Abstract
Medication non-adherence and non-persistence in type 2 diabetes (T2D) are common and associated with poor outcomes. Medication attributes have a major influence on adherence in chronic disease but it is not clear which medication classes are associated with better adherence and persistence in T2D. A systematic review and cohort analyses of comparative adherence and persistence across medication classes in T2D was performed. MEDLINE, Embase, The Cochrane Library, The Cochrane Register of Controlled Trials, PsycINFO, and CINAHL were searched for studies comparing class adherence and persistence. Where n>5 studies provided the same comparison a meta-analysis was performed. Retrospective cohort analyses used the Royal College of General Practitioners Research and Surveillance database to compare adherence and persistence with oral medications. The systematic review synthesised 66 studies (38 in meta-analyses). Adherence was better with thiazolidinediones (TZDs) and sulphonylureas than metformin. TZDs had slightly better adherence than sulphonylureas. Limited data suggest low adherence with alpha-glucosidase inhibitors (AGIs) and meglitinides. Insulin analogues had longer persistence than GLP1 analogues and human insulins. There was little data on comparative persistence with oral medications and adherence with injectable medications. Adherence and persistence were measured in cohorts of 60,327 and 145,546 people with T2D respectively (55,728 and 76,593 oral medications). After adjustment in regression models metformin and SGLT2 inhibitors had the longest persistence, AGIs and meglitinides the shortest. TZDs, DPP4 inhibitors, and SGLT2 inhibitors had the highest adherence, and metformin, AGIs and meglitinides the lowest. Sulphonylureas had intermediate adherence and persistence. Younger age, female gender, and non-white ethnicity were also associated with reduced persistence and adherence. Adherence and persistence vary considerable between classes of medication for T2D. Class switching may improve adherence in people with low adherence e.g. metformin to DPP4 inhibitors. Pre-emptive selection of classes which promote adherence in those with non-adherence risk factors may also improve outcomes.