Does A1c consistently reflect mean plasma glucose?

Background: A1c, a surrogate measure of glycemic control, is known to have a strong linear correlation with mean plasma glucose (MPG) when analyzed in populations of patients. However, clinically significant intersubject variability in this relationship exists, which suggests that A1c measurements may not reflect actual glycemic control in some patients. In the present study we explored the extent to which A1c accurately represents glycemic control, as measured by MPG, for individual patients. Methods: Data were pooled from randomized clinical trials in which A1c and self-monitored plasma glucose (SMPG) profiles were collected by patients with Type 2 diabetes treated with insulin analog regimens. MPG levels were calculated from SMPG profiles. Distributions of MPG were analyzed for patients within similar ranges of A1c (< 6.5%, 6.5%- < 7.5%, 7.5%- < 8.5%, 8.5%- < 9.5%, and >= 9.5%) and distributions of A1c were analyzed in patients within similar ranges of MPG (< 6.1, 6.1- < 7.8, 7.8- < 9.4, 9.4- < 11.1, and >= 11.1 mmol/L). Results: Substantial proportions of patients had clinically significant differences between A1c and MPG. For example, among 260 patients with A1c between 6.5% and 7.5%, 10% had MPG levels < 6.4 mmol/L, whereas 10% had MPG > 9.5 mmol/L. Among the 224 patients with MPG levels >= 6.1 mmol/ L and < 7.8 mmol/ L, 10% had A1c < 6% and 10% had A1c > 8.1%. Conclusions: In the absence of SMPG, A1c may inadequately represent glycemic control for many diabetic patients.