Investigation of elective surgery in Barking, Dagenham and Havering in relation to demand management. Public Health Research Report No 151

Transcription

1 Investigation of elective surgery in Barking, Dagenham and Havering in relation to demand management Public Health Research Report No 151 Author: Dr F Perlman Issued: January 2002 ISSN No ISSN X Issued by the The Clock House, East Street, Barking, Essex IG11 8EY Telephone: Fax: Further reports issued by the Public Health Directorate are available on this web site.

2 Contents Page No Executive summary...4 Abbreviations Introduction Background and task definition Timescale Local picture Scientific and political picture Aims Objectives Personal involvement and that of others Competencies and evidence for having addressed them Critical literature review Introduction Search strategy Factors affecting demand for elective surgery Methods of increasing clinically effective practice Using routine hospital data to plan health services Using the literature review to inform planning Data analysis Data analysis by SEIPH Further analysis and interpretation Selecting procedures for discussion Low operative rates Coding and classification discrepancies Day case rates Discussion Other information Private health care Conclusions Using this information for planning Issues Taking the work forward Implementing change Outcome The present situation Discussion

3 5. Final conclusions Conclusions Wider implications for public health Lessons learnt Personal contribution as a public health physician...40 References

4 Executive summary Background There were concerns about high rates of elective surgery and long waiting times in Barking, Dagenham and Havering. Aim To investigate the appropriateness of elective surgery locally using routine data in order to increase appropriate practice. Methods I reviewed the literature critically using a clearly defined search strategy, exploring factors affecting demand for elective surgery and effective methods of changing clinical practice. I used a commissioned data analysis which measured rates of surgical HRGs against two comparators. I identified key procedures contributing to HRGs with high rates, exploring reasons for the differences. The findings were discussed with local clinicians. Results Numbers of several common procedures were higher than elsewhere. Some were clinically ineffective. Reductions were negotiated for tonsillectomy, grommet insertion, cystoscopy, circumcision, arthroscopy, and varicose vein and minor skin operations. Plans for change were agreed, including clinical guidelines, audit, altered service delivery and reductions in commissioning. Follow up showed reductions in some procedures, although these may have been part of wider trends. Discussion Some elective surgery locally seems clinically inappropriate. This method helped target initiatives for clinical change. Operative reductions were limited, however. I discuss barriers to change, alternative approaches and the effects of recent NHS restructuring. Key words Demand management, waiting lists, HES data, elective surgery. Competencies I address the following competencies: To assemble, review critically and interpret the published literature (and, where appropriate, other sources) on a particular topic To identify and obtain relevant information and show how it can be used to plan health services or other activities aimed at improving health 4

5 Abbreviations BHHA FCE HHT HES HRG LMC PCG PCT RHCT SEIPH SLA Barking and Havering Health Authority Finished Consultant Episode Havering Hospitals Trust Hospital episode statistic (data) Healthcare resource group Local Medical Committee Primary Care Group Primary Care Trust Redbridge Health Care Trust South East Institute of Public Health Service level agreement 5

6 1. Introduction 1.1 Background and task definition This project was undertaken following concerns about waiting times and levels of elective surgery in Barking, Dagenham and Havering. I was asked by the Director of Public Health to lead on a project initiated before I joined BHHA. A data analysis of comparative rates of elective surgery by HRG had been commissioned from SEIPH 1 since it was considered more feasible to reduce elective than emergency surgery. I was asked to use this information to identify high rates of elective surgery in Barking, Dagenham and Havering which might be inappropriate and where reduction might be possible. I was asked to present my findings at discussions between BHHA and Havering Hospitals Trust and to lead the implementation of any recommendations. 1.2 Timescale SEIPH produced their data analysis 1 in May Meetings with clinicians were held in July 1998 and the resultant plans were implemented over the following months. The guidelines were published early in Local picture BHHA (Figure 1.1) has a population of approximately 380,000. Barking and Dagenham is a deprived urban borough, whereas Havering is more affluent. Ethnic minorities constitute 4.58% of the population, concentrated in Barking. Most are South Asian. Rates of smoking-related diseases are high, for example the SMR for CHD is 106. Smoking rates are 33.2% in men and 28.8% in women. 2 A quarter of general practices are single-handed. The generic prescribing rate in 1998 varied between 56.9% in Barking and 63.9% in Romford, lower than the then national rate of 69%. 3 From April 1999 until April 2001 there were five PCGs. Approximately 75% of elective surgery for local residents, including all ENT surgery, takes place at Havering Hospitals Trust (HHT) and most of the remainder at Redbridge Health Care Trust (RHCT). 6

7 This project formed part of BHHA s demand management strategy to address rising rates and costs of hospital treatment. It was guided by the Waiting List Management Group, consisting of hospital consultants and managers, GPs, the Director of Public Health and BHHA commissioners. 1.4 Scientific and political picture Demand management The NHS faces increasing costs and patient expectations together with limited resources. Rising demand for health care is shown by increased GP consultation rates. 4 Waiting times and list sizes for elective surgery and outpatient appointments 5 are also increasing despite the introduction of waiting time targets with financial penalties for failing to meet them, and initiatives to clear lists by increasing the volume of surgery. Demand has been poorly managed previously, for example by failing to curb public expectations. Now it needs to be addressed to use limited resources effectively. 6 Pencheon defines demand management as Curtailment of demand for ineffective or inappropriate services, coping with appropriate demand for effective and cost effective clinical services, and creating demand where there is genuine unmet need. 7 Demand management has essentially utilitarian objectives: the maximisation of total met need for the greatest number within available resources. 7 This is distinct from rationing. In the UK, elective surgical rates vary widely between districts, whether measured by GP referral rates, operative rates or surgical admission rates. 15 Variations are apparently unrelated to levels of pathology and include procedures of limited effectiveness, for example tonsillectomy. 7 The relationship between waiting times and demand is complex. Longer waiting times are associated with fewer referrals 8 and influence patients to choose private care 9 that may constitute 8% of elective surgery, more in affluent areas. 10 Increased admissions for elective surgery are correlated with lower waiting times, but not with lower list size, since additions to the list increase concurrently. 11 This is of concern since Government initiatives aim to reduce list size by increasing the volume of surgery. 11 Waiting lists do not manage demand as directly as previously supposed, since most patients eventually receive surgery. 12 I explore the demand for elective surgery further in the critical literature review Healthcare resource groups HES data is routinely produced by NHS trusts. HRGs, adapted from the American diagnosis-related groups (DRGs), are groups of surgical procedures or medical diagnoses with similar care and resource requirements. 13 Medical diagnoses are coded by ICD-10 codes and surgical procedures by OPCS-4 codes. HRGs take into account age and any complicating factors. Adjusting for casemix, the proportion of patients in each HRG helps compare efficiency and costs between hospitals or clinicians. I explore the use of HRG use in healthcare planning in the critical literature review. 7

8 1.5 Aims I aimed to see how routine hospital data could be used to explore the appropriateness of elective surgery in Barking, Dagenham and Havering in order to influence clinical practice and service provision, thus helping to manage demand. 1.6 Objectives My objectives were to: Review critically relevant literature Use the data analysis by SEIPH to identify HRGs where rates of elective surgery were higher in Barking, Dagenham and Havering than the comparators Identify key procedures contributing to these HRGs Identify unexpectedly low rates of surgery Work with clinicians to explore reasons for high operative rates, and find areas where more appropriate care might reduce these rates Agree feasible reductions in procedures with clinical directors Work with clinicians and managers to formulate plans for reducing these rates Implement these plans Monitor the impact of these plans on surgical activity 1.7 My involvement and that of others The data analysis commissioned from SEIPH 1 measured rates of surgical HRGs in BHHA and two comparators. I analysed this data further, adding a clinical interpretation and requesting further data where necessary. I presented the data analysis with a commentary and recommendations at discussions between BHHA and HHT (Section 4.1) and led the implementation of many of the recommendations. I commissioned monitoring data from SEIPH and the BHHA information department. 1.8 Competencies and evidence for having addressed them To assemble, review critically and interpret the published literature (and, where appropriate, other sources) on a particular topic I present the literature review and search strategy in Section 2. To identify and obtain relevant information and show how it can be used to plan health services or other activities aimed at improving health In Section 3 I show how I obtained relevant information and consider this critically. In Section 4 I identify issues requiring consideration in planning and show how data analysis informed this. I describe implementation of these plans and the results of follow up. In the discussion I also consider how NHS changes would affect the use of this information in planning. 2. Critical literature review 2.1 Introduction To understand how best to use this data to assess and influence the appropriateness of surgery I formulated the following search questions:- What factors affect demand for elective treatment in the National Health Service (NHS)? 8

9 What approaches have been tried in order to reduce demand for treatment by increasing clinically effective care? How has HES data been used to plan health services, especially elective surgery, and were any problems encountered? I modelled the care process to identify factors affecting demand for elective surgery at each stage and to identify relevant interventions which could influence demand (Table 2.1). I used this to guide my literature search. Table 2.1 Factors influencing demand for elective surgery and relevant interventions Stage in healthcare Influencing factors Interventions process Patient consults GP Knowledge Health beliefs Patient education Decision aids GP decides to refer ( gatekeeper role ) Surgeon decides to operate Patient agrees to referral or intervention Patient chooses private sector Other factors in the system (not related to any particular stage) Clinical knowledge Uncertainty Other (e.g. doctor-patient relationship, style of consultation) Fear of medicolegal consequences Clinical knowledge and beliefs Other factors Time for operating Knowledge Belief in clinician Doctor-patient relationship Length of NHS waiting list Insurance Affluence Waiting list size Financial incentives Supply of surgeons Private sector Cultural beliefs Referral guideline Educational intervention Feedback referral data Clinical audit Consultation skills training as above Patient education or decision aid Address at structural level 2.2 Search strategy I searched the following sources: Ovid Medline (1966 to the present) HELMIS and King s Fund databases NHS publications (from the Department of Health website and the BHHA public health department library). Grey literature (Work of others, particularly that presented by other London health authorities and trusts at a demand management conference in May 1998). References quoted in papers When searching Medline and the Kings Fund and HELMIS databases, I combined and used search terms from MeSH headings in Table Table 2.2 Search terms used Question Factors affecting healthcare demand (Table 2.8 and Section 2.3) Search terms Healthcare demand and need Healthcare rationing 9

10 Methods to increase clinically effective practice (Table 2.9 and Section 2.4) HRG data (Section 2.5) State medicine Great Britain Cost control Physician s role Practice guidelines Clinical protocols Physicians practice patterns Critical pathways Decision making Patient Diagnosis-related groups Healthcare planning I limited the searches to: English language Humans I describe my criteria for accepting or rejecting papers in the individual sections. Table 2.3 Hierarchy of evidence 14 I. Strong evidence from at least one systematic review of multiple welldesigned randomised controlled trials II. III. IV. Strong evidence from at least one properly-designed RCT of appropriate size Evidence from well-designed trials without randomisation, single group pre-post, cohort, time series or matched case-controlled studies Evidence from well-designed, non-experimental studies from more than one centre or research group V. Opinions of respected authorities, based on clinical evidence, descriptive studies or reports of expert committees 2.3 Factors affecting demand for elective surgery I found a limited number of papers. I considered all levels of evidence 14 for factors I considered applicable to elective NHS surgery. I summarise these papers in Table 2.8. Factors other than clinicians knowledge, including patient or clinician characteristics and the doctor-patient relationship, may affect practice, although further research is required. 17 GP referral rates increase steadily in line with consultation rates and vary three- or four-fold between practices. 15 This may indicate correct clinical agreement, but could also indicate a collusion of anonymity. 16 GPs believe that prior knowledge of a patient helps their clinical decision-making, although the effect on practice is untested. 18 Multidisciplinary panels may consider fewer patients appropriate for surgery than individual clinicians. 19 GPs have a key role as gatekeepers in managing demand. 15 Differing referral thresholds may reflect varied ability to tolerate uncertainty 20 which could be reduced by knowledge and risk sharing with secondary care

11 Patients decisions to seek care may be influenced more by social networks and personal beliefs about a symptom than actual need. 21 Psychological distress does not affect help-seeking. 22 Level of education may also be important. 23 Greater deprivation is associated with higher operative rates for varicose vein surgery, with a U-shaped relationship for tonsillectomy and cataract surgery. 24 Reasons for this are uncertain. Factors in the consultation affect clinical decision-making. Shared decisionmaking can affect the uptake of interventions. 25 Most patients feel satisfied with their involvement in a decision for surgery, but they may not influence the decision. 26 Societal factors may influence decision-making. Surgeons decisions to operate do not correlate with their country s operative rates that may be affected by economic and social forces. 27 The clinician s interaction with his profession and the health care system 17 could have an effect but has not been tested empirically. Removing system incentives could affect practice, 28 but others believe that this approach is ineffective. 29 Much of this research is from overseas and some is theoretical, so its applicability is uncertain. 2.4 Methods of increasing clinically effective practice For specific interventions that appeared relevant I used systematic reviews. For less measurable factors and local studies I accepted lower levels. I summarise these papers in Table 2.9. Evidence-based information in the form of a properly implemented clinical guideline can change practice. 31 Factors in guidelines development can affect uptake including appropriate choice of topics 32 (Table 2.4) and guideline quality 31 (Table 2.5). Simple and more testable recommendations are more likely to be followed. 33 Controversy exists over whether clinicians are more likely to use nationally or locally developed guidelines. 31 Guidelines may increase demand where there is unmet need

12 Table 2.4 Criteria for selecting topics for guidelines and protocols 32 Prevalence of a condition or Burden of a condition on an individual frequency of a procedure and community Economic costs of condition and Potential improvement in health interventions outcomes if guidelines are followed Undesirable variation in clinical Potential for consensus practice Implementation possible Table 2.5 Criteria for a good quality guideline 31 Validity Applicability (to the local population) Reproducibility Clinical flexibility Reliability Clarity Representative development Meticulous documentation group Clear review procedure Cost-effectiveness Utilisation review However, effective implementation methods are key in affecting their uptake (Table 2.6). These may interact, since multifaceted methods are more effective. 34,35 Most methods are effective in some, but not all, circumstances. 35 For some methods the effective component is hard to identify or measure. 36,37 More research is required. The few trials specifically addressing implementation of GP referral guidelines showed a combination of training, referral cards and joint consultation sessions helped. 38 Interventions affecting patients and the doctor-patient relationship can influence practice, such as guidelines available in the consultation, 39 patient-mediated interventions, 34 shared and informed decision-making 25 and better doctor-patient communication. 40 Training clinicians in communication skills may help. Other factors are clearly important, but harder to measure. Changes in organisations and individuals are important. Diagnostic analysis can identify particular barriers and ways to overcome them. 35 Barriers include scarce resources, conflicting priorities, lack of skills and poor research evidence. 41 Table 2.6 Effectiveness of implementation methods Intervention Effectiveness Printed educational materials 42 Little effect Computerised information 39,43 Physician and patient reminder and patient education can make significant difference Audit and feedback 44,45 Modest effect. Better when combined with other (unspecified) interventions Patient specific reminders 31 Help change practice Educational outreach visits 31,36 Effective, but? which component Continuing medical education 46 Broad category. Outreach visits and physician reminders most effective. Local opinion leaders 37 Mixed effects, unknown what they do Physician profiling (peer comparison Statistically effective but minimal effect feedback) 47 Mass media 48 May be effective (study quality poor) Multifaceted 34,35 More effective than single interventions 12

13 2.5 Using routine hospital data to plan health services I found few papers. Work, mostly in the USA, relates more to treatment costs 52 than appropriateness. Limitations in research include poor quality or incomplete data, lack of concurrent controls, inability to ascertain important study outcomes and the lack of prospective studies. 53 Coding errors are widespread (Table 2.7). Table 2.7 Coding inaccuracies in HRGs or DRGs Country Coding error rate USA % USA % Australia 56 22% UK % (less for surgical HRGs). UK 58 15%(surgical) 20%(medical) Giuffrida and Gravelle 59 criticise the use of yearly hospital admission rates as NHS performance indicators due to data instability, and suggest a moving average of two or three years data. Rates are significantly affected by age and sex standardisation, socio-economic variation and, to a lesser extent, the supply of hospital care. Standardisation and regression analysis may eliminate these confounding factors. Hospital admission rates should be used to target further investigation rather than as NHS performance indicators Using the literature review to inform planning Several issues were taken forward for consideration (Section 4.1). 13

14 Table 2.8 Papers addressing factors affecting demand (discussed in Section 2.3) Author Type of paper Source Relevant findings (Level I evidence 14 ) Anderson 25 Systematic review (Level III evidence 14 ) Cochrane 3 studies, prostate surgery and prostate cancer screening. Interactive videos. 3 studies (1011 participants). Shared decision making led to significantly less PSA screening requests, patients were more satisfied. Kahan 19 Prospective Medline 9 member panels (surgeons and GPs) rated appropriateness of surgery individually and together. Fewer patients considered suitable for surgery than individual clinicians. Larsson 26 Prospective 666 patients awaiting surgery. Questionnaire.73% satisfied with involvement in decision (less in women and immigrants), 41% thought it was a joint decision Berkanovic 21 Prospective 6 interviews in a year. 57% of variance explained by a set of variables of which need was 12% and network influences and personal beliefs about a symptom were 42%. Berkanovic 22 Prospective Baseline interview and 6 interviews in a year. No relationship between psychological distress and frequency of consultations Rutkow 27 Observational 4687 surgeons from 3 countries assessed case scenarios. Results from surgeons did not vary, despite variations in rates between the countries. (Level IV evidence 14 ) Chaturvedi 24 Observational Medline GP morbidity survey and HES data showed differences in GP consultation rates and rates of related procedures for common conditions in differently deprived groups. Kuh, Stirling 23 Observational Better educated British women of higher social class have lower hysterectomy rates Lowy 28 Observational Secondary reference 20 presumed due to financial incentives without a corresponding reduction in the hospital workload. Hjortdahl 18 Observational Medline 133 GPs evaluated 3918 consultations retrospectively. In 2/3, prior patient knowledge was thought useful 14

15 Author Type of paper Source Relevant findings (Level V evidence 14 ) Borowitz 29 Review/opinion Medline Controlling demand by increasing appropriate practice requires micro-interventions aimed at clinicians rather than system incentives. Gillam, Review/opinion Patient-, doctor- or service-related factors can affect GP referral rates. Referral thresholds and ability to Pencheon 20 tolerate uncertainty important. Low referral rates may not indicate good practice. Few incentives to avoid referral. Coulter 15 Review/opinion The GP s gatekeeping role is arguably the NHS s most important mechanism for demand management. Referral rates increase steadily in line with consultation rates and vary three- or four-fold between practices. Feeding back referral data to GPs is recommended to improve understanding, but does not influence referral rates. GPs show scepticism about the data and its quality, and concerns that under-referral may be concealed. Pencheon 6 Review/opinion Gatekeeping role of GPs could be enhanced by more support by knowledge sharing and risk sharing (including medicolegal risks) with secondary care. Vickery, Lynch 30 Review/opinion Four components of health service demand: morbidity, non-health motives, perceived need and patient preference. The latter two are probably most amenable to change (by patient empowerment and education). Eisenberg 17 Theory/opinion Sociological factors affect clinicians decision-making: clinician and patient characteristics, clinician-patient relationship, clinician s relationship with profession and health system. Theoretical 15

16 Table 2.9 Effectiveness of interventions in changing practice (discussed in Section 2.4) Author Type of paper Source Intervention Relevant findings (Level I evidence 14 ) Balas 43 Systematic review Cochrane Library Computerised information Thomson O Brien 44 Audit and feedback vs. alternative strategies Thomson Audit and O Brien 45 feedback Thomson Educational O Brien 36 outreach visits Thomson Local opinion O Brien 37 leaders 100 RCTs. Physician and patient reminder, treatment planner and patient education can all make a significant difference. Size of studies not accounted for. 12 randomised studies. Reminders more effective (not significantly). No complementary method improves effectiveness of audit and feedback. Characteristics of intervention not studied. 37 randomised studies. Method descriptions poor. Sometimes effective in improving practice, especially prescribing and test ordering. Moderate but worthwhile effects. Other approaches (unspecified) should be added. 18 randomised studies. Several components e.g. written materials, conferences, feedback, audit. Effective aspects unclear. May modify behaviour if combined with social marketing, especially prescribing. 8 randomised studies, various clinical areas. Mixed effects. Unclear what they do and if replicable 16

17 Author Type of paper Source Intervention Relevant findings Oxman trials, random or quasi-random allocation. Interventions included educational materials, conferences, outreach visits, opinion leaders, patient-mediated interventions, audit and feedback, reminders, multifaceted interventions, local consensus. All showed an effect some of the time, but the size of effect was limited for any. Stewart 40 Physicianpatient communicatio n Balas 47 Meta-analysis of RCTs (Level III evidence 14 ) Conroy, Systematic Shannon 49 review Physician profiling 10 RCTs, 11 observational studies. Most showed correlation between doctorpatient communication and patient outcomes (emotional and physiological). Peer-comparison feedback interventions. 12 studies, 10 were positive. Statistically significant but minimal effect on frequency of intervention. Medline Development Local credibility possibly enhanced by explicit referral to a national guideline. Local ownership is also important, involving end-users in development and approval by local opinion leaders. Belief that a guideline uses sound evidence and development by clinicians from the same discipline influence clinicians to use them. Grimshaw, Russell 50 Development Guideline criteria (Table 2.4). Systematic evidence review is the most valid development method. Where evidence is limited formal or informal consensus may be required. Local groups are recommended to adapt national guidelines which are better resourced. Grimshaw, Freemantle 5 1 Various Over 90% of 87 studies reported improved adherence to recommended practice. Over 70% of 17 studies measuring patient outcome reported an improvement. 17

18 Author Type of paper Source Intervention Relevant findings EHCB: Implementing clinical practice guidelines studies, various designs. 43 of 44 studies with Grade 1 evidence showed significant change in clinical process, 8 out of 11 in outcome. Development, implementation and monitoring affect adherence. Guidelines change practice more if they take into account local circumstances, are disseminated by an active educational intervention and implemented using patient-specific reminders in the consultation Grilli 48 Mass media 17 studies, various types. Media could influence use of health care interventions. Freemantle 42 Printed educational materials Shiffman 39 Computerbased guideline implementatio Davis 46 Systematic review Cochrane Library n systems Continuing medical education Grilli 33 Topic of guideline Primary study quality variable. 11 studies, various designs. Small effect compared to controls, but poor analyses in primary studies. 25 studies, 10 controlled trials (9 randomised), 10 time-series correlational studies. All systems had patient-specific recommendations, 19 were available in the consultation. Guideline adherence improved in 14 of 18 systems. 99 studies, 145 interventions measured physician performance of which 70% showed change. Effective strategies were outreach visits, opinion leaders, patient-mediated interventions and physician reminders. Limited detail of original studies 23 studies, (143 recommendations). Low average compliance. Better with less complex and more testable recommendations, but these only accounted for 47% in the variability in compliance. 18

19 Author Type of Source Intervention Relevant findings paper Grimshaw 38 Secondar y reference 35 Interventions to change referrals 4 studies (hard to draw firm conclusions). Training plus referral cards plus joint consultation sessions effective. Developing and disseminating local consensus guidelines ineffective. (Level IV evidence 14 ) Raine 41 Grey literature Local barriers to implementation include poor perception of relevance, lack of resources, short-term outlook, conflicting priorities, difficulty measuring outcomes, lack of necessary skills, limited research evidence on effectiveness, perverse incentives and a high intensity of intervention required to change practice 39. (Level V evidence 14 ) Grimshaw, Review/opi Medline Development Criteria for selecting topics for guidelines and protocols (Table 2.3) Feder 31 nion criteria (Much of review is Level I, but the aspect I chose is level V 14 ) EHCB: Getting evidence into practice 35 Summary of systematic reviews Medline Various Individual and organisational change important. Diagnostic analysis identifies factors influencing change and can guide interventions chosen. Multifaceted interventions targeting different barriers to change are most effective. Adequate resources, knowledge and skills are required. Monitoring and evaluation should be included. 19

20 3. Data analysis 3.1 Data analysis by SEIPH The Director of Public Health commissioned a data analysis from SEIPH, 1 an academic unit. As with a previous project 60 he believed that their contribution would increase data credibility in the eyes of clinicians. Casemix data for elective surgery in six specialties was analysed: General surgery Urology Orthopaedics Ophthalmology Ear, nose and throat (ENT) surgery Oral surgery Operative rates were provided for residents of: Barking, Dagenham and Havering Comparator areas (Figure 1.1): o Bexley and Greenwich o South London (Bexley and Greenwich; Croydon; Bromley; Kingston and Richmond; Merton, Sutton and Wandsworth and Lambeth, Southwark and Lewisham) Bexley and Greenwich was chosen as demographically similar to BHHA. South London is a large urban area where operative rates are believed to be high and whose data completeness (1996-7) was greater than Greater London. Procedures (measured as FCEs) were subdivided into inpatient and day cases. Directly standardised age specific admission rates using HRG data were provided. SEIPH believed that HRGs were useful to trawl the data initially to identify areas requiring further investigation. They produced a full report including the additional information I requested Further analysis and interpretation I compared the five HRGs with the highest rates of surgery for each specialty in Barking, Dagenham and Havering with Bexley and Greenwich and South London. I considered additional HRGs if numbers and rates seemed sufficiently large. Table 3.1. Reasons for variation in operative rates Differences in: Rates of pathology Features of the local population (e.g. socioeconomic 24 ) Surgical practice GP referral thresholds 20 Health authority policies (funding of procedures) Coding practice

21 Table 3.2 Factors determining the appropriateness of high HRG rates and feasibility of reduction Seriousness of the underlying condition Acceptability of reducing the operative rate to clinicians, patients and others Increased operative rate merely reflecting different local pathology Evidence of effectiveness of the procedure Evidence for alternative (non-surgical) management I requested a breakdown into procedures for HRGs with higher rates in BHHA than the comparators (Table 3.3) also considering whether numbers of FCEs in Barking, Dagenham and Havering would be sufficient to warrant targeting resources for clinical change. I excluded HRGs where reduction seemed inappropriate, having considered reasons for variation in rates (Table 3.1) and factors determining the appropriateness of high rates and scope for intervention (Table 3.2). Table 3.3 HRGs broken down into individual procedures HRG HRG Title HRG HRG Title J37 Minor skin procedures H10 Arthroscopy Q11 Varicose vein procedures H13 Hand procedures Categ 1 F95 Anus minor procedures C24 Mouth and throat procedures Categ 3 L39 Minor open procedures penis C22 Nose procedures Categ 3 L21 Bladder minor endosc procedures C01 Ear procedures Categ 1 L30 Prostate/bladder neck minor endoscopic procedures B06 Other ophthalmology procedures Categ 4 C14 Mouth and throat procedures Categ Selecting procedures for discussion Rates of some HRGs were statistically significantly higher in Barking, Dagenham and Havering (Table 3.4), so differences were unlikely to be by chance. The total number of FCEs and the main constituent procedures are also shown. I calculated 95% confidence intervals (Tables 3.4 and 3.5) by converting rates into proportions (SEIPH had not done so). I compared individual procedure rates where the procedures of interest comprised less than 100% of the HRG (Table 3.5). These procedure rates were higher than the comparators, statistically significantly so for all but grommet insertion. I used crude rates since subsequently obtained data were not broken down by age and sex. This probably reduced accuracy only slightly since the population structures were similar. 21

22 Table 3.4 Rates of HRGs selected for further discussion in BHHA, South London and Bexley and Greenwich (BGHA), with principal procedures in BHHA ( HES data) HRG Area Cases Standardised rate Principal procedures in BHHA per 100,000 population (95% confidence intervals) J37 Q11 L39 L21 H10 C24 C01 BHHA ( ) Assorted excisions of skin lesions, and nail procedures (100%) S London ( ) BGHA ( ) BHHA ( ) Assorted varicose vein operations (not injections) (100%) S London ( ) BGHA ( ) BHHA ( ) Circumcision (65.5%) S London ( ) BGHA ( ) BHHA ( ) Diagnostic cystoscopy (87%) S London ( ) BGHA ( ) BHHA ( ) Diagnostic arthroscopy knee (21.6%) S London ( ) BGHA ( ) BHHA ( ) Tonsillectomy (70.3%) S London ( ) BGHA ( ) BHHA ( ) Grommet insertion (62.9%) S London ( ) BGHA ( ) 22

23 Table 3.5 Comparing crude rates of selected procedures in BHHA, South London and Bexley and Greenwich (BGHA) ( HES data) Procedure Area Number of cases Rate per 10,000 population (95% confidence intervals) Diagnostic cystoscopy Circumcision Arthroscopy Tonsillectomy Grommet insertion BHHA ( ) S London ( ) BGHA ( ) BHHA ( ) S London ( ) BGHA ( ) BHHA ( ) S London ( ) BGHA ( ) BHHA ( ) S London ( ) BGHA ( ) BHHA ( ) S London ( ) BGHA ( ) Some HRGs with higher rates were not selected for discussion (Table 3.6). 3.4 Low operative rates Three NHS performance indicators for clinical effectiveness, 71 total hip replacement, total knee replacement and cataract surgery, had similar rates to the comparators. I found no other high volume HRGs with lower rates. 23

24 Table 3.6 HRGs with higher rates in BHHA not selected for discussion ( HES data) HRG Reason not selected No of FCEs in HRG in BHHA residents BHHA % of South London rate General surgery F73 Inguinal/umbilical/femoral % No alternative management hernia repair >69 G14 Biliary tract major procedures > % For serious pathology J04 Intermediate breast surgery % <.49 Urology L19 Bladder intermed. endoscopic. procs % For suspected serious pathology L17 Bladder major endoscopic procs % L45 Lithotripsy % No alternative management L13 Ureter intermediate endoscopic procs % For suspected serious pathology L12 Ureter major endoscopic % procs Orthopaedics H17 Soft tissue/bone procs categ 1 < % Many diverse procedures in HRG H22 Minor musculoskeletal % procedures H12 Foot procedures categ % ENT C22 Nose Categ 3 (sinus, turbinate, septoplasty) C02 Nose Categ 1(turbinates, cautery, antral drainage) C31 Ear Categ 4 (mastoid, tympanoplasty, ossicles) % Evidence poor for indications for surgery. Other areas in ENT prioritised % % 3.5 Coding and classification discrepancies Coding and classification discrepancies gave some artificially high HRG rates (Table 3.7). HHT classified some outpatient procedures (not part of HES data) as day cases to increase the apparent day case rate, an NHS efficiency measure. Table 3.7 HRGs and procedures with coding discrepancies HRG Procedure (% of HRG) F95 Haemorrhoid procedures (injection and surgery). (63.1%) L30 Rectal needle biopsy prostate (53.2%) Dilatation outlet female bladder (45.5%) B05 Cautery of lesion of retina (47.3%) L21 Caverject injections (35%) 3.6 Day case rates 24

25 Day case rates were significantly lower in BHHA than the comparators in orthopaedics (Table 3.8, Figure 3.1) but not other specialties. Table 3.8. Orthopaedic day case rates (1996-7: HES) HRG % Day cases (total number of cases) BHHA South London BGHA Arthroscopy 46.3% (765) 62.2% (1800) 68.4% (512) Hand procs. Categ 1 (carpal tunnel) 59.1% (90) 88.7% (338) 83.6% (60) Figure 3.1 Day case rates in orthopaedics (Source: HES data) Percentage of procedures performed as day cases Barking & Havering arthroscopy (H10) South London arthroscopy (H10) Bexley and Greenwich arthroscopy (H10) Barking & Havering carpal tunnel (H22) South London carpal tunnel (H22) Bexley and Greenwich carpal tunnel (H22) Procedure and health authority 3.7 Discussion Methodology The data was trawled using HRGs to identify high operative rates. Comparing individual procedure rates would be more accurate since the proportion of procedures in an HRG could vary between comparators. However, most HRGs with significantly higher rates locally (Table 3.4) also contained key procedures with significantly higher rates (Table 3.5), except of grommet insertions. Trawling with HRGs therefore probably sufficed as a focus for further investigation 59 and avoided the complexity of looking at a very large number of procedures. The single year s data used here may fluctuate. 59 SEIPH did not use Giuffrida and Gravelle s proposed moving average of two or three years data. 59 However, differences in rates identified the need for further investigation. 59 The commissioning of data analyses is discussed further in Section 6.2. Choosing procedures for further discussion involved clinical discretion, which is subject to error. 25

26 3.7.2 Data quality The Department of Health stated that data completeness was good and checked the coding and internal consistency of patient records. Coding inaccuracies concur with findings elsewhere despite standardised manuals. Incentives to classify outpatient procedures as day cases are of concern Comparators Bexley and Greenwich forms part of South London. However, it seemed reasonable to use both as comparators since Bexley and Greenwich forms only one-sixth of the South London population, their operative rates were similar and South London was the principal comparator used in planning. Bexley and Greenwich was chosen as demographically similar to Barking, Dagenham and Havering, having a mixture of deprived and better off areas. 1 The Jarman and Townsend scores of both are similar and intermediate to the other health authorities, making comparisons seem reasonable (Figs 3.2 and 3.3). The effect of variation in ethnicity between districts on these procedures is unclear. London admission rates were said to be high. 1 admissions, 64 but not elective surgery. This has been shown for acute Figure 3.2 Townsend scores by health authority Townsend score Bromley Kensington and Richmond Croydon Barking and Havering Bexley and Greenwich Merton Sutton and Wandsworth Lambeth Southwark and Lewisham -6 Health authority Source: ONS 26

27 50 Figure 3.3 Jarman scores by health authority Jarman score Bromley Kingston and Richmond Croydon Barking and Havering Merton, Sutton and Wandsworth Bexley and Greenwich Lambeth, Southwark and Lewisham -30 Health authority Source: ONS 3.8 Other information Other information could have supplemented the data analysis. Many of the procedures identified are performed for minor pathology. However, bladder cancer, whose incidence in Barking, Dagenham and Havering is higher than average (figure 3.4) is diagnosed and monitored by cystoscopy. A higher operative rate might be appropriate. Figure 3.4 Standardised registration rates for bladder cancer (pooled data ) Source: ONS SMR Barking and Havering Bexley and Greenwich Bromley Croydon Kingston and Richmond Lambeth, Southwark and Lewisham Merton, Sutton and Wandsworth Health authority High levels of smoking locally 2 could be related to an increased incidence of glue ear, itself smoking related

28 I found no evidence in the literature about appropriate rates of these procedures. Clinical audit data at this stage could give more information about appropriateness of care. 3.9 Private healthcare Information about widely performed private elective surgery 10 is important, but the range of accessible providers made finding information complex. Local NHS patients have not received these procedures at private providers Conclusions Rates of several HRGs and procedures (Tables 3.4, 3.5) were significantly higher locally. For some there appeared to be scope for reduction. There were no important procedures whose rates were lower. Coding anomalies were found. 4. Using this information for planning I considered several issues from the preceding chapters in planning, identifying those requiring change and exploring how the data analysis could inform planning. 4.1 Issues Pressure to use limited resources effectively meant that demand needed to be curtailed for ineffective treatment and created for effective care. 6 Operative rates vary widely. 7 The data analysis highlighted several procedures with higher rates than comparator districts, some of which could be clinically inappropriate. Discussion with clinicians was needed to assess the feasibility of reductions. I calculated the potential reductions in FCEs if BHHA rates fell to those of South London (Table 4.1) using Method (A) since only this information was available at the time of negotiation. Method A (from Table 3.4) South London HRG rate x FCEs in BHHA BHHA HRG rate I subsequently used Method B for comparison which is probably more accurate (although only crude rates were available), since comparing procedure rates removes the effect of the varying proportion of procedures in HRGs between comparators. Method B (from Table 3.5) South London crude procedure rate x BHHA crude procedure rate FCEs in BHHA Method A overestimated the potential for reducing some procedures, particularly grommet insertions compared to Method B (Table 4.1). 28

29 Table 4.1 Excess FCEs in BHHA if South London rate applied (1996-7) Procedure Excess FCEs in BHHA Method A. South London HRG rate applied Method B. South London crude procedure rate applied Minor skin procedures 437 N/A Varicose vein 534 N/A operations Diagnostic cystoscopy Circumcision Arthroscopy Tonsillectomy Grommet insertion I identified no clinically effective procedures with unexpectedly low rates requiring action. The Director of Public Health decided that BHHA should lead the implementation of change, an issue discussed further in Section Initiatives for change needed to be targeted at GPs, surgeons and patients, addressing their decision-making, behaviour and beliefs (Table 2.1). Using evidence-based clinical guidelines was considered since good quality guidelines can change practice 31 (Table 2.5), developing them from others systematic reviews where possible, otherwise using consensus 31 and adapting national guidelines to increase ownership. Effective multifaceted implementation was required to change practice, 34,35 especially clinician reminders, 31 outreach visits, 31,36 and patient information to help shared decision-making. 25 Awareness that demand can increase if unmet need exists was important. 20 Policies not to treat certain conditions needed to be developed through BHHA s agreed framework covering cosmetic procedures and to involve the public, since these might be interpreted as rationing. Evidence-based commissioning is important and may increase appropriate practice 29,31 but limitations include implementation costs, guidelines possibly increasing activity, and the uncertain relationship between rates and appropriateness of referral. 70 Changing service structure and pathways can reduce costs, although little research relates to the primary-secondary care interface. 78 HHT considered staffing reductions in ENT surgery to be unacceptable. Forcible implementation methods include referral or operative quotas, filtering referrals or returning inappropriate referrals. These were rejected due to concerns about denying effective treatment, medico-legal concerns and awareness that GPs refer for many reasons. Gaining commitment from primary and secondary care was important, as were identifying and addressing barriers to implementation

30 Awareness of wider patient, clinician, system and societal influences (Section 2.6) is important. Caution was needed in using this data due to annual fluctuation in HRG 59 and coding inaccuracies Obvious reasons for differences in operative rates and day case rates (e.g. waiting list blitzes, theatre closures), together with coding and classification problems (Table 3.7), needed to be explored with HHT. Pressure to reduce waiting times existed. The effect of reducing surgery on these was uncertain so monitoring was important. 11,12 Involvement of RHCT was considered, but since they only perform a small proportion of the chosen procedures (none in ENT), this was limited to consultation in district guideline development. 4.2 Taking the work forward I presented the issues and information (Section 4.1) at meetings at HHT with clinical directors, directorate business managers, the medical director and senior hospital managers. The Director of Public Health, the Director of Acute Commissioning and myself represented BHHA. Discussions aimed to get commitment and to: Explore reasons for high operative rates and scope for reduction (Table 4.2) Agree targets for evidence-based reductions in commissioning based on excess procedures over South London rate (Table 4.1) Address coding and classification discrepancies Formulate feasible plans to reduce surgery aimed at GPs, hospital clinicians and patients and agree implementation, including: o Evidence-based clinical guideline developed using methods above and multifaceted implementation o Patient guidelines. o Policies o Service restructuring Agree monitoring of activity and waiting times 30

31 Table 4.2 Reasons for high operative rates postulated by clinical director with my comments Procedure Reasons for increased rate given by Comment clinical director Tonsillectomy Inappropriate GP referral. High local rate of pathology No evidence for high pathology. Inappropriate referrals do not require surgery. Limited range of indications. 80 Grommet insertion 81 Cystoscopy Circumcision Varicose vein operations Minor skin procedures Arthroscopy GPs refer dipstick haematuria without a confirmatory MSU result (some false positives). Bladder cancer patients followed up more frequently than urological guidance. Inappropriate GP referrals. (Religious circumcisions not funded for some time.) Many done for cosmetic reasons Not enough done by GPs Unlikely 28 Underuse of MRI for suspected meniscal tears Over-referral possible. 82,83 Audit needed to confirm this and follow-up intervals. Urologists and general surgeons do them, some surgeons do one or two per year (inappropriate selection of cases) Probably true Likely 61 Table 4.3 shows procedures with apparently inappropriately high and potentially reducible rates, together with negotiated reductions in SLAs and plans to achieve them. Before PCGs commenced BHHA and NHS Trusts reached SLAs about the total volume and cost of surgery commissioned for each specialty. These whole specialty SLAs were reduced in line with the agreed operative reductions in Increased day case rates for certain orthopaedic procedures were also agreed. Coding and classification corrections (Table 3.8) were agreed. 31