The prevalence of Type 2 diabetes is predicted to rise over the next few decades which will pose a major public health challenge [1]. Epidemiologic data have shown that the prevalence of Type 2 diabetes varies with ethnic origin. People originating from the Indian subcontinent ('South Asians') who have settled in other parts of the world seem to be particularly susceptible to Type 2 diabetes [2–6]. Type 2 diabetes is up to four times more common in British South Asians than in the indigenous white population; in the UK one in five older South Asians have diabetes [7]. It is recognized that good education is a keystone in diabetes management, and it is worrying that some studies of different South Asian communities in Britain have shown that they generally know less about diabetes and its management than comparable white subjects [8–10]. Our initial aim was to assess baseline levels of knowledge, attitudes, and some aspects of self-management of diabetes care in a representative sample of South Asian people with Type 2 diabetes in Glasgow, UK. Then we used the baseline results to help develop a culturally appropriate (competent) educational intervention programme for these groups. We then investigated whether this intervention could produce an improvement, and finally whether any improvement was greater than background changes in knowledge in comparison groups.

Of the 145 (including the white group), only one hundred and one could be re-interviewed at 6 months. At the time of follow-up, out of 145 patients six (4.1%) were dead, four were hospitalised, twelve refused to participate in the second questionnaire, and 25 (17.2%) patients were unavailable. Some of these patients had returned to their home country but the proportion was unknown. Details are given in figure 1. To assess the possibility of a selective bias introduced by loss of patients the dropped out group and the group who completed the study were compared for their baseline findings. No differences were found.

Overall, the results of educational interventions aimed at patients with Type 2 diabetes are difficult to interpret. The diabetes education intervention for ethnic minority groups in Glasgow demonstrated that South Asians with Type 2 diabetes could improve their knowledge and change their attitudes about Type 2 diabetes. The results showed an increase in diabetes knowledge and changes of attitudes, particularly regarding seriousness of Type 2 diabetes in the intervention group. However there was also an improvement in the ethnic control group during the period of the study, and also in the white comparison group. This finding is not unusual, as other studies have shown similar effects in control groups [12–14].

There are several reasons why the knowledge, attitudes, and practice of diabetes in the control group have improved. We looked at potential biases. Those who dropped out from the ethnic group might have been different from those who stayed within the study. However, our analysis showed no difference in starting scores in those members of the ethnic control group who stayed in the study compared to those who dropped out. Some increased knowledge during the study in the control group subjects, and the white comparison group, might have resulted from merely completing the study questionnaires. Subjects' questions were answered frankly during the interview, regardless of whether subjects were in the intervention or the control group. Completing the questionnaire may have had an effect on its own. It was not considered ethical to withhold information from the control group subjects during data collection and all questions were answered. In addition, the control groups had frequent subsequent interactions with their health care providers (such as their GP or practice nurse) following the initial assessment and may have been stimulated by the study to ask further questions from their carers.

One of the issues which arises in this kind of study is contamination. This might happen particularly in day care centres where people are used to meeting weekly. The day care centres provided a good opportunity for people to discuss and exchange information with each other. Therefore, after the education sessions for the intervention groups, subjects in the intervention groups may talk to their friends in the control group and may have discussed some of the issues raised by the questionnaire. We could not control this, but in the GP practices such contamination was much less likely.

Another possible explanation might be the 'reactive (Hawthorne) effect'. This refers to the effect of being studied upon those who know they are in a study. Their knowledge of the study may influence their behaviour (they may become more interested in the topic, pay more attention to it and become biased), or they may change their behaviour simply because someone (the investigator) is taking an interest in them. When people feel that they are being tested, they feel the need to create a good impression. Alternatively if the study stimulates interest not previously felt for the topic under investigation, this may distort the results.

Finally translation is a significant issue for interviewing with non-English speakers. Although the translators were asked not to interpret any question and add much from their own views, the accuracy and validity of this procedure was difficult to measure. Translation for ethnic minority groups with different languages is not just a word by word translation. Therefore, the effect of the translator on the subject's answers and subsequent knowledge might be a further confounding factor.

The major question that must be asked is whether the intervention tested in this study was effective. It was for all four scores. However there was 'background' improvement in the ethnic control group which may have had contamination in the day care settings from the intervention group. There was also improvement in the white comparison group, all from a general practice setting where confounding was less likely, indicating that there was some improvement in knowledge in the wider population with diabetes in Glasgow at least. Intervention studies of this type are always subject to influences outwith the control of the investigators. The relative effect of the intervention might not be significant, but the absolute effect was that knowledge improved in the ethnic population with diabetes, regardless of how it was produced or induced. That improvement in understanding should lead in the end to better diabetic control.

This study has provided useful information about Type 2 diabetes amongst South Asians in Glasgow. Based on the findings and experience gained through the study the following recommendations are made for diabetes care and for further research. Since the UK is a multi-cultural country, culturally competent behavioural interventions should be the focus of major national initiatives. Future research should be aimed at developing culturally appropriate outcome measures, addressing translation issues for non-English speaking populations, and exploring motivating factors and strategies for diabetes self-management. Consideration needs to be made about: how to involve people who would benefit most from diabetes education interventions in community gathering places and in the home; optimal intervention intensity; ideal providers; and how to integrate diabetes education with primary care. Further research is needed to determine the effect of diabetes educational interventions on other outcomes such as complication rates, lifestyle, cost, and quality of life.