of hospital and the different age groups. Subgroups The above-mentioned cost calculations were also performed for different groups of admissions and for specific reasons for admission within the sample of 331 admissions, which were often related to medication. The most common reasons for medication related hospitalization were gastrointestinal tract problems (15%) such as gastrointestinal bleeding, constipation and diarrhea. Other common problems were cardiovascular symptoms (11%), respiratory symptoms (8%), and poor glycemic control (6%). Furthermore, the costs were evaluated for admissions of people younger than 65 and older than 65 years of age, separately. Result s Medical costs The 331 potentially preventable medication-related hospital admissions included in this study accounted for 3571 normal-care inpatient days, with a total cost of which appeared to be the main cost driver. Twenty-four of the 331 patients were also admitted to an ICU, accounting for an extra 82 days at an ICU corresponding to a cost of The cost of emergency room visits for the preventable admissions was Costs were calculated for every HARM- 99

100 Table 1 Cost outcomes in of potentially preventable hospital admissions related to medication, divided by type of hospital and age group All hospitals combined University hospital General Hospital < 65 years Medical costs one admission Productivity loss costs one admission Total costs one admission Total costs per year in the Netherlands years Medical costs one admission Productivity loss costs one admission Total costs one admission Total costs per year in the Netherlands Total costs Per admission Per year in the Netherlands admission including normal-care inpatient days, ICU stay and emergency room visits and summed, resulting in or US$ a in total. The average costs for one preventable admission was 5071 or US$ 7367 a, prior to applying the multipliers. In the subset of 153 cases, more detailed costs were retrieved. This resulted in additional costs amounting to approximately 20% of the total admission costs. These costs consisted of transportation by ambulance to the hospital at the time of admission ( ), specialist consultation during admission ( 8409), specialist consultation at admission ( ) and medical procedures (including diagnostic tests) ( ). The detailed cost estimate of the subset was used to estimate the multiplier at 1.22 for the admissions to a general hospital and 1.18 for the admission to a university hospital. Applying these multipliers to the cost estimates of every admission resulted in total medical costs for 331 admissions of or US$ a. The average more detailed medical costs for one preventable admission was 5461 or US$ 7934 a, inclusive of the application of the multipliers. Production loss The total costs of production loss were estimated at or US$ a for all 331 studied admissions. The average production loss costs for one admission were 1712 for a person below 65 years of age. The total production loss costs for a 1.00 is US$ 1.45; exchange rate January

101 Cost analysis of the HARM- study Chapter 2.4 one admission varied between 61 for a 19 year old man who was admitted for one day to for a 37 year old man who was admitted for 38 days to hospital (excluding those aged 65 years and over with theoretical costs of production losses at 0). Extrapolation Combining the medical costs and the costs of production losses resulted in an average of 6009 for one potentially preventable, medication-related hospital admission. Extrapolating this to the Dutch health care system, taken into account the different type of hospitals, resulted in total costs of over 94 million or US$ 137 million a in one year. Of this total 86 million is estimated to be due to medical costs. These direct medical costs reflect 0.49% of the total hospital care budget in the Netherlands (see Table 1). Subgroups Costs of a medication-related hospital admission in a university hospital were estimated to be higher ( 8453) than in a general hospital ( 5748) due to higher inpatient day costs in university hospitals. Yet, the total costs of medication-related hospital admissions in one year were less in university centers (almost 14 million) than the admission costs in general hospitals (almost 81 million) as the total amount of admissions to university hospitals is less than to general hospitals. The average total costs of one admission for patients 65 years and older ( 5637) were estimated to be lower than for younger patients ( 6800). Taking into account the medical costs only, the admission costs of an elderly patient were higher ( 5637) than the costs of a younger patient ( 5088), reflecting the different impacts of production losses in both age groups. The costs of the most common potentially preventable reasons for admission to hospital related to medication are presented in Table 2. The total costs of admissions due to problems of the gastro-intestinal system were estimated to be the highest (over 17 million), followed by cardiovascular problems and respiratory tract problems (both over 8 million) and admissions related to the endocrine system ( 5 million). C omment Extrapolation of the results of this study shows that the total costs associated with preventable medication-related hospital admissions in the Netherlands are more 101

102 Table 2 Cost outcomes of potentially preventable hospital admissions related to medication, divided by most common reason for admission Reason for admission number of admissions direct medical costs production loss costs total costs total costs per year n (%) a per admission per admission per admission in the Netherlands Gastro-intestinal (GI) system GI tract bleeding 48 (14.5%) Other GI tract symptoms (e.g., diarrhea, constipation) 22 ( 6.6%) Circulatory system Cardiovascular symptoms (e.g., dysrhythmias, heart failure) 35 (10.5%) Respiratory system Respiratory symptoms (e.g., dyspnea) 26 ( 7.8%) Endocrine system Hypoglycemia or hyperglycemia 20 ( 6.0%) a) Percentage of the total number of potentially preventable hospital admissions related to medication (n=332). 102

103 Cost analysis of the HARM- study Chapter 2.4 than 94 million. Eighty six million euro of the 94 million is due to medical costs. This reflects 0.21% of the total health care costs and 0.49% of the hospital costs in the Netherlands. The main cost driver is bed occupancy and therefore costs are highly dependent on length of stay in hospital, which in our framework also largely determined other cost components, such as production loss. The median length of hospital stay in our patient group at eight days 2 is similar to the UK study by Pirmohamed et al. 1 but the average costs per inhabitant are lower in our study: 5.9 per person in the Netherlands per year versus 9 per person in the UK per year. This difference can be explained by the selection of admissions. We only calculated the costs of potentially preventable admissions, while all medication-related admissions were taken into account in the UK study. The total costs of one hospitalization of 8453 in a university hospital and 5748 in a general hospital, are within range of previously published smaller studies. 8 The estimated total annual cost of more than 94 million reflects a considerable amount and justifies investments in patient safety that might not only prevent such adverse events but might even be cost saving. Our study has a number of limitations. Firstly, our cost estimation may be too conservative. The frequency of medication-related hospitalizations may be underestimated because of the conservative assessment of admissions using a 3-step approach. On the other hand, this approach is likely to result in high specificity, adding to the reliability of the results. Secondly, we only accounted for short-term costs: medical costs during the hospital admission and production loss costs incurred from the time in hospital only. Costs related to referral to a tertiary care center or outpatient health care after discharge, were not taken into account, neither were non-medical direct costs such as travel costs hence and forth the hospital. Also productivity loss after discharge was not taken into account. It might be expected that days of absence from work extend beyond the actual period of the admission only. All this may have led to an underestimation of the costs. Furthermore, although a thorough search was performed of the medical charts of the included patients, non-invasive procedures are often underreported, whereas surgical interventions are well-documented. Since non-invasive procedures are not cost-drivers, we do not expect that this has led to major distortions in our results. We also do note that production loss costs may have been slightly overestimated. The included patients in the HARM-study had relatively many co-morbidities and are therefore more likely to be chronically ill and more likely to be less productive. On the other hand the production loss accounts for only 8% of the total costs and the overestimation of the total costs is therefore only a few percent. 103

104 The design of the initial HARM-study was such that admissions to a psychiatric ward were excluded, as well admissions of children and pregnancy-related admissions. Especially the frequency of medication-related admissions to a psychiatric hospital or hospital ward can be considerable (10% 24 to 23% 25 ) and therefore exclusion of these admissions may result in an underestimation of true costs. With this costing study based on the HARM-study, the calculated costs are limited to medication related problem that arised before admission to hospital. The calculated costs do not include the costs from adverse drug event that occurred during the admission and might have prolonged stay in hospital or transfer to the ICU. Our study was done for the Netherlands. Obviously, our cost estimates may not be straightforwardly extrapolated to other countries with different health care systems, different relative costs between resource-use components and different use of medications. Despite the limitations in our study, the data used in this costing study may be considered reliable as they are obtained from a large representative sample of Dutch hospitals, with screening a large number of admissions from many patient groups and wards, thus providing reliable information on the burden of the problem and on potentially preventable costs. Furthermore, the thorough method of medical chart review of a large unbiased sample of admissions resulted in a consistent and robust multiplier to determine the actual costs of the hospital admissions. Based on the findings from the HARM-study, combined with this costing study, several recommendations can be made. First, reviewing the medication use of highrisk patients (e.g., elderly patients with polypharmacy) for potential medicationrelated problems. The focus in a review should be on the medication errors identified in the HARM-study to prevent these admissions and save costs. Therefore reducing overprescription, improving compliance, monitor drug-therapy and preventing drug-drug interactions may save costs if lowering the frequency of medicationrelated hospitalizations. 26 Second, when analyzing the most common reason for admission combined with the costs, some interventions might be considered to prevent these costs. The provision of gastroprotection for NSAID users is effective to prevent gastro-intestinal events 27,28 and is also cost effective although this is mostly dependent on the price of the protective treatment, which can differ in different health care settings. 29,30 Monitoring of blood glucose levels in diabetic patients can prevent hypoglycemia or hyperglycemia 31 and will be cost effective in certain patient groups. 32,33 Several laxatives are effective to treat constipation 34,35 but there are not sufficient data on cost-effectiveness of different laxatives and treatment strategies in the management of constipation or opiate related constipation. Medication related constipation was detected as an important problem in the HARM-study. Further study is needed into 104

105 Cost analysis of the HARM- study Chapter 2.4 the cost-effectiveness of these recommendations in reducing the risks of medicationrelated hospitalizations for certain patient groups. In addition, more information is needed on direct medical costs and indirect costs, related to the medication-related admissions, after discharge from hospital. C onclusions The cost estimates of potentially preventable hospital admissions related to medication are considerable. Insight into the subclasses of medication-related hospitalizations that are related with the highest costs, offers a starting point for patient safety interventions, which may be cost-effective or even cost saving. References Pirmohamed M, James S, Meakin S, et al. Adverse drug reactions as cause of admission to hospital: prospective analysis of patients. BMJ. 2004;329:15-9. Leendertse AJ, Egberts AC, Stoker LJ, van den Bemt PM; HARM Study Group. Frequency of and risk factors for preventable medication-related hospital admissions in the Netherlands. Arch Intern Med Sep 22;168: Bates DW, Spell N, Cullen DJ, et al. The costs of adverse drug events in hospitalized patients. Adverse Drug Events Prevention Study Group. JAMA. 1997;277: Hoonhout LH, de Bruijne MC, Wagner C, et al. Direct medical costs of adverse events in Dutch hospitals. BMC Health Serv Res. 2009;9:27. Patel KJ, Kedia MS, Bajpai D, et al. Evaluation of the prevalence and economic burden of adverse drug reactions presenting to the medical emergency department of a tertiary referral centre: a prospective study. BMC Clin Pharmacol. 2007;7:8. Giesbers H. Hospital bed capacity 2003 [in Dutch: Beddencapaciteit ziekenhuizen 2003]. In: Volksgezondheid Toekomst Verkenning, Nationale Atlas Volksgezondheid. Bilthoven: RIVM; June Available from: Zorg\ Ziekenhuiszorg\ Algemene en academische ziekenhuizen\ Aanbod] [Accessed December 2009]. Van Hilten O, Mares AMHM. Figures in Health and Healthcare 2007 [in Dutch: Gezondheid en zorg in cijfers 2007]. Voorburg/Heerlen: CBS; Senst BL, Achusim LE, Genest RP, et al. Practical approach to determining costs and frequency of adverse drug events in a health care network. Am J Health-Syst Pharm. 2001;58: Rodriguez-Monguio R, Otero M, Rovira J. Assessing the economic impact of adverse drug effects. Pharmacoeconomics. 2003;21:

106 HM Treasury. Table 5.4 public sector current and capital expenditure on services by function, to [online]. London, United Kingdom; 29 May Available from: [Accessed November 2009]. Wasserfallen J, Livio F, Buclin T, et al. Rate, type and cost of adverse drug reactions in emergency department admissions. Eur J Intern Med. 2001;12: Dartnell JGA, Anderson RP, ChohanV, et al. Hospitalisation for adverse events related to drug therapy: incidence, avoidability and costs. Med J Aust. 1996;1634: Bloom BS. Direct medical costs of disease and gastrointestinal side effects during treatment for arthritis. Am J Med. 1988;84 Suppl. 2A:20-4. Ramesh M, Pandit J, Parthasarathi. Adverse drug reactions in a south Indian hospital--their severity and cost involved. Pharmacoepidemiol Drug Saf. 2003;12: Moore N, Lecointre D, Noblet C, et al. Frequency and cost of serious adverse drug reactions in a department of general medicine. Br J Clin Pharmacol. 1998;45: Kramer MS, Leventhal JM, Hutchinson TA, Feinstein AR. An algorithm for the operational assessment of adverse drug reactions, I: background, description, and instructions for use. JAMA. 1979;242: Schumock GT, Thornton JP. Focusing on the preventability of adverse drug reactions. Hosp Pharm. 1992;27:538. Oostenbrink JB, Bouwmans CAM, Koopmanschap MA, Rutten FFH. Dutch guidelines for pharmacoeconomic research. [in Dutch: Handleiding voor kostenonderzoek; methoden en standaard kostprijzen voor economische evaluaties in de gezondheidszorg. Geactualiseerde versie 2004]. Diemen: College voor zorgverzekeringen; Oostenbrink JB, Koopmanschap MA, Rutten FF. Standardisation of costs: the Dutch Manual for Costing in economic evaluations. Pharmacoeconomics. 2002;20: College voor zorgverzekeringen (CVZ). Richtlijnen voor farmaco-economisch onderzoek. Diemen: CVZ; Koopmanschap MA, van Ineveld BM. Towards a new approach for estimating indirect costs of disease. Soc Sci Med. 1992;34: Jacob-Tacken KH, Koopmanschap MA, Meerding WJ, Severens JL. Correcting for compensating mechanisms related to productivity costs in economic evaluations of health care programmes. Health Econ May;14: Koopmanschap MA, Rutten FFH, van Ineveld BM, et al. The friction cost method for measuring indirect costs of disease. J Health Econ. 1995;14: Salem RB, Keane TM, Williams JG. Drug-related admissions to a Veterans Administration psychiatric unit. Drug Intell Clin Pharm. 1984;18:74-6. Stewart RB, Sprinker PK, Adams JE. Drug-related admissions to an inpatient psychiatric unit. Am J Psychiatry. 1980;137: Zermansky AG, Silcock J. Is medication review by primary-care pharmacists for older people cost effective?: a narrative review of the literature, focusing on costs and benefits. Pharmacoeconomics. 2009;27: Yeomans ND, Tulassay Z, Juhasz L, et al. A comparison of omeprazole with ranitidine for ulcers associated with nonsteroidal anti-inflammatory drugs. Acid Suppression Trial: Ranitidine vs. Omeprazole for NSAID-associated Ulcer Treatment (ASTRONAUT) Study Group. N Engl J Med. 1998;338:

107 Cost analysis of the HARM- study Chapter Hawkey CJ, Karrasch JA, Szczepanski L, et al. Omeprazole compared with misoprostol for ulcers associated with nonsteroidal anti-inflammatory drugs. Omeprazole vs. Misoprostol for NSAID-Induced Ulcer Management (OMNIUM) Study Group. N Engl J Med. 1998;338: Brown TJ, Hooper L, Elliott RA, et al. A comparison of the cost-effectiveness of five strategies for the prevention of non-steroidal anti-inflammatory drug-induced gastrointestinal toxicity: a systematic review with economic modelling. Health Technol Assess. 2006;10: Vonkeman HE, Braakman-Jansen LM, Klok RM, Postma MJ, Brouwers JR, van de Laar MA. Incremental cost effectiveness of proton pump inhibitors for the prevention of non-steroidal anti-inflammatory drug ulcers: a pharmacoeconomic analysis linked to a case-control study. Arthritis Res Ther. 2008;10:R Coster S, Gulliford MC, Seed PT, Powrie JK, Swaminathan R. Monitoring blood glucose control in diabetes mellitus: a systematic review. Health Technol Assess. 2000;4: Gray A, Raikou M, McGuire A, et al. Cost effectiveness of an intensive blood glucose control policy in patients with type 2 diabetes: economic analysis alongside randomised controlled trial (UKPDS 41). United Kingdom Prospective Diabetes Study Group. BMJ. 2000;320: Tunis SL, Minshall ME. Self-monitoring of blood glucose (SMBG) for type 2 diabetes patients treated with oral anti-diabetes drugs and with a recent history of monitoring: costeffectiveness in the US. Curr Med Res Opin. 2010;26: Miles CL, Fellowes D, Goodman ML, Wilkinson S. Laxatives for the management of constipation in palliative care patients. Cochrane Database Syst Rev Oct 18;(4): CD Mihaylov S, Stark C, McColl E, et al. Stepped treatment of older adults on laxatives. The STOOL trial. Health Technol Assess. 2008;12:

108

109

110

111

112 Abstract Background Medication can be effective but can also be harmful and even cause hospital admissions. Medication review or pharmacotherapy review has often been proposed as a solution to prevent these admissions and to improve the effectiveness and safety of pharmacotherapy. However, most published randomised controlled trials on pharmacotherapy reviews showed no or little effect on morbidity and mortality. Therefore we designed the PHARM (Preventing Hospital Admissions by Reviewing Medication)-study with the objective to study the effect of the total pharmaceutical care process on medication related hospital admissions and on adverse drug events, survival and quality of life. Methods/Design The PHARM-study is designed as a cluster randomised, controlled, multi-centre study in an integrated primary care setting. Patients with a high risk on a medication related hospital admission were included in the study with randomisation at GP (general practitioner) level. We aimed to include patients, 7100 in each arm, from at least 142 pharmacy practices. The intervention consisted of a patient-centred, structured, pharmaceutical care process. This process consisted of several steps, was continuous and occurred over multiple encounters of patients and clinicians. The steps of this pharmaceutical care process were a pharmaceutical anamnesis, a review of the patient s pharmacotherapy, the formulation and execution of a pharmaceutical care plan combined with the monitoring and follow up evaluation of the care plan and pharmacotherapy. The patient s own pharmacist and GP carried out the intervention. The control group received usual care. The primary outcome of the study was the frequency of hospital admissions related to medication within the study period of 12 months of each patient. The secondary outcomes were survival, quality of life, adverse drug events and severe adverse drug events. The outcomes will be analysed by using mixed-effects Cox models. C onclusi ons The PHARM-study is one of the largest controlled trials to study the effectiveness of the total pharmaceutical care process. The study should therefore provide evidence as to whether the entire pharmaceutical care process should be implemented in the primary care setting. 112

113 PHARM- study design Chapter 3.1 Background Drug regulatory authorities give a medicine marketing authorisation if the balance between efficacy and safety is considered positive, based on the available evidence in the studied population. On the level of the individual patient treated with one or more drugs in daily clinical practice, it is, however, hard to predict whether the positive effects outweigh the negative effects. Individual (patho)physiological and psychological characteristics of the patient, concomitantly used medication as well as the treatment setting, can largely modulate pharmacokinetics, pharmacodynamics and behaviour of the patient and thereby treatment outcomes. Unfortunately for some patients the net effect of pharmacotherapy is not positive but harmful. For example, several studies have shown that 3 5% of all hospital admissions is medication related. 1 The Dutch multi-centre HARM-study on medication-related hospital admissions, showed that almost half (46%) of these admissions were considered as potentially preventable admissions. Patients who were especially at risk of medication-related admissions were elderly patients with multiple drug use, who are cognitive impaired or who are non adherent to their pharmacotherapy. 1 5 Medication review or pharmacotherapy review has often been proposed as a solution to improve the effectiveness and safety of pharmacotherapy. 2 However, most published randomised controlled trials on pharmacotherapy reviews showed no or little effect on morbidity and mortality. 6,7 These studies differed largely with respect to the nature and extensiveness of the review techniques, the outcomes studied, setting and follow-up time and results are therefore difficult to compare. One study reported an increase in emergency readmissions, 8 while in the other studies there is no suggestion that patients were harmed by the interventions, 9,10 and even some consistency in suggesting that falls 11 and hospital admissions 12,13 might be reduced. Potentially relevant elements in these studies were a review of a full available medical and drug history, structured pharmaceutical care plan approach, combined effort of pharmacist and GP and involvement and commitment of the patient. Hence there is still a need for large studies to evaluate the effectiveness of a so called patient centered pharmaceutical care process, including reviewing pharmacotherapy, formulating a pharmaceutical care plan and monitoring and follow up evaluating of pharmacotherapy, in daily practice of an integrated primary care setting. Therefore we designed the PHARM-study with the objective to study the effect of the total pharmaceutical care process on medication related hospital admissions and on adverse drug events, survival and quality of life in patients with a high risk of medication related hospital admissions. This study is a practice based, 113

114 cluster randomised controlled intervention study in an integrated primary care setting in the Netherlands. The study design with its explanation and definition of the different steps in of the pharmaceutical care process will be described in this paper. The results of the study will be presented in a separate paper. Methods/Design Par ticipants Only patients at high risk for a medication-related hospital admission were included in the study. To identify these patients we used risk factors from the HARM study as inclusion criteria. 2 This resulted in four criteria regarding age, polypharmacy, type of drug class used and non-adherence. 14 (See Table 1 for detailed information on inclusion criteria) Patients could be included if they met all four of these criteria The number of included patients was limited to the capacity of the GP and the pharmacist. If too many patients were eligible for inclusion, random selection by Table 1 Criterion Definition of the inclusion criteria of the PHARM-study Definition 65 years or older 65 years of age or older, at the time of inclusion. Polypharmacy Five or more drugs with different chemical substances. Every separate drug is prescribed and dispensed, at least 3 times in the 12 months before inclusion and dispensed at least once in the 6 months before inclusion. ATC A or ATC B drug One or more drugs from the Anatomical Therapeutic Chemical (ATC) 25 class A: alimentary tract and metabolism or ATC class B: blood and blood forming organs. At least one of each drug prescribed and dispensed in the 12 months before inclusion. Non-adherence At least one drug with a refill rate below 0.8 or above 1.2. This refill rate was calculated by dividing the number of dispensed daily doses, by the number of days between the first and the last prescription dispensed. The refill rate was calculated for all chronically prescribed drugs, as defined above under polypharmacy, of which at least 60 daily doses were dispensed in the 12 months before inclusion. The number of dispensed daily doses was calculated by summing the dispensed daily doses between the first to the last prescription date and multiplying it by factor 1.1 to correct for irregular drug use and early collection of the prescription. The refill rate was only calculated for oral and inhalation drugs with a clear prescribed dose regimen. It was therefore not calculated for PRN (Pro Re Nata = as needed) drugs. 114

115 PHARM- study design Chapter 3.1 computer based on patient serial number, was used within the eligible group of one GP and pharmacist, to comply with the limit. The pharmacist and GP both could include patients in 2008 and 2009, with a follow-up period of 12 months. Patients were excluded if they were resident in a nursing home, if their life expectancy was less than three months or if they did not give informed consent. Patients had to be withdrawn from the study when they were no longer a patient of the participating GP or pharmacist. This is possible when a patient moves to another area or to a nursing home or when a patient withdraws his/her informed consent. The study was conducted in agreement with the principles of the Declaration of Helsinki (Edinburgh 2000). The patient s pharmacist explained the procedure, possible benefit and burden of participation in the study to each patient and provided an informed consent form approved by the ethical review board of the METOPP (Medisch-Ethische Toetsing Onderzoek Patiënten en Proefpersonen). The patient was asked to sign the form prior to inclusion into the study. Patient data from this study was coded by the patient s own pharmacist. Analysing and publication of the results of this study will be performed anonymously. Study setting The study was carried out in 2008 and 2009 in an integrated primary care setting, by the patient s own GP, pharmacist and practice nurse if available. All Dutch GPs, practice nurses and pharmacists working in primary care were informed about the study and invited to participate by mail, by websites, by several articles in pharmacy and GP journals and by several presentations at regional meetings or at national symposia. They were asked to cooperate as a group of at least one pharmacist and at least two GPs. This cooperation included sharing data from electronic medical records required for the study. The GPs were allocated as an intervention GP or a control GP, by random selection. Suppor t Before and during the intervention period the GP and the pharmacist were offered support by training, education and a web based community of practice with a helpdesk on pharmacotherapy. The web-based community of practice offered a place of sharing information and experiences within the group of participants, provided all materials for the study and established for the researchers a place to encourage and support the participants and monitor the progress of the study. The pharmacist received training on communication skills focussing on the pharmaceutical anamnesis, motivation of patients and communicating with the GP. 115

116 Figure 1 Design PHARM study ATC = Anatomical Therapeutic Chemical; GP = general practitioner; EQ5D = EuroQuol5Dimensions; VAS = Visual Analogue Scale 116

117 PHARM- study design Chapter 3.1 The GP and the pharmacist participated together in at least three workshops on the study protocol, assessment of pharmacotherapy and the assessment of adverse drug events. The helpdesk was managed by a clinical pharmacist with support of experts on pharmacotherapy, medication safety and care of the elderly. The GP from the control patients did not received any additional training or support, other than generally in usual care. Inter venti on The intervention consisted of a patient-centred, structured, pharmaceutical care process. This process consisted of several steps, was continuous and occurred over multiple encounters of patients and clinicians. The steps of this pharmaceutical care process were a pharmaceutical anamnesis, a review of the patient s pharmacotherapy, the formulation and execution of a pharmaceutical care plan combined with the monitoring and follow up evaluation of the care plan and pharmacotherapy (see Figure 1). Phar mace utical anamnesi s The purpose of the pharmaceutical anamnesis was to gather information from the patient and its drug therapy. The anamnesis was performed by the pharmacist who met the patient at home or in a private consultation room at the pharmacy or at the GP practice. During the pharmaceutical anamnesis the patient s medication use and experiences therewith were discussed, including the patient s current pharmacotherapy, drug history, drug use, possible allergies or intolerabilities and concomitant use of OTC drugs or other health products, but also the patient s beliefs, perceptions, understandings, attitudes, and concerns of the pharmacotherapy. Pharmacotherapy review The purpose of the pharmacotherapy review was to identify potential drug therapy related problems, and pharmaceutical care issues. A pharmaceutical care issue is defined as an element of a patient s drug related need, which can lead to a drug therapy related problem. We defined a drug therapy related problem as any undesirable event or risk thereof, experienced by the patient that involves or is suspected to involve pharmacotherapy and that actually or potentially interferes with the desired patient outcome. 15,16 The pharmacist identified these possible drug therapy related problems (DTPs) by combining sociological, patho-physiological and pharmacological knowledge of the patient in a systematic way. This structure was derived from the classification by Strand et al. 15,16 as defined and described in Table 2 and in Table 3. The pharmaceutical care issues are described in Table

118 Table 2 Classification and definition of drug therapy problems Category of drug therapy problem Question to identify drug therapy problem Drug therapy problem Definition Indication Does the patient have an indication for each of his/her drug therapies, and is each of the patient s indications being treated with drug therapy? The patient has a medical condition or is experiencing symptoms that Additional drug therapy required Unnecessary drug therapy requires the initiation of new or additional drug therapy or is at a high-risk to develop a new medical condition for which additional drug therapy is indicated. The patient is taking drug therapy that is unnecessary given his or her present condition or the patient is at risk to develop a new medical condition that is a result of taking an unnecessary drug for which there is no valid medical indication. Effectiveness Are the drug therapies effective for his/her medical condition? Are the intended outcomes of drug therapy reached? The drug product is not being effective at producing the desired response. Ineffective drug therapy The patient is not experiencing the intended positive outcome from a certain drug regimen or the intended outcome is not reached. Or an alternative drug therapy has a higher probability of producing the desired outcome, or an alternative drug therapy is equally effective but less expensive. Dosage too low The patient has a medical condition for which too little of the correct drug is being taken to produce the desired beneficial outcome or the patient is at risk to develop a new medical condition because too little of the correct drug is being taken to expect a beneficial outcome. The patient s drug concentration in the body can be below the desired therapeutic range or the timing of prophylaxis can be inadequate for the patient or dose and interval can be inadequate for the patient or drug, dose, route or formulation conversions were inadequate for the patient. 118

119 PHARM- study design Chapter 3.1 (Table 2 continued) Category of drug therapy problem Question to identify drug therapy problem Safety Are the drug therapies as safe as possible? Is everything done to keep them as safe as possible? Drug use Is the patient able and willing to comply with the drug therapies as prescribed? Are the drug therapies as convenient as possible to the patient? Drug therapy problem Definition The patient has a medical condition or is experiencing symptoms or is at risk Adverse drug event of developing a medical condition which is undesired effect and is related to the drug therapy. This can be an idiosyncratic reaction to the drug, an allergic reaction to the drug or a pharmacologically expected reaction to the drug, possible due to a medication error. Dosage too high The patient has a medical condition for which too much of the correct drug is being taken or the patient is at risk to develop a new medical condition because too much of the correct drug is being taken. The patient s drug concentration in the body can be above the desired therapeutic range or the drug dose escalating can be too rapidly or there can be drug accumulation from chronic administration or dose and interval can be inadequate for the patient or drug, dose, route or formulation conversions were inadequate for the patient. Drug use problem Drug use problem is defined as the patient s inability or unwillingness to take a drug regimen that the general practitioner, pharmacist or other health care provider has clinically judged to be appropriately indicated, adequately efficacious and able to produce the intended outcomes without any undesired effects. 119

120 Table 3 Description of drug therapy problems a Drug therapy problem Common causes of drug therapy problem Examples Additional drug therapy required A medical condition requires the initiation of drug therapy. Preventive drug therapy is required to reduce the risk of developing a new condition (according to the national guidelines). A medical condition requires additional pharmacotherapy to produce an additive of synergistic effect. The patient is suffering from pain with no analgesic therapy. A patient with chronic heart failure due to left ventricular systolic dysfunction, without an ACE-inhibitor or an angiotensin receptor blocker. A patient with atrial fibrillation without antithrombotic therapy. Calcium and vitamin D supplements for a patient with osteoporosis who is already taking a bisphosphonate. Unnecessary drug therapy There is no current valid medical indication for the drug therapy for the individual patient. Multiple drug products are being used for a medical condition that requires single drug therapy. The medical condition is more appropriately treated with non-drug therapy or lifestyle changes. Drug therapy is being taken to treat an avoidable adverse event associated with another medication. Lifestile ( e.g. drug abuse, alcohol use, diet, smoking) is causing the problem. A patient is using a low dose of aspirin without a high risk of cardiovascular disease or any signs of a cardiovascular disease. A patient is using three different laxative products in an attempt to treat his constipation. A patient is using a benzodiazepine every night as a hypnotic drug for three years while it is better to recommend alternative sleeping patterns, sleep hygiene and exercise. A patient is using furosemide to prevent swollen ankles. A patient is using paracetamol combined with codeine (500/10) and is suffering from constipation which is treated with lactulose and bisacodyl occasionally. A patient uses a protonpump inhibitor to treat dyspepsia associated with alcohol abuse. Ineffective drug therapy The drug is not effective for the medical problem. The drug product is not the most effective for the indication being treated. The formulation of the drug products is inappropriate. The drug is not effective because of the characteristics of the patient. (e.g. renal impairment, hepatic function) A patient is using an antibiotic for a common cold (viral infection). A patient with benign prostatic hyperplasia uses doxasozine for more than four years. A patient with severe COPD uses salbutamol in a turbuhaler. A patient with renal impairment uses a thiazide to lower the blood pressure. 120

121 PHARM- study design Chapter 3.1 (Table 3 continued) Drug therapy problem Common causes of drug therapy problem Examples Dosage too low The dose is too low to produce the desired outcome. The dosage interval is too long to produce the desired outcome. A drug-drug interaction reduces the amount of active drug available and the dose is not adjusted too produce the desired outcome. The duration of the drug therapy is too short to produce the desired outcome. A patient is prescribed simvastatine 10 mg every other day after a myocardial infarction. A patient uses 500 mg paracetamol, only twice a day, to control chronic pain in osteoarthritis. A patient uses 375 mg amoxicilline, only once a day, to treat an airway infection. A patient uses acenocoumarole and vitamin K. A patient uses paroxetine for 4 days to treat anxiety. Adverse drug event The drug causes an undesirable reaction that is not doserelated. A safer product is required due to risk factors. A drug interaction, with another drug or food, causes an undesirable reaction that is not dose-related. The drug is contraindicated due to risk factors or other diseases.the drug causes an allergic reaction. A drug dosage was increased or decreased too fast. A drug alters the patient s laboratory test results due to interference from a drug he/she uses. A patient on a low dose of aspirin is experiencing bruises. An elderly patient uses flurazepam to sleep and is experiencing drowsiness at day time. A patient who uses methotrexate gets prescribed cotrimoxazole for an infection (increase of anti-folate effect which can result in haematopoietic suppression). A patient gets prescribed indometacin to control chronic pain, which is contraindicated because of his/her history with a peptic ulcer. A patient is prescribed flucloxacillin for a dermal infection and develops a rash after the second dose. A patient who uses prednisolone 20 mg every day for the last 6 months for arthritis symptoms iss instructed to take 10 mg for 2 more days and then discontinues the medication. A patient had high blood glucose levels, due to the start of prednisolone therapy. Positive ketone test in urine due to captopril use. 121

122 (Table 3 continued) Drug therapy problem Common causes of drug therapy problem Examples Dosage too high The drug causes an undesirable reaction due to too high dose. The dosing frequency of the drug is too short. The duration of drug therapy is too long. The drug dose is too high in the patient because of its characteristics (excretion). A drug-drug interaction occurs resulting in a toxic reaction to the drug. The dose of the drug was administered too rapidly. A patient develops bradycardia resulting from a high (0.5 mg) daily dose of digoxine.hyperkaleamia after a dose of amiloride 10 mg three times a day. A patient who experienes nasal congestion uses a nasal spray with xylometazoline for four weeks. A patient with impaired renal function (creatinine clearance 20 ml/min) is prescribed a normal dose of 300 mg allopurinole a day, which causes nausea. A patient has an increased International Normalized Ratio (INR) after given metronidazole while also using acenocoumarole. Cardiac arrest after infusion of a bolus of potassium phosphate (5 ml intravenously) instead of slow infusion. Drug use problem The patient does not understand the instructions. The patient prefers not to take the medication. The patient forgets to take the medication. The patient cannot administer the drug appropriately him/herself. The drug therapy does not comply with the lifestyle of the patient. The patient has no access to the medication. The patient uses naproxen only when pain is unbearable while it is prescribed three times a day. The patient is afraid of taking fluvoxamine because of possible side-effects. The patient forgets to take his antihypertensive medication. The patient is unable to administer the timolol eyedrops for her glaucoma. A patient does not take furosemide because of attending activities at home. A patient is not able to fetch the medication at the pharmacy. a) See Appendix I for translation in Dutch. 122

123 PHARM- study design Chapter 3.1 Table 4 Classification and definition of pharmaceutical care issues Pharmaceutical care issue Monitoring Drug-drug interaction Contra-indicated drug Lifestyle Double medication Definition An intermittent (regular or irregular) series of observations in time, carried out to determine the effectiveness, safety and adherence of the drug therapy. A combination of two or more drugs, administered by one patient that can result in a modification of the effect of at least one drug. The effect may be an undesired effect or a lack of effect of the drug. A drug that is undesired because of the medical condition of the patient, which can lead to an adverse drug event or a lack of effect of the drug or a worsening of the medical condition of the patient. The lifestyle of the patient that could interfere with effective and safe drug therapy or that could result in non-adherence. Therapeutic duplication is defined as the use of two or more drugs in the same ATC classification and with similar pharmacodynamic properties, which can lead to adverse drug events. Phar mace utical care plan The purpose of the pharmaceutical care plan was to organize all of the pharmaceutical care agreed upon by the pharmacist, the GP, the practice nurse and the patient to achieve the goals of pharmacotherapy by addressing, resolving and preventing drug therapy problems. Together they formulated a care plan and the pharmacist specifically was responsible for the monitoring and the follow-up evaluation of the care plan and pharmacotherapy. After establishing the drug therapy problems they were categorized based on a medical condition and on the drug therapy. When multiple drug therapy problems were present, they were prioritized based on the patient s perspective, focussing on those that caused the most concern and which one the patient was willing to address. The next step was to set the goals of the drug therapy and to agree on the interventions to achieve these goals. The interventions included start, stop or switch drug therapy, adjusting the dosage regimens or formulation, monitoring of drug therapy, referral to other clinician and individualized patient counselling and education. All these interventions were documented in the pharmaceutical care plan with detailed information on who was responsible for a certain action (GP, pharmacist or practice nurse), when the action would be taken and when it would be evaluated. 123

124 Monitoring and follow-up evaluation of care plan and phar macothe rapy The focus of the monitoring and follow-up evaluation of pharmacotherapy is on the implementation of the pharmaceutical care plan. The pharmacist, the GP and if needed the patient, verified if all planned interventions were completed and they examined the outcomes of the documented drug therapy problems. They also assessed if the goals of drug therapy were achieved. At this evaluation moment it was possible to plan new interventions to achieve previously documented goals of therapy or to solve previously identified drug therapy problems. The pharmacist documented the outcomes and possible new interventions in the care plan. Control group Patients in the control group received usual care from their GP, pharmacist, practice nurse and other primary health care staff. This care consisted of repeat prescriptions, medication surveillance according to the current clinical guidelines. Outcomes The primary outcome of the study was the frequency of hospital admissions related to medication within the study period of 12 months of each patient. Two independent clinical pharmacists assessed all admissions for a causal relationship between the reason for admission and the medication use, prior to the admission. This was done by reviewing the, blinded, discharge letter combined with the medical and medication data from an electronic Case Report Form (CRF), according to the adjusted Kramer algorithm. 17 The secondary outcomes were survival, quality of life, adverse drug events and severe adverse drug events. The quality of life was measured by the EuroQol EQ5D combined with the VAS (visual analogue scale) questionnaire at the start of the inclusion period and at the end of the inclusion period, after 12 months. 18 The pharmacist and the GP asked the patient at the end of the study period about symptoms of the past three months. They reported and assessed the adverse drug events according to the adjusted Kramer algorithm 17 combined with National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP) scheme for categorisation of the severity of the events. 19 Other outcomes were the frequency and type of drug therapy problems (Table 2) as documented in the pharmaceutical care plans over the intervention period. Two independent pharmacists coded from the care plans, all the drug therapy problems, the proposed and executed intervention combined with the drug therapy. They met to reach consensus when needed. 124

125 PHARM- study design Chapter 3.1 Sample size We expect to achieve an effect of 50% reduction of medication-related hospital admissions. 2,20,21 It seemed possible to include 50 intervention patients from one intervention GP and 50 control patients from one control GP in one pharmacy practice, in a period of 12 months with a follow-up period of 12 months. To show a statistically significant difference between the intervention and the control arm with an expected prevalence of p 0 = 0.01, we planned to include patients, 7100 in each arm, from at least 142 pharmacy practices and at least 284 GP practices to participate in the PHARM-study. 22 This is based on an alpha of 0.05, a power (1-beta) of 0.8. It should be noted that the prevalence of HARM is expected to be higher in the group eligible for inclusion in the study. Distribution of inclusion criteria To determine the size of the population at risk and therefore eligible for inclusion, we searched the Kring-kubus-database. 23 This database contains anonymous drug dispensing data from 100 (5.5%) community pharmacies, spread throughout the Netherlands, belonging to the franchise organization Kring-apotheek (n=330). In the Dutch community pharmacies, the drug history of individual patients can be considered as nearly complete because most patients visit only one single pharmacy. 24 This database includes drug-dispensing data from patients with patient related information (age, sex, unique anonymous identifier), information on the dispensed drug (chemical substance coded according to the Anatomical Therapeutic Chemical (ATC) system 25 and dispensed product), date of dispensing, amount dispensed and prescribed dosage regimen. Patients who complied with the defined risk factors and therefore with the inclusion criteria as well (Table 1), were selected from the database by using Microsoft Access. The use of the drugdispensing data was performed in compliance with the Dutch privacy regulations. We found that in an average pharmacy practice 6.4% (range %) of the patients had a high risk of being admitted to hospital with a medication-related cause, according to the HARM-risk model. In an average Dutch pharmacy practice the size of this high-risk population was around 450 patients, enough eligible patients for participating in the PHARM-study. The type of drug (ATC class A; alimentary tract and metabolism or ATC class B; blood and blood forming organs) risk factor was the most common risk factor in the population of a pharmacy: 37.2% (range %), while non-adherence was the smallest risk factor in the population of a pharmacy: 22.4% (range %). The relative numbers of high risk patients per risk factor in the Dutch pharmacies and the distribution of these risk factors are presented in the box plots in Figure 2. The numbers of high-risk 125

126 Figure 2 Proportion of the patients at HARM risk in a pharmacy with the distribution of the risk factors in the pharmacies in the Kring-kubus database HARM = Hospital Admissions Related to Medication; ATC = Anatomical Therapeutic Chemical patients per risk factor and the overlap between the risk factors are presented in the Venn diagram in Figure 3. R andomisation Cluster randomisation took place after informed consent of the participating GPs and pharmacists on a GP level and before the selection of patients. For every pharmacist a random selection of the intervention GP was made. The other participating GP became the GP who included control patients. 126

127 PHARM- study design Chapter 3.1 Figure 3 The number of high-risk patients per risk factor and the overlap between the factors (total patient population: ) Data analysis From each included patient data was collected in an electronic CRF containing: patient data, medical history and complaints, clinical data, quality of life and visual analogue scale, drug history with drug-allergies and drug-intolerabilities and a part with information on hospital admissions and adverse drug events. 127

128 Detailed information on the data collected in the CRF can be found in Table 5. The pharmacist and the GP completed the CRF with data from the pharmacy database, the GP database and from the interview with the patient. The collected data were based on literature and present guidelines. 16,26 28 For the intervention patients the CRF also contained a separate part for documenting the pharmacotherapy review, the pharmaceutical care plan and monitoring and follow up evaluation of pharmacotherapy. In this plan information was organised according to drug therapy related problems with the Subjective-Objective- Assessment-Plan (SOAP) notes method. 29 The assessment and evaluation of the different drug therapy problems were subsequently documented and prioritised and the subsequent interventions were documented with specific information on who of the clinicians is taking the action, when the action is taken and when the action is evaluated. Data from the digital Excel CRF and from a Microsoft Access database was combined and further analysed by using R, version In R mixed-effects Cox models 21 were designed to study the effect of the intervention on hospital admissions related to medication and the effect on survival, quality of life and adverse drug events. All available patient data were included. The pharmacist and the GP were integrated as random effects in the models. P-values < 0.05 were regarded as statistically significant. The two-sided 95% bootstrap percentile confidence intervals were computed using 1000 replications. Bootstrap samples were obtained by random sampling GP s or patients with replacement from the population. To assess the model s goodness of fit we created plots of outcome versus follow-up time, versus number of diseases and examined residuals. The influence of the baseline characteristics as age, gender, number of physicians, number of diseases, number of drug prescribed, refill rate and follow-up time were also analysed using linear mixed-effects models or generalized linear mixed-effects models. Discussion The PHARM-study studies the effect of the complex pharmaceutical care process on medication related hospital admissions. This patient centred pharmaceutical care process was highly patient individualized, included several steps (Figure 1) and was conducted in a primary care setting. This care process can only be partly structured in a study protocol because it is dependent on patients autonomy, the performance of the clinicians and on the cooperation between them. These issues cannot be completely restrained to a protocol and will be a reflection of real clinical 128

129 PHARM- study design Chapter 3.1 practice. In the PHARM-study the effectiveness of the pharmaceutical care process will be studied in daily clinical practice and due to this the study is a pragmatic trial. Such a trial is suitable to evaluate effectiveness rather than to measure efficacy. 30 A key methodological issue in pragmatic trials is balancing internal and external validity. 30 This issue was addressed for the PHARM-study in the study design and evaluation protocol. External validity, or generalizability, was achieved by allowing the GPs and the pharmacists to implement the intervention in their own manner within their daily practice. External validity was also addressed by having very few exclusion criteria and general inclusion criteria. The included high-risk patients are a heterogeneous group, with multiple and varying morbidities and medication usage. Internal validity was maintained by selecting intervention patients from another general practitioner than the control patients, to reduce contamination bias by ensuring that GPs did not apply strategies and knowledge used in the intervention patients to control patients. We expect this study to have several other strengths besides internal and external validity. First, we have gained experience during several previous pilot projects and because of this we were able to optimize the intervention by providing a structure for the pharmacotherapy review. We have also optimized the CRF for documentation of the pharmaceutical care and the planned intervention in drug therapy to guarantee the continuity of care. Since pharmacotherapy and medical services become more complex, creating a comprehensive documentation is required to facilitate collaboration between members of the health care team and ensure continuity of care. Second, the pharmacist and the GP were offered a training regarding the study protocol, the documentation and how to perform a structured pharmacotherapy review. A third strength is the large group of patients, which makes it possible to measure a possible effect on medication related hospitalisations. A fourth strength is the extensive intervention during a period of 12 months in an interdisciplinary setting in cooperation between the patient and the own GP and pharmacist, which makes it more likely to solve all major drug therapy problems. This study also has some limitations. First, selection bias may occur by just including patient who are willing to cooperate in the intervention. However, the same selection will take place in the control group because they also have to cooperate in completing the questionnaires. Second the selection of the participating pharmacist and GPs was based on voluntary participation. Only motivated couples of GPs and a pharmacist who established a good cooperation would participate in the study, which hampers the generalizability. But in case the intervention shows an effect, these participating GPs and pharmacists can be a role model for their colleagues in the region. A third limitation is the extensive and therefore demanding intervention. 129

130 Table 5 Part Data collection; content of case report form (CRF) Data content Patient data patient ID a age a,b gender a,b pharmacist a,b general practitioner a,b weight c,d lengths c,d diet d alcohol consumption d smoking habits d cognition c abilities/disabilities d daily activities d former education\language d religion d home situation d home help d carer d medication aids d other relevant information (free text) d Medical data acute medical conditions (menu, coded with ICPC-code 31 ) c chronic medical conditions (menu, coded with ICPC-code 31 ) a,c current complaints or symptoms (menu, coded with ICPC-code 31 ) c significant past medical conditions or procedures c other relevant health information (free text) c Clinical data Available clinical data is collected from 12 months before inclusion till the end of the inclusion period. c laboratory values c blood pressure c temperature c weight c other relevant measured clinical data c 130

131 PHARM- study design Chapter 3.1 (Table 5 continued) Part Data content Medication data Full drug history: all dispensed drugs from 12 months before inclusion till the end of the inclusion period, b containing information on: chemical substance (menu, coded with ATC-code 25 ) b dispensed product (extracted from pharmacy database) a,b dispensed amount (extracted from pharmacy database) a,b dispensing date (extracted from pharmacy database) a,b prescribed dose (extracted from pharmacy database) a,b over the counter (OTC) drugs b,d vitamins, dietary or nutritional supplements b,d other health products b,d medical devices b,d drug allergies b,d intolerabilities or adverse drug events b,d Study outcomes hospital admissions (discharge letter as appendix) a date of each admissions a reason for each admission a quality of life; EQ5D score (at t=0 a and t=12months) d visual analogue scale score (at t=0 a and t=12months) d ADE with symptom, related drug and timing information (at t=12 months) c,d severity of the ADE c,d Pharmaceutical care plan e DTPs (according to the SOAP notes) 29 therapy goals priorities interventions clinician who is executing the action date of actiondate of evaluation of the action ID = number to identify the patient; ICPC = International Classification of Primary Care; ATC = Anatomical Therapeutic Chemical; EQ5D = EuroQuol5Dimensions; ADE = adverse drug event; DTP = drug therapy related problem; SOAP = Subjective-Objective-Assessment-Plan a) Obligatory. b) Obtained from medication record. c) Obtained from medical record. d) Obtained from the patient during pharmaceutical anamnesis or questionnaire. 131

132 It is thinkable that it will not be implemented as described in the protocol because this is not possible or feasible in daily practice. A fourth limitation is the selection of intervention and control patients on a general practice level but from the same pharmacy which implies that control patients are cared for by the same pharmacist as the intervention patient. This possibly leads to contamination bias, but we do not expect that the pharmacist will provide other care than usual care to the control patients. A fifth limitation is the assessment of the ADEs which will be done by the patient s own pharmacist and GP who are not blinded to the assignment to intervention or control group. Knowing to which group the patient is assigned, may possibly affect their judgement of ADEs, which can potentially lead to the detection of more ADEs in the control group. On the other hand an advantage may be that they know their patients so they can make a better judgement if the symptoms are related to the medication or to a disease. C onclusion The PHARM-study is one of the largest controlled trials to study the effectiveness of the total pharmaceutical care process, including a pharmaceutical anamnesis, followed by a pharmacotherapy review, a pharmaceutical care plan, monitoring and follow up evaluation of pharmacotherapy, in an ordinary integrated primary care setting on medication related hospital admissions. Few controlled trials in this area have been performed and none have investigated the extensive intervention of the whole pharmaceutical care process, performed in cooperation of the patient, the pharmacist, the GP and the practice nurse, in this group of high risk patients, at the scale of this study either in terms of the number of practices or the number of patients. The PHARM-study should provide evidence as to whether the entire pharmaceutical care process should be implemented in a primary care setting in the Netherlands. Acknowlegdement - The authors would like to thank Svetlana V. Belitser, MSc for providing statistical support, Marjolein M.M. Geleedst-de Vooght, PharmD, Matthijs M. Engering, PharmD and Frank A. van Wees, PharmD for participating in the pilot study of the PHARM-study. 132

133 PHARM- study design Chapter 3.1 References Leendertse AJ, Visser D, Egberts AC, van den Bemt PM. The relationship between study characteristics and the prevalence of medication-related hospitalizations: a literature review and novel analysis. Drug Saf. 2010;33: Leendertse AJ, Egberts AC, Stoker LJ, van den Bemt PM: HARM Study Group. Frequency of and risk factors for preventable medication-related hospital admissions in the Netherlands. Arch Intern Med. 2008;168: Aparasu RR, Fliginger SE. Inappropriate medication prescribing for the elderly by officebased physicians. Ann Pharmacother. 1997;31: Walker J, Wynne H. Review: the frequency and severity of adverse drug reactions in elderly people. Age Ageing. 1994;23: Horne R, Weinman J, Barber N, et al. Concordance, adherence and compliance in medicine taking: report for the National Co-ordinating Centre for NHS Service Delivery and Organisation R&D. London: NCCSDO; Holland R, Desborough J, Goodyer L, Hall S, Wright D, Loke YK. Does pharmacist-led medication review help to reduce hospital admissions and deaths in older people? A systematic review and meta-analysis. Br J Clin Pharmacol. 2008;65: Zermansky AG, Silcock J. Is medication review by primary-care pharmacists for older people cost effective?: a narrative review of the literature, focusing on costs and benefits. Pharmacoeconomics. 2009;27: Holland R, Lenaghan E, Harvey I, et al. Does home based medication review keep older people out of hospital? The HOMER randomised controlled trial. BMJ. 2005;330: Richmond S, MortonV, Cross B, et al. RESPECT trial team. Effectiveness of shared pharmaceutical care for older patients: RESPECT trial findings. Br J Gen Pract. 2010;59: Krska J, Cromarty JA, Arris F, et al. Pharmacist-led medication review in patients over 65: a randomized, controlled trial in primary care. Age Ageing. 2001;30: Zermansky AG, Alldred DP, Petty DR, et al. Clinical medication review by a pharmacist of elderly people living in care homes: randomised controlled trial. Age Ageing. 2006;35: Schnipper JL, Kirwin JL, Cotugno MC, et al. Role of pharmacist counseling in preventing adverse drug events after hospitalization. Arch Intern Med. 2006;166: Stewart S, Pearson S, Luke CG, Horowitz JD. Effects of home-based intervention on unplanned readmissions and out-of-hospital deaths. J Am Geriatr Soc. 1998;46: Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol. 1997;50: Strand LM, Morley PC, Cipolle RJ, Ramsey R, Lamsam GD. Drug-related problems: their structure and function. DICP. 1990;24: Cipolle RJ, Strand LM, Morley PC. Pharmaceutical care practice: the clinician s guide. 2nd ed. New York: McGraw-Hill; Kramer MS, Leventhal JM, Hutchinson TA, Feinstein AR. An algorithm for the operational assessment of adverse drug reactions, I: background, description, and instructions for use. JAMA. 1979;242: The EuroQol Group. EuroQol-a new facility for the measurement of health-related quality of life. Health Policy. 1990;16:

134 19. National Coordinating Council for Medication Error Reporting and Prevention (NCC_ MERP). NCC MERP Index for categorizing medication errors. Available from: neccmerp.org/ [Accessed November 2009]. 20. Stichting Farmaceutische Kengetallen. Polyfarmacie. Pharm Weekbl. 2005;32: Campbell MK, Thomson S, Ramsay CR, MacLennan GS, Grimshaw JM. Sample size calculator for cluster randomized trials. Comput Biol Med. 2004;34: Brown H, Prescott R. Multi-Centre Trials and Meta-Analyses. In: Applied Mixed Models in Medicine. 2nd ed. West Sussex: JohnWiley & Sons Ltd; Geerts AF, De Koning FH, De Smet PA, Van Solinge WW, Egberts TC. Laboratory tests in the clinical risk management of potential drug-drug interactions: a cross-sectional study using drug-dispensing data from 100 Dutch community pharmacies. Drug Saf. 2009;32: Buurma H, Bouvy ML, De Smet PAGM, et al. Prevalence and determinants of pharmacy shopping behaviour. J Clin Pharm Ther 2008;33: WHO Collaborating Centre for Drug Statistics Methodology. Complete ATC index [online]. Available from URL: [Accessed November 2009]. Floor-Schreudering A, De Smet PA, Buurma H, Egberts AC, Bouvy ML. Documentation Quality in Community Pharmacy: Completeness of Electronic Patient Records After Patients First Visits. Ann Pharmacother. 2009;43: Royal Dutch Association for the Advancement of Pharmacy (KNMP). Dutch Pharmacy Standard (NAN). The Hague, the Netherlands; 26 May Koda-Kimble MA, Young LL, Kradjan WA, Guglielmo BJ. Applied therapeutics. The clinical use of drugs. 9th ed. Philadelphia: Lippincott Williams & Wilkins; August Weed LL. The problem oriented record as a basic tool in medical education, patient care and clinical research. Ann Clin Res. 1971;3: Godwin M, Ruhland L, Casson I, et al. Pragmatic controlled clinical trials in primary care: the struggle between external and internal validity. BMC Med Res Methodol. 2003;3: Lamberts H, Wood M. International Classification of Primary Care. Oxford: Oxford University Press;

135 PHARM- study design Chapter 3.1 Appendix I Hulpmiddel bij de analyse van farmacotherapie voor de individuele patiënt I ndicatie zijn alle aandoeningen behandeld? zijn alle geneesmiddelen geïndiceerd? E ffectiviteit hebben de geneesmiddelen het beoogde effect? (zie ook lab) is de dosering effectief? (sterkte, frequentie, duur, aanwijzingen, lab) is de toedieningsweg en de toedieningsvorm effectief? is er sprake van een interactie met andere geneesmiddelen? V eiligheid is de dosering juist? (sterkte, frequentie, duur, aanwijzingen, lab) zijn er mogelijke bijwerkingen? zijn er geneesmiddelen gecontraïndiceerd? is de benodigde monitoring uitgevoerd? is er sprake van een allergie? G ebruik gebruikt de patiënt zijn/haar geneesmiddelen volgens voorschrift? wat vindt de patiënt van zijn/haar geneesmiddelen? kan het gebruik van de geneesmiddelen gemakkelijker voor de patiënt? 135

136

137

138 Abstract Background Previous studies have shown an increased risk of hospital admissions with a medication related cause especially for elderly patients. Regular review of pharmacotherapy has been recommended to prevent hospital admissions and to improve pharmaceutical care. The pharmaceutical care process serves as a model for medication review, involving collaboration between general practitioners (GPs), pharmacists, patients, and carers. Its use is advocated with older patients who are prescribed several drugs. However, it has yet to be thoroughly evaluated. Therefore we designed the PHARM (Prevention of Hospital Admissions by Reviewing Medication) study to investigate the effectiveness of the complete pharmaceutical care process on medication-related hospital admissions, survival, adverse drug events and quality of life in collaboration with the GP, the pharmacist and the patient in an ordinary integrated primary care setting. Methods The PHARM-study is designed as an open, controlled, multi-centre study in an integrated primary care setting. Patients with a high risk on a medication related hospital admission based on old age, non-adherence, type of medication used and polypharmacy, were included in the study. The intervention consisted of a patient-centred, structured, pharmaceutical care process. This process consisted of several steps, was continuous and occurred over multiple encounters of patients and clinicians. The steps of this pharmaceutical care process were a pharmaceutical anamnesis, a review of the patient s pharmacotherapy, the formulation and execution of a pharmaceutical care plan combined with the monitoring and follow up evaluation of the care plan and pharmacotherapy. The patient s own pharmacist and GP carried out the intervention. The control group was included by another GP than the intervention GP and received usual care. The primary outcome of the study was the frequency of hospital admissions related to medication within the study period of 12 months of each patient. The secondary outcomes were survival, quality of life, adverse drug events and severe adverse drug events. The outcomes were analysed by using mixed-effects Cox models. Resu lts 364 intervention and 310 control patients were included in 42 primary health care settings of at least one pharmacist and at least two GP s. More medication related hospital admissions were found in the control group than in the intervention group; respectively ten and six admissions. The effect was dependent on the 138

139 PHARM- study results Chapter 3.2 number of diseases. We calculated the effect as a hazard ratio (HR). The effect of the intervention for three diseases was HR 0.77 (95%CI ); for four diseases HR 0.43 (95%CI ); for five diseases HR 0.28 (95%CI ); for six diseases HR 0.19 (95%CI ); for seven diseases HR 0.14 (95%CI ); and for eight diseases HR 0.11 (95%CI ). Between the intervention and control group no statistically significant differences were found in the secondary outcomes survival, adverse drug events and quality of life. C onclusion With the PHARM study we demonstrated that the patient s own pharmacist together with its own GP and the patient may prevent hospital admissions with a medication related cause. 139

140 Introduction The Western world is facing an unprecedented growth in the number of elderly people, 1 who are often frail, have progressive degenerative health problems and take multiple medication. 2 Combined with physiological change, frequent cognitive impairment, poor adherence to prescribed drugs and interactions between these drugs makes them vulnerable to adverse effects of drug use. 3 5 Previous studies have shown an increased risk of hospital admissions with a medication related cause especially for elderly patients. 6,7 Regular review of pharmacotherapy has been recommended to prevent hospital admissions and to improve pharmaceutical care. 8 The central aim of pharmaceutical care is to ensure that medication is prescribed and used appropriately to improve a patient s health and quality of life and to reduce the burden caused by adverse drug events. This is achieved through a patient centred approach, of a collaborative and iterative process of pharmaceutical care according to the dynamic situation of a person s health. Pharmaceutical care, which originated in the US, leaves physicians ultimately responsible for patients care, but gives pharmacists the role of managing pharmacotherapy in collaboration with the treating physician and the patient. 9 Drug therapy problems as well as potential drug therapy problems related to indication, effectiveness, adverse drug events and adherence are to be reviewed by collaborating health care professionals in the pharmaceutical care process. Although pharmaceutical care is advocated in many European countries, it has yet to be thoroughly evaluated. Several studies have been conducted on the effectiveness of pharmaceutical care 10 or on a part of this care process: the pharmacotherapy or medication review. 11,12 Extracting meaningful conclusions from the data of the studies is difficult because the studied outcome is limited to the clinically less relevant outcome drug therapy problems or the studies are limited by the interventions proposed on the basis of the pharmacotherapy or medication review. 13,14 Other studies did look into clinically more relevant outcomes such as morbidity, hospital admissions or mortality but these studies generally did not involve the patient, 15 or the intervention was not performed in daily practice by the patients own general practitioner (GP) and own pharmacist, 16 or not in close collaboration between the GP and the pharmacist,10 or there was no monitoring and evaluation of the pharmacotherapy on a regular basis as part of the intervention. 17 One study reports an increase in emergency readmissions, 16 while in the other studies there is no suggestion that patients were harmed by the interventions, 11,12 and even some consistency in suggesting that falls 18 and hospital admissions 19,20 might be reduced. No studies reported a clear benefit in terms of mortality, mental capacity or activities of daily living. It is likely 140

141 PHARM- study results Chapter 3.2 that the patients health will benefit from a close collaboration between the patient, the GP, the pharmacist and the practice nurse but this has not yet been the setting of studies on pharmaceutical care or pharmacotherapy review. Therefore we sought to investigate the effectiveness of the complete pharmaceutical care process on medication-related hospital admissions, survival, adverse drug events and quality of life in collaboration with the GP, the pharmacist and the patient in an ordinary integrated primary care setting. In addition, the discrepancy between the intended and actual number of included patients is discussed (Appendix). Methods Study design The effect of the pharmaceutical care process on medication related hospital admissions was studied in an open controlled design. The design is described in detail in another paper (Chapter 3.1). 21 The study was approved by the ethical review board of the METOPP (Medisch-Ethische Toetsing Onderzoek Patiënten en Proefpersonen; a medical ethics committee for clinical trials in patients and in healthy volunteers). All patients were asked for informed consent before participating in the study. Par ticipants The pharmaceutical care process was carried out in an integrated primary care setting, by the patient s own GP, pharmacist and primary care practice nurse if available. All Dutch GPs and pharmacists, working in primary care were eligible to participate in the study. In every setting at least one pharmacist was working with at least one GP who included intervention patients and at least one GP who included control patients. Only the intervention GP participated in the pharmaceutical care process. The control GP provided data for the control patients. Before and during the study period the intervention GP and the pharmacist were offered support by training, education, a web-based community of practice and a helpdesk on pharmacotherapy. We included only patients with a high risk of a medication related hospital admission in the study, based on the findings from the HARM-study. 6 This study resulted in four criteria for high risk: old age, non-adherence, type of medication used and polypharmacy. Patients could be included of 65 years of age or older, who had five or more drugs on repeat prescription of which at least one was filled with a 141

142 refill-rate 22 of less than 80% or more than 120% as a measure of non-adherence, and who were dispensed one or more drugs from the Anatomical Therapeutic Chemical (ATC) Classification System 23 class A (therapeutics that act on the alimentary tract and metabolism) or ATC class B (therapeutics that act on blood and blood forming organs). The pharmacist selected patients who complied with the inclusion criteria, from the dispensing records in the pharmacy. If too many patients were eligible for inclusion, random selection was used to comply with the limit. Patients were not included in the study if they were resident in a nursing home, if their life expectancy was less than three months or if they did not give their informed consent. Inter venti on The intervention, the pharmaceutical care process, consisted of taking a pharmaceutical anamnesis, executing a pharmacotherapy review, formulating and agreeing on a pharmaceutical care plan and the monitoring and follow up evaluation of pharmacotherapy as documented in the pharmaceutical care plan. The pharmaceutical care process was a continuous process of different steps and occurred over multiple encounters of patients and clinicians during the 12 months study period. The GP and the pharmacist both had access to the medication history and medical history of the included patients including recent laboratory data. The pharmaceutical care process started with a pharmaceutical anamnesis: an encounter of the patient and the pharmacist. The pharmacist met with the patient at home or in a private consultation room, where they discussed the patients medication experience in general, discussed all medicines with their use, discussed each condition being treated and asked about relevant symptoms or possible adverse drug effects. The pharmacist used the acquired information combined with the information from the patients medical history and medication history to perform a structured pharmacotherapy review to identify possible drug therapy related problems (DTPs) based on the concept by Hepler and Strand 9 and pharmaceutical care issues (CIs). This implied assessing if all indications were treated appropriately, if the pharmacotherapy was appropriately indicated, effective, safe, if monitoring was appropriate and if the patient was willing and able to adhere to the pharmacotherapy and to the pharmaceutical care instructions. The pharmacist documented CIs as an element of a patient s drug related need, which can lead to a DTP; and documented DTPs as any undesirable event or risk thereof, experienced by the patient that involves or is suspected to involve pharmacotherapy and that actually or potentially interferes with the desired patient outcome. 24,25 After this assessment the pharmacist and the GP met to discuss the most important DTPs or potential 142

143 PHARM- study results Chapter 3.2 DTPs, to prioritise these DTPs and to set goals for the pharmacotherapy. This was documented in the pharmaceutical care plan together with the agreed interventions and their evaluation. The GP, the pharmacist or the practice nurse carried out the monitoring and follow-up evaluation of the pharmacotherapy as agreed in the pharmaceutical care plan. At least two follow-up visits of the pharmacist and the patient were scheduled after three and six months, to assess whether there were any further pharmaceutical care issues to address with the GP. C ontrol group Patients in the control group received usual care from their GP, pharmacist, practice nurse and other primary health care staff. This care included repeat prescriptions and medication surveillance according to the current clinical guidelines. Outcome measures The primary outcome of the study was the frequency of hospital admissions related to medication within the intervention period. The secondary outcomes were survival, quality of life and adverse drug events. Other outcomes were the frequency and type of the identified DTPs, the proposed and executed interventions and the related drug therapy. Data collection For both intervention and control patients we collected information on patient characteristics, medical history, drug history, hospital admissions, possible adverse events, death and quality of life in a digital Excel CRF (case report form). Pharmaceutical care plans were obtained from all intervention patients at the end of the study period. Hospital admissions were assessed for the causal relationship between the reason for admission and the medication use, prior to the admission by two independent clinical pharmacists. This was done by reviewing the discharge letter combined with the medical and medication data from the CRF, according to the adjusted Kramer algorithm. 26 The assessments were entered into a Microsoft Access database. The quality of life was measured by the EuroQol EQ5D combined with the VAS (visual analogue scale) questionnaire 27 at the start of the inclusion period and at the end of the inclusion period. The pharmacist and the GP asked the patient at the end of the study period about any symptoms possibly related to their medication. They reported and assessed these adverse drug events according to the adjusted Kramer algorithm 26 in order to assess causality. 143

144 Two independent pharmacists coded all the care plans with respect to type of drug therapy problem (DTP), type of care issue (CI) and the proposed and executed interventions. They met to reach consensus when needed. The information on the DTPs, intervention and related drug therapy was entered into a Microsoft Access database. Data analysis We expect to achieve an effect of 50% reduction of medication-related hospital admissions. 2,28,29 It seemed possible to include 50 intervention patients from one intervention GP and 50 control patients from one control GP in one pharmacy practice, in a period of 12 months with a follow-up period of 12 months. To show a statistically significant difference between the intervention and the control group with an expected prevalence of p 0 =0.01, we planned to include patients, 7100 in each arm, from at least 142 pharmacy practices and at least 284 GP practices to participate in the PHARM-study. 30 This is based on an alpha of 0.05, a power (1 beta) of 0.8. Data from the digital Excel CRF and from a Microsoft Access database was combined and further analysed by using R, version In R mixed-effects Cox models were designed to study the effect of the intervention on hospital admissions related to medication and the effect on survival, quality of life and adverse drug events. All available patient data were included. The pharmacist and the GP were integrated as random effects in the models. P-values < 0.05 were regarded as statistically significant. The two-sided 95% bootstrap percentile confidence intervals were computed using 1000 replications. Bootstrap samples were obtained by random sampling GP s or patients with replacement from the population. To assess the model s goodness of fit we created plots of outcome versus follow-up time, versus number of diseases and examined residuals. The influence of the baseline characteristics gender, number of diseases, number of drug prescribed, refill rate and follow-up time were analysed using linear mixed-effects models. The influence of age and number of physicians were analysed using generalized linear mixedeffects models. Result s Intervention and control patients were included in 42 primary health care settings of at least one pharmacist and at least two GP s. They included 364 patients in the intervention group and 310 patients in the control group. In both groups 17 patients 144

145 PHARM- study results Chapter 3.2 were lost to follow up due to a move and being cared for by another physician (see Figure 1). The median follow up time was 239 (95% confidence interval [95%CI] ) days in the intervention group. Patients in the control group had a Figure 1 Patient flow PHARM study ATC = Anatomical Therapeutic Class; GP = general practitioner a) Intervention patients without pharmaceutical care plan if no relevant drug therapy problems were identified or if there was no plan to deliver pharmaceutical care. b) Loss to follow up due to death or move to other GP and or pharmacist. 145

146 longer follow up with a median of 274 (95%CI ) days. 112 patients in the intervention group and 85 patients in the control group were followed less than six months while 127 patients in the intervention group and 118 patients in the control group were followed for at least nine months. Both groups were similar in age (average 76 years old), similar in gender (44% male in intervention group and 40% male in the control group). Also the same number of prescribing medical specialists was found in both groups with a median of 2.5 in both groups. The number of drug therapies on repeat prescription was almost similar in both groups with a mean of 7.8 drug therapies in the intervention group and in the control group (Table 1). Both groups differed in the number of documented medical conditions, with a mean of 5.2 in the intervention group and 3.3 in the control group (Table 1). This Table 1 Characteristics of included patients in the PHARM-study Intervention (n=364) Control (n=310) p mean (95%CI) a mean (95%CI) a Age, in years, at the time of inclusion ( ) ( ) p = b Gender (male) in % 43.7% 40.0% p = b No. of prescribing physicians (incl. GP) 2.50 ( ) 2.63 ( ) p = b No. registered diseases 5.24 ( ) 3.39 ( ) p < b,c No. of chronically prescribed drug therapies 7.82 ( ) 7.87 ( ) p = b 95%CI = 95% confidence interval; GP = general practitioner a) All two-sided 95% bootstrap percentile confidence intervals, computed using 1000 replications. b) P-value of the effect of the intervention compared to the control in the generalized linear mixed-effect model. c) Significant difference in number of diseases between intervention and control group. difference is also seen in specific diseases like cardiovascular diseases (83% of the patients in the intervention group had a cardiovascular disease versus 57% of the patients control group) and rheumatology diseases (42% versus 27%). More medication related hospital admissions were found in the control group than in the intervention group; respectively ten and six admissions. The uncorrected effect of the pharmaceutical care process on medication related hospital admissions could be described in a mixed-effects Cox model with a hazard ratio (HR) of 0.50 (95%CI ; p = 0.20). The effect seemed to be dependent on the number of diseases therefore we incorporated the number of diseases in a mixed-effects 146

147 PHARM- study results Chapter 3.2 Figure 2 Effec t of the pharmaceutical care process for the number of diseases mean; median and two-sided 95% bootstrap percentile confidence intervals, computed using 1000 replications Cox model with resulting in an effect, HR, of the intervention of 0.30 (95%CI ; p = 0.033). The best fitted model incorporated besides the intervention, log(number of diseases) and also the interaction of intervention log (number of diseases), which implies an effect of the intervention depending on the number of diseases. Therefore we calculated the effect as a HR and a number needed to treat (NNT) for 3, 4, 5, 6, 7 and 8 diseases. The effect of the intervention for three diseases is HR 0.77 (95%CI ) and NNT 231; for four diseases HR 0.43 (95%CI ) and NNT 57; for five diseases HR 0.28 (95%CI ) and NNT 30; for six diseases HR 0.19 (95%CI ) and NNT 19; for seven diseases HR 0.14 (95%CI ) and NNT 14; and for eight diseases HR 0.11 (95%CI ) and NNT 11 (see Figure 2). Between the intervention and control group no statistically significant differences were found in the secondary outcomes survival, adverse drug events and quality of life (see Table 2). Eleven patients died in the intervention group and eight in the control group. The effect of the pharmaceutical care process on survival was analysed using mixed-effects Cox models with log number of diseases in the model, and the pharmacy and the GP as a random effects. The HR of the pharmaceutical care process (0.78 [95%CI ]) showed no statistically significant effect on 147

148 Table 2 Outcome Effects of secondary outcomes, analysed with linear/cox mixed-effects model, including pharmacy, GP level and corrected for the number of diseases Value ofintervention (95% CI) a survival 0.78 b ( ) ADE c d ( ) QOL e 0.16 d ( ) VAS f 1.73 d ( ) GP = general practitioner; 95%CI = 95% confidence interval; ADE = Adverse Drug Event; QOL = quality of life; VAS = visual analogue scale a) Two-sided 95% bootstrap percentile confidence intervals, computed using 1000 replications. b) Hazard Ratio, analysed with Cox mixed effects model. c) Number of ADEs. d) Mean effect, analysed with lineair mixed effects model. e) Difference in quality of life measurement on EQ5D scale (min 5 to max 15) between end of the inclusion period minus start of the inclusion period. f) Difference in visual analogue scale on health status (min 0 to max 100) between end of the inclusion period minus start of the inclusion period. survival. The number of adverse events seemed slightly lower in the intervention group (-0.06 [95%CI ]) but this was also not significant. The difference in quality of life, measured with the EuroQol EQ5D questionnaire (0.16 [95%CI ]) and the VAS (visual analogue scale) (1.73 [95%CI ]) 27 between the end of the inclusion period and at the start of the inclusion period also showed a slightly positive effect of the pharmaceutical care process but this was neither clinically nor statistically significant. In 332 (91%) patients a total of 1025 DTPs and 242 care issues (CI) were identified by the GP and the pharmacist (see Figure 3). This is an average of 2.8 DTPs and 0.7 CIs per included patient. The most commonly identified DTPs were Additional drug therapy required (n = 241) in 47% of the patients and Drug taking problems (n = 202) in 42% of the patients. Other identified DTPs were: Unnecessary drug therapy (n = 179), Adverse drug event (n = 140), Dosage too low (n = 106), Dosage too high (n = 89) and Ineffective drug therapy (n = 68). Monitoring was the most common care issue identified (n = 164) in 38% of the patients. Other care issues were not often described in the pharmaceutical care plans: Lifestyle (n = 37), Contra-indicated drug (n = 16), Drug-drug interaction (n = 13) and Double medication (n = 12). Start of new drug therapy was the most commonly occurring intended intervention as described in the pharmaceutical care plan by the GP and the pharmacist (see Figure 3). This intervention was established in 59% of the interventions. This 148

149 PHARM- study results Chapter 3.2 Figure 3 Number of DTPs and CIs with their most common intended en established intervention DTP = drug therapy related problem; CI = care issues 149

150 Table 3 Drug therapy problems and care issues with their interventions and related drugs Total no. of DTPs or CIs No. of patients with at least one of this DTP or CI Proportion of patients with the DTP or CI (%) a Most common intended intervention No. of intended interventions intended interventions / no. of DTPs or Cis No. of established interventions Established interventions / intended intervention Most common combination of DTP + intervention + drug therapy; no. of patients with this combination (%) Indication additional drug therapy required? % start new drug therapy % ,8% statins; 13 (10,9%) unnecessary drug therapy % stop drug therapy % 67 54,5% benzodiazepines; 8 (11,9%) Effectiveness ineffective drug therapy % switch drug therapy % 20 66,7% parasympatholytics; 4 (20%) dosage too low % adjust dosage of drug % 71 78,9% oral antidiabetic drugs; 14 (20%) Safety adverse drug event % switch drug therapy % 16 51,6% statins; 4 (25%) dosage too high % adjust dosage of drug % 45 63,4% benzodiazepines; 6 (13%) betablockers; 6 (13%) Drug taking drug-taking % advise to patient about drug use % ,1% inhaler devices with corticosteroids; 14 (11%) or with beta2- agonists; 14 (11%) Total b % start new drug therapy % ,8% statins; 14 (11%) 150

151 PHARM- study results Chapter 3.2 Table 3 continued Total no. of DTPs or CIs No. of patients with at least one of this DTP or CI Proportion of patients with the DTP or CI (%) Most common intended intervention No. of intended interventions intended interventions / no. of DTPs or Cis No. of established interventions Established interventions / intended intervention Most common combination of DTP + intervention + drug therapy; no. of patients with this combination (%) Care issues monitoring % monitoring % ,5% statins; 10 (8.8%) drug-drug interaction % advise to patient about drug use % 3 75,0% NE c contraindicated drug % stop drug therapy % 5 100,0% NE c lifestyle % advise to patient about drug use % 17 85,0% NE c double medication % stop drug therapy % 1 25,0% NE c DTP = drug therapy related problem; CI = care issue; NE = onot estimable a) Proportion calculated with regard to total number of patienst (364). b) In 32 patients no drug therapy problems nor care issues were identified. c) Numbers to small (one or two) to report involved drugs or drug classes. 151

152 intervention was often proposed to treat the DTP Additional drug therapy required in 206 of the 241 problems. Adjust dosage of drug was intended to treat Dosage too low in 90 of the 106 problems or Dosage too high in 71 of the 89 problems. Switch drug therapy was intended to treat Ineffective drug therapy in 30 of the 68 problems and Adverse drug event in 31 of the 140 problems. The detected problems around Drug taking were followed by Advice to patient about drug use (see Table 3 and Figure 3). Discussion The PHARM study is the first controlled trial to show an effect of the total pharmaceutical care process, including a pharmacotherapy review, a pharmaceutical care plan and monitoring and follow-up evaluation of pharmacotherapy, on preventing medication-related hospital admissions in an integrated primary care setting in the Netherlands. By studying almost 700 patients it showed that pharmaceutical care delivered by the patient s own GP and own community pharmacists to vulnerable elderly patients may prevent medication related hospital admissions. The pharmaceutical care process showed no significant effect on the number of adverse drug events, the quality of life or survival. Previous studies on the effect of pharmacotherapy reviews or pharmaceutical care have also shown little or no benefit to patients on the number of adverse drug events, the quality of life or survival. 10,11,15 20 However, our study did show a decrease in medication-related hospital admissions, which has not been shown before. The HOMER study showed the opposite effect, namely an increase in the number of hospital admissions. 16 A potential explanation for these contradictory findings may lie in the type of intervention: in the HOMER study the intervention was carried out by a special review pharmacist who reported possible adverse drug reaction and drug interaction to the GP, the review was done after discharge from the hospital including one follow-up visit from the review pharmacist and the review was done with a discharge letter. While in our study the intervention was a close collaboration of the patient s own GP and pharmacist who were responsible for the patient s drug therapy, both had access to the full medication history and medical history and was carried out over numerous encounters with the patient in his stable condition. Schnipper et al found a similar reduction in medication related emergency department visits and hospital admission (8 in the control group versus 4 in the intervention group) but this was not statistically significant. Stewart et al also showed statistically significantly fewer 152

153 PHARM- study results Chapter 3.2 unplanned readmissions (154 vs. 197: p = 0.022) in the intervention group, which seems a smaller difference but with no information on possible medication related readmissions. The difference between the number of the included patients and the intended number of included patients is remarkable high. This may be the explanation of the lack of effect of the pharmaceutical care process on the secondary outcomes survival, adverse drug events and quality of life. Several factors might explain these low numbers of included patients. These concern project management, collaboration between the pharmacists and the GPs, knowledge, skills and time, fit of the intervention with the present working process and information management. These are further discussed in the Appendix. The positive effect on the quality of prescribing by pharmacotherapy review 12,31 and pharmaceutical care 10 is confirmed in our study. In our study we found an average of almost three DTPs per patient which is in line with a smaller Dutch study of a cooperation of the GP and the pharmacist 13 while other trials find higher number of problems such as an average of four DTPs, 14 or more than seven pharmaceutical care issues. 31 This difference can be explained by the various definitions of the concept of DTP and by the great variety of settings in which the pharmaceutical care process takes place. We registered only DTPs that were actual problems to the patient in which the GP and the pharmacist agreed upon, while in other trials the pharmacist identified problems or care issues by himself, 14,31 which can result in a higher number of DTPs. In the SMOG trial 14 interventions were suggested by the pharmacist of which 28% were implemented. This percentage is much lower than the 73% found in our study. This can also be explained by the integrated care setting in which the intended intervention was agreed upon by the GP and the pharmacist after meeting the patient during the pharmaceutical anamnesis. The most common DTPs such as Additional drug therapy required and Drug taking problems such as non-adherence are also seen in other studies as the most common or important problems. 33,34 These problems appear to be relevant in preventing hospital admissions because the medication errors Drug not indicated and Non-adherence are among the most common errors which lead to medicationrelated hospital admissions. 6 One of the strengths of this trial is the generalizability and the applicability of the results. The study was carried out in daily practice and the pharmaceutical care process was performed during daily clinical care. The pharmaceutical care process could only be partly structured in protocols. Protocolling in the pharmaceutical care process is difficult because decision making in this process is mostly based 153

154 on therapeutic reasoning in stead of pharmacotherapeutic or pharmacological considerations. The performance of the clinicians, the patient autonomy and the cooperation between them played an important role in the outcome of the pharmaceutical care process for the individual patient. Therefore the pharmaceutical care process will determine the benefit of a treatment within routine clinical care rather than under ideal conditions. The large amount of participating practices throughout the entire country also contributes to the generalizability. Another strength of this trial is the studied population of ambulant elderly patients on multiple repeat prescriptions: a group at high risk of medication related problems, who in comparison to nursing home patients have greater control over their medicines. 3 This group is likely to benefit from the pharmaceutical care process and therefore an important group to study. The integrated care setting, including the patient, is also a strength. The identified DTPs are problems related to the drug therapy of a patient and not only related to the drug. The interventions are planned as a result of the cooperation between the GP, the pharmacist, the patient and sometimes the practice nurse. This ensures that the identified DTPs are indeed relevant, medically as well as with respect for the wellbeing of the patient. A limitation of this trial is the omission of randomisation. However, except for number of diseases the intervention and the control group look similar in most patient characteristics. Randomisation on a patient level was not preferable because the risk of contamination of the control patients. The planned cluster randomisation on GP level was abandoned because it was not feasible to find enough GPs who wanted to be randomly allocated as an intervention GP or as a control GP. A difference is seen in the number of diseases between the control and the intervention group. This cannot be explained by the prevalence of a certain disease or group of diseases in the intervention group. The difference can partly be explained by the incomplete medical history in the medical chart of the GP which was experienced during the intervention period. From the pharmacotherapy review, questions about the medical history would rise and previous documentation was studied which often led to completion of the medical history. This implies that diseases were missing in the control group. But this cannot explain a mean difference of almost two extra diseases in the intervention group. A true difference in number of diseases between intervention and control patients seems likely. Self-selection of general practices and pharmacies may have excluded those practices associated with poorer standards of prescribing and medication surveillance, which could have affected the results of this study and resulted in less medication-related hospital admissions than in the total population and less DTPs to intervene and 154

155 PHARM- study results Chapter 3.2 therefore less effect. Furthermore, by selecting intervention and control patients from the same pharmacy the contrast between both groups may also decrease, as the pharmacist will be inclined to provide better care to the control patients as well. Therefore the true effect of the pharmaceutical care process might have been higher. The assessment of the ADEs by the patient s own pharmacist is also a limitation. Although the pharmacists were trained to assess the ADEs, a protocol and comprehensive score forms were provided, the pharmacists did not always assess in a uniform way and did not always provide sufficient information. The open design of the study might have led to more ADE reports in the control group leading to a false positive effect for the pharmaceutical care process. It is also likely that the pharmacist had a more intense therapeutic relationship with the intervention patient than with this control patient, which might have resulted in more ADE reports in the intervention group than in the control group. As we eventually found no effect of the pharmaceutical care process on the number of ADEs, these effects may have counterbalanced each other. The follow up period may have been too short, especially for showing an effect on quality of life, although a trend towards a positive effect was seen. A power problem may be another explanation, as the calculated power could not be reached due to inclusion problems of pharmacists and GPs willing to participate, which is further discussed in the Appendix. Studies using the pharmaceutical care process as we defined it, but with longer follow-up periods and including more patients may show an effect on quality of life. Future studies should also include an economic evaluation of the pharmaceutical care process to determine if the process can be cost effective. In this way the pharmacist and the general practitioner can be reimbursed for performing the time consuming pharmaceutical care process. Also more information is needed on how to implement the pharmaceutical care process in daily practice. Comparison involving more countries can be useful for the development of international guidelines for the practice of pharmaceutical care. Finally, it is important to consider training needs. Most undergraduate courses teach basic clinical and communication skills. The level of these skills that would be required for a clinical pharmacotherapy review in primary care are significantly higher. This would mean that relevant postgraduate training in clinical pharmacy would be necessary before a pharmacist could take up this role. With the PHARM study we demonstrated that the patient s own pharmacist together with the patient and its own GP may prevent their patients from being admitted to hospital for a medication related cause. This can be done by following the structure 155

156 of the pharmaceutical care process. A suitably trained pharmacist and GP with full access to the patient, the medical record, drug history and other members of the primary health care team can conduct a pharmaceutical care plan and perform the monitoring en follow-up evaluation of the pharmacotherapy of the patient. Funding: The PHARM-study was part of a larger study program, entitled Implementation of interventions for preventing adverse drug events in high risk patient populations in primary care and care institutions, by a team of doctors and hospital pharmacists. This study program was financially supported by the Patient Safety Program of the Netherlands Organisation for Health research and Development (ZonMw). An unrestricted grant was given by Merck Sharp & Dohme and GlaxoSmithKline. Achmea and Zorg en Zekerheid health insurances provided a fee, per patient, to the participating pharmacists and GPs. References Demographics of older persons. United Nations Department of Public Information, September 1999 [online]. Available from: [Accessed 21 May 2010]. Data en feiten Stichting Farmaceutische Kengetallen 2009 [online]. Available from URL: [Accessed 21 May 2010]. Aparasu RR, Fliginger SE. Inappropriate medication prescribing for the elderly by officebased physicians. Ann Pharmacother. 1997;31: Walker J, Wynne H. Review: the frequency and severity of adverse drug reactions in elderly people. Age Ageing. 1994;23: Horne R, Weinman J, Barber N, et al. Concordance, adherence and compliance in medicine taking: report for the National Co-ordinating Centre for NHS Service Delivery and Organisation R&D. London: NCCSDO; Leendertse AJ, Egberts AC, Stoker LJ, van den Bemt PM: HARM Study Group. Frequency of and risk factors for preventable medication-related hospital admissions in the Netherlands. Arch Intern Med. 2008;168: Pirmohamed M, James S, Meakin S, et al. Adverse drug reactions as cause of admission to hospital: prospective analysis of patients. BMJ. 2004;329:15-9. Hepler CD. Pharmaceutical care. Pharm World Sci. 1996;18: Hepler CD, Strand LM. Opportunities and responsibilities in pharmaceutical care. Am J Hosp Pharm. 1990;47: Richmond S, MortonV, Cross B, et al.; RESPECT trial team. Effectiveness of shared pharmaceutical care for older patients: RESPECT trial findings. Br J Gen Pract. 2010;60:e

157 PHARM- study results Chapter Holland R, Desborough J, Goodyer L, Hall S, Wright D, Loke YK. Does pharmacist-led medication review help to reduce hospital admissions and deaths in older people? A systematic review and meta-analysis. Br J Clin Pharmacol. 2008;65: Zermansky AG, Silcock J. Is medication review by primary-care pharmacists for older people cost effective?: a narrative review of the literature, focusing on costs and benefits. Pharmacoeconomics. 2009;27: Stuijt CC, Franssen EJ, Egberts AC, Hudson SA. Appropriateness of prescribing among elderly patients in a Dutch residential home: observational study of outcomes after a pharmacist-led medication review. Drugs Aging. 2008;25: Vinks TH, Egberts TC, de Lange TM, de Koning FH. Pharmacist-based medication review reduces potential drug-related problems in the elderly: the SMOG controlled trial. Drugs Aging. 2009;26: Coleman EA, Grothaus LC, Sandhu N, Wagner EH. Chronic care clinics: a randomized controlled trial of a new model of primary care for frail older adults. J Am Geriatr Soc. 1999;47: Holland R, Lenaghan E, Harvey I, et al. Does home based medication review keep older people out of hospital? The HOMER randomised controlled trial. BMJ. 2005;330: Zermansky AG, Petty DR, Raynor DK, Freemantle N, Vail A, Lowe CJ. Randomised controlled trial of clinical medication review by a pharmacist of elderly patients receiving repeat prescriptions in general practice. BMJ. 2001;323: Zermansky AG, Alldred DP, Petty DR, et al. Clinical medication review by a pharmacist of elderly people living in care homes: randomised controlled trial. Age Ageing. 2006;35: Schnipper JL, Kirwin JL, Cotugno MC, et al. Role of pharmacist counseling in preventing adverse drug events after hospitalization. Arch Intern Med. 2006;166: Stewart S, Pearson S, Luke CG, Horowitz JD. Effects of home-based intervention on unplanned readmissions and out-of-hospital deaths. J Am Geriatr Soc. 1998;46: Leendertse AJ, de Koning FH, Goudswaard AN, et al. Preventing hospital admissions by reviewing medication (PHARM) in primary care: design of the cluster randomised, controlled, multi-centre PHARM-study. [Chapter 3.1; submitted] Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol. 1997;50: WHO Collaborating Centre for Drug Statistics Methodology. Complete ATC index [online]. Available from: [Accessed Nov 2009]. Strand LM, Morley PC, Cipolle RJ, Ramsey R, Lamsam GD. Drug-related problems: their structure and function. DICP. 1990;24: Cipolle RJ, Strand LM, Morley PC. Pharmaceutical care practice: the clinician s guide. 2nd ed. New York: McGraw-Hill; Kramer MS, Leventhal JM, Hutchinson TA, Feinstein AR. An algorithm for the operational assessment of adverse drug reactions, I: background, description, and instructions for use. JAMA. 1979;242: The EuroQol Group. EuroQol-a new facility for the measurement of health-related quality of life. Health Policy. 1990;16: Stichting Farmaceutische Kengetallen. Polyfarmacie. Pharm Weekbl. 2005;32:

158 Campbell MK, Thomson S, Ramsay CR, MacLennan GS, Grimshaw JM. Sample size calculator for cluster randomized trials. Comput Biol Med. 2004;34: Brown H, Prescott R. Multi-Centre Trials and Meta-Analyses. In: Applied Mixed Models in Medicine. 2nd ed. West Sussex: JohnWiley & Sons Ltd; Crotty M, Halbert J, Rowett D, et al. An outreach geriatric medication advisory service in residential aged care: a randomised controlled trial of case conferencing. Age Ageing. 2004;33: Krska J, Cromarty JA, Arris F, et al. Pharmacist-led medication review in patients over 65: a randomized, controlled trial in primary care. Age Ageing. 2001;30: Rao D, Gilbert A, Strand LM, Cipolle RJ. Drug therapy problems found in ambulatory patient populations in Minnesota and South Australia. Pharm World Sci. 2007;29: Kuijpers MA, van Marum RJ, Egberts AC, Jansen PA; OLDY (OLd people Drugs & dysregulations) Study Group. Relationship between polypharmacy and underprescribing. Br J Clin Pharmacol. 2008;65:

159 PHARM- study results Chapter 3.2 Appendix The discrepanc y between intended and ac tual number of included patients in the PHARM- study When designing the PHARM study we aimed, based upon the sample size calculation, to include patients, 7100 in each arm, from 142 centres of at least one pharmacist and two general practitioners (GPs). During a period of two years we anticipated to include 50 intervention patients and 50 control patients in every centre that were to be evaluated during a follow-up time of 12 months. Initially hundreds of pharmacists expressed interest in participating in the PHARM study, either by making enquiries about the study or by attending a meeting in which we provided information on the study, its intervention and workload. This resulted in 200 pharmacists (several accompanied by GP s from the same centre) participating in the first workshop (as described in Chapter 3.1 under support). At the start of the study they were confident that they would be able to include the requested number of patients within the two-year study period. However, even after extending the study period by 4 months (from 24 to 28 months) the total number of included patients was only 674, from 42 centres, of which only 18% had a follow up time of 12 months, according to the protocol. Several factors might explain these low numbers of included patients. These concern project management, collaboration between the pharmacists and the GPs, knowledge, skills and time, fit of the intervention with the present working process and information management. We did not identify any patient related barriers, such as willingness to participate or providing informed consent. Project management In the HARM-study the result in terms of number of included patients (Chapter 2.2) was much higher than in the PHARM-study. Although the studies are noncomparable in design, observational study versus intervention study, some factors may explain the difference in success regarding to output of patients, which should be taken into account when designing new research. Collecting retrospective data is less complicated than initiating a complex intervention that demands a lot of effort from the participants. In the HARMstudy a full time researcher was available to perform most of the work. The researcher was full time available for the project and had no other responsibilities, while each pharmacist and GP couple in the PHARM-study had to perform the time consuming intervention on top of their normal workload. While the project budget of the HARM study allowed the hospital pharmacy department to employ 159

160 Appendix continued a researcher, the pharmacist and the GP in the PHARM study were not or only partly reimbursed for executing the intervention. The researcher in the HARMstudy had a clear protocol to follow, while each GP and pharmacist in the PHARMstudy had to rely on his or her own knowledge and skills to perform an optimal intervention. In the HARM-study the final assessment was performed centrally while in the PHARM-study the GP and the pharmacist had to assess the patients themselves. The intervention in the PHARM-study required close collaboration with the patient, while in the HARM-study the only contact with the patient was to apply for informed consent and checking the medication history. C ollaboration between GP and pharmacist The collaboration between pharmacists and GPs is often poor. This can be explained by the differences between those professionals in responsibilities, focus, background, personality and accommodation. Both professionals have their own professional autonomy, often work in solo practice and there is no culture of holding each other accountable for clinical decisions. Physical barriers are common, since they have separated practices, which do not facilitate regular contact and collaboration. The main responsibility of the pharmacist lies in dispensing the right drug to the right patient while the main responsibility of the GP, with regard to the pharmacotherapy, is prescribing the right drug to the right patient at the right moment. The focus of the pharmacist is on the product and also on managing the pharmacy business process with its staff. Moreover, during the period in which the PHARM-study was implemented the pharmacists also faced new budgetary restrictions, which may have induced them to focus more on managing the product logistics. The focus of the GP differs from that of the pharmacist and is primarily patient focussed. The pharmacist is trained in analytical techniques, likes to make decisions on detailed information, aims to rule out uncertainty completely and has a more linear way of reasoning. The GP is educated in recognizing patterns trained in making decisions on incomplete information and in risk calculation, is used to deal with uncertainty and has a more inductive way of reasoning. These fundamental differences in professional approach may have made it difficult to collaborate on a complex intervention and to make a joined pharmaceutical care plan than we had anticipated. Fit with the working process The pharmaceutical care process as intended in the PHARM study may not have aligned sufficiently with the day-to-day processes in pharmacy and the 160

161 PHARM- study results Chapter 3.2 Appendix continued GP practices. The present way of practicing as a pharmacist is mainly reactive while the pharmaceutical care process is proactive including the follow-up. Both practices have different organisations in different locations in which the professional processes do not overlap. The pharmaceutical care process is a nonlinear process that requires a joint organisation of the care. The organisation of the process should be patient centred while this does not comply with the present collaboration of the different organisations. Therefore to embed the pharmaceutical care process in the regular working process, both professionals should organise this patient centred care together across both practices. Another difficulty was the time consuming nature of the pharmaceutical care process. This adds to the already considerable workload of the GP and the pharmacist. The practice learning curve requires seeing and caring for a sufficient number of patients to become proficient. With the very limited number of patients that were included per setting in the PHARM-study, the pharmaceutical care process could not become part of daily practice and the pharmacist was unable to progress on the learning curve to becoming an integral part of patient care. Thus, the practitioners were less likely to experience any successes in adding quality to life. This dismisses the possible incentives: satisfaction of professional growth or of adding quality of life to patients. In most settings there were no financial incentives in participating in the PHARM-study. Unfortunately, because of the new budgetary restrictions during the PHARM-study, the pharmacist had to spend more time on managing the logistics. With these new restrictions each health insurance company now decides which generic brand will be reimbursed and therefore need to be dispensed to the insured patient. Knowledge and skills For the participating pharmacists and GPs practicing the pharmaceutical care process could have been too difficult with their current skills and knowledge. As researchers, we may have overestimated the ability of GP/pharmacist partnerships to proactively manage the risk to patients associated with medication use. We observed several limitations, which could have led to unsatisfactory pharmaceutical care processes, leading to frustration, which may have contributed to discontinuation of participation in the PHARM-study. The information a pharmacist has about medicines and their pharmacological action seems inadequate to respond to the patients needs. The pharmacist needs clinical understanding of medicines to address diseases, needs competencies to manage uncertainty and needs responsive engagement with the patient, aiming to achieve 161

162 Appendix continued effective two-way communication of information, to provide good pharmaceutical care with the intention to add quality to a patient s life. From several GPs we experienced evasive behaviour and reluctance to change the therapy. The GP needs more knowledge on pharmacotherapy and needs to trust on the knowledge of the pharmacist. Besides clinical skills it takes organisational skills to conduct the complex intervention of the PHARM-study. In retrospect, the research team should have place more emphasis on the organisational management of the project instead of only offering clinical pharmacy support. Information management For a successful pharmaceutical care process it is essential to document the inquiry process and associated actions into a care plan. The GPs experienced problems in retrieving the right information on the past medical history from their patients. This added to the time consuming nature of the intervention. The pharmacists were not skilled in documenting their pharmaceutical care and needed to develop clinical skills in documentation, problem identification and problem solving. This also added to the time consuming nature of the intervention and also hampers the continuity of the pharmaceutical care. Within the PHARM-study it was required to document the medical history, medical data, drug history and the pharmaceutical care plan in a separate IT application designed specifically for the purpose of the study. This application could not exchange information with the current information systems in the GP-practice or in the pharmacy. Therefore the pharmacist was obliged to fill in all the forms. Unfortunately, the pharmaceutical care plan was not designed for sharing data, which would have helped the organisation of the care among the professionals. It was also not possible to incorporate it in the information system of the pharmacy or of the GP-practice. This can lead to frustration and was mentioned as a reason to discontinue the participation in the PHARM-study. These factors might have contributed to the low number of included patients. Recommendations following these implementation problems are discussed in Chapter 4 of this thesis. 162

163

164 In this thesis several studies have been presented providing information about the prevalence, nature and determinants of medication related hospital admissions and information on a patient centered method to reduce the burden of these medication related hospital admissions. These studies especially focused on patients outside the hospital and on the pharmaceutical care provided in an integrated primary care setting of the pharmacist, the GP and the practice nurse involving the patient in the process of pharmaceutical care. They add to our knowledge and show which patients are at risk of a medication related hospital admission, which medication errors can lead to medication related hospital admissions and which processes can be initiated to prevent medication related hospital admissions. In Chapter 2.1, a literature review is given that shows that reported frequencies of medication-related hospital admissions vary as a function of the setting (all admissions or only acute admissions), studied population (entire hospital, specific wards, selected population and age group), outcome (ADR/ADE), the method of data collection and the continent in which the study is performed. Higher frequencies for hospital admissions related to medication are reported to occur in the elderly, in psychiatry, when studying ADEs and when using a medical chart review method. Extrapolations from these studies to the Dutch situation are difficult. This is the reason the HARM (Hospital Admissions Related to Medication) was performed. The HARM-study (Chapter 2.2) identified a considerable frequency (5.6%) of all acute hospital admissions in the Netherlands as medication related which is comparable to some of the large studies included in the literature review in Chapter 2.1. Almost half (46.5%) of these admissions were related to a medication error and therefore potentially preventable. Lack of a clear indication for the medication, nonadherence and inadequate monitoring of the drug therapy, were the most common medication errors found. The HARM Study identified anticoagulant and antiplatelet drugs as major causes of medication-related hospital admissions. Impaired cognition, number of comorbidities, dependent living situation, nonadherence and polypharmacy were the most important risk factors identified. Another important risk factor that was identified was impaired renal function. Of the 714 medication-related hospital admissions from the HARM-study, 70 admissions (10%) were related to medication and impaired renal function. In these patients one or more medication errors (lack of monitoring, dosing errors, drugdrug interactions and drug-disease interactions), were deemed to contribute to the reason for admission. Adjusting pharmacotherapy according to renal function is relevant and may help to prevent these hospital admissions. 164

165 General discussion Chapter 4 The average medical costs for one preventable medication-related hospital admission were The average production loss costs for one admission were 1712 for a person below 65 years of age. Combining the medical costs and the costs of production losses resulted in average costs of 6009 for one, potentially preventable, medication-related hospital admission. The HARM-study has given extensive insight into the frequency, risk factors and costs of medication related hospital admissions in The Netherlands. After the HARM-study several recommendations were made, among which the regular review of medication for high risk patients was an important one. However, it was unclear how to predict the effectiveness of such a structured clinical medication review in every day practice. Therefore, the PHARM-study, a controlled intervention study in integrated primary care, was designed to determine whether this recommendation on regular medication review could actually prevent medication related hospital admissions (Chapter 3.1). The intervention consisted of a pharmaceutical intake, a review of the pharmacotherapy and conducting, executing and monitoring a pharmaceutical care plan. The inclusion criteria were based on the risk factors from the HARM-study. The PHARM-study showed that the intervention was indeed able to reduce the number of medication related hospital admissions in the high risk population. Unfortunately we were not able to include 14,000 patients as we intended but only 674 patients (see Appendix to Chapter 3.2). In this final Chapter 4 the presented studies will be put into broader perspective. The implications of the findings and the implementation problems will be discussed further. But first some methodological issues will be dealt with. Method olo gical issues with respect to studies on medication rel ated hospitalizations Causality Assessment In this thesis outcomes were measured by assessing the relationship between every event, the medication and sometimes a medication error. As seen in Chapter 2.1 the findings in different studies vary due to the different settings, populations and outcomes but also due to the different methods used to assess the relationship between the adverse event and the medication. 1 In our studies the results might also deviate from the true outcome due to the complexity of the causality assessment. Although we used the same systematic approach of causality assessment in all studies, 2 a triggerlist and a standardized form to collect and present the information, 165

166 the assessment remains complex because several factors can cause harm besides the medication, like disease and patient related factors. It is often not clear what the exact contribution is of the medication or of other disease related factors. Therefore the assessment will involve a subjective interpretation of the rater, leading to different causality assessments, dependent on the rater s own clinical experience and background. Also the differences in events c and the lack of information can cause variability in our outcomes. Causality assessment of adverse drug events and its variability is a well-known methodological problem in clinical research. Also in clinical practice the identification of adverse drug events is difficult because they mimic other diseases and many of the symptoms of adverse drug events occur commonly in healthy individuals taking no medication. Thus clinicians fail to recognize the features of an adverse drug event as experienced in our studies, 25% of the HARM-cases were not recognized by the physician before the researcher discussed the possible relationship with the medication as was found in the triggerlist. The systematic approach, like the one we used in our studies, of a simplified algorithm combined with a triggerlist can therefore be of value as a screening tool to focus on potential adverse drug events besides other types of clinical events. The implementation of such a tool can be of help in finding the right balance between effectiveness and safety and can then improve the decision-making about the optimization of a patient s pharmacotherapy. Assessment of pre ventability Preventability is also difficult to assess because the actual circumstances in which doctors had to make their decisions on prescribing medication are never known in retrospect. It seems easy to say that a prescribing error was made based on medical chart review, but the doctor may have included factors that are valid and did make the specific prescription necessary. Therefore we speak of potential preventability in all studies. An event was assessed as potentially preventable when a medication error was related to the event. This is a limitation in medication safety practice research and should be taken into account when interpretating the results. Medical char t re view Medical chart review was used to determine the occurrence of medication related hospitalizations and of medication errors. This method relies on the documentation of treatments and events by doctors and nurses. Several studies have shown that 166

167 General discussion Chapter 4 this documentation is often not optimal. 4 6 Yet, alternative methods of identifying events may suffer from even greater disadvantages, as we have shown in our literature review. Database methods do not only rely on proper documentation but on valid coding systems as well. 4,7 In general, studies using data base methods result in much lower frequencies of hospital admissions related to medication. The second alternative method, voluntary reporting, is known to suffer from severe underreporting, resulting in much lower frequencies as well. 8 In conclusion, medical chart review is the most optimal method, but may result in some bias due to incomplete documentation of treatments and events. Inter vention studies Studies on interventions aimed at improving medication safety are often difficult to design. The optimal design, a randomized controlled double-blind study, is rarely suitable to study a medication safety intervention. This is mainly caused by the fact that medication safety interventions generally deal with behavioral and educational interventions aimed at the health care professionals. Therefore, randomizing patients at the patient level would lead to improved treatment of control patients as well. A solution would be cluster randomization (at the health care professional level). But even then cross contamination on improved treatment can t be ruled out completely, as health care professionals often learn from each other. 9 This is the reason that interventions on medication safety are often studies using beforeafter designs. Such designs have a number of sources of potential bias, especially concerning the comparability of patients before and after the intervention and the influence of time trends. Using multivariate statistical analysis and time series analysis may partly correct for these biases, but unknown confounders remain a concern. In our PHARM-study randomizing at the patient level, as was in the original study protocol, was abandoned because this would lead to contamination of the control group. Because both GPs and pharmacists will learn the relevant techniques necessary for the pharmaceutical care process cumulatively, with the result that those who use these techniques for experimental patients cannot readily withhold them from control patients. Randomization at a patient level would be possible if the intervention would be carried out by a third person who cannot intervene in the care of the control patients. This was done in the HOMER-trial 10 with an independent pharmacist, other than the patient s own community pharmacist. Using a comparable design for the pharmaceutical care process will hamper generalizability. It follows that randomisation must be by clusters of patients, for example patients associated with a practice or pharmacy. The use of a before-after 167

168 study design, as in the RESPECT-trial. 11 would not have been appropriate in our setting after the presentation of the HARM-study results and the growing awareness of medication safety and the focus on improving pharmacotherapy by reviewing medication a significant influence of time was expected. At first we planned cluster randomization at a GP level. This design was abandon because of practical obstacles: most participating GPs had a strong preference of providing the pharmaceutical care process or providing usual care. If we kept to our cluster randomization design their would be insufficient participants to show a possible effect. Therefore, in studying the pharmaceutical care process intervention in the PHARM study we have chosen for a controlled design, comparing intervention GPs (whose patients receive the intervention) with control GPs (whose patients receive usual care). Embracing uncer tainty In the PHARM study the objective was the prevention of medication related hospital admissions by designing an intervention for clinicians working with patients in a primary care setting. We expected that the well-established methods to improve patient safety would be less suitable to prevent HARM. By applying the core tools to ban errors in healthcare (variation of so-called system - or root-cause analyses in combination with incident reporting systems) we expected to uncover relative simple problems; like wrongly issued prescription, medication not delivered or software not working as expected. 12,13 While we aimed to improve of the more complex situation of several HARM-medication errors combined in one patient. Also by implementing the well-established To-err-is-human - logic, uncertainties will be controlled and banned, 14 while the uncertainties, the medication errors leading to HARM, in our study can be a desirable part of the drug therapy. For example if a patient is not following the recommended use by the prescriber, he or she will be non-adherent which is a medication error. But this error can be beneficial to the patient if he or she is suffering from diarrhoea and do not want to take his or her oral anti-diabetic drug because not feeling well. This feeling of the patient may prevent the harm of a hypo-glycaemia. Banning and controlling certain uncertainties might not only be a mission impossible it can also be counterproductive and cause HARM. 15,16 Therefore we like to differentiate between two logics of patient safety that differ in how they deal with uncertainty. Apart from the banning and controlling uncertainties we prefer the approach of embracing uncertainty 17 as a possible way to improve patient outcomes and therefore prevent HARM in primary care. In this theory of embracing uncertainty, we recognize and acknowledge the complexity of drug therapy and pharmaceutical care in primary 168

169 General discussion Chapter 4 care. The complexity is embraced in the pharmaceutical care process by counting on knowledge and expertise of the clinicians, while the clinicians will still be held accountable. C omplex inter vention The pharmaceutical care process is a complex intervention because it includes several components. The intervention consisted of four steps, which were adjusted to the individual patient and to individual clinicians who were taking part in the pharmaceutical care process. This is the strength of the pharmaceutical care process to optimise it to the clinical situation and the needs of the patient. But it makes it more difficult to define, develop, document and reproduce the intervention and it includes more variation than intervening with one single drug. In our trial we depended on the expertise, knowledge and skills of various clinicians as well as the clinical investigations, pharmacotherapy, treatment guidelines, and arrangements for evaluation and follow up. The different GP practices and pharmacies may also vary in terms of cooperation, organisation, management, and skill mix. The active components of the pharmaceutical care process may be difficult to specify, making it difficult to replicate the intervention. It also implies that there will be a large variation between pharmaceutical care interventions of other studies and this variation of interventions make it difficult to combine and extrapolate the effect of these studies. Drawing conclusions from of such an extrapolation should be done with great caution. 18 An example of another complex intervention is a stroke unit. 19 A trial to evaluate the benefits of specialist stroke units would also have to consider the expertise of various health professionals as well as diagnostic tests, drugs, protocols, and arrangements for discharge and follow-up. Stroke units may also vary in terms of organisation, management, and skill mix. This complexity of the intervention is comparable to our intervention. We followed several steps to define, develop, document and reproduce the intervention, to make it suitable for a clinical study. To define the pharmaceutical care process as an intervention we first observed similar practices in the UK and searched the literature. This allowed us to identify which components of the intervention might have the desired effect: full access to clinical notes and medication history, involvement of the patient and follow-up time. The next step in defining the intervention was a pilot study to explore the possibilities of the intervention in daily practice and to improve understanding of the components of an intervention and their interrelationships. This qualitative research was also used to show how the intervention works and to find potential barriers to implement change in trials 169

170 that seek to alter patient or professional behaviour. 20 The information gathered was used to develop the best acceptable and feasible intervention and study design and to define the intervention as specific as possible. The intervention therefore secured the autonomy of the clinician to decide what would be the best pharmacotherapy for their patient and secured the autonomy of the patient to decide what pharmacotherapy would address their needs best with the support of the clinicians. Therefore it would be safe to participate in the study. The GPs and the pharmacists were only given a structure to follow, as described in Chapter 3.1. This implies the risk of permissiveness, which can be prevented by the pressure of a third party. We used peer pressure of the research team and in some centres it was possible to use a financial pressure by reimbursement of insurance companies. This implies the risk of checking boxes to obtain the financial support without practicing the pharmaceutical care plan as it was intended in the intervention. This was prevented by reasoning about the objective of the study and moreover by the intrinsic motivation of the GPs and the pharmacists to care for their patients. Which implies the risk of lack of time, which will be discussed later in this chapter with possible remedies. 21 This makes the complex intervention of the pharmaceutical care process a divers intervention and therefore more difficult to reproduce the intervention and locate the components which are effective. On the other hand the findings, as presented in Chapter 3.2, would be more generalisable or applicable if these are performed in an ordinary primary care setting in which they are most likely to be implemented. Implications of the HARM study In the years before 2006 data on hospital admissions related to medication were derived from international studies and extrapolated to the Dutch situation. For example, based on a review of literature Beijer and De Blaey estimated the number of hospital admissions due to medication for The Netherlands as high as in total. 22 As we have shown in our literature review, such extrapolations based on computing mean frequencies of all studies combined (irrespective of the definition of events, methodology, patient population and setting) can result in invalid frequencies. Indeed, the HARM-study has shown that the number of medication related hospitalizations is lower ( admissions yearly). In the same period as the HARM-study, the IPCI study was performed, 23 which showed very comparable results both with respect to the frequency as to the type of medication involved. 170

171 General discussion Chapter 4 Both studies have greatly added to the sense of urgency on medication safety. They have received considerable attention in the media and in healthcare journals, leading to parliamentary inquiries to the Minister of Health. Subsequently, both studies were used as input for an expert committee, that was formed by the Ministry of Health in order to reduce the burden of medication related hospitalizations. This expert committee published their recommendations on quick wins in the form of the HARM wrestling report in The quick wins were based on the most commonly involved medication in the HARM study and concerned improvement measures that were estimated to be implemented relatively easily (e.g. gastroprotection with NSAID use in high risk patients). The HARM wrestling report is now used for example for the development of algorithms within clinical decision support systems (= medisch-farmaceutische beslisregels ) in hospital and primary care, aimed at reducing medication related harm. Numerous other initiatives by healthcare workers have been started based on the results of the HARM-study. The recommendation on optimization of medication reconciliation on transfer to and from hospital has stimulated projects on safe medication transfer. Furthermore, the HARM results have been used to stimulate the national project on electronic patient data transfer. The recommendation on sharing clinical laboratory values between hospital and community pharmacies was embedded in an agreement statement between the KNMP (Royal Dutch Association for the Advancement of Pharmacy) and the NVKC (Dutch Society of Clinical Chemistry). Projects have been started in which laboratory data are being exchanged between hospital and community pharmacies and are being used to improve individual pharmacotherapy. Software is being developed to share this information and more clinical data between hospital, community pharmacies and GP practices and share a pharmaceutical care plan. The recommendation on reviewing medication of patients with the provoking risk factors identified in the HARM-study, has resulted in several regional and national projects on implementation and effect of pharmacotherapy review, one of which is the PHARM project described in this thesis. Many of these project focus on elderly patients, patients with polypharmacy or patients living in a care or nursing home. For example the Dutch Institute for Rational Use of Medicine (IVM) started a project with GPs and community pharmacists to review the pharmacotherapy of elderly patients living in care homes. Several healthcare insurers have developed a reimbursement scheme for GPs and community pharmacists performing such reviews. Their materials are often based on the materials used in the PHARMstudy. Health care insurance companies also started projects in GP practices, which targets elderly patients over 75 with involvement of the practice nurse. 171

172 Another implication is the development of a multidisciplinary clinical guideline around the care of elderly patients with polypharmacy. The objective of this guideline is to improve the pharmacotherapy and improve clinical outcomes of this vulnerable group. In conclusion, the HARM study had many implications, both societal and within the health care system. It has given rise to many projects to improve pharmacotherapy, which are expected to reduce the burden of medication related hospitalizations. Whether this will be actually the case, needs to be the topic of further studies. One of the studies aimed at determination of the effect of such an improvement project is the PHARM study. The results of this study are presented earlier in this thesis and are positive with respect to a reduction in medication related hospital admissions. Lessons of the PHARM study Besides the effect of medication review on the primary outcome of medication related hospitalizations, the PHARM study has given much more lessons as to implementation of such a multidisciplinary intervention, with respect to the level of education, with respect to the compatibility of the intervention to the present working processes (and thus the time consuming nature) with respect to the role of the patient and with respect to the policymakers. These aspects will be discussed in more detail in this last section of this discussion. The discrepanc y between intended and actual number of included patients in the PHARM-study Medication safety is a recognized problem by doctors and pharmacists and already much has been done to reduce medication errors. With the publication of the HARM-study medication safety became a public concern and many clinicians were motivated to prevent medication related hospital admissions. Because we wanted to prevent medication related admissions to hospital we designed an intervention for doctors and pharmacists working with patients outside the hospital, in a primary care setting. Although there was a sense of urgency created with the HARM-study, we experienced many problems when implementing the pharmaceutical care process in primary care. Hundreds of pharmacists were interested in participating in the PHARM study and made enquiries about the study or went to a meeting in which we provided information on the study, its intervention and workload. Over 200 pharmacists (several together with a GP), from more than 200 settings, 172

173 General discussion Chapter 4 participated in the first workshop (as described under support in Chapter 3.1). At that time they were all confident to carry out the intervention with one or more GPs within their practice. They all agreed to include 40 to 50 intervention patients and the same number of control patients within the 12 months period of the study. Even after prolonging the inclusion period with three months, only 42 settings (of a pharmacist and at least two GPs) included a total of 674 patients. The number of included patients varied from 1 to 61 with a median of 13 for all patients per setting, and varied from 1 to 31 with a median of 8 for intervention patients per setting. This was much less than we expected to include. Based upon the sample size calculation, we aimed to include patients, 7100 in each arm, from 142 centres of at least one pharmacist and at least two GPs (Chapter 3.1). Therefore we investigated what caused this low number of included patients, during and after the study period, by interviewing pharmacists and GPs, by studying the behavior of the pharmacists and the GPs. Several factors might explain these low numbers of included patients as was discussed in the Appendix to Chapter 3.2. These concern project management, collaboration between the pharmacists and the GPs, knowledge, skills and time, fit of the intervention with the present working process and information management. We experienced no barriers in the number of eligible patients or in obtaining consent from patients. In this section we will focus on possible recommendation for implementing the pharmaceutical care process and future studies. Project management In the HARM-study the results in terms of included patients was much higher than in the PHARM-study. Although the studies are non-comparable in design, observational study versus intervention study, we discussed some reasons to explain the difference in success regarding to output of patients, which should be taking into account when designing new research (Appendix to Chapter 3.2). Observing retrospectively is of course less complicated than initiating a complex intervention that demands a lot of effort from the participants. In the HARM-study a full time researcher performed most of the workload. Therefore in future studies it is recommended that the professionals who are performing the intervention would have this as a separate task and would be able to dedicate time to do this can be done by a separate budget. The professionals should also be fully equipped with knowledge, skills and support. This will be discussed separately regarding to the pharmaceutical care plan. 173

174 C ollaboration between GP and pharmacist As discussed in the Appendix to Chapter 3.2 the collaboration between a GP and a pharmacist is often poor. This can be explained by the differences between the professionals in responsibilities, focus, background, personality and accommodation. The pharmaceutical care process can only be successful if both the GP and the pharmacist invest in working together, in a structural way. Both clinicians also should be able to receive feedback on the pharmacotherapy of the patient from each other. Especially because the problems in the pharmaceutical care process setting were no acute problems and they were not always as clear and obvious since there is limited evidence in the elderly patient with several co-morbidities or there are no clear relationships between a symptom and lack of effect. Most of the teams that did participate in the study often had to intensify and improve their level of collaboration in order to make the pharmaceutical care process work. Centres were they managed to include several intervention patients, in general, already have had experiences in multidisciplinary projects of patient care and were often health centres with several GPs and several pharmacists therefore they were used to cooperate and share pharmaceutical care. Moreover younger pharmacist who used the participation in the PHARM-study as part of their ongoing education project, were more successful in the number of included patients. In these settings the GP was appointed as a tutor and was experienced as a role model. To implement the pharmaceutical care process we recommend a structural meeting between the GP, the pharmacist and a practice nurse or specialised pharmacy technician if available. If such multidisciplinary meetings would exist once a week new pharmaceutical care plan can be agreed upon and the follow-up of the care plans can be secured. If they would be working in the same place, e.g. a health centre, it would be easier to establish a good professional relationship and diminish distrust. This would also help in lowering the threshold for all involved professionals to consult each other in the pharmaceutical care of their joint patient. In this way they can experience what the additive value is of the knowledge and skills of the other professional. If the pharmacist would be separated from the dispensary, he can easier focus on the pharmaceutical care of patients. In addition the pharmacist must also invest time to establish new professional relationships with consultants from the nearby hospitals, with nurses, with physiotherapist, with home care, with care-home staff and other health care providers who will need to be aware of the services and provided care in order to communicate possible drug therapy problems and in order to refer patients. 174

175 General discussion Chapter 4 Fit with the working process The pharmaceutical care process does not comply with the present regular working process in the pharmacy or the GP practice as discussed in the Appendix to Chapter 3.2. A major part of the pharmacists who first intended to participate later dropped out due to the time consuming nature of the pharmaceutical care process, as they stated. By coincidence the pharmaceutical care process was implemented in the same time that new budgetary restrictions had to be implemented in the pharmacies. These cost reducing measures involved major logistic changes in the working process of community pharmacies, leaving them with less time for patient oriented tasks. The time consuming nature of the pharmaceutical care process is partly due to the lack of compatibility of the pharmaceutical care process with the present working processes. In the PHARM-study we observed a considerable time investment of the pharmaceutical care process for both community pharmacists and GPs, approximately four hours per patient for the pharmacist and one hour per patient for the GP. Even though the pharmaceutical care process succeeded in reducing the number of medication related hospital admissions and even though these HARMs are associated with considerable costs, it is unlikely that the pharmaceutical care process will be cost saving in its present form. Although it is thinkable that there will be willingness to pay to improve patient outcomes and it can therefore be cost effective. Furthermore, cost-effectiveness studies are required in order to determine whether the time involved with the pharmaceutical care process is counterbalanced by the results with respect to reduction of HARMs and possible improvement in quality of life. To implement the pharmaceutical care process into practice, significant organisational changes need to be taken. Because of the two separated organisations and the linear way of working, assistance is needed to make this organisational change. The professionals should start thinking from the unique needs of the patient and organise the care around the patient. This operational change is outside the scope of daily practice but needs to become daily practice, which implies readjusting daily practice. The help of an organisation manager might be helpful. Professionals should make choices on the care they provide. In a joint care process the professionals should all agree on the provided care. To separate the pharmacist from the dispensary and bring him into the organisation, or closer to, the GP practice might be beneficial. The role of the practice nurse and the pharmacy technician in the pharmaceutical care process should be explored and defined to secure continuity of care, improve outcomes and reduce costs. 175

176 In general people do not like to change. Any organisational change like described above will create resistance. To overcome resistance incentives are needed. Several incentives can be thought of for implementing the pharmaceutical care process: a financial incentive; the urgency to change to survive, this can be the case for the pharmacists to be of added value to the health care system; professional development by training and education; and satisfaction through adding quality of life to a patient. We thought the last two would give sufficient intrinsic motivation to perform the pharmaceutical care process. Further research is needed to explore the possible incentives and their contribution to organisational change. Future research is needed on successful organisation management of the pharmaceutical care process. In future research of a complex intervention, more emphasis should be on the organisation of the intervention and planning assistance is needed. Knowledge and skills For the participating pharmacists and GPs practicing the pharmaceutical care process could have been too difficult with their current skills and knowledge. As researchers our expectations of the clinicians in general were too high regarding managing the risk to patients associated with medication use, as discussed in the Appendix to Chapter 3.2. To improve the level of knowledge and skills on clinical pharmacy, an elaborate training program was developed to train community pharmacists in communication skills and documentation skills, and to train both pharmacists and doctors in pharmacotherapy within the high-risk elderly patient population, as described in Chapter 3.1. During the study we also experienced that several pharmacists had insufficient competence to manage uncertainty and to act in the patient s interest. As mentioned in the Appendix to Chapter 3.2, the pharmacist focuses on the true solutions and lack of prove brings uncertainty. This wisdom cannot be taught in a workshop of several hours but has to be developed through past experience, both individual and shared with GPs. We also experienced that the level of education of the GPs on pharmacotherapy was too low for the GP to make an assured decision about changing the pharmacotherapy of an individual patient. This resulted in no interventions in the pharmacotherapy of the patients, which was especially seen in the missing intervention: stop drug. The knowledge of pharmacotherapy of practice nurses can also be improve to be of better support in the pharmaceutical care practice. An important lesson of the PHARM-study is that education and training of pharmacists at the university level needs to be improved with respect to clinical 176

177 General discussion Chapter 4 pharmacy. The knowledge about medicines and the understanding of their action has to be put in the context of the disease with the aim to improve quality of life for individual patients and groups of patients. Parts of these improvements have already been implemented in curricula at the universities, but only in recent years and pharmacists graduating with that improved program are still relatively scarce. A second way to improve the level of education is to develop postgraduate programs. Currently the postgraduate training program for community pharmacist is professionalized. This professionalized program should be more patient centred and encompass further training in skills in clinical pharmacy, pharmaceutical care, communication and documenting pharmaceutical care. This should lead to advanced practitioners, who can serve as the role models needed to develop the same in others. By combining postgraduate education of pharmacists and GPs they can not only improve their knowledge on pharmacotherapy but also they can learn skills from each other, in their way of reasoning and therefore improve the quality of their cooperation in a non-threatening learning environment. They can experience the possible contribution of the other professional and this can create a culture in which they can benefit from the other professional, can to count on the other professional and ultimately lead to better sharing of problem solving. A new learning curve in practice is also recommended, for both the pharmacist and the GP. This learning curve applies not only to the provision of the pharmaceutical care but also to the understanding and participating in the joint process. The practice learning curve requires seeing and caring for a sufficient number of patient to become proficient. Information management The gathering and documentation involved in the PHARM-study took too much time and energy what might have led to frustration. Therefore more emphasis should be on support of the clinicians in the accessibility of the information and on sharing the information. They can be supported by improved IT systems in which they can document drug therapy related problems and care issues of the patient, the pharmaceutical care with its intended intervention, follow up and evaluation and share relevant drug therapy related information, medical history, medication history including intolerabilities and clinical laboratory data, between clinicians. Sharing this pharmaceutical care plan with all the information is required for the management organisation of the care. The clinicians can also be supported by multi disciplinary guidelines and local procedures, and by supporting staff to relieve them from administrative and non-medical duties and by improved reimbursement systems for pharmaceutical care. 177

178 Role of the patient The patient has a central role in medication safety. He needs to be fully informed in order to use the medication as planned and in order to be convinced of the necessity of its use. This last aspect is important in order to achieve optimal adherence and thus to reduce the risk of medication related hospital admissions (as non-adherence was an important risk factor identified in the HARM-study). To improve a patient s drug therapy and therefore a patient s medication safety, changes in the existing drug therapy are often unavoidable. But in the PHARM study several difficulties were experienced by the patient or by the clinician to make these changes. A major difficulty was the resistance of patients to change, especially if they did not experience any problems or if they did not experience direct improvement. The patient s autonomy could also lead to difficulties with elderly patients who were not able to manage their own medication due to impaired cognition, hearing or vision, or lack of support of carers. The elderly patients included in the study are within an age group of patients who put the physician on a pedestal and are restrained in their communication with the physician about their complaints and medication experiences. The resistance to change was experienced by the clinicians but can not be concluded from a questionnaire that was send out at the end of the study period to 119 patients. Eighty (67%) patients from 10 settings returned the questionnaire. 87% of the patients (28 of 32) felt comfortable in the change in medication the GP or pharmacist suggested to them. We might conclude from this questionnaire that the patients were satisfied with the pharmaceutical care process and changes in their medication. Most of the 70 patients that did recognize the pharmaceutical care process, were satisfied with this service: 59 (84 %) of the patients. We recommend to emphasize on the pharmaceutical care of the vulnerable group of elderly patients with multimorbidities, polypharmacy and non-adherence and focus on the communication between the clinicians and the patient and its carers about symptoms, medication experience and medication use. It is important to actively involve the patients and their carers in the pharmacotherapy of the patient. By improving the quality of life of an individual patient the intrinsic motivation of the involved clinician will improve which makes it possible to provide an intervention such as the pharmaceutical care process. Role of the polic ymaker From this discussion on implementation factors and recommendations, we can extract recommendations for policymakers. They can improve medication safety and patient outcomes by supporting the clinicians and creating external pressure, 178

179 General discussion Chapter 4 for instance with legislation or budgetary constrains. Support can be financially or by creating the right environment. To be more specific, support can be given in building a health care centre with several disciplines, financially support for the care that is supplied or for other staf, support in administrative duties or other nonmedical duties, support in organisational management, support educational change, emphasize on local procedures and support the development of clinical guidelines. C onclusion In conclusion, this thesis has presented the results on studies on medication related hospitalizations and on the effect of a multidisciplinary intervention to reduce these hospitalizations. The HARM study has had major impact on society and healthcare and has resulted in several improvement projects in the Netherlands. This PHARM intervention has shown to be effective, but much can be learned from the difficulties experienced while implementing this intervention. With these lessons the intervention and the organization of the intervention can be optimized in order to achieve a HARM-reducing intervention that will be cost effective. Future studies should especially look into implementation of the pharmaceutical care process, which can also make this effective for large group of patients and might be cost effectiveness. References Van Doormaal JE, Mol PG, van den Bemt PM, et al. Reliability of the assessment of preventable adverse drug events in daily clinical practice. Pharmacoepidemiol Drug Saf. 2008;17: Kramer MS, Leventhal JM, Hutchinson TA, et al. An algorithm for the operational assessment of adverse drug reactions, I: background, description, and instructions for use. JAMA. 1979;242: Kunac DL, Reith DM, Kennedy J, Austin NC, Williams SM. Inter- and intra-rater reliability for classification of medication related events in paediatric inpatients. Qual Saf Health Care. 2006;15: Jha AK, Kuperman GJ, Teich JM, et al. Identifying adverse drug events: development of a computer-based monitor and comparison with chart review and stimulated voluntary report. J Am Med Inf Assoc. 1998;5: Thomas EJ, Lipsitz SR, Studdert DM, Brennan TA. The reliability of medical record review for estimating adverse event rates. Ann Intern Med. 2002;136:

180 Murff HJ, Patel VL, Hripcsak G, Bates DW. Detecting adverse events for patient safety research: a review of current methodologies. J Biomed Inform. 2003;36: Honigman B, Light P, Pulling RM, Bates DW. A computerized method for identifying incidents associated with adverse drug events in outpatients. Intern J Med Inf. 2001;61: O Neil AC, Petersen LA, Cook EF, Bates DW, Lee TH, Brennan TA. Physician reporting compared with medical-record review to identify adverse medical events. Ann Intern Med. 1993;119: Edwards S, Braunholtz D, Stevens A, et al. Ethical issues in the design and conduct of cluster RCTs. BMJ. 1999;318: Holland R, Lenaghan E, Harvey I, et al. Does home based medication review keep older people out of hospital? The HOMER randomised controlled trial. BMJ. 2005;330: Wong I, Campion P, Coulton S, et al. Pharmaceutical care for elderly patients shared between community pharmacists and general practitioners: a randomised evaluation. RESPECT (Randomised Evaluation of Shared Prescribing for Elderly people in the Community over Time). BMC Health Serv Res 2004;4:11. Bont de A, Jerak-Zuiderent S, Roland B. Safety in health care-background study on the state of healthcare 2009 (In Dutch: Veiligheid in de zorg-achtergrondstudie bij de Staat van de Gezondheidszorg 2009). Rotterdam: Institute of Health Policy & Management Erasmus University; Jerak-Zuiderent S, Zwart DLM, de Bont A. Exploring the logics & practices of patient safety in primary care - banning or embracing uncertainties? [submitted] Kohn L, Corrigan J, Donaldson M; Committee on Quality of Health Care in America, Institute of Medicine. To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press; Law J. Ladbroke Grove, or how to think about failing systems. Lancaster: Centre for science studies, Lancaster University; Amalberti R. What is the future and which safety strategy for an ultrasafe system? Transfus Clin Biol. 2009;16:80-5. Bijker WE. American and Dutch coatal engineering:differences in risk conception and differences in technological culture. Social Studies of Science. 2007;37: Holland R, Desborough J, Goodyer L, Hall S, Wright D, Loke YK. Does pharmacist-led medication review help to reduce hospital admissions and deaths in older people? A systematic review and meta-analysis. Br J Clin Pharmacol. 2008;65: Stroke Unit Trialists Collaboration. Organised inpatient (stroke unit) care for stroke. Cochrane Database Syst Rev Oct 17;(4):CD Haynes B, Haines A. Barriers and bridges to evidence based clinical practice. BMJ. 1998;317: Leistikow IP. Sturen op patiëntveiligheid SAFER [dissertation]. Delft: Delft University of technology; [in press] Beijer JHM, de Blaey CJ. Hospitalisations caused by adverse drug reactions (ADR): a metaanalysis of observational studies. Pharm World Sci. 2002;24: Van der Hooft CS, Dieleman JP, Siemes C, et al. Adverse drug reaction-related hospitalisations: a population-based cohort study. Pharmacoepidemiol Drug Saf. 2008;17:

181 General discussion Chapter HARM-wrestling; Een voorstel van de Expertgroep Medicatieveiligheid mbt concrete interventies die de extramurale medicatie veiligheid op korte termijn kunnen verbeteren. Den Haag: Ministerie van Volksgezondheid, Welzijn en Sport;

182

183

184

185 Medication is one of the most commonly applied medical interventions in health care and it has shown to improve health and quality of life. Unfortunately it can also do harm and can cause severe adverse events like hospital admissions. These adverse events can be caused by the drug product itself but can also be caused by a human factor: medication error. Patients, clinicians and health policy makers are requesting more information on the burden of medication errors. Moreover, there is an urge for an effective method to prevent these errors and to improve patient outcomes. The focus in this thesis will be on medication related hospital admissions: the magnitude of the problem, the provoking factors associated with and the prevention of these admissions. To design an effective method to reduce these admissions, information is needed on which patients are at risk, which drugs are hazardous and which medication errors cause hospital admissions. Therefore, the objective of this thesis is to gain more information on medication-related hospital admissions and to use this information to improve patient outcomes. The first part (Chapter 2) focuses on the prevalence of medication-related hospital admissions, the burden to society of these admissions and which factors can provoke these admissions. Based on the findings in the first part we designed an intervention to prevent medication-related hospital admissions: the pharmaceutical care process. The design and the results of the study on the effect of the pharmaceutical care process are described in the second part (Chapter 3). In Chapter 2.1 we explored the influence of study characteristics on the prevalence of medication-related hospitalizations. After a literature search we categorized the retrieved studies, based on study setting, studied population, outcome of medication-related problem, method of data collection and continent in which the study took place. We then examined the relationship between these factors and reported prevalence of medication-related hospital admissions. Ninety-five studies were analysed, with a range of reported prevalence of medication-related 185

186 hospitalizations from 0.1% to 54%. Higher prevalences were found in the studies examining all hospital admissions than in the studies examining only acute hospital admissions, in the elderly population than in children, in studies examining ADEs than in studies examining only ADRs, using medical chart screening than database methods or spontaneous reporting and lower prevalences were found in Europe than on other continents. Because of this variation in reported prevalences, we concluded that extrapolation using national hospital admission data and the prevalence identified by pooling international studies should be carried out with great caution. Given this variation in prevalences and the need for information on potential risk factors associated with medication-related hospital admissions, we conducted the multicenter HARM (Hospital Admissions Related to Medication) study as described in Chapter 2.2. This study aimed to assess the prevalence and preventability of medication-related hospitalizations in the Netherlands and to identify risk factors for the preventable hospitalizations. The results of this study show that a considerable proportion (5.6%) of all unplanned admissions are medication related of which almost half (46.5%) are potentially preventable. The main determinants of preventable medication-related hospital admissions were impaired cognition (odds ratio [OR] 11.9; 95% confidence interval [95%CI] ), four or more comorbidities (OR 8.1; 95%CI ), dependent living situation (OR 3.0; 95%CI ), impaired renal function (OR 2.6; 95%CI ), non-adherence to the medication regimen (OR 2.3; 95%CI ), and polypharmacy (OR 2.7; 95%CI ). These identified risk factors provide a starting point for preventing medication-related hospital admissions. In Chapter 2.3 we further analyzed the HARM-cases to determine if medication errors and renal impairment contribute to hospital admission and to gain more information on these errors. We analyzed the 714 medication-related hospital admissions and divided them into three groups based on the availability of creatinine levels: Group A, the home-monitored group (n=227), group B, a remarkably large, in-hospital-monitored group (n=420) and group C, the nonmonitored group (n=67). After assessment 70 admissions (10%) were related to a medication error and renal impairment (A: 29, B: 41, C: none). A dosing error was found in 46 patients (A: 14, B: 32), a drug-drug interaction in 22 patients (A: 13, B: 9) and a drug-disease interaction in 17 patients (A: 10, B: 7). From these results we conclude that renal impairment and medication may lead to medication-related hospital admissions. From the admissions in the hospital-monitored group B we 186

187 Summar y Chapter 5.1 conclude that regular monitoring of renal function is relevant and could have prevented these admissions had it been carried out. Although renal function was monitored in group A, relevant medication errors still occurred in this group. This supports the conclusion that besides monitoring of renal function, knowledge on the adaptation of pharmacotherapy to the renal function is important and should be improved in order to prevent hospital admissions. Hospital admissions related to medication are not only a burden to patients and their relatives but also to society, potentially involving high costs. To provide more information on the economic burden of preventable adverse drug events of outpatients, we present in Chapter 2.4 a cost study on the data collected in the HARM-study. We calculated the average costs for a preventable medication-related hospital admission by summing the direct medical costs and the production losses of all the preventable admissions, taking into account the different types of hospitals (academic and general) and the age of the admitted patients. The average medical costs for one preventable medication-related hospital admission were The average production loss costs for one admission were 1712 for a person below 65 years of age. Combining the medical costs and the costs of production losses resulted in average costs of 6009 for one, potentially preventable, medicationrelated hospital admission for all ages. We therefore concluded that the costs of potentially preventable hospital admissions related to medication are considerable. Patient safety interventions to prevent ADEs and hospital admissions may therefore be cost-effective or even cost saving. Medication review or pharmacotherapy review has often been proposed as a solution to prevent hospital admissions related to medication and to improve the effectiveness and safety of pharmacotherapy. However, most published randomized controlled trials on pharmacotherapy reviews showed no or little effect on morbidity and mortality. These studies differed largely with respect to the nature and extensiveness of the review techniques, the outcomes studied, setting and follow-up time and results are therefore difficult to compare. Therefore we designed the PHARM (Preventing Hospital Admissions by Reviewing Medication)-study with the objective to study the effect of the total pharmaceutical care process on medication related hospital admissions and on adverse drug events, survival and quality of life. The study design with its explanation and definition of the different steps in of the pharmaceutical care process is described in Chapter 3.1 while the results of the study are presented in Chapter

188 The PHARM-study was designed as a cluster randomized, controlled, multi-centre study in an integrated primary care setting. We included only patients with a high risk of a medication related hospital admissions in the study, based on the findings from the HARM-study. This study resulted in four criteria for high risk: old age, non-adherence, type of medication used and polypharmacy. The intervention, the pharmaceutical care process, consisted of taking a pharmaceutical anamnesis, executing a pharmacotherapy review, formulating and agreeing on a pharmaceutical care plan and the monitoring and follow up evaluation of pharmacotherapy as documented in the pharmaceutical care plan. The pharmaceutical care process was a continuous process of different steps and occurred over multiple encounters of patients and clinicians during the 12 months study period. Randomisation took place at GP (general practitioner) level. The patient s own pharmacist and GP carried out the intervention. The control group was included by another GP than the intervention GP and received usual care. The primary outcome of the study was the frequency of hospital admissions related to medication within the study period of 12 months of each patient. The secondary outcomes were survival, quality of life, adverse drug events and severe adverse drug events. The outcomes were analysed by using mixed-effects Cox models. In the final study protocol we abandoned the randomized design, because of the difficulties in finding cooperating GPs. Instead we used a controlled design, selecting control patients of one general practitioner and the intervention patients of another general practitioner (both from the same community pharmacy). We aimed to include patients, 7100 in each arm, from at least 142 pharmacy practices but experienced many difficulties in including patients. Eventually 364 intervention and 310 control patients were included in 42 primary health care settings of at least one pharmacist and at least two GPs. More medication related hospital admissions were found in the control group than in the intervention group; respectively 10 and 6 admissions. The effect was dependent on the number of diseases. We calculated the effect as a hazard ratio (HR). The effect of the intervention for three diseases was HR 0.77 (95%CI ); for four diseases HR 0.43 (95%CI ); for five diseases HR 0.28 (95% CI ); for six diseases HR 0.19 (95% CI ); for seven diseases HR 0.14 (95% CI ) and for eight diseases HR 0.11 (95% CI ). Between the intervention and control group no statistically significant differences were found in the secondary outcomes survival, adverse drug events and quality of life. With these results we demonstrated that the patient s own pharmacist together with its own GP and the patient may prevent hospital admissions with a medication related cause. 188

189 Summar y Chapter 5.1 In Chapter 4 some methodological aspects, implications of the HARM-study and lessons learned for the PHARM-study are discussed in a broader context. The difficulties in the assessment of the causality between an adverse event and the medication as well as between an adverse event and the medication error are discussed. Furthermore the design of the PHARM-study is discussed. The HARM-study has already had many implications, both societal and within the healthcare system. It has given rise to many projects to improve pharmacotherapy, which are expected to reduce the burden of medication related hospitalizations. The PHARM-study is one of these projects and besides the effect of the pharmaceutical care process on the primary outcome of medication related hospitalizations, the PHARM study offered more lessons to learn. The implementation of the multidisciplinary pharmaceutical care process did not succeed in several primary care because of the difficult cooperation between different groups of clinicians, the pharmaceutical care process was not compatible to the present working processes (and thus the time consuming nature) and the insufficient knowledge and skills from the participants. With these lessons learned the pharmaceutical care process can be optimized in order to achieve a HARM-reducing intervention, which does not only prevent negative effects of pharmacotherapy but also assuring positive effects and therefore improve quality of life and moreover will be feasible in an integrated primary care setting. 189

190

191 Geneesmiddelen zijn de meest gebruikte medische behandeling in Nederland. Geneesmiddelen kunnen de gezondheid van patiënten verbeteren en de kwaliteit van leven verhogen. Helaas kunnen geneesmiddelen ook bijwerkingen hebben en schade veroorzaken zoals een ziekenhuisopname. Zo n ongewenst gevolg kan veroorzaakt worden door het geneesmiddel zelf of de wijze waarop mensen het geneesmiddel gebruiken. Dit laatste noemen we dan een medicatiefout. Patiënten, behandelaren en beleidsmakers zijn gebaat bij meer informatie over de omvang en de gevolgen van deze medicatiefouten. Maar het is nog belangrijker om een effectieve manier te vinden om deze medicatiefouten te voorkomen en de behandeling met geneesmiddelen te verbeteren. In dit proefschrift besteden we aandacht aan geneesmiddelgerelateerde ziekenhuisopnames: de omvang van het probleem, factoren die van invloed zijn en hoe de opnames te voorkomen zijn. Om een effectieve manier te ontwikkelen om deze opnames te voorkomen is meer informatie nodig over welke groepen patiënten een hoog risico hebben om te worden opgenomen, welke geneesmiddelen een ziekenhuisopname kunnen veroorzaken en welke medicatiefouten leiden tot ziekenhuisopname. In het eerste deel (Hoofdstuk 2) beschrijven we onderzoeken naar de omvang van geneesmiddelgerelateerde ziekenhuisopnames, welke gevolgen dit heeft voor de maatschappij en welke factoren een rol spelen bij deze opnames. Gebaseerd op de uitkomsten van het eerste deel ontwikkelden we een manier om deze opnames te voorkomen: het farmacotherapeutische zorgproces. Het ontwerp en de resultaten van dit onderzoek beschrijven we in het tweede deel van dit proefschrift (Hoofdstuk 3). In Hoofdstuk 2.1 beschrijven we de resultaten van een literatuur onderzoek. Hiervoor hebben we artikelen verzameld over studies naar geneesmiddelgerelateerde ziekenhuisopnames. Door deze naast elkaar te leggen, konden we het verband bestuderen tussen het voorkomen van dergelijke opnames (de prevalentie ) en 191

192 bepaalde kenmerken: land en plaats van de studie, de bestudeerde groep patiënten, de gemeten uitkomsten en de manier van gegevens verzameling. Vervolgens bekeken we per categorie de prevalentie. We vonden in de totaal 95 studies een prevalentie van geneesmiddelgerelateerde ziekenhuisopnames die varieerde van 0,1% tot 54%. De hogere prevalenties vonden we in studies die alle ziekenhuisopnames bestudeerden in plaats van alleen niet-geplande ziekenhuisopnames, die oudere patiënten bestudeerden in plaats van kinderen, die alle ongewenste gevolgen van geneesmiddelen bestudeerden in plaats van alleen bijwerkingen van het geneesmiddel zelf, die de dossiers van patiënten bestudeerden in plaats van een computerbestand of een spontane melding van een bijwerking, en in onderzoeken die buiten Europa hadden plaatsgevonden in plaats van in Europa. Vanwege de grote verschillen in prevalentie is het niet mogelijk gebleken om op basis hiervan conclusies te trekken over de Nederlandse situatie. Om inzicht te krijgen in de Nederlandse situatie en om inzicht te krijgen in welke patiënten een verhoogd risico hebben op een opname, voerden we het zogeheten HARM-onderzoek uit: Hospital Admissions Related to Medication = ziekenhuisopnames gerelateerd aan geneesmiddelen. Dit onderzoek staat beschreven in Hoofdstuk 2.2. We wilden in kaart brengen hoeveel geneesmiddelgerelateerde ziekenhuisopnames er zijn in Nederland, of deze te voorkomen zouden zijn en welke factoren het risico op een dergelijke opname vergroten. We keken in 21 ziekenhuizen naar alle niet-geplande opnames. We keken bij deze opgenomen patiënten naar de reden van hun opname en naar hun geneesmiddelgebruik thuis. Als de reden van opname een ongewenst gevolg was van de gebruikte geneesmiddelen dan beschouwden we dit een geneesmiddelgerelateerde ziekenhuisopname en verzamelden we informatie van de patiënt. We ontdekten dat 5,6% van alle niet-geplande ziekenhuisopnames gerelateerd was aan een geneesmiddel en dat bij bijna de helft (46%) van deze ziekenhuisopnames een medicatiefout gemaakt was en daarmee de ziekenhuisopname potentieel vermijdbaar was. De gemiddelde leeftijd van deze potentieel vermijdbare opnames lag hoger (68 jaar) dan van alle niet-geplande ziekenhuisopnames (60 jaar). Andere kenmerken van patiënten die een hoog risico hebben waren een verminderde cognitie (ongeveer twaalfmaal hoger risico), vier of meer aandoeningen (ongeveer achtmaal hoger risico), niet zelfstandig wonend bijvoorbeeld in een verzorgings- of verpleeghuis (ongeveer driemaal hoger risico), verminderde nierfunctie (ongeveer twee en een half maal hoger risico), gebruik van geneesmiddelen niet volgens de instructies van de arts (ruim tweemaal hoger risico) en het gebruik van vijf of meer geneesmiddelen (ruim twee en een half maal hoger risico). Deze kenmerken of 192

193 Samenvatting Chapter 5.2 risicofactoren geven handvaten voor het voorkomen van geneesmiddelgerelateerde ziekenhuisopnames. In het HARM-onderzoek ontdekten we dat verminderde nierfunctie het risico op een geneesmiddelgerelateerde ziekenhuisopname verhoogt. In Hoofdstuk 2.3 onderzoeken we of de verminderde nierfunctie zelf ook leidt tot een geneesmiddelgerelateerde ziekenhuisopname. We analyseerden de opnames uit het HARM-onderzoek, achterhaalden de nierfunctie van deze patiënten, bepaalden of er een medicatiefout gemaakt was met betrekking tot de nierfunctie en indien dat het beval was, of die heeft bijgedragen aan de reden van opname. We deelden de HARM-patiënten in op basis van de beschikbaarheid van de creatininewaarde in het bloed (een maat voor de nierfunctie): Groep A (bestaande uit 227 patiënten) met creatininewaardes van voor de opname, groep B (een opvallend grote groep van 420 patiënten) met alleen creatininewaardes gemeten tijdens de opname en groep C (bestaande uit 67 patiënten) zonder creatininewaardes. Na beoordeling vonden we zeventig patiënten (10%) waarbij de ziekenhuisopname gerelateerd was aan een medicatiefout en aan de verminderde nierfunctie (29 in groep A; 41 in groep B; en geen in groep C). Bij 46 patiënten (14 in groep A; 32 in groep B) was er sprake van een doseerfout, bij 22 patiënten vond er ongewenste interactie plaats tussen verschillende geneesmiddelen (13 in groep A; 9 in groep B), bij 17 patiënten (10 in groep A; 7 in groep B) vond er ongewenste interactie plaats tussen een geneesmiddel en de aandoening van de patiënt. Met deze resultaten concluderen we dat een verminderde nierfunctie in combinatie met geneesmiddelen kan leiden tot een geneesmiddelgerelateerde ziekenhuisopname. Daarnaast concluderen we dat het zinvol is om de nierfunctie te controleren en om de behandeling met geneesmiddelen hierop aan te passen om zo een mogelijke opname te voorkomen. Geneesmiddelgerelateerde ziekenhuisopnames zijn niet alleen last voor de patiënt en zijn omgeving maar ook voor de maatschappij, omdat er hoge kosten mee gemoeid zijn. Om hier meer zicht op te krijgen hebben we de kosten van de HARMpatiënten bestudeerd en beschreven in Hoofdstuk 2.4. We berekenden de kosten van de vermijdbare opnames door per opname, de directe medische kosten en de arbeidskosten van de patiënt bij elkaar op te tellen. Hierbij hielden we rekening met het type ziekenhuis (academisch of perifeer) en de leeftijd van de patiënt. Dit resulteerde in gemiddelde medische kosten voor één geneesmiddelgerelateerde ziekenhuisopname van 5461 Euro en gemiddelde arbeidskosten tijdens de opname, van een patiënt tot 65 jaar, van 1712 Euro. Gecombineerd komen de gemiddelde kosten voor één vermijdbare geneesmiddelgerelateerde ziekenhuisopname op ruim 193

194 6000 euro, ongeacht de leeftijd van de patiënt. Hieruit concluderen we dat de kosten van vermijdbare geneesmiddelgerelateerde ziekenhuisopnames substantieel zijn en er kosten te besparen zijn door deze opnames te voorkomen. Een medicatiebeoordeling, ook wel medicatiereview genoemd, wordt vaak als oplossing gezien om geneesmiddelgerelateerde ziekenhuisopnames te voorkomen en om de effectiviteit en veiligheid van de behandeling met geneesmiddelen te verbeteren. Echter, de meeste onderzoeken naar medicatiebeoordeling tonen geen of nauwelijks effect aan op ziekte of overleven. Daarom combineerden wij de medicatiebeoordeling met een farmaceutische anamnese, een farmacotherapeutisch behandelplan en het controleren, evalueren en opvolgen van dit behandelplan, tot het farmacotherapeutisch zorgproces (= pharmaceutical care process). Om het effect te meten van dit zorgproces op geneesmiddelgerelateerde ziekenhuisopnames, ontwikkelden wij het PHARM-onderzoek (Preventing Hospital Admissions by Reviewing Medication = preventie van ziekenhuisopnames door medicatiebeoordeling). Daarnaast onderzochten we ook het effect van het farmacotherapeutisch zorgproces op bijwerkingen van geneesmiddelen, op de kwaliteit van leven en op overleven. Het ontwerp van het onderzoek en de details van het farmacotherapeutisch zorgproces staan beschreven in Hoofdstuk 3.1. De resultaten van het PHARM-onderzoek staan beschreven in Hoofdstuk 3.2. Het PHARM-onderzoek is ontworpen als een gecontroleerd onderzoek met een interventie uitgevoerd door de huisarts en de apotheker in de eerstelijnsgezondheidszorg. Alleen patiënten die een hoog risico hadden om te worden opgenomen in het ziekenhuis door een geneesmiddel, werden gevraagd om mee te werken. Deze risicofactoren werden gebaseerd op de resultaten van het HARM-onderzoek en resulteerden in vier inclusie-criteria. Patiënten moesten voldoen aan alle criteria: 65 jaar of ouder, gebruik van vijf of meer geneesmiddelen, bepaalde geneesmiddelen én een hoger of lager verbruik van geneesmiddelen dan voorgeschreven door hun arts. De interventie, het farmacotherapeutisch zorgproces, was een continue proces dat bestond uit verschillende contacten tussen de patiënt en de eigen apotheker en de eigen huisarts. Na twaalf maanden werden per patiënt gegevens verzameld over geneesmiddelgerelateerde ziekenhuisopnames, bijwerkingen van geneesmiddelen, kwaliteit van leven aan het begin en aan het einde van het proces en eventueel overlijden. Deze gegevens werden ook verzameld van de controlepatiënten die de gebruikelijke zorg kregen van hun huisarts en hun apotheker. We vergeleken deze 194

195 Samenvatting Chapter 5.2 gegevens van de interventie groep met die van de controle groep met behulp van een statisch model: Mixed-effects Cox-model. Het plan was om in totaal patiënten te betrekken in het onderzoek: 7100 controlepatiënten en 7100 interventiepatiënten uit 142 apotheken. Maar uiteindelijk hadden wij maar gegevens van 364 interventiepatiënten en 310 controlepatiënten uit 42 apotheken om te vergelijken. In de interventiegroep waren minder patiënten met een geneesmiddelgerelateerde ziekenhuisopname dan in de controle groep; zes tegenover 10. Het farmacotherapeutische zorgproces maakte het risico op een opname voor patiënten kleiner. Als een patiënt drie aandoeningen had, werd dit risico 1,3 maal kleiner. Voor een patiënt met vier aandoeningen vonden we dat het risico 2,3 keer kleiner werd, met vijf aandoeningen 3,6 keer kleiner, met zes aandoeningen 5,3 keer kleiner, met zeven aandoeningen 7,1 keer kleiner en voor acht aandoeningen 9,1 keer kleiner. Vanaf vijf aandoeningen wordt dit effect statistisch significant. Tussen de interventie- en de controlegroep vonden geen statistisch significant effect op bijwerkingen van geneesmiddelen, kwaliteit van leven en overleven. In Hoofdstuk 4 bespreken we enige methodologische aspecten, de gevolgen van het HARM-onderzoek en wat we kunnen leren van het PHARM-onderzoek. We bediscussiëren de moeilijkheden in het beoordelen van de relatei tussen het geneesmiddelgebruik en de ziekenhuisopname en het verband tussen de medicatiefout en de ziekenhuisopname. Een probleem waar een patiënt mee in het ziekenhuis komt kan veroorzaakt worden door een geneesmiddel maar dit probleem kan ook ontstaan zonder het geneesmiddel. De relatie tussen het geneesmiddel en het probleem is daarom niet altijd helemaal zeker. Deze onzekerheid wordt bediscussieerd in Hoofdstuk 4. Ook bediscussiëren we in dit hoofdstuk het ontwerp van het PHARM-onderzoek en de moeilijkheid bij het vastleggen van de exacte interventie, het farmacotherapeutische zorgproces. Het HARM-onderzoek heeft al verschillende gevolgen gehad, binnen de maatschappij en de gezondheidszorg. De resultaten van het onderzoek hebben geleid tot verschillende projecten om de behandeling met geneesmiddelen te verbeteren en om de gevolgen van de geneesmiddelgerelateerde ziekenhuisopnames te verminderen. Het PHARM-onderzoek was één van die projecten. Naast het gemeten effect van het farmacotherapeutisch zorgproces was ook het geringe aantal patiënten dat meewerkte in het onderzoek opmerkelijk. Ondanks alle inspanningen van de onderzoekers, apothekers en huisartsen bleek het multidisciplinaire farmacotherapeutische zorgproces moeilijk te implementeren. Redenen die wij hiervoor ervaren waren de moeizame samenwerking tussen de verschillende 195

196 zorgverleners, het slecht aansluiten van het proces bij de huidige werkprocessen in de apotheek en de huisartsenpraktijk en daarmee de tijdsinvestering, en het gebrek aan kennis en kunde. Tenslotte doen we enkele aanbevelingen voor implementatie van het farmacotherapeutisch zorgproces en voor toekomstig onderzoek naar het voorkomen van geneesmiddelgerelateerde ziekenhuisopnames. 196

197 Promoveren doe je niet alleen! Alle mensen die ik ben tegengekomen in de afgelopen jaren maakten het promotietraject interessant. In het HARM-onderzoek onderzochten we uiteindelijk in 21 ziekenhuizen hoeveel mensen er werden opgenomen door bijwerkingen van geneesmiddelen. In het PHARM-onderzoek wilden we deze geneesmiddel gerelateerde ziekenhuisopnames voorkomen. Het vinden van deze opnames, de zogenaamde HARMs, het ontwikkelen van een methode en de uitvoering van de uiteindelijke interventie lukte met veel inspirerende mensen in mijn omgeving. Dankbaar ben ik voor al die hulp en inspiratie. Een aantal personen wil ik hier graag met naam noemen. In de eerste plaats bedank ik mijn promotie team van promotoren en copromotoren voor de directe begeleiding. In het begin werd ik begeleid door Toine Egberts en Patricia van den Bemt. Later werd het team versterkt met Han de Gier, Fred de Koning en Lex Goudswaard. Toine Egberts bedank ik voor het bieden van de mogelijkheid om onderzoek te doen, voor het blijven geloven in het succes van de mission impossibles, zoals hij het in presentaties noemde, en voor het out of the box -denken en zijn vernieuwende ideeën. Patricia van den Bemt voor haar meedenken, meewerken en haar altijd snelle en concrete respons op al mijn vragen en stukken. Patricia, dank voor jouw enorme inspanningen bij het samen beoordelen van alle HARM-dossiers. Ik heb veel van je geleerd! Han de Gier voor zijn altijd opbouwende opmerkingen en het stellen van de goede vragen die mij aan het denken zetten. Dit leverde weer nieuwe inzichten op. Fred de Koning voor zijn bijstand in alles, voor zijn enorme betrokkenheid en gedrevenheid en voor zijn geloof in de apotheker als zorgverlener. Lex Goudswaard voor het doktersperspectief, zijn prikkelende stellingen en vragen omtrent de farmacie, voor de ervaring die ik mocht opdoen in zijn praktijk en voor zijn aandacht voor de promovenda zelf. 197

198 De leden van de beoordelingscommissie: Cor Kalkman, Niek de Wit, Chiel Hekster, Marcel Bouvy en Steve Hudson, wil ik bedanken voor het beoordelen van het manuscript en voor hun complimenten, commentaar en suggesties. Cor Kalkman wil ik tevens bedanken voor de plek die hij mij bood in de stimulerende omgeving van het kenniscentrum patiëntveiligheid van het UMC Utrecht. Chiel wil ik tevens bedanken voor zijn aandeel in mijn clinical pharmacy training en onze gedachtewisselingen over de toekomst van een clinical pharmacist in Nederland. Mijn interesse voor het onderzoek werd gewekt door Marcel Bouvy en Henk Buurma met hun inspirerende SIR Masterclass. De enthousiaste reactie van Henk bij mijn sollicitatie op de Universiteit Utrecht en later de stelligheid waarmee hij verkondigde dat Anne Leendertse moest promoveren gaven mij vertrouwen om verder te gaan. I would have never started this research project if it wasn t for Ruud Dessing who recommended the MSc Clinical Pharmacy course at Strathclyde University and who introduced me to Steve Hudson. Steve Hudson continued the education that Han de Gier started in clinical pharmacy and pharmaceutical care, in Scotland. With his nasty questions he taught me a patient centred way of thinking and acting. Thank you Steve for your support during the MSc, the PhD and for our lively discussions. De NVZA wil ik bedanken voor het initiatief, de mogelijkheid en de ondersteuning bij het HARM-onderzoek en PHARM-onderzoek. Martin Schuitenmaker voor zijn inbreng bij de financiering van het HARM-onderzoek en voor de Noord-Hollandse gezelligheid. Anne de Roos voor haar ondersteuning bij de website en het NVZAkennisplein. Eric van Roon voor het bedenken van het briljante acroniem: HARM. Lennart Stoker voor het samen ontdekken hoe het HARM-onderzoek uitgevoerd kon worden in het Diakonessenhuis te Utrecht. Zonder jouw motiverende en prettige samenwerking was het niet zo n succes geworden! Alle HARM-onderzoekers: Dana Appelo, Mireille Toet, Zina Brkic, Francine Prak, Liselotte Soeting, Christine Evertse, Djoek Vogel, Martijn Brummer, Nielka van Erp, Jeroen Diepstraten, Ebby Ruiz, Jeroen Doodeman, Christien Schmitz, Gert de Marie, Jan Maarten Langbroek, Danielle de Keizer, Milly Attema, Vincent Tan, Bram van Arkel, Rixt Nynke Eggink, Koen Oosterhuis, Moniek Boekweit, Marrit van Buuren, Maartje Wijnhoven, Marjan Bouma, Karen van Loon, Mariet Heins, en Yvonne Scheffer, bedank ik voor hun doorzettingsvermogen bij het speuren naar de HARMs, het verzamelen en verwerken van alle gegevens en hun collegialiteit. Ik vond het een genoegen om met een ieder van jullie samen te werken in jullie 198

199 Dank woord Chapter 5.3 ziekenhuizen. De ziekenhuisapothekers: Marjo Janssen, Paul van der Linden, Pieter Knoester, Paul le Brun, Hans Harting, Frank Jansman, Rob Moss, Irene Twiss, Sjoukje Troost, Paul Kloeg, Ingrid van Haelst, Hans Overdiek, Eric van Roon, Eric Franssen, Erik Meijer, Ed Wiltink, Patricia van den Bemt, Marian Laseur, Piet Nauta, Liesbeth van Dijk, Elsbeth Helfrich en Martin Schuitenmaker, voor hun inzet, betrokkenheid, begeleiding van de onderzoekers en gastvrijheid in hun ziekenhuis. Het HARM-onderzoek had niets opgeleverd zonder de medewerking van alle dokters, verpleegkundigen, ziekenhuismedewerkers, openbare apothekers, huisartsen en patiënten in de deelnemende ziekenhuizen. Dank ook voor jullie medewerking. Youssef Chahid wil ik bedanken voor zijn hulp bij het valideren van de data van het HARM-onderzoek. Patrick Souverein bij het managen van de database van het HARM-onderzoek. De KNMP wil ik bedanken voor hun uitstekende werk in het communiceren van de HARM-onderzoeksresultaten en Peter de Smet voor het tot stand komen van de concrete aanbevelingen in het HARM-wrestling rapport. Liesbeth van Dijk wil ik bedanken voor de fijne samenwerking bij het beoordelen van de relatie tussen nierfunctie en HARM en medicatiefouten. Jouw enthousiasme maakte het papierwerk levendig. Maarten Postma en Bart Poolman wil ik bedanken voor hun hulp bij het in kaart brengen van de kosten van de HARMs in Nederland. Bart voor het rigoureuze speurwerk en Maarten voor de ondersteuning. Kring Apotheek wil ik bedanken voor de ondersteuning bij het werven van de apothekers voor het PHARM-onderzoek. In Den Bosch bij Kring ontmoette ik ook vele enthousiaste collega s die actief meedachten in het omgaan met alle barrières die ik tegenkwam bij het implementeren. Met name wil ik noemen: Arjen Geerts, Maarten Voesten, Thijs Vinks en Jeroen Derijks als mede onderzoekers voor het delen van onze onderzoeksproblemen en successen; Pim Poels, Maayke Fluitman, Marja Coelewij, Peter de Jong, Roel Majoor, en Marinke Vegter voor het promoten van de MC en de collegiale gezelligheid en Eric Nieboer voor het leggen van contact met Achmea en voor het parachute springen. Alle onderzoekers van de NVZA CAREFUL groep uit het LUMC, UMCG en AMC wil ik bedanken voor het delen van onze onderzoekservaringen. 199

200 Het PW wil ik bedanken voor het bieden van een podium om apothekers en huisartsen te enthousiasmeren om HARMs te gaan voorkomen en te participeren in het PHARM-onderzoek. Het NHG wil ik bedanken voor hun goedkeuring en vertrouwen in het PHARMonderzoek. Dit heeft geholpen bij het enthousiasmeren van de huisartsen. Mijn oom Haiko Jonkhoff voor het filosoferen over het niet kunnen starten van een pilot voor het PHARM-onderzoek en voor zijn tips voor het mogelijk wel slagen van een interventie door huisarts en apotheker. Andries Jonkhoff voor zijn cruciale rol in het uitvoeren van een pilot voor het PHARM-onderzoek. Door je groeiend enthousiasme en je praktische inbreng was het mogelijk om het pharmaceutical care proces, wat in mijn hoofd leefde, uit te voeren in de huisartsenpraktijk. Zonder jouw hulp was dit niet gelukt! Ik ben blij dat je mij ook als paranimf wilt ondersteunen. Mira Holt voor het verzamelen van de gegevens en de achtergrond informatie van jullie patiënten die ik mocht bezoeken. Frank van Wees voor zijn gastvrijheid, vertrouwen en zijn ondersteuning in de farmacotherapeutische behandeling van de proefpatiënten. Marjolein Geleedst-de Vooght en Matthijs Engering wil ik bedanken voor het meewerken en meedenken bij de pilot voor het PHARM-onderzoek. Henny Otten voor haar enthousiasme, haar praktische invulling, het kritisch meedenken bij de interventie en voor het inzicht in het prachtige vak van praktijkondersteuner. Theo Bollerman voor zijn zeer bruikbare adviezen in de communicatie met huisartsen, apothekers en patiënten, en voor de inspirerende samenwerking bij het ontwikkelen van de communicatie cursus voor apothekers. Tijl Koenderink en Berry Kriesels voor het ontwerpen van IT-applicaties waarin we gegevens verzamelden voor het HARM- en het PHARM-onderzoek. Ook ben ik jullie dankbaar voor de kennismaking met de wereld van hoogbegaafdheid die daardoor minder eng is geworden. Ik bedank alle apothekers en huisartsen die meewerkten aan het PHARMonderzoek, ook diegene die het niet is gelukt om patiënten aan te leveren. Allen 200

201 Dank woord Chapter 5.3 hebben een belangrijke bijdrage geleverd. Met name wil ik noemen: Aafke Donker, Annemarie Leuverman, Anniek Ligtenberg, Barend Dam, Bas Klok, Bert Kwakernaak, Bert Storm, Birgit van Tuijl, Caroline van der Steeg-van Gompel, Cecilia Kerkhof-Demjén, Claudia Blokland, Cobie Otten-Hoogeboom, Cor van Otterloo, Cora Jonker, Debby Steinmeijer, Elles Sanders-Bruininx, Elvire ten Berge-de Ridder, Emiel van der Pijl, Erik Evers, Erna van den Broek, Eward Melis, Frank Schroor, Helma Wolters, Henk-Frans Kwint, Hugo de Wit, Inge Coehorst, Jacqueline Maas, Jan Bernaards, Jan Oomen, Jeanine Latour, Joke Haasnoot, Judie Hulsbosch, Karin Grote, Mandy Pelzer, Marc van Asten, Marianne van den Berg, Marielle Verboom, Marjolein Geleedst-de Vooght, Marjolein Laterveer, Martine Ruitenburg, Dorine Hoogdalem, Nanda Croes, Paul Cornips, Maud Verhulst, Gudy Meijvis, Peter Chen, Piet Ooms, Renate v Uffelen, Ricky Smith, Rikie Elling, Serdar Araz, Shafak Firdausi-Ikram, Tara van der Linden en Tom van Daal. En natuurlijk de patiënten die wilden participeren in het onderzoek. Achmea en Zorg en Zekerheid wil ik bedanken voor de financiële ondersteuning van de apothekers en huisartsen. Clementine Stuijt en Adrianne Faber voor het meedenken in de praktische invulling van de interventie en voor het inbedden van het PHARM-onderzoek in de PIAFcursus. De collega s bij FenF wil ik bedanken voor hun interesse en het thuisgevoel in het Went. Met name wil ik bedanken: Svetlana Belitser, mijn fijne kamergenoot, voor alle statistische ondersteuning. Addy, Ineke, Suzanne en Marije voor het beantwoorden van bijna al mijn vragen, de ondersteuning en gezelligheid. Willem Rekvelt voor de IT ondersteuning en voor eerste hulp vervangende machines. Ingeborg, Maarten en later Helga voor onze Toine-AIO-lunches. Worden het dan nu de niet-meer-toine-aio-familie-diners? Annemieke Floor, Henk-Frans Kwint, Arjen Geerts en Marlies Geurts voor het delen van hun onderzoekservaring en resultaten. Annemieke wil ik speciaal bedanken voor de vruchtbare en prettige samenwerking in het Medicijncheck-project met Zorg en Zekerheid. Mijn collega s bij het kenniscentrum patiëntveiligheid (KPV) in het UMC Utrecht voor de inspirerende omgeving. Ik had geen betere werkplek kunnen bedenken! 201

202 In het bijzonder wil ik Marcella Brun bedanken voor haar geweldige organisatie van het PHARM-onderzoek. Jij nam een enorme last van mijn schouders en regelde perfect de cursussen, bezoeken door het hele land, de administratie en stuurde de apothekers aan. Ik kijk terug op een goede tijd samen. Ian Leistikov wil ik bedanken voor zijn gastvrijheid en voor het inzicht dat hij mij heeft gegeven in het analyseren van fouten en het sturing geven aan patiëntveiligheid. Astrid Koffrie voor haar hulp en regelwerk. Astrid zo fijn dat ik altijd op je kon rekenen. Dorien Zwart door wie ik de dokters beter ging begrijpen en voor de inspirerende gesprekken en Maartje Swennen voor de goede gesprekken, kennis en inzicht bij de implementatie van evidence based medicine. Mijn andere kamergenoten: Petra Gademan, Feikje Stiphout, Maurice Pouw, Emily Thieme Groen, Chantal de Bree, Karine Groen voor onze brainstormsessies, het uitwisselen van goed en slecht nieuws en het delen van onderzoekservaringen. Yvonne van der Tuijn, Liesbeth van Rensen, Kiek Tates, Jantien de Loor, Loes Pijnenborg, Sandra Numan, Karien den Ridder, Bas de Vries voor hun collegialiteit en een luisterend oor. Paul Barach for his inspiring questions. De collega s bij Mediportaal voor de ondersteuning en gezelligheid. Alle studenten, jonge dokters, bezoekers en KPV-lunch bezoekers die er een inspirerende omgeving van maakten. I would like to thank Pam Kato for our philosophy sessions on research and teaching, for your help with the English writing and for your trust and friendship. De ziekenhuisapothekers van het UMC Utrecht wil ik bedanken voor het contact met de ziekenhuisfarmacie tijdens de interessante bijeenkomsten. Nellie en Jolanda voor jullie geweldige ondersteuning, antwoord op mijn vragen, de band met de apotheek en de goede grappen. Ruud Weeda voor de IT ondersteuning overal en nergens. De studenten die mij hielpen met allerlei klussen en onderzoeken en die het vooral heel gezellig maakten: Marijn Verhoef voor het samen coderen van alle behandelplannen, Valérie Meijvis voor het ophalen en invoeren, Djurre Visser voor het verzamelen van de literatuur, Pieter van Iren voor zijn frisse blik en invoerwerk in de PHARM-database en met Sander van den Bogert als onze Access koning! 202

203 Dank woord Chapter 5.3 Vivian de Gier voor het verwoorden van wat ik de afgelopen jaren heb gedaan, zodat het ook voor de buurt is te begrijpen. Francis te Nijenhuis voor een schitterend boekje ondanks alle last-minute stress. Mijn nieuwe collega s bij de SIR die hielpen met het relativeren van de promotie stress in die moeilijke eindfase. Mijn broeders en zusters van het Apostolisch Genootschap voor de inspiratie, voor de motivatie en voor het geloof in mij. Mijn jaarclubgenoten, familie, vrienden en buurtgenoten die interesse bleven tonen, afleiding gaven en hielpen relativeren. Mijn broers Jaap en Roel door er gewoon altijd te zijn als het nodig was. Mijn ouders Jet en Matthijs voor hun altijd onvoorwaardelijke liefde en steun, bij al mijn plannen. Lennart voor de afleiding, mooie lego bouwsels en kritische vragen over apothekers, pillen en ziekte. En Ulrich, voor alles. Zonder jou had het niet gekund en had ik het niet gekund. Ik wil je bedanken voor al je steun, je zorgen en je liefde. En natuurlijk ook voor zo n mooie voorkant en voor je ondersteuning als paranimf. 203

204

205 Aff iliations during the conductance of the research Svetlana V B elitser Division of Pharmacoepidemiology & Clinical Pharmacology of the Utrecht Institute for Pharmaceutical Sciences (UIPS), Faculty of Science, Utrecht University, Utrecht, the Netherlands Patricia ML A van den B emt Division of Pharmacoepidemiology & Clinical Pharmacology of the Utrecht Institute for Pharmaceutical Sciences (UIPS), Faculty of Science, Utrecht University, Utrecht, the Netherlands Department of Hospital Pharmacy, Erasmus University Medical Center, Rotterdam, the Netherlands Sander CA van den B oger t Division of Pharmacoepidemiology & Clinical Pharmacology of the Utrecht Institute for Pharmaceutical Sciences (UIPS), Faculty of Science, Utrecht University, Utrecht, the Netherlands L iesbeth A van Dijk Department of Clinical Pharmacy, VieCuri Medical Center, Venlo, the Netherlands Toine CG Egber ts Department of Clinical Pharmacy of the University Medical Center Utrecht, Utrecht, the Netherlands Division of Pharmacoepidemiology & Clinical Pharmacology of the Utrecht Institute for Pharmaceutical Sciences (UIPS), Faculty of Science, Utrecht University, Utrecht, the Netherlands 205

206 Johan J de Gier Department of Pharmacotherapy and Pharmaceutical Care, University of Groningen, Groningen, the Netherlands Alex N G oudswaard Dutch College of General Practitioners (NHG), Utrecht, the Netherlands GP Practice Wernaar, Houten, the Netherlands Andries R Jonkhoff Jonkhoff huisartsen, GP Practice Koningshoed, Haarlem, the Netherlands Fred HP de Koning Kring Pharmacies, s-hertogenbosch, the Netherlands Division of Pharmacoepidemiology & Clinical Pharmacology of the Utrecht Institute for Pharmaceutical Sciences (UIPS), Faculty of Science, Utrecht University, Utrecht, the Netherlands Bar t JB Poolman Unit of PharmacoEpidemiology & PharmacoEconomics, Departement of Pharmacy, University of Groningen, Groningen, the Netherlands Maar ten J Postma Unit of PharmacoEpidemiology & PharmacoEconomics, Departement of Pharmacy, University of Groningen, Groningen, the Netherlands Peter AGM de Smet Department of Clinical Pharmacy, Radboud University Nijmegen Medical Centre, Nijmegen, the Netherlands The Royal Dutch Association for the Advancement of Pharmacy (Koninklijke Nederlandse Maatschappij ter bevordering der Pharmacie, KNMP), Den Haag, the Netherlands L ennar t J Stoker Department of Clinical Pharmacy of the Diakonessenhuis Utrecht-Zeist-Doorn, Utrecht, the Netherlands 206

207 List of co -authors Chapter 5.4 Marijn Verhoef Division of Pharmacoepidemiology & Clinical Pharmacology of the Utrecht Institute for Pharmaceutical Sciences (UIPS), Faculty of Science, Utrecht University, Utrecht, the Netherlands Djurre Visser Division of Pharmacoepidemiology & Clinical Pharmacology of the Utrecht Institute for Pharmaceutical Sciences (UIPS), Faculty of Science, Utrecht University, Utrecht, the Netherlands 207

208

209 Publications related to the thesis Stoker LJ, Leendertse AJ, Prak F, Rynja FJ, van den Bemt PMLA. Een vermijdbaar probleem? Pilot-studie naar incidentie en determinanten van geneesmiddelgerelateerde ziekenhuisopnames. Pharm Weekbl. 2006;37: Leendertse AJ, van Dijk EA, van der Velde RY. Boedglucose eerder meten. 89-jarige patiënt krijgt hyperglykemie bij prednisolon. Pharm Weekbl. 2006;43: Van den Bemt PMLA, Leendertse AJ, Egberts ACG.Het HARM-onderzoek: een observationeel multicenteronderzoek naar geneesmiddelgerelateerde ziekenhuisopnames. Bijblijven. 2007;23: Leendertse AJ, Egberts ACG, Stoker LJ, van den Bemt PM; HARM Study Group. Frequency of and risk factors for preventable medication-related hospital admissions in the Netherlands. Arch Intern Med. 2008;168: Leendertse AJ, Visser D, Egberts ACG, van den Bemt PMLA. The relationship between study characteristics and the prevalence of medicationrelated hospitalizations: a literature review and novel analysis. Drug Saf. 2010;33:

210 Leendertse AJ, van den Bemt PMLA, Poolman JB, Stoker LJ, Egberts ACG, Postma MJ. Preventable Hospital Admissions Related to Medication (HARM): Cost analysis of the HARM-study. Value in Health. (in press) 210

211 Anne Leendertse was born on 17 May 1971 in Amsterdam the Netherlands. She studied pharmacy at Utrecht University from 1990 to 1999 and registered as a community pharmacist in 2001 in Apotheek de Gors in Purmerend. She worked in the Netherlands as a community pharmacist till 2002 in Apotheek Zuiderhaaks, Den Helder, and also worked in this period as a teacher and on the development of pharmaceutical care at the Health Base Foundation in Houten. In 2007, she obtained the Master s of Science in Clinical Pharmacy, with distinction, at the University of Strathclyde, Glasgow, United Kingdom. During this period she was given the opportunity to obtain clinical experience as a clinical pharmacist in mainly Edinburgh Hospitals. In this period ( ) she also worked part-time for the franchise organisation Service Apotheek as a pharmaceutical care manager. Since 2005 she worked as a researcher in patient safety at the department of clinical pharmacy of the University Medical Center Utrecht and at the division of pharmacoepidemiology and clinical pharmacology, Utrecht Institute for Pharmaceutical Sciences, Utrecht University where she conducted the HARM-study and the PHARM-study. The results of this research are presented in this thesis. The author is a member of the General Committee of the European Society of Clinical Pharmacy and also plays an active role in several other committees. From 2010 onwards, she works at the SIR Institute for Pharmacy Practice and Policy. She lives in Utrecht with Ulrich and their son Lennart. 211