A1c Levels: Is Lower Always Better?

Gregory A. Nichols, PhD


April 23, 2010

Survival as a Function of HbA1c in People With Type 2 Diabetes: A Retrospective Cohort Study

Currie CJ, Peters JR, Tynan A, et al.
Lancet. 2010;375:481-489.

Study Summary

This study assessed the association between all-cause mortality and hemoglobin A1c (A1c) levels. Using electronic data from the General Practice Research Database in the United Kingdom, 27,965 patients who had recently switched from oral monotherapy to sulfonylurea plus metformin combination therapy were included in cohort 1 (SUMET), and 20,005 patients who initiated insulin after previous treatment with oral agents comprised cohort 2 (INS). The index date was the date of initiation of therapy that qualified the individual for the cohort. Patients were followed until death or until the qualifying therapy was changed. Because data were obtained from 1986 through 2008, some patients (n = 5588) were included in both cohorts. The cohorts were divided into deciles on the basis of the mean of all postindex A1c values. The primary outcome was all-cause mortality. Among patients with no record of previous cardiovascular disease, the occurrence of a major cardiovascular event was assessed as a secondary outcome. Cox regressions were used to determine the hazard ratios of mortality by A1c deciles (using the fourth decile of about 7.5% as the reference group) and to adjust for other risk factors, including age, sex, smoking, cholesterol, previous large-vessel disease, and the Charlson comorbidity index. Cox models were constructed for each cohort individually and for all patients combined.

Mean follow-up was 4.5 years in the SUMET cohort and 5.2 years in the INS cohort. A1c values in the lowest decile (median, 6.4%) were associated with an increased risk for mortality for all patients (hazard ratio [HR], 1.52; 95% confidence interval [CI], 1.40-1.64). This finding was stronger in the INS cohort (HR, 1.79; 95% CI, 1.45-1.22) than in the SUMET cohort (HR, 1.30; 95% CI, 1.07-1.58). Only the 10th decile (median, 10.4%) was also associated with increased mortality risk in the SUMET cohort (HR, 1.93; 95% CI, 1.55-2.42); but in the INS cohort, deciles 2 (median, 6.95%; HR,1.45; 95% CI, 1.17-1.80), 3 (median, 7.3%; HR, 1.35; 95% CI, 1.09-1.67), 9 (median, 9.4%; HR, 1.46; 95% CI, 1.21-1.77), and 10 (median, 10.6%; HR, 1.80; 95% CI, 1.49-2.17) were all associated with greater risk. The combined model yielded results similar to the INS cohort, and the inclusion of a variable for membership in the INS cohort was significantly associated with increased mortality (HR, 1.49; 95% CI, 1.39-1.59). The adjusted risk for progression to large-vessel disease had the same general U-shaped association as for all-cause mortality, and insulin treatment was associated with an increased risk for a first large-vessel disease event.


For many years, the American Diabetes Association has recommended that patients with diabetes attempt to keep their A1c levels below 7%. However, in 2008 the ACCORD study was stopped early because of an increased risk for mortality in the intensively treated group relative to the standard group (target A1c < 6% vs 7%-7.9%),[1] raising concerns that the goal of A1c < 7% might be too low. However, 2 other large clinical trials of intensive control did not confirm ACCORD's findings,[2,3] leading to the consensus that major changes in glycemic control targets were not needed.[4]

The results from Currie and associates further question the necessity of a lower limit for optimal glycemic control. The findings were strongest among insulin users, probably because most patients with type 2 diabetes will ultimately require insulin therapy as their disease progresses. Despite the investigators' efforts to control for confounders, patients in the insulin cohort were probably at greater risk for death. Unfortunately, the study could not determine cause of death, or even whether patients with lower A1c levels who died had experienced more frequent or more severe episodes of hypoglycemia. This, of course, is the hypothesized reason for the findings. Because that hypothesis cannot ethically be tested in a clinical trial, replicating these findings in an observational dataset that could account for hypoglycemia or other causes of death could be extremely important.



Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.