A comparison of educational interventions to improve prescribing by junior doctors

Transcription

1 Q J Med 2015; 108: doi: /qjmed/hcu213 Advance Access Publication 16 October 2014 A comparison of educational interventions to improve prescribing by junior doctors J.S. THOMAS 1, D. GILLARD 2, M. KHOR 2, P. HAKENDORF 3 and C.H. THOMPSON 4 From the 1 School of Medicine, University of Adelaide South Australia, 5000, 2 Pharmacy Department, Level 2 East Wing, Royal Adelaide Hospital, 50 North Tce, Adelaide, South Australia 5000, Australia, 3 Flinders Centre for Epidemiology and Biostatistics, Flinders University, Bedford Park, South Australia 5042 and 4 School of Medicine, University of Adelaide, South Australia 5000, Australia. Address correspondence to Dr J.S. Thomas, 1-4 EHB, Royal Adelaide Hospital, 50 North Tce, Adelaide, South Australia 5000, Australia. Josephine.thomas@health.sa.gov.au Received 24 July 2014 and in revised form 4 September 2014 Summary Background: Prescribing is a complex task with potential for many types of error to occur. Despite the introduction of a standard national medication chart for Australian hospital inpatients in 2006, simple prescribing errors are common. Aim: To compare the effect of quality improvement initiatives on the rate of simple prescribing errors. Design: A prospective, multisite comparison of prescribing education interventions. Methods: Using three hospital sites, we compared site-specific changes in prescribing error rates following use of an online education module alone (low intensity) with prescribing error rates following a high-intensity intervention (comprising the same online education module plus nurse education and academic detailing of junior prescribers). The study period was 4 months between May and August Results: Full completion of the adverse drug reactions field did not improve after either intervention; however, there was better documentation of some elements following high-intensity intervention. Prescriber performance improved significantly for more elements in the regular prescription category than any other category of prescription. Legibility of medication name improved across all categories following interventions. Clarity of frequency, prescriber name and documentation of indication improved following both high- and low-intensity intervention. Conclusions: Improvements were seen in several prescription elements after the intervention but the majority of elements that improved were affected by both low- and high-intensity interventions. Despite targeted intervention, significant rates of prescribing breaches persisted. The prevalence of prescription breaches partially responds to an online education module. The nature of any additional intervention that would be effective is unclear. Introduction Medication-related harm has many potential sources including errors or omissions in prescribing, dispensing or administration of drugs. Junior medical staff working in hospitals are likely to make prescribing errors 1,2 which usually relate to breaches/omissions in documentation, rather than more complex cognitive errors of drug choice or dosing calculation. These are termed simple prescribing errors. 3,4 These errors do not often translate to an actual adverse event however, they may be a preventable source of harm. Furthermore, errors that arise within the hospital setting can be perpetuated following discharge of the patient into the community with further potential for harm.! The Author Published by Oxford University Press on behalf of the Association of Physicians. All rights reserved. For Permissions, please journals.permissions@oup.com

2 370 J.S. Thomas et al. The training of medical staff in the skills necessary for safe prescribing starts in the medical school and should continue after graduation. The type of training required is not well established. Simple prescribing errors are common in hospitals in Australia and New Zealand and there is site-specific prevalence of particular errors. 5 This suggests a one-size-fits-all approach to training is inappropriate. The introduction of the National Inpatient Medication Chart (NIMC) in Australia in 2006 was an attempt to standardize hospital prescribing and allow better training of doctors in prescribing. 6,7 The tools available for training prescribers include online learning and face-to-face teaching. Online prescribing education has been shown to be effective in improving prescribing skills and confidence NPS MedicineWise has developed an online educational module to support prescribers in the use of the NIMC. 11 This online module has not been a mandatory part of junior medical staff training in the hospitals studied, and was not widely advertized. Academic detailing is an educational approach that aims to provide clinicians with unbiased, evidencebased, practical and non-commercial information on the comparative effectiveness of medical therapies to facilitate good clinical decision making. It involves a personal visit by a trained person to health professionals in their own setting 12, and has been shown to be effective in changing clinical behaviour and reducing prescribing error However, this approach has largely been used in community practice settings rather than targeting junior prescribers in hospitals. Medication orders should be legible, unambiguous and contain sufficient information to support the clinical intent of use. 16 If the Inpatient Medication Chart is completed correctly, nurses and pharmacists should be able to clearly understand each intended medication, dose, route, frequency and indication and determine the prescriber and their contact details. The patient s known history of adverse reactions should be clear, and there should be no ambiguity regarding the patient for whom the prescription is intended. The study aimed to compare interventions of different intensity on the rate of simple prescribing errors. Methods Three hospital sites in Australia participated in the study between May and August, 2011: Royal Perth Hospital, WA; Royal Adelaide Hospital SA; Flinders Medical Centre, SA. Each participating site chose the level of intervention. All sites utilized the NIMC for inpatient medications. All sites completed pre and post-chart sampling and their intervention during a single junior doctor rotation period. At both intervention sites all junior medical officers performing admissions in the acute medical unit (including night staff) participated in the interventions. Medical staff ranged between postgraduate years 2 and 5 and were unaware of the audit but were required to participate in the interventions as a quality improvement exercise. Interventions The low-intensity site used an online education module for prescribers as the intervention. The online education module details the correct completion of all fields of the NIMC and highlights potential harm from breaches in documentation. The high-intensity site employed a multidisciplinary approach to intervention, utilising the same online education module for prescribers plus nurse education and academic detailing of admitting junior doctors by specifically trained clinical pharmacists these were all clinical pharmacists (with at least 5 years experience) who attended an intensive 2-day workshop on academic detailing. The focus of the academic detailing was correct documentation of adverse drug reactions (ADRs); correct documentation of slow release (SR) medications and use of generic medication names in all prescriptions. These issues were also promoted on highly visible wall posters in the ward. The other elements of highintensity interventions were nursing in-services encouraging challenge of poor prescribing by junior doctors. The focus was on clarity and completeness of prescriptions. The third site employed no intervention (Table 1). Data collection Clerical staff photocopied the medication charts of patients within 24 h of admission. This ensured minimal impact of consultant input and allowed us to ensure prescribing sampled was by those junior doctors participating in the interventions. We aimed to collect 100 consecutive Acute Medical Unit admissions at each site. Patient medication charts were excluded from the study if: no medications were ordered on NIMC; patients were aged less than 16 years; or medication orders were completed by a doctor not included in the intervention. A further period of chart collection immediately followed the relevant interventions at each site, again aiming to collect 100 consecutive patient admissions. Two pharmacists examined all photocopied charts, with data collected according to predetermined criteria. These pharmacists were not blinded

3 A comparison of educational interventions 371 Table 1 Summary of interventions at each site Low-intensity site High-intensity site Control site Completion of the online education module by junior medical staff Completion of the online education module by junior medical staff Academic detailing of junior medical staff by pharmacists Nursing staff education with training in challenge by questioning techniques Promotional material No intervention Targets correct completion of all fields Targets correct completion of all fields Targeted SR orders, ADR documentation, use of generic name Targeted clarity and completeness of prescriptions Targeted SR meds, ADR documentation, use of generic name Table 2 Elements specific to each prescription category were assessed Regular medication orders As required (prn) orders Once only (stat) orders Prescriber name was printed Prescriber contact number was present Clarity of medication dose Medication frequency was clear No inappropriate duplication of orders Indication was documented Prescriber name was legible Administration time was charted Administration time correlated with frequency Correct documentation of slow release preparations Prescriber name was printed Prescriber contact number was present Clarity of medication dose Medication frequency was clear No inappropriate duplication of orders Indication was documented Prescriber name was legible Maximum dose was charted Administration date/time charted to the intervention and were not involved in the academic detailing. Patient characteristics and general documentation on chart, including: age and gender; time of admission (whether during usual working hours); was patient Identification (ID) label present; was patient name printed below ID label; was height and weight recorded. ADR documentation: was any information documented in the ADR field; were generic names of drugs used; were details of the ADR provided; did a clinician sign the entry. The NIMC has several categories of prescription for medication orders including three major categories: regular, as required (prn) and once only (stat). These three different categories differ slightly in the fields available for completion but share common elements (Table 2). For each medication order, the following elements (common to all categories) were assessed 6 : was the date of prescription entered; was the order signed; was the generic medication name used; was the medication name legible; was the medication spelt correctly; was the route of administration clear; were any error-prone abbreviations used. Statistical analysis We focused our analysis upon elements of the prescription where performance was poor before the intervention. Good baseline performance was deemed to have occurred for any individual element when the completion rate for that element was adequate or correct in 95% or more of prescriptions. The effect on the individual elements at each site was compared before and after intervention using two-sided Fisher s exact test. Poisson regression analysis was used to compare the relative change at intervention sites with the control site for each element studied. P values < 0.05 were taken as significant. Power calculations were not performed prior to the study due to the large number of individual

4 372 J.S. Thomas et al. Table 3 Patient and medication chart characteristics Low-intensity intervention site High-intensity intervention site Control site Patients total no. (before/after 214 (118/96) 139 (77/62) 196 (83/113) intervention) Mean age in years (SD) 71.6 (19.2) 69.8 (16.9) 67.4 (17.3) Gender (% female) 56% 46% 45% Regular orders (before/after 1407 (824/583) 661 (338/323) 963 (357/606) intervention) As required (prn) orders (before/after 232 (122/110) 155 (89/66) 303 (118/185) intervention) Stat orders (before/after intervention) 511 (262/249) 418 (223/195) 599 (371/228) elements examined. Breach prevalence differed between elements at any one site and for the same element between sites. It was accepted that some elements might not reach a sufficient number of prescriptions to show a statistically significant change. Results Patient characteristics Details of patients and charts analysed at each site are listed in Table 3. Mean age was similar at each site. The medication charts of 29 patients at the high-intensity site were excluded from analysis because a junior doctor who was not included in the intervention had completed them. No patient charts were excluded at other sites. General chart and ADR documentation The rates of general and ADR documentation at each site before and after intervention are summarized in Table 4. There was improvement in the use of patient ID labels at both low- and high-intensity sites after intervention, although only statistically significant at the low-intensity site, and no significant change at the control site. The baseline rate of compliance with this element was much lower at the low-intensity site compared with other sites. Prescribers rarely completed height and weight fields, and infrequently printed the patient name below the ID label. There was no significant change in completion of height, weight or patient name elements at any site after intervention. Although some form of ADR documentation occurred in the majority of charts (between 81% and 92%), full completion of all elements of the ADR field on the chart was performed poorly at baseline at all sites (52 61% of charts) and did not improve after either low- or high-intensity intervention. The high-intensity intervention significantly improved two elements of ADR documentation ( nature of the adverse reaction documented and ADR field signed by prescriber ) when compared with control site. There was a trend to improvement for these elements following the low-intensity intervention. Prescription elements The outcomes for all prescription elements, across categories of charts (regular, stat and prn) are summarized in Table 5. In all these categories of prescriptions, there were certain elements with good prescriber compliance at baseline at all sites, with no significant change following intervention. Many of these elements were common to all categories of prescription. For some elements, prescribers performance differed in the one site across these three different categories of prescription. Following either intervention, performance significantly improved for more elements in the regular prescriptions category than it did in either of the other two categories. However, legibility of medication name improved across all three categories of prescription after low- and high-intensity intervention; and when compared with performance changes at the control site (Tables 5 and 6). Low- vs. high-intensity intervention For the majority of elements that showed improvement, both low- and high-intensity interventions seemed equally efficacious (Table 5). Many elements were unaffected by either intervention, even those specifically targeted by the academic detailing component of the high-intensity intervention, and elements where there was a poor baseline performance by prescribers. Poisson analysis of elements that showed improvement at intervention sites is also represented in Table 6. This allows the comparison of these relative improvements with contemporaneous changes at the control site. At both lowand high-intensity sites, a significant improvement relative to control site was seen in the prescriber

5 A comparison of educational interventions 373 Table 4 General and ADR field documentation on chart Before intervention % (n/total no charts) After intervention %(n/total no. charts) Relative risk compared with control site (CI) ID label was present on chart Height was recorded on chart Weight was recorded on chart Patient name was printed below label ADR drug was documented by generic name ADR field was dated by prescriber Any form of ADR documentation was present Nature of the adverse reaction was documented ADR field was signed by the prescriber ADR field documentation was complete Low intensity 62% (73/118) 82% (79/96)* 1.28 ( )** High intensity 84% (65/77) 94% (58/62) 1.07 ( ) Control site 84% (70/83) 88% (99/113) 1 Low intensity 2% (2/118) 1% (1/96) Not applicable*** High intensity 1% (1/77) 0% (0/62) 0.93 ( ) Control site 1% (1/83) 0% (0/113) 1 Low intensity 2% (2/118) 1% (1/96) Not applicable*** High intensity 1% (1/77) 3% (2/62) Not applicable*** Control site 8% (7/83) 7% (8/113) 1 Low intensity 3% (3/118) 2% (2/96) 0.98 ( ) High intensity 4% (3/77) 8% (5/62) 2.47 ( ) Control site 8% (7/83) 7% (8/113) 1 Low intensity 88% (104/118) 90% (86/96) 0.98 ( ) High intensity 94% (72/77) 94% (58/62) 0.97 ( ) Control site 94% (78/83) 97% (110/113) 1 Low intensity 67% (79/118) 74% (71/96) 1.31 ( )** High intensity 70% (54/77) 71% (44/62) 1.20 ( ) Control site 82% (68/83) 69% (78/113) 1 Low intensity 92% (108/118) 94% (90/96) 2.68 ( ) High intensity 88% (68/77) 98% (61/62)* 0.50 ( ) Control site 81% (67/83) 95% (107/113)* 1 Low intensity 73% (86/118) 71% (68/96) 1.10 ( ) High intensity 77% (59/77) 85% (53/62) 1.27 ( )** Control site 84% (70/83) 74% (84/113) 1 Low intensity 79% (93/118) 89% (85/96) 1.21 ( )** High intensity 83% (64/77) 94% (58/62) 1.22 ( )** Control site 83% (69/83) 77% (87/113) 1 Low intensity 52% (61/118) 50% (48/96) 1.07 ( ) High intensity 48% (37/77) 50% (31/62) 1.15 ( ) Control site 61% (51/83) 56% (63/113) 1 *P < sided Fisher s exact test. **P < 0.05 (Poisson regression analysis, 2 ). ***Numbers too small for analysis. name being printed on that part of the chart reserved for regular prescriptions. The most dramatic relative change was in documentation of the medication indication on the regular prescription section of the chart. For legibility of medication name, significant changes were noted for all categories of prescriptions at low- and high-intensity sites. However, only the regular chart showed a significant change at both intervention sites relative to the control site (Table 6). Discussion There were several elements, across different categories of prescription, which indicated good prescriber performance at baseline and this was sustained following intervention. These elements included: date, signature, correct spelling of medication, clarity of route, no inappropriate duplication. This is not surprising and these elements reflect prescription tasks that may make intuitive sense to the prescriber or may reflect good chart design. These elements are consistent across different categories of prescriptions despite these differing sections of the chart not being identical in their layout/design of fields. These aspects do not merit further input or educational effort. For some elements, where baseline performance was poor, a significant improvement occurred relative to the site where no intervention occurred. Generally, when improvement was seen, it occurred irrespective of intervention intensity. This suggests that either intervention would reduce error rates

6 374 J.S. Thomas et al. Table 5 Outcomes for prescription elements according to type of intervention Prescription element Regular PRN stat No inappropriate duplication of medication A A A Correct spelling of medication A A A Clarity of route of administration A A A Date of prescription was recorded A A A Order was signed A A B* Medication name was legible B*, C* C B, C No error prone abbreviations used B D D Clarity of medication dose D C D Medication was ordered with generic name D D B Medication frequency was clear B, C* B Prescriber name was printed B*, C* D Prescriber contact number was documented B, C D Indication was documented B*, C* D Prescriber name was legible A D Correct documentation of SR orders D Administration time correlates with frequency A Administration time was charted D Maximum dose was charted D Date and time of dose was recorded D A, Good performance at baseline (>95% correct), no change with intervention. B, Statistically significant improvement at low-intensity site (two-sided Fishers exact test P < 0.05). C, Statistically significant improvement at high-intensity site (twosided Fishers exact test P < 0.05). D, No significant effect at intervention sites. Shading indicates elements targeted by academic detailing. Statistically significant change relative to control site (Poisson analysis, 2 *P < 0.05). despite large differences in implementation costs. Low-intensity interventions are worthwhile because they are easy to implement with minimal cost and have a beneficial effect on error rates. The cost benefit of more intensive initiatives requires further examination. Changing the design of the medication chart itself in order to reduce duplication and to align the chart with the workflow of the prescriber might reduce the prevalence of simple prescription errors. There were more significant improvements in prescribers performance when charting regular medications than in the other categories of prescription. Regular orders are a more frequently used category, comprising over 50% of orders at each site (Table 3), thus greater numbers of orders may improve ability to demonstrate improvement. Additionally, chart design may be relevant. Differing error rates for the same element across the three prescribing categories may reflect the circumstances in which these three different categories of prescription are written. For example, stat orders are often documented in more urgent clinical settings (in response to crisis) and may be a new therapy whereas a regular medication, often a longstanding prescription, is usually transcribed in a less time-critical situation. Legibility of medication name was the only element that improved for all three categories of prescription following intervention. Workflow can be defined as the organization of tasks for the clinician in a time-efficient manner. Workflow may be relevant when considering prescriptions on the prn section of the chart as distinct to those on the regular section of the chart. The indications for medications are often not known at the time of admission to hospital. 17 At baseline, prescribers rarely documented the indication for a medication in either the regular or the prn chart; this poor compliance has been documented previously 18 and might be explicable from a workflow perspective. Prescribing performance in this element improved in the regular medication category after intervention, but performance was unchanged by the intervention when charting prn medications. This improvement in compliance for the regular medication category and not the prn category suggests other factors such as chart design may be culpable. There are multiple fields in the NIMC that are duplications or are difficult to complete at the time of prescribing. Without penalty for non-completion, this can be seen as promoting complacency. The user becomes desensitized to non-completion of fields and violations therefore seem less concerning. There were elements that were poor at baseline and did not show any improvement following either intervention including: correct documentation of SR medication, use of generic names and ADR

7 A comparison of educational interventions 375 Table 6 Elements of prescriptions where significant improvement occurred Before intervention % (n/total no charts) After intervention %(n/total no. charts) Relative risk compared with control site (CI) Order was signed (stat Medication name was legible (regular Medication name was legible (prn Medication name was legible (stat No error prone abbreviation was used (regular Medication dose was clear (prn Medication was ordered with generic name (stat Medication frequency was clear (regular Medication frequency was clear (prn Prescriber name was printed (regular Prescriber contact number was present (regular Indication was documented (regular Low intensity 93% (243/262) 99% (247/249)* 1.07 ( )** High intensity 97% (216/223) 98% (191/195) 1.02 ( ) Control site 100% (371/371) 100% 228/228) 1 Low intensity 95% (780/824) 98% (573/583)* 1.04 ( )** High intensity 89% (302/338) 98% (316/323)* 1.10 ( )** Control site 99% (355/357) 99% (600/606) 1 Low intensity 95% (116/122) 97% (107/110) 1.02 ( ) High intensity 93% (83/89) 100% (66/66)* 1.07 ( ) Control site 98% (116/118) 99% (183/185) 1 Low intensity 91% (239/262) 97% (242/249)* 1.06( ) High intensity 93% (207/223) 98% (191/195)* 1.05( ) Control site 99% (141/143) 100% (228/228) 1 Low intensity 55% (453/824) 60% (354/583)* 1.24 ( ) High intensity 49% (164/338) 47% (151/323) 1.09 ( ) Control site 59% (211/357) 52% (318/606 1 Low intensity 88% (107/122) 94% (103/110) 1.03 ( ) High intensity 82% (73/89) 94% (62/66)* 1.10 ( ) Control site 94% (111/118) 98% (181/185) 1 Low intensity 73% (192/262) 82% (203/249)* 1.10 ( ) High intensity 83% (184/223) 84% (163/195) 1.00 ( ) Control site 80% (114/143) 81% (184/228) 1 Low intensity 97% (796/834) 99% (578/583)* 1.01 ( ) High intensity 92% (310/338) 99% (320/323)* 1.07 ( )** Control site 95% (340/357) 97% (585/606) 1 Low intensity 62% (76/122) 88% (97/110)* 1.31 ( )* High intensity 74% (66/89) 71% (47/66) 0.89 ( ) Control site 81% (96/118) 88% (163/185) 1 Low intensity 78% (643/824) 87% (509/583)* 1.18 ( )** High intensity 67% (225/338) 90% (291/323)* 1.43 ( )** Control site 98% (349/357) 93% (561/606)** 1 Low intensity 12% (97/824) 18% (104/583)** 0.90 ( ) High intensity 30% (102/338) 67% (218/323)** 1.33 ( ) Control site 2% (7/357) 3% (20/606) 1 Low intensity 2% (20/824) 7% (43/583)** 8.84 ( )** High intensity 6% (19/338) 10% (32/323)** 5.13 ( )** Control site 3% (12/357) 1% (7/606)** 1 *P < 0.05 two-sided Fishers exact test. **P < documentation. This lack of improvement occurred despite academic detailing specifically targeting these areas (Table 5). This may be due to chart design but workflow might also affect the documentation of ADR. For SR medication documentation and generic name of medications, workflow is also likely to be implicated as the main barrier. If a junior doctor is unsure or unaware about the generic name or SR status of a drug at the time of admission, they would need to consult a reference text/website

8 376 J.S. Thomas et al. which may add additional time pressure. The skill of the pharmacists performing the academic detailing may also play a role. All participants were experienced and completed training specific to the task during a 2-day workshop and used the same materials in the intervention. Although weight is required to determine dosing of some medications, both weight and height were rarely recorded on the medication chart in our study, and no significant change occurred in this practice with any intervention. In contrast, handwriting a patient s name below the printed label is intended to be an important safety step that ensures the correct patient label has been applied and should not be seen as optional. Unfortunately, this step is infrequently completed and represents a missed opportunity to reduce risk. There would be a role for prescriber education to target the relevance of this element. This study was restricted to examining the completion of fields and did not assess their accurate completion, which is a limitation. The use of a control site in this study is also open to debate because site differences in the prevalence of different types of errors makes intersite comparisons of intervention efficacy difficult. Significant baseline differences in prevalence of a variety of prescribing breaches were expected 5, so we did not expect to demonstrate consistent changes relevant to the control site for all elements. Differences in culture and training between sites may also affect the ability of prescribers to improve following intervention. The study was underpowered to show a difference in all elements, particularly in categories where prescription numbers were less (i.e. prn and stat orders) and in elements where breaches were seldom. The safe administration of medications is the consequence of a complex system that requires contributions from many health disciplines and a culture of safety. Correct completion of the NIMC is the focus of this study, but only one part of this system. To optimize chart completion, any intervention beyond that of an online education module needs careful consideration for both its content and its mode of delivery. An electronic prescribing method will diminish, but not totally eradicate, the problems we have identified, and may unleash a new series of concerns. Conclusion Where interventions reduced the frequency of prescribing breaches, both the low- and high-intensity interventions had significant effect. Despite the improvement, significant prescribing breaches persisted at an unacceptable frequency. Given the small but measurable effect of the online module, this module could be expanded to address a significant number of important and persistent prescribing breaches, including patient and prescriber identification and ADR documentation. Improved training of prescribers, better design of the medication chart itself and a better understanding of the workflow of the prescriber might reduce the prevalence of simple prescription errors. Ethics Approval This study was reviewed by the Human Research Ethics committee of the Royal Adelaide Hospital. Acknowledgements We thank Dr Sepehr Shakib for constructing the database, Olimipia Nigro and Karen Macolino for pharmacy support and advice, Dr. Angela Graves for medication chart collection, Sue Roberts for medication chart collection and Dr Leslie Jackowski for invaluable advice. Conflict of interest: Dr Thomas has been a paid consultant for Adelaide Health Technology Association reviewing Pharmaceutical Benefits Scheme applications. She is also an expert reviewer for Australian Medicines Handbook and Therapeutic Guidelines. References 1. Dean B, Schachter M, Vincent C, Barber N. Causes of prescribing errors in hospital inpatients: a prospective study. Lancet 2002; 359: Roughead EE, Semple SJ. Medication safety in acute care in Australia: where are we now? Part 1: a review of the extent and causes of medication problems Aust NZ Health Policy 2009; 6: Aronoson JK. Medication errors: definitions and classifications. Br J Clin Pharmacol 2009; 67: Williams DJP. Medication errors. JR Coll Physicians Edinb 2007; 37: Barton L, Futtermenger J, Gaddi Y, Kang A, Rivers J, Spriggs D, et al. Simple prescribing errors and allergy documentation in medical hospital admissions in Australia and New Zealand. Clin Med 2012; 12: National Inpatient Medication Chart. au/internet/safety/publishing.nsf/content/80a0ef37f281a8 D7CA25718F000CCC2F/$File/chguide.pdf (February 2011, date last accessed). 7. Coombes ID, Stowasser DA, Reid C, Mitchell C. Impact of a standard medication chart on prescribing errors: a beforeand-after audit. Quality Safety Health Care 2009; 18:

9 A comparison of educational interventions Khanal S, Buckley T, Harnden C, Koo M, Peterson G, Ryan A, et al. Effectiveness of a national approach to education for multiple disciplines. Br J Clin Pharm 2012; 75: Smith A, Tasioulas T, Cockayne N, Misan G, Walker G, Quick G. Construction and evaluation of a web-based interactive prescribing curriculum for senior medical students. Br J Clin Pharm 2006; 62: Cattling F, Williams J, Baker R. A prescribing e-tutorial for medical students. Clin Teach 2014; 11: NIMC online training module. nimc (July 2013, date last accessed). 12. O Brien MA, Rogers S, Jamtvedt G, Oxman AD, Odgaard-Jensen J, Kristofferson DT, et al. Educational outreach visits: effects on professional practice and health care outcomes (Review). The Cochrane Library 2008; 3: Shaw J, Harris P, Keogh G, Graudins L, Perks E, Thomas PS. Error reduction: academic detailing as a method to reduce incorrect prescriptions. Eur J Clin Pharmacol 2003; 59: Avorn J, Soumerai SB. Improving drug-therapy decisions through educational outreach a randomized controlled trial of academically based detailing. N Engl J Med 1983; 308: Soumerai S, Avorn J. Principles of education outreach ( Academic Detailing ) to improve clinical decision making. J Am Med Assoc 1990; 263: Australian Pharmaceutical Advisory Council. Guiding principles to achieve continuity in medication management Pippins JR, Gandhi TK, Hamann C, Ndumele CD, Labonville SA, Diedrichsen EK, et al. Classifying and predicting errors of inpatient medication reconciliation. J Gen Intern Med. 2008; 23: Thomas JS, Koo M, Shakib S, Wu J, Khanal S. Impact of a compulsory final year medical student curriculum on junior doctor prescribing. Intern Med J 2014; 44: