Rapid Responses to:

GENERAL PRACTICE: Simon Griffin and Trisha Greenhalgh Diabetes care in general practice: meta-analysis of randomised control trials • Commentary: Meta-analysis is a blunt and potentially misleading instrument for analysing models of service delivery BMJ 1998; 317: 390-396 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

SUMMATIVE ASSESSMENTS ARE NOT HELPFUL

Ray Jones   (11 August 1998)

Diabetes in general practice; Resources should follow the patient.

Ian Quigley   (11 August 1998)

Central registers could coordinate services

Somdutt Prasad   (21 August 1998)

Like must be compared with like in meta-analyses

Kamlesh Khunti   (7 September 1998)

Untitled

Kamlesh Khunti   (11 September 1998)

SUMMATIVE ASSESSMENTS ARE NOT HELPFUL 11 August 1998
Ray Jones, Senior Lecturer in Health Informatics Send response to journal: Re: SUMMATIVE ASSESSMENTS ARE NOT HELPFUL	Griffin's meta-analysis of randomised controlled trials of diabetes care in general practice [1] provides an interesting and informative summary of these studies. However, I share Greenhalgh's scepticism of the use of meta-analysis. As she points out there were major differences in the interventions. It would have been useful, as well, to know more about the homogeneity of the hospital control groups. Both hospital care and general practice care for diabetes have changed over the last 20 years, mainly as a result of implementation of various computer systems to support registers, clinical records, and other forms of 'order' into the chaos of diabetes clinics and general practice of the 1970s. For example, although the evidence is only from 'before and after' studies, the computer-supported diabetes clinics at Nottingham University Hospital from 1979 onwards delivered improved care, compared to its Nottingham General Hospital predecessors, in the years before. Records in computer-supported care were more complete [2], non-attendance [3] and losses to follow-up were reduced [4], and simulation of these effects suggested that this would lead to long term reducton in diabetes related morbidity [5]. Such changes, although not measured, may have been happening in other hospital clinics. For example, the Aberdeen diabetes clinics in Griffin's analysis were early users of computers to support clinical records. In General Practice also, the number of practices with the ability to maintain accurate computer-based records has increased dramatically from the time (1982) of the first study included in Griffin's analysis. Both Griffin and Greenhalgh seem convinced that organized care is better than chaotic care, but I would agree with Greenhalgh that we need to know more about what mix of competencies is needed and how they work. Studies comparing service models are likely to have a limited 'shelf-life'. In the early years of the 21st century we might see more remote contact between diabetes patient and professional (whether hospital or general practice) using telemedicine, more selfcare with expert system support, more mobility of patients between different carers through the use of pervasive computer networks or smart cards. The only certainty is that, we are unable to say once and for all, that either general practice care or hospital care is 'best'. Yours sincerely Dr Ray Jones Senior Lecturer in Health Informatics University of Glasgow 1. Griffin S. Diabetes care in general practice: meta-analysis of randomised control trials. (Greenhalgh T. Commentary: Meta-analysis is a blunt and potentially misleading instrument for analysing models of service delivery.) BMJ 1998;317:390-6. 2. Jones RB, Hedley AJ. A computer in the diabetic clinic: completeness of data in a clinical information system for diabetes.. Practical Diabetes 1986;3:295-296. 3. Jones RB, Hedley AJ. Non-attendance in an outpatient clinic. Public Health 1988;102:385-391. 4. Jones RB, Hedley AJ. Methods of estimating losses to follow-up from a diabetic clinic. Practical Diabetes 1989;6:129-133. 5. Jones RB, Hedley AJ. Evaluation of a diabetes register and information system.. In: Bryant J, Roberts J, Windsor P, eds. In Current Perspectives in Health Computing. Manchester: British Journal of Healthcare Computing, 1986;80-87.
Diabetes in general practice; Resources should follow the patient. 11 August 1998
Ian Quigley, Principal in General Practice Western Road Medical Centre, 99 Western Rd, Romford, RM11 1HH Send response to journal: Re: Diabetes in general practice; Resources should follow the patient.	The meta analysis by Griffin and the comments by Greenhalgh show there is no reason why structured diabetes care should not be provided by GPs. Many GPs do provide an excellent service but few get paid for it beyond the small Chronic Disease Management payments. I have set up a diabetic clinic in a 6 partner practice with 13,500 patients of whom 238 are diabetic. I have been approved by the health authority to do this under HSG (96) 31;A National Framework For The Provision of Secondary Care Within General Practice. I know of only one other practice which has done this. I have described my clinic and its organisation on the practice web site. Http://www.westernroad.co.uk/diabetic.htm I write to encourage other GPs to fund diabetes care this way as we move from Fundholding to Primary Care groups.
Central registers could coordinate services 21 August 1998
Somdutt Prasad Send response to journal: Re: Central registers could coordinate services	Editor- We have mixed feelings about Griffin's meta-analysis of diabetes care in general practice [1]. While the technique of meta-analysis has its limitations as highlighted by Greenalgh [1] and Jones [2], we agree that regular prompted recall and review of people with diabetes would result in improved outcomes. Griffin's analysis however does not take into account important outcome measures like the use of and ease of access to services like chiropody and diabetic retinopathy screening which are essential to good long-term results for people with diabetes. Developing technologies will in the near future see some of this care being provided by telemedicine; for instance, it will be possible for retinal images to be analysed remotely by computers to grade retinopathy [3]. The increasing use of such technologies will mean that central diabetes registers will need to be maintained which will enlist all diabetics in an area, whether they are cared for by their general practitioners, hospital physicians or are under some form of 'shared care'. Such a central register would co-ordinate the provision of ancillary services and also prompt recall for both hospital and general practice clinics. We have established such a central diabetes register in the Wirral, which has enrolled 7519 diabetics. One of the main benefits is that this has allowed a formal diabetic retinopathy surveillance scheme to be implemented, with the register generating call and recall for all diabetics in the area an excellent uptake of the service has been achieved. Sixty percent of diabetics have had their eyes screened for sight threatening retinopathy in the first year of operation and the trend indicates that over 80% will have had this done by the time the register completes 18 months of operation. The evolving changes in healthcare delivery for this very important chronic disease means that the issue is no longer hospital versus general practice care. A holistic approach needs to be taken integrating hospital, general practice and the myriad of ancillary services that are needed to ensure good long-term outcomes for people with diabetes. The establishment of central computerised registers will go a long away in ensuring that optimum care is made available to all diabetics. Somdutt Prasad MS FRCSEd Vitreo-Retinal Fellow Department of Ophthalmology Arrowe Park Hospital, Wirral somprasad@bigfoot.com Karen Jones Diabetes Register Manager The Wirral Diabetes Register Arrowe Park Hospital, Wirral References: 1. Griffin S. Diabetes care in general practice: meta-analysis of randomised control trials. (Greenhalgh T. Commentary: Meta-analysis is a blunt and potentially misleading instrument for analysing models of service delivery.) BMJ 1998;317:390-6. 2. Jones R. Summative assessments are not helpful. eBMJ . 11 August 98. 3. Williamson TH, Keating D. Telemedicine and computers in diabetic retinopathy screening. Br J Ophthalmol 1998;82:5-7.
Like must be compared with like in meta-analyses 7 September 1998
Kamlesh Khunti, Clinical Lecturer Department of General Practice and Primary Health Care, University of Leicester Send response to journal: Re: Like must be compared with like in meta-analyses	General practitioners are now playing a greater role in the management of patients with diabetes. However, there are wide variations in performance between general practitioners. The effectiveness of systems of care in general practice is of great interest and Simon Griffin’s meta-analysis of diabetes care in general practice(1) is therefore timely. However, like Trish Greenhalgh (Commentary(2)), I have considerable concerns about some of the conclusions of Griffin’s meta-analysis. Only five studies met the inclusion criteria for the meta-analysis and all patients were receiving care from hospital. Griffin recommends that prompted care should be provided to a selected group of diabetics. However, this type of prompting would need to be delivered to all the diabetics in the practice as it would be difficult to organise it for a small select proportion of diabetic people. Our recent study has shown that just over half of diabetic people are under general practice care, 19% under hospital care and approximately 30% are under shared care(3), although I agree with Griffin that the taxonomy of shared care is not fully developed1. To make valid conclusions of a meta-analysis, like must be compared with like. Other factors such as the practice population and practice organisation (besides a recall system) may also be associated with the quality of care. Although the proportion of local practices involved did not explain interstudy heterogeneity(1), the size of the practice was only reported in two trials(4),(5). Only one study reported whether the practices ran diabetic mini-clinics(5). Furthermore, as Greenhalgh comments(2), the population in the studies included in the meta-analysis are subject to selection bias and cannot therefore be representative of the general population. The care of patients is very complex and many potential confounders are not taken into account in the meta-analysis. I recently conducted a literature search focused on quality of care of patients with diabets in primary care and found over 37 potential factors. It was therefore inappropriate to use meta-analytical techniques for such a complex and varied group of practices. I am currently involved in a study to determine which factors are related to good diabetic care. The study involves data from 169 practices (639 general practitioners) in three different health authorities that had conducted an audit of patients with diabetes. Preliminary results show that there are wide variations in the process and outcome measures. Regression analysis indicates that having a recall system is not related to the process or outcome of care of patients with diabetes. Basing our decisions on evidence from randomised controlled trials is becoming increasingly acceptable in general practice. A randomised controlled trial with randomisation at practice level is required to assess the effectiveness of structured care with a prompted recall system. In the meantime the debate over who should deliver care to a selected group of diabetic patients will continue. Yours sincerely Dr Kamlesh Khunti Clinical Lecturer Department of General Practice and Primary Health Care Leicester General Hospital Gwendolen Road Leicester LE5 4PW Tel: 0116 2584873 Fax: 0116 2584982 Conflict of interest None References (1) Griffin S. Diabetes care in general practice: meta-analysis of randomised control trials. BMJ 1998;317:390-5. (2) Greenhalgh T. Commentary: Meta-analysis is a blunt and potentially misleading instrument for analysing models of service delivery. BMJ 1998;317:395-6. (3) Hurwitz B, Goodman C, Yudkin J. Prompting the clinical care of non-insulin dependent (type II) diabetic patients in an inner city area: one model of community care. BMJ 1993;306:624-30. (4) Diabetes Integrated Care Evaluation Team. Integrated care for diabetics: clinical, psychological, and economic evaluation. BMJ 1994;308:1208-12. (5) Khunti K, Baker R, Lakhani M, Rumsey M. Quality of care of patients with diabetes: Collation of data from multi-practice audits of diabetes in primary care. Family Practice (in press).
Untitled 11 September 1998
Kamlesh Khunti Send response to journal: Re: this article	Editor General practitioners are now playing a greater role in the management of patients with diabetes. However, there are wide variations in performance between general practitioners. The effectiveness of systems of care in general practice is of great interest and Simon Griffin’s meta- analysis of diabetes care in general practice(1) is therefore timely. However, like Trish Greenhalgh (Commentary(2)), I have considerable concerns about some of the conclusions of Griffin’s meta-analysis. Only five studies met the inclusion criteria for the meta-analysis and all patients were receiving care from hospital. Griffin recommends that prompted care should be provided to a selected group of diabetics. However, this type of prompting would need to be delivered to all the diabetics in the practice as it would be difficult to organise it for a small select proportion of diabetic people. Our recent study has shown that just over half of diabetic people are under general practice care, 19% under hospital care and approximately 30% are under shared care(3), although I agree with Griffin that the taxonomy of shared care is not fully developed(1). To make valid conclusions of a meta-analysis, like must be compared with like. Other factors such as the practice population and practice organisation (besides a recall system) may also be associated with the quality of care. Although the proportion of local practices involved did not explain interstudy heterogeneity(1), the size of the practice was only reported in two trials(4),(5). Only one study reported whether the practices ran diabetic mini-clinics(5). Furthermore, as Greenhalgh comments(2), the population in the studies included in the meta-analysis are subject to selection bias and cannot therefore be representative of the general population. The care of patients is very complex and many potential confounders are not taken into account in the meta-analysis. I recently conducted a literature search focused on quality of care of patients with diabets in primary care and found over 37 potential factors. It was therefore inappropriate to use meta-analytical techniques for such a complex and varied group of practices. I am currently involved in a study to determine which factors are related to good diabetic care. The study involves data from 169 practices (639 general practitioners) in three different health authorities that had conducted an audit of patients with diabetes. Preliminary results show that there are wide variations in the process and outcome measures. Regression analysis indicates that having a recall system is not related to the process or outcome of care of patients with diabetes. Basing our decisions on evidence from randomised controlled trials is becoming increasingly acceptable in general practice. A randomised controlled trial with randomisation at practice level is required to assess the effectiveness of structured care with a prompted recall system. In the meantime the debate over who should deliver care to a selected group of diabetic patients will continue. Yours sincerely Dr Kamlesh Khunti Clinical Lecturer Department of General Practice and Primary Health Care Leicester General Hospital Gwendolen Road Leicester LE5 4PW Conflict of interest None References (1) Griffin S. Diabetes care in general practice: meta-analysis of randomised control trials. BMJ 1998;317:390-5. (2)Greenhalgh T. Commentary: Meta-analysis is a blunt and potentially misleading instrument for analysing models of service delivery. BMJ 1998;317:395-6. (3) Hurwitz B, Goodman C, Yudkin J. Prompting the clinical care of non-insulin dependent (type II) diabetic patients in an inner city area: one model of community care. BMJ 1993;306:624-30. (4) Diabetes Integrated Care Evaluation Team. Integrated care for diabetics: clinical, psychological, and economic evaluation. BMJ 1994;308:1208-12. (5)Khunti K, Baker R, Lakhani M, Rumsey M. Quality of care of patients with diabetes: Collation of data from multi-practice audits of diabetes in primary care. Family Practice (in press).