Long-term follow up of older people on diabetes medications: observational study using linked health databases.

There is uncertainty about the long-term benefits and risks of diabetes medications in older people. We investigated differences in hypoglycaemia, cardiovascular disease (CVD) or mortality in older people according to diabetes medication, using linked national hospitalisation and mortality data from New Zealand. Adults aged ≥65 years dispensed diabetes medication in 2010 with a baseline glycated haemoglobulin (HbA1c) level (n=18099, mean age 73 years, 50% female) were included and stratified into four groups: metformin-only (42%); metformin-plus-other-oral-hypoglycaemic/s (27%); other-oral/s-only (11%); and any-insulin (20%). Time to first event was analysed with Cox models adjusted for sociodemographic factors; clinical history (prior hospitalisation for diabetes or CVD, and comorbidities); glycated haemoglobin; and CVD medications. Over 7-year follow up, 16% of participants experienced hypoglycaemia, 36% a CVD event and 31% died. Compared with metformin-only, insulin and other oral hypoglycaemic/s were associated with five- to 10-fold long-term increased risk of hypoglycaemia, and increased risk of CVD and death although adjusted survival curves showed no important separation between medication groups for CVD and death with the possible exception of insulin. Although confounding by indication is unable to be eliminated, this study raises further questions about the use of second-line diabetes medications in older people.