Trends and predictors in all-cause and cause-specific mortality in diabetic and reference populations during 21 years of follow-up

Background Patients with diabetes mellitus (DM) have a markedly higher overall mortality from coronary heart disease (CHD), as well as many other causes of death like cancer. Since diabetes is a multisystem disease, this fact together with the increased lifespan among individuals with diabetes may also lead to the emergence of other diabetes-related complications and ultimately to diversification of the causes of death. Methods The study population of this observational historic cohort study consisted of subjects with DM, who had purchased for at least one insulin prescription and/or one oral antidiabetic between January 1, 1997 and December 31, 2010 (N=199 354), and a reference population matched by age, sex and hospital district (N=199 354). Follow-up was continued until December 31, 2017. All-cause and cause-specific mortality (cancer, CHD and stroke) was analysed with Poisson and Cox's regression. Associations between baseline medications and mortality were analysed using LASSO (Least Absolute Shrinkage and Selection Operator) models. Results The mortality rates were significantly elevated among the patients with DM. However, the relative risk of all-cause mortality between the DM and reference populations tended to converge during the follow-up. The lowering trend was most apparent in CHD mortality. The difference between DM and reference populations in stroke mortality vanished with a later entrance to the follow-up period. There were a few differences between DM and no-DM groups with respect to how baseline medications were associated with mortality. Conclusions The gap between the mortality of patients with diabetes compared to subjects who are non-diabetic diminished markedly during the 21-year period. This was driven primarily by the reduced CHD mortality.