In older people, the association between diabetes medication group and hypoglycaemia, cardiovascular disease, and mortality: prospective primary care-based cohort study 2010-2016

Problem

There is uncertainty about the long-term benefits and risks of diabetes treatment in older people. We aimed to determine whether in older people there are differences in hypoglycaemia, cardiovascular disease (CVD) events or mortality according to diabetes medication regimen.

Approach

Prospective cohort study. People in the Auckland and Northland regions of New Zealand, aged 65 years or older dispensed insulin and/or oral diabetes medications in 2010 with a baseline HbA1c measurement. The cohort was stratified into four sub-cohorts by medication group: metformin only, metformin plus other oral hypoglycaemic agents, other oral hypoglycaemic agents only, and any insulin. Participants were followed to the end of 2016 using linked national hospitalisation and mortality data. ICD-10 coded outcomes were hypoglycaemia-associated hospitalisation, fatal and non-fatal CVD, and all-cause mortality. The time to first event was analysed with Cox models adjusted for age, sex, ethnicity, socioeconomic deprivation, baseline HbA1c, prior CVD and prior diabetes hospitalisation, and modified Charlson comorbidity index.

Findings

Of the 18,099 participants, at baseline 7669 (42%) were on metformin only, 4842 (27%) were on metformin and other oral agent/s, 1922 (11%) were on other oral agent/s only, and 3666 (20%) were on insulin. During follow-up, 16% experienced hypoglycaemia, 36% CVD and 31% died (of whom half died from CVD causes). The risk of hypoglycaemia was associated with a high baseline HbA1c 70+mmol/mol, increasing age, Māori and Pacific ethnicity, increasing deprivation, prior CVD, comorbidity burden and all medication groups compared to metformin-only with insulin having an adjusted hazard ratio 11.9, 95% CI 10.3-13.8. The risk of the other outcomes was also associated with age, Māori ethnicity, deprivation, comorbidity burden, and all medication groups compared to metformin-only although the magnitude of the difference between insulin and other oral/s was much less. There was marked separation in the adjusted survival curves by medication group for hypoglycaemia but not for the other outcomes.

Consequences

In older people, after adjusting for multiple risk factors, insulin and any other oral medications compared to metformin-only were associated with long-term increased risk of hypoglycaemia-associated hospitalisation with insulin being the strongest predictor. While associated with increased risk of fatal and non-fatal CVD and all-cause mortality, the impact of medication group compared to metformin-only was similar. These findings lead us to question the use of these second-line diabetes medicines in older people.