Background:
Few studies have examined the natural history of prediabetes, specifically, regression to normoglycemia as well as progression to clinical diabetes. Furthermore, there are several definitions of prediabetes in current use.
Methods:
We conducted a prospective cohort analysis of 8214 participants without diabetes who attended visit 2 (1990-1992) and visit 4 (1996-1998) of the Atherosclerosis Risk in Communities (ARIC) Study. We used multinomial logistic regression to compare cumulative incidence of diabetes and regression to normoglycemia among persons meeting different definitions of prediabetes (American Diabetes Association criteria): fasting glucose (FG) 100-125 mg/dL; A1C 5.7-6.4%; or elevations in both tests (single-sample confirmatory). Incident diabetes was defined as a self-reported diagnosis, medication use, or both FG ≥ 126 mg/dL and A1C ≥ 6.5%; normoglycemia was defined as both FG < 100mg/dL and A1C < 5.7% and no diabetes diagnosis or medication use. Analyses were adjusted for age, sex, and race-center.
Results:
The prevalence of prediabetes at visit 2 (age range, 46-69; 57% female; 17% black) was 37.6% (n=3089) based on FG criteria, 17.4% (n=1427) based on A1C criteria, and 10.6% (N=867) by both FG and A1C criteria. The 6-year cumulative incidence of diabetes was 7.2% in participants with FG-defined prediabetes, 17.1% in participants with A1C-defined prediabetes, and 22.4% in participants meeting both criteria. Among participants with prediabetes by FG criteria, regression to normoglycemia was 1.9 times as likely as progression to diabetes (14.0% vs. 7.2%). However, when prediabetes was defined by A1C and by confirmatory criteria, progression to diabetes was 1.5 times and 3.5 times more likely than regression to normoglycemia, respectively.
Conclusions:
Risks of progression to diabetes and regression to normoglycemia differed by prediabetes definition. Prediabetes identified by A1C criteria and single-sample confirmatory criteria identified people at highest risk for diabetes.