Population stratification in type 2 diabetes mellitus: A systematic review

Aims There is increasing interest in using stratification in type 2 diabetes to target resources, individualise care and improve outcomes. We aim to systematically review and collate literature that has utilised population stratification methods in the study of adults with type 2 diabetes; and to describe and compare stratification methodologies, population characteristics, variables used to stratify and outcome variables. Methods The MEDLINE, EMBASE, CINAHL and Cochrane databases were searched from inception to July 2020. Studies included adults with type 2 diabetes using population stratification methods. The review protocol was registered on PROSPERO (ID: CRD42020206604) and conducted in line with PRISMA guidance. Extracted data included study aims; study setting (primary or secondary care); population characteristics; stratification variables and outcomes; and methodological approach to stratification. Results Across 348 included studies, there were a total of 10,776,009 participants with a mean age of 61.0 years (SD 5.94). 6.7% of studies used data‐driven methods and the rest employed expert‐driven approaches using pre‐defined stratification criteria. The commonest variable used to stratify populations was HbA1c (n = 57, 16.4%); few studies stratified using clinically important non‐traditional variables such as health behaviours and beliefs. Conclusions Most studies performing population stratification in type 2 diabetes used expert‐driven approaches with the aim of predicting outcomes in glycaemic control, mortality and cardiovascular complications. We identified relatively few studies using data‐driven approaches, which offer opportunities generate hypotheses beyond current expert knowledge. We describe important research gaps including stratification with regard to disease remission.