Poor adherence to treatment is very prevalent in patients with diabetes, and varies according to the type of nonadherence being measured, and across the range of self-care behaviours that are components of treatment. Thus prevalence rates should be assessed by type of behaviour. In addition, prevalence rates may vary by diabetes subtype (i.e. type 1, type 2 or gestational), and also appear to be influenced by other factors such as age, gender and level of complexity of the treatment regimen. The rate of adherence, or the variables affecting adherence, may vary according to nationality, culture or subculture. Therefore, these factors should also be taken into account when assessing the prevalence of adherence in populations of patients with diabetes.

The lack of standard measurements prevents comparison being made between studies and across populations. Much work needs to be done to develop standardized, reliable and valid measurement tools.

Data from developing countries concerning the prevalence and correlates of adherence in patients with diabetes are particularly scarce. The pressing need to undertake more research in developing countries is emphasized by the WHO estimates indicating that by 2025 the largest absolute increase in prevalence rates of diabetes worldwide will occur in developing countries. Patients and providers of care in developing nations face additional barriers to achieving adequate diabetes self-care because of poverty, inadequate systems for delivering health care, and a host of other priorities that compete for national and individual attention.

More research is needed on adherence in women with gestational diabetes, and in study populations that include minorities and ethnic groups. Also, cross-cultural comparison studies should be encouraged. However, when making comparisons between different ethnic groups or countries, a number of aspects should be taken into account and controlled for, including types of health care system, health care coverage and socioeconomic macro- and micro-factors, as well as language and cultural differences. Adequate translation and validation of study measurements are required when using questionnaires developed in another country.

It is also important to point out not only the large number of factors that affect adherence behaviours in patients with diabetes, but also that the complex interactions that take place between them affect both adherence and metabolic control. Multivariate approaches to data are required to obtain more accurate representations of the relevant predictors and correlates.