Severe hypoglycaemia and absolute risk of cause-specific mortality in individuals with type 2 diabetes: a UK primary care observational study

Abstract Aims/hypothesis Several pathophysiological mechanisms would suggest a causal link between hypoglycaemia and cardiovascular death; conversely, current knowledge would not support a causal relationship with other causes of death. To clarify the nature and the magnitude of the association between hypoglycaemia and death, we investigated the 5 year mortality risks for cardiovascular disease, cancer and other causes in individuals with type 2 diabetes admitted to hospital for a severe hypoglycaemic episode. Methods We defined in the UK Clinical Practice Research Datalink database a prevalent cohort of adults with type 2 diabetes diagnosed between 1 January 1998 and 1 January 2011 (index date), with available linkage to the Office for National Statistics (ONS) and the Hospital Episode Statistics (HES). A hospital admission reporting hypoglycaemia as the underlying cause was identified before the index date in the HES; date and underlying cause of death were obtained from the ONS. We quantified the 5 year risk of cause-specific death in people with and without admission to hospital for severe hypoglycaemia, adjusting for potential confounders and accounting for competing risk. Results Of the 74,610 subjects included in the cohort, 388 (0.5%) were admitted at least once for a severe hypoglycaemic episode; subjects admitted were older, with higher HbA 1c and a greater prevalence of multimorbidity. During a median follow-up of 7.1 years, 236 (60.8%) and 18,539 (25.0%) deaths occurred in subjects with and without a previous severe hypoglycaemia, respectively. Non-cardiovascular causes accounted for 71% of all deaths in both subjects with and without hypoglycaemia. In a 60-year-old person with severe hypoglycaemia, the 5 year absolute risk of death, adjusted for age, sex, ethnicity, systolic blood pressure, total cholesterol, HbA 1c , BMI, eGFR, smoking status, alcohol consumption and deprivation (Townsend score), was 6.6%, 1.1% and 13.1% for cardiovascular, cancer and other causes, respectively, while the 5 year absolute risk difference compared with a subject without severe hypoglycaemia was 4.7% (95% CI 1.0, 8.3) for cardiovascular, −1.4% (−4.1, 1.4) for cancer and 11.1% (6.1, 16.1) for other causes of death. Results were consistent in models further adjusted for medications and comorbidities (myocardial infarction, stroke, peripheral artery disease, heart failure, atrial fibrillation, cancer), with sulfonylurea and insulin associated with increased mortality rates (from cause-specific hazard ratio of 1.06 [95% CI 0.99, 1.14] for cancer death with use of sulfonylurea to 1.42 [1.29, 1.56] for cardiovascular death with use of insulin). Results were robust to missing data. Conclusions/interpretation The results of this study indicate severe hypoglycaemia as a marker of, rather than causally linked to, an increased risk of long-term mortality. Regardless of the nature of the association, a severe hypoglycaemic episode represents a strong negative prognostic factor in patients with type 2 diabetes.