Is HbA1c a valid surrogate for mortality in type 2 diabetes? Evidence from a meta-analysis of randomized trials

Abstract

Aims Hemoglobin A1c (HbA1c) has been repeatedly questioned as a valid surrogate marker, especially for patient-relevant outcomes. The aim of this study was to validate the HbA1c value as a surrogate for all-cause mortality in people with type 2 diabetes. Methods The effect estimates for HbA1c lowering after treatment as well as reductions in all-cause mortality of randomized trials were extracted from a systematic review and updated. For the measurement of actual surrogacy, weighted linear regression models with a random intercept for the study effect were used with the all-cause mortality estimate (risk difference and log relative risk) as the outcome and the estimate for HbA1c difference as the covariate. Surrogacy was assessed according to the criteria of Daniels and Hughes. Results A total of 346 HbA1c-mortality-pairs from 205 single randomized trials were included in the analysis. Regarding the risk difference of all-cause mortality, there was no evidence for surrogacy of the HbA1c value. For the log relative risk, a small positive association between HbA1c and the all-cause mortality estimate (slope 0.129 [95% confidence interval −0.043; 0.302]) was observed. However, there was no sign of valid surrogacy. Conclusions Based on the results of more than 200 randomized trials, HbA1c is not a valid surrogate marker for all-cause mortality in people with type 2 diabetes.