Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Selection bias in cluster trial
- David J Torgerson (19 February 2008)
-
Participation in research should be part of a GP's contract of employment
- Shaun Treweek, Frank Sullivan (19 February 2008)
-
DESMONDally disappointing
- Roger A Fisken (22 February 2008)
-
Informed diabetes care needs for the disadvantaged elderly in India
- R. K. Bansal, Naresh C Jain, Mukesh Kumar (26 February 2008)
-
Effects of Structured Educational Programme in diabetes: Transient Benefits?
- Sujoy Ghosh, Tarik Elhadd, Iqbal Malik (1 March 2008)
-
Better Regulation Using Carbohydrate and Insulin Education (B.R.U.C.I.E) - an intensive education programme for individuals with type 1 diabetes.
- Jane Cook, Ciara Heverin, Gail Blockley, Anne Reid, Susan Diamond, Sujoy Ghosh (1 March 2008)
-
DESMOND programme is not about effectiveness. Just a phase II study.
- Alain Braillon (1 March 2008)
-
Self-management interventions in type 2 diabetes: the need to look beyond HbA1c to integrated care
- Xavier Debussche (22 March 2008)
-
DESMOND randomised controlled trial - authors' reply
- Melanie J Davies, Simon Heller, Kamlesh Khunti, Chas Skinner, Marian Carey, Helen Dallosso (14 April 2008)
|
|
|||
|
David J Torgerson, Director, York Trials Unit University of York YO10 5DD
Send response to journal:
|
This study is a large cluster randomised controlled trial but suffers from recruitment bias due to poor allocation concealment. Allocation concealment is deemed to be vital in individually randomised trial and the same applies to cluster trials. In this study there was no allocation concealment from the people recruiting the participants, consequently there is a danger of recruitment bias. Indeed, the nature of the intervention - an educational package - would be likely to increase recruitment bias, which is a form of selection bias. In this study it is clear that this possibility has occurred. More of the intervention practices recruited patients and recruited more participants than the control practices. Even had they recruited similar numbers we could never be sure that the participants were similar in unknown characteristics. This design flaw has been pointed out in the past (Torgerson, 2001) and there are methods of dealing with this - such as using someone who is blind to the allocation and study hypothesis recruiting participants (Puffer et al, 2003). Given the major design flaws in this study we can only, at best, treat data from this study as good observational data. Unfortunately a systematic review of cluster trials published in the BMJ and other leading medical journals some years ago found that 40% of them suffered some form of bias, due to poor design (Puffer et al 2003). It seems that this poor design practice is still ongoing in BMJ reported cluster trials. Torgerson DJ. Contamination in trials: Is cluster randomisation the answer? British Medical Journal 2001;322:355-7. Puffer S, Torgerson DJ, Watson J. Evidence for Risk of Bias in Cluster Randomised Trials: A Review of Recent Trials Published in Three General Medical Journals. British Medical Journal 2003;327:785. Competing interests: None declared |
|||
|
|
|||
|
Shaun Treweek, Research Fellow Tayside Centre for General Practice, University of Dundee, DD2 4BF, Frank Sullivan
Send response to journal:
|
Davies and colleagues emphasise the generalisability of their structured group education programme for type 2 diabetes [1]. The trial was indeed large and recruited to target for its primary outcome; this is no mean feat. The authors do not, however, say how many practices were eligible for involvement in the trial, how many of these were approached, nor why 45 of the 207 that said yes subsequently referred no participants. The sample size calculation with its average of 18 participants per practice suggests that the intention was to involve far fewer practices; confirmation perhaps of the well-known observation that eligible patients have a tendency to evaporate once a trial begins [2]. And was the decision to give control practices money for longer consultations an acknowledgement that a standard care control would lead to very few referrals? Even with an ‘enhanced standard care’ control, these practices referred 133 patients fewer than the intervention practices. Studies looking at recruitment have identified some strategies that might help [2, 3]. Financial incentives can be effective, as can simple protocols and moving the consent process from clinicians to the research team. Despite this, such studies have struggled to find methods that can be generalised and most do not consider primary and secondary care separately. The update of the Cochrane recruitment review [3], which will consider primary care as a subgroup may help. Until and if this and other reviews provide a clear, empirical basis for designing recruitment strategies perhaps the UK should consider linking the potential recruitment benefit of financial incentives to the undoubted ability of the QoF to change priorities and behaviour [4]. Although there is much debate as to whether QoF actually improves quality [5] it does seem clear that if we want general practices to prioritise an activity then it needs to be part of the their contract of employment. The trick is in choosing the activities. Participation in research, especially trials, should be one of them. References 1. M J Davies, S Heller, T C Skinner, M J Campbell, M E Carey, S Cradock, H M Dallosso, H Daly, Y Doherty, S Eaton, C Fox, L Oliver, K Rantell, G Rayman, K Khunti on behalf of the Diabetes Education and Self Management for Ongoing and Newly Diagnosed Collaborative. Effectiveness of the diabetes education and self management for ongoing and newly diagnosed (DESMOND) programme for people with newly diagnosed type 2 diabetes: cluster randomised controlled trial. BMJ 2008 0: bmj.39474.922025.BEv1. 2. van der Woudena JC, Blankenstein AH, Huibers MJH, van der Windt DAWM, Stalman WAB, Verhagena AP. Survey among 78 studies showed that Lasagna’s law holds in Dutch primary care research. Journal of Clinical Epidemiology 2007; 60: 819-824. 3. Treweek S, Sullivan F, Pitkethly M, Jackson C, Wilson S, Kjeldstrøm M, Johansen M, Jones R, Cook J. Strategies to improve recruitment to randomised controlled trials. Update of Mapstone J, Elbourne D, Roberts I. Strategies to improve recruitment to research studies. Cochrane Database of Systematic Reviews 2007, Issue 2. Art. No.: MR000013. DOI: 10.1002/14651858.MR000013.pub3. 4. Roland M. Linking physician pay to quality of care: a major experiment in the UK. N Engl J Med 2004;351:1448-54. 5. Heath I, Hippisley-Cox J, Smeeth L. Measuring performance and missing the point? BMJ 2007; 335:1075-1076. Competing interests: None declared |
|||
|
|
|||
|
Roger A Fisken, Consultant Physician/Diabetologist Friarage Hospital, Northallerton
Send response to journal:
|
The authors and investigators of the DESMOND study are to be congratulated on their enthusiasm, diligence and integrity in the performance and reporting of the DESMOND study. It is very disappointing, though perhaps not surprising, that the intervention and control groups showed no difference in HbA1c or in any quality of life measure at one year. The difference of 1.1kg in weight, though statistically significant, is not, frankly, terribly impressive. Many of us were extremely alarmed when the Department of Health began to exert considerable pressure on diabetes services to adopt and implement DESMOND before any proper evaluation had been carried out; we now see that this approach is very unlikely to be cost effective in the management of the majority of patients with type 2 diabetes and the case for its universal introduction has effectively collapsed. The lesson here is that, just like drugs or surgical procedures, educational interventions need rigorous piloting and assessment, as was performed by the DESMOND group, before politicians jump on the bandwagon and insist that they are the answer to everybody's prayers. Competing interests: None declared |
|||
|
|
|||
|
R. K. Bansal, Professor Surat Municipal Inst. of Medical Education and Research, Surat- 395010, Gujarat, India, Naresh C Jain, Mukesh Kumar
Send response to journal:
|
The article by MJ Davies et al. raises important issues in context of India, where the number of diabetic patients and elderly are increasing. The edifice of modern medicine is based on the notions of a fully informed patient, who is an active partner in decision-making. The healthcare marketplace is typically characterised by a huge information asymmetry between patients and the providers (1). In a preliminary study among 50 disadvantaged people with diagnosed diabetes, seeking medical care in public facilities we found that none had received significant information on healthy lifestyles; education about nutrition, weight and exercise; regular monitoring and medications; and avoidance of complications such as blindness, amputation, heart disease, heart attack and stroke. These patients were taking irregular medications and were not undergoing regular monitoring. The preliminary results of another study among aged living in slums of Surat city has revealed undiagnosed diabetes among 13 of 121 surveyed elderly, whereas only 2 cases were diagnosed. These elderly avoid seeking medical care due to monetary constraints and when they did seek care diabetes had remained undiagnosed in busy outpatient settings of public hospitals. The 2 diagnosed diabetic patients reported receiving of cursory health advice limited to diet and injury prevention and spent 2 to 5 minutes with their doctor. They had avoided a follow up visit due to problems of long queues; referral to different investigation sites; high out of pocket expenses; wage loss; and rude or insensitive behaviour in public hospitals whereas private hospitals were too costly. The situation was somewhat better among those living in old age homes in Surat city. 13 (19.1%) of these 68 residents had diagnosed diabetes. They had reported of more satisfaction with their in-house medical facilities though with absence of any regular screening and health education services. They also avoided visiting public hospitals for similar reasons. Those seeking care at private facilities also reported receiving of scant health advice and cursory examination. The importance of patient information, shared care and decision- making is already well established in diabetes management minimise complications(2). Randomised Controlled Trials (3) suggest that patient education may reduce foot ulcerations and amputations and foot care knowledge and behaviour seem positively influenced by patient education in the short term. An important problem is the wide gap between practice recommendations and actual delivery of diabetes care in India(4). The drugs used in elderly diabetics could preclude drug safety, compromise quality of life and increase costs unnecessarily(5). Our study had revealed that expensive medical care was an important reason of neglect of elderly by their family members necessitating their stay at a home for the aged. Such a scenario has grim implications for the elderly, who comprise a disadvantaged group with restricted resources and limited physical mobility and need intervention. We feel that medical and health curriculum and training need modification to enhance health professional’s skill in communicating and advising elderly diabetic patients. The culture of informed patients and health education should be made a desirable trait rather than a task to be looked down upon and relegated to subordinate staff. There is a need to develop education and counselling services by nurses and dieticians for diabetic patients. Perhaps, the developing countries could gain from the present study of MJ Davies et al. References: 1. Bansal R.K., Kumar P. Informed patient care: the edifice of modern care. Indian Practitioner 2007; 60 (10): 641- 644. 2. Lenz ER, Mundinger MO, Hopkins SC, Lin SX, Smolowitz JL. Diabetes care processes and outcomes in patients treated by nurse practitioners or physicians. Diabetes Educ 2002; 28 (4):590-8. 3. Valk GD, Kriegsman DM, Assendelft WF. Patient education for preventing foot ulceration. Cochrane Database Syst Rev 2001; (4): CD 001488. 4. Nagpal J, Bhartia A. Quality of diabetes care in the middle- and high- income group populace: the Delhi Diabetes Community (DEDICOM) survey. Diabetes Care 2007; 30: e27. 5. Palatty PL, Lakshmi P, Revankar M, Udaykumar P. Demographic profile and drug utilisation study in geriatric patients with diabetes mellitus and hypertension, attending a tertiary care centre in Dakshina Kannada district. Antiseptic 2003; 100: 125-7. Competing interests: None declared |
|||
|
|
|||
|
Sujoy Ghosh, Clinical Teaching & Clinical Research Fellow The Ayr Hospital, Ayr, KA6 6DX, Scotland, United Kingdom, Tarik Elhadd, Iqbal Malik
Send response to journal:
|
Sujoy Ghosh, Tarik Elhadd, Iqbal Malik The Ayr Hospital, Scotland, United Kingdom We have read with interest the article by Davies et al. [1] Firstly we would like to congratulate the authors for undertaking a study like this which looks at structured group education for patients with newly diagnosed type 2 diabetes. It is indeed encouraging to note that patients were more likely to quit smoking and lose weight if they underwent intensive structured education and that the benefit persisted even at 12 months. However we note that at 12 months there was no (sustained) difference in levels of physical activity and triglyceride levels (? surrogate marker of dietary pattern) in the 2 groups. Now that is worrying! The main purpose of a structured educational programme is to lead to not only a change in the attitude and perception of the patient towards diabetes but also to lead to sustained behavioural changes (including changes in levels of physical activity and dietary habits). We are concerned that the intensive education probably failed to produce any long term changes in levels of physical activity and dietary habits and that the benefits of the education programme is probably only transient. It would be interesting to note what happens to the two groups in the years to follow and we would encourage the researchers to report the same in a follow up study. Reference: [1] M J Davies, S Heller, T C Skinner, M J Campbell, M E Carey, S Cradock, H M Dallosso, H Daly, Y Doherty, S Eaton, C Fox, L Oliver, K Rantell, G Rayman, K Khunti on behalf of the Diabetes Education and Self Management for Ongoing and Newly Diagnosed Collaborative.Effectiveness of the diabetes education and self management for ongoing and newly diagnosed (DESMOND) programme for people with newly diagnosed type 2 diabetes: cluster randomised controlled trial. BMJ 2008; 0: bmj.39474.922025.BEv1 Competing interests: None declared |
|||
|
|
|||
|
Jane Cook, Diabetes Specialist Nurse Diabetes Centre, The Ayr Hospital, Ayr, KA6 6DX, Scotland, United Kingdom, Ciara Heverin, Gail Blockley, Anne Reid, Susan Diamond, Sujoy Ghosh
Send response to journal:
|
Jane Cook, Ciara Heverin, Gail Blockley, Anne Reid, Susan Diamond, Sujoy Ghosh Diabetes Centre, The Ayr Hospital, Ayr, KA6 6DX, Scotland, United Kingdom We have found it interesting and encouraging to read the article by Davies et al. [1] We would like to thank the authors for highlighting the need for intensive education for people living with diabetes. It has been well documented and recognised that people with diabetes benefit from structured education and intensive treatment plans. [2] [3] In line with other structured education programmes we have developed Better Regulation Using Carbohydrate and Insulin Education (B.R.U.C.I.E) a one day training programme on carbohydrate counting. This is a randomised control study of 52 weeks of patients attending the diabetes centre at Ayr Hospital. A total of 36 participants will be recruited for the treatment arm of the study and 36 participants will be recruited as a control group. Both structured education and intensive treatment require many hours of training and education for health care professionals and individuals with diabetes. The DAFNE trial (Dose Adjustment for Normal Eating) involved a 5 day programme of education. Consequently the Diabetes Control and Complications trial involved many clinic and telephone contacts to intensify treatment. [2] [3] We have restricted resources locally and following focus groups within our clinic it was recognised that a one day training programme would be both feasible and well accepted by our clinic population. BRUCIE is a one day training programme for groups of 4-6 individuals with follow-up for one year (telephone and clinic contacts). Individuals will be seen as part of a group it has been recognised that diabetes education delivered as a group session is equally effective as individualised education (Rickheim et al 2002). Group-based education can invite greater interaction than individual approaches. [4] We are nearing the end of recruitment for the study and so far have provided education for 30 individuals with type 1 diabetes. Informal feedback so far has been positive. Data collection for the study will be complete March 2009. Reference: [1] MJ Davies, S Heller, TC Skinner, MJ Campbell, ME Carey, S Cradock, HM DAllosso, H Daly, Y Doherty, S Eaton, C Fox, L Oliver, K Rantell, G Raymen, K Khunti on behalf of the Diabetes Education and Self Management for Ongoing and Hely Diagnosed Collaborative. Effectiveness of the diabetes education and self management for ongoing and newly diagnosed (DESMOND) programme for people with newly diagnosed type 2 diabetes: cluster randomised control trial. BMJ 2008; 0: bmj.39474.922025.BEv1 [2] DAFNE Study Group. Training in flexible, intensive insulin management to enable dietary freedom in people with type 1 diabetes: dose adjustment for normal eating (DAFNE) randomised control trial. British Medical Journal, 2002 325, pp746. [3] DCCT Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependant diabetes mellitus. New England Journal of Medicine, 1993 329, pp 977-86 [4] Mensing CR, Norris SL. Group education in diabetes: effectiveness and implementation. Diabetes Spectrum, 2003 16, pp96-103. Competing interests: None declared |
|||
|
|
|||
|
Alain Braillon, Public Health University hospitals. 80000 Amiens. France
Send response to journal:
|
Efficacy and effectiveness relate to one of the most important questions in medicine, they express distinctly different concepts that are often confused. Efficacy is high on internal validity but at the expense of generalisability; effectiveness is high on external validity but at the expense of careful controls. A treatment is efficacious when it proves to be superior to placebo or best to another treatment of known efficacy. Whether an intervention works in routine clinical care relates to effectiveness.1 DESMOND programme is limited to: a) to short term (12 months) and sustaining adherence in behavioural interventions over a long term is more than challenging; 2) surrogate endpoints without change in glycated hemoglobin levels.2 Moreover, the control group was a placebo group which is ethically and scientifically unjustified : Non specific interventions may have more effects than behavioural interventions. Indeed, a trial testing adherence to specific lifestyle interventions among Pima Indians of Arizona showed than structured activity and nutrition interventions resulted in more weight gain (2.5 kg vs 0.8 kg, p = 0.06) than unstructured activities emphasizing history and culture.3 This contrast with Steno2 (intensive-therapy targeting a glycated hemoglobin level less than 6.5%): a 8 years follow up and an effect on mortality and morbidity.4 Comparing DESMOND programme and Steno2, I find a negative correlation between clinical pertinence of the results and marketing aggressiveness of the title. In conclusion, don’t care diabetic patients, cure them! 1 Fletcher RH, Fletcher SW, Wagner EH. Clinical epidemiology – the essentials. Baltimore: Williams & Wilkins, 1982. 2 Davies MJ, Heller S, Skinner TC et al. Effectiveness of the diabetes education and self management for ongoing and newly diagnosed (DESMOND) programme for people with newly diagnosed type 2 diabetes: cluster randomised controlled trial. BMJ, doi:10.1136/bmj.39474.922025.BE (published 14 February 2008) 3 Narayan KM, Hoskin M, Kozak D, Kriska AM et al. Randomized clinical trial of lifestyle interventions in Pima Indians: a pilot study. Diabet Med. 1998;15:66-72. 4 Gæde P, Lund-Andersen H, Parving HH, Pedersen O. Effect of a multifactorial intervention on mortality in Type 2 diabetes. N Engl J Med 2008,358:580-591. Competing interests: None declared |
|||
|
|
|||
|
Xavier Debussche, Diabetologist, Head of Department CHR de la Réunion, Saint Denis, Reunion Island
Send response to journal:
|
The study reported by Davies et al. of a practicable, structured self -management education programme offered at the primary care level to newly diagnosed type 2 diabetes patients raises some crucial questions: - The principal benefit is a significant reduction in the risk of cardiovascular complications as measured by the UKPDS score. Unfortunately, this score was not chosen as an outcome measure in the study. In any case, efforts that focus only on changing HbA1c in type 2 diabetes do not offer an adequate strategy for the effective prevention of complications (Turner et al, 1996). In this study, as in most other clinical trials, the choice of HbA1c as a primary outcome measure meant that the effects on the risk of complications were not assessed. - Although the study does not address the long-term effects of the intervention, the results are sufficiently encouraging to justify continuing the programme. They include significantly increased weight loss, overall reduction of cardiovascular risk, increased assumption of personal responsibility, improved health beliefs, and a decrease in depression scores in the intervention group. - The absence of a significant positive effect on Quality of Life may be a function of the measure employed (WHOQOL-BREF). This has been validated for use with people who suffer from type 2 diabetes, but its sensitivity as an outcome measure in prospective studies is questionable. - The integration of self-management into the care process at the primary care level (van Dam et al, 2003) is essential, as is the concomitant engagement of health care professionals and their input to planned care based on a systemic approach (Kimura et al, 2008). The results of this study should be seen in the context of integrated approaches to ongoing care in the UK. This helpful study makes a valuable contribution in that it draws attention to the divergence between the need to provide practicable, well- structured education as soon as the diabetes is first recognised and the fact that the vast majority of patients do not benefit from it until the onset of complications or the first hospitalisation. In the early stages, the effectiveness of self-management programmes cannot be measured only in terms of the observed effects on HbA1c, primarily because the anti-diabetic treatment usually achieves effective glycaemic control in the short term. Management of diabetes by means of a narrow prescriptive approach does not prevent complications in the long run. An adaptive response by health systems and health professionals to the burden of chronic diseases, particularly diabetes, must combine and integrate cure and care. Kimura J, DaSilva K, Marshall R. Population management, systems-based practice, and planned chronic illness care: integrating disease management competencies into primary care to improve composite diabetes quality measures. Disease Management, 2008, 11, 13-22. Turner, R., Cull, C., Holman, R. United Kingdom Prospective Diabetes Study 17: A 9-Year Update of a Randomized, Controlled Trial on the Effect of Improved Metabolic Control on Complications in Non-Insulin-dependent Diabetes Mellitus. Ann Intern Med, 1996, 124, 136-145. van Dam, H. A., van der Horst, F., van den Borne, B., Ryckman, R., Crebolder, H. Provider-patient interaction in diabetes care: effects on patient self-care and outcomes: A systematic review. Patient Education and Counseling, 2003, 51, 17-28 Competing interests: None declared |
|||
|
|
|||
|
Melanie J Davies, Professor of Diabetes Medicine University of Leicester LE1 5WW, Simon Heller, Kamlesh Khunti, Chas Skinner, Marian Carey, Helen Dallosso
Send response to journal:
|
Dear Editor In response to recent correspondence related to the DESMOND randomised controlled trial (1), we are well aware of the issues in recruitment bias in cluster trials and that Professor Torgerson's issues (Rapid Response 19 February 2008) have been discussed previously (2). In all trials there is a balance between practicality and perfection, particularly in recruiting patient participants. The DESMOND trial successfully recruited to target and is representative of eligible patients, unlike many other similar trials. Since patients in DESMOND were recruited through practices which were intervention or control it is hard to see how one could use a recruiter who was blind to the study hypothesis and allocation. We noted the differential recruitment in the trial and carried out sensitivity analyses to check our conclusions were robust. Since the outcomes were remarkably similar in both groups, and the results were unchanged by the sensitivity checks, we remain confident in our conclusions. As investigators of the DESMOND study, we share Dr Fisken’s disappointment (Rapid Response 22 February 2008) that the greater actual drop in HbA1c in the DESMOND intervention group did not survive the adjustments of the cluster analysis. However, we agree with Dr Debussche’s comments (Rapid Response 22 March 2008) on the limitations of HbA1c as an outcome measure, and welcome those of Dr Sean Dinneen (3) on the effect of the UK’s Quality and Outcomes Framework (QoF) in Primary Care on HbA1c levels in people newly diagnosed people with diabetes. His pertinent observations highlight the need for researchers to look beyond HbA1c as the traditional marker of patient improvement. All the more reason to regard the greater levels of weight loss and smoking cessation, and lower levels of depression, all sustained to 12 months, seen as a result of the DESMOND intervention, to represent a satisfactory conclusion, particularly as these are areas which treatment to QoF targets does not affect. Ghosh and colleagues (Rapid Response 1 March 2008) question the sustainability of the DESMOND programme on behavioural and biomedical outcomes. The DESMOND trial reflects evidence from systematic reviews that self-management education improves outcomes in the short term, but that regular reinforcement is required to sustain these.(4) A rolling programme of group education delivered to individuals with Type 2 diabetes in a secondary care setting has led to sustained improvements in biomedical outcomes.(5) Though our published DESMOND trial only refers to initial education for people newly diagnosed with Type 2 diabetes, we have envisaged that these patients will require ongoing education and support in the long term. We have recently completed a Foundation Module for those with established diabetes, and are currently developing the ongoing education for long term support which will be integrated into routine primary care. Meanwhile, a 3-year follow-up of patients in this randomised controlled trial will give us additional information on the effects to be expected of a single intervention in newly diagnosed patients. Whatever opinions on the DESMOND trial might be, it addresses our knowledge gap on the effect of structured education on people newly diagnosed with Type 2 diabetes, and has shown the acceptability of, and positive results such education can have, when offered as part of routine diabetes care. (1) Davies MJ, Heller S, Skinner TC, Campbell MJ et al, on behalf of the DESMOND Collaborative. Effectiveness of a structured group education programme on individuals newly diagnosed with Type 2 diabetes: a cluster randomised controlled trial of the DESMOND programme. BMJ, 2008;336:491- 495 (2) All that clusters...' MJ Campbell Society for Social Medicine and the International Epidemiological Association European Group, Oxford 2001. (3) Dinneen SF. Structured education for people with type 2 diabetes. BMJ 2008; 336, 459-460 (4) Norris SL Engelgau MM, Narayan KM. Effectiveness of self-management training in type 2 diabetes: a systematic review of randomized controlled trials. Diabetes Care 2001;24:561-587 (5) Trento M, Passera P, Borgo E, Tomalino M, Bajardi M, Cavallo F, et al. A 5-year randomized controlled study of learning, problem solving ability, and quality of life modifications in people with Type 2 diabetes managed by group care. Diabetes Care 2004; 27:670-675. Competing interests: None declared |
|||