Association of HbA 1c levels with vascular complications and death in patients with type 2 diabetes: Evidence of glycaemic thresholds

Aims/hypothesis: There is conflicting evidence regarding appropriate glycaemic targets for patients with type 2 diabetes. Here, we investigate the relationship between HbA _1c and the risks of vascular complications and death in such patients. Methods: Eleven thousand one hundred and forty patients were randomised to intensive or standard glucose control in the Action in Diabetes and Vascular disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE) trial. Glycaemic exposure was assessed as the mean of HbA _1c measurements during follow-up and prior to the first event. Adjusted risks for each HbA _1c decile were estimated using Cox models. Possible differences in the association between HbA _1c and risks at different levels of HbA _1c were explored using linear spline models. Results: There was a non-linear relationship between mean HbA _1c during follow-up and the risks of macrovascular events, microvascular events and death. Within the range of HbA _1c studied (5.5-10.5%), there was evidence of 'thresholds', such that below HbA _1c levels of 7.0% for macrovascular events and death, and 6.5% for microvascular events, there was no significant change in risks (all p>0.8). Above these thresholds, the risks increased significantly: every 1% higher HbA _1c level was associated with a 38% higher risk of a macrovascular event, a 40% higher risk of a microvascular event and a 38% higher risk of death (all p<0.0001). Conclusions/interpretation: In patients with type 2 diabetes, HbA _1c levels were associated with lower risks of macrovascular events and death down to a threshold of 7.0% and microvascular events down to a threshold of 6.5%. There was no evidence of lower risks below these levels but neither was there clear evidence of harm. Trial Registration:: ClinicalTrial.gov NCT00145925 Funding:: Servier and the National Health and Medical Research Council of Australia (project grant ID 211086 and programme grant IDs 358395 and 571281)