Selectivity of beta-blockers, cardiovascular and all-cause mortality in people with hypoglycaemia: An observational study.

ABSTRACT

BACKGROUND AND AIMS: The association of beta-blockers and their selectivity with mortality and cardiovascular events in patients with and without hypoglycaemia is unknown. METHODS AND RESULTS: Insulin-treated patients with diabetes were identified within the UK CPRD database. All-cause deaths, cardiovascular events, and hypoglycaemic episodes were captured to assess the interaction between beta-blocker therapy and selectivity with hypoglycaemia. 13,682 patients, of which 2036 (14.9%) with at least one hypoglycaemic episode, were included; 3148 deaths and 1235 cardiovascular events were recorded during a median of 2.3 and 4.7 years in patients with and without incident hypoglycaemia, respectively. Treatment with any beta-blocker was not associated with risk of death in both patients with and without hypoglycaemia, without significant interaction. Compared to no therapy, non-selective beta-blockers were associated with higher risk of death in patients without hypoglycaemia (hazard ratio (HR) 2.93 [1.26-6.83] in the fully adjusted model) but not in those with hypoglycaemia; interactions was not significant. For beta1-selective beta-blockers, there was no association with mortality in both patients with and without hypoglycaemia, without significant interaction. After missing data imputation, results were consistent for non-selective beta-blockers (HR in patients without hypoglycaemia 1.59 [1.22-2.08]) while indicated a reduced risk of death for beta1-selective beta-blockers in patients with hypoglycaemia (HR 0.76 [0.61-0.94]). Due to few cardiovascular events, complete-case analysis compared only any vs no beta-blocker therapy and indicated no associations with therapy or interaction by hypoglycaemia. CONCLUSION: In patients with hypoglycaemic episodes, treatment with beta1-selective beta-blockers may potentially reduce the risk of death. These explorative findings and the potential role of confounding by indication need to be evaluated in other studies.