What does make the difference?

In this issue, Deakin and colleagues [1] report the data from a randomized controlled trial assessment of the patient education programme developed by them for people with Type 2 diabetes. The authors are to be commended for a number of features of this trial that are not usually seen on this topic. Assessment of outcome was undertaken by individuals blind to group allocation, the final outcome measure is a year after completion of the programme (not a year after entry into the study) and the authors attempt to address issues around patient blinding by the control group receiving an additional 55 min of contact time with primary care staff. Furthermore, the effects seen, although small, were achieved in individuals whose glycaemic control was not particularly poor. This design is noticeably more robust than other studies undertaken recently and which have been used to argue for the wholesale adoption of a programme, such as the dose adjustment for normal eating (DAFNE) trial in Type 1 diabetes [2], which had no control contact, and where advantages in the primary outcome at 6 months regressed between 6 and 12 months of follow-up assessment. Indeed, the diabetes education for self-management: ongoing and newly diagnosed (DESMOND) programme for Type 2 diabetes patients has, as yet, not even been compared with the usual treatment. Therefore, it would seem reasonable for Deakin and colleagues to argue that the X-PERT programme they have developed makes a difference for individuals with Type 2 diabetes. However, we must continue to apply the same critical analysis to self-management education research as we do to pharmacological interventions. The first problem that arises is that of the control group. The authors raise this issue themselves, and refer to a paper which they claim shows that, with equivalent contact time, a group programme does better than one-to-one education [3]. Unfortunately, in that study the one-to-one section had 5 h of contact and the group 7 h. The authors of that paper indicate that the extra 2 h was required to manage group dynamics and processes, which of course is one of the potential key mechanisms by which groups may achieve change. This is a perennial issue in self-management education which Deakin and colleagues have made an attempt to address, which is more than can be said for the DAFNE trial. However, without a psychologically equivalent placebo, we have no way of knowing that the effects of these group programmes is no more than the impact of intensive contact with highly motivated, charismatic and enthusiastic educators, and the well-documented therapeutic effects of groups through increased affiliation, social support and social comparison. Furthermore, there is an abundance of empirically supported programmes already available, such as the chronic disease self-management programme [4] the empowerment programmes of Anderson and colleagues [5] or, in Type 1 diabetes, the blood glucose awareness training course [6], or a simple-patient activating consultation [7]. Thus, in reality there is no longer an excuse for not having an adequate psychologically equivalent control group, especially with the advent of tools to check for expectancy effects now being argued as essential in psychological studies [8]. One way to argue against the contact and group affiliation mechanism of groups is to show that the intervention is working as it is theoretically modelled to work, phase 2 of the guidance issued by the Medical Research Council [9]. Deakin and colleagues have indicated that they have used some theory and a battery of measures to assess intermediate variables. Theories generate specific testable hypotheses about how change is achieved, but the empowerment model that is stated to underpin the programme is not in fact a theory but a philosophical standpoint based on existential truths [10] and so does not generate specific hypotheses about mechanisms of action. The second theory mentioned is that of discovery learning, but with no reference, for this it is not possible to know how this theory has guided the content or process of the intervention. Although more fundamentally is the fact that the authors provide us with no evidence that changes in the process measures, analysis of which is dubious because of the low numbers completing follow-up assessment, are related to concurrent outcome measures or change in outcome measures. With no previous modelling data to refer to, we are left with no evidence of the mechanism of action of the intervention. Thus, it would seem that, although Deakins and colleagues suggest that the X-PERT programme makes a difference, the evidence they produce is much the same as for other group programmes, including DAFNE and DESMOND. The data only provide compelling evidence that, when enthusiastic and motivated health-care professionals invest more time with patients in a structured way to help them manage their diabetes, we see improved biomedical outcomes.