Prediction of Type 2 Diabetes by Hemoglobin A1c in Two Community-Based Cohorts

Abstract

Objective: Hemoglobin A_1c (HbA_1c) can be used to assess type 2 diabetes (T2D) risk. We asked whether HbA_1c was associated with T2D risk in four scenarios of clinical information availability: 1) HbA_1c alone, 2) fasting laboratory tests, 3) clinic data, and 4) fasting laboratory tests and clinic data.

Research design and methods: We studied a prospective cohort of white (N = 11,244) and black (N = 2,294) middle-aged participants without diabetes in the Framingham Heart Study and Atherosclerosis Risk in Communities study. Association of HbA_1c with incident T2D (defined by medication use or fasting glucose [FG] ≥126 mg/dL) was evaluated in regression models adjusted for 1) age and sex (demographics); 2) demographics, FG, HDL, and triglycerides; 3) demographics, BMI, blood pressure, and T2D family history; or 4) all preceding covariates. We combined results from cohort and race analyses by random-effects meta-analyses. Subsidiary analyses tested the association of HbA_1c with developing T2D within 8 years or only after 8 years.

Results: Over 20 years, 3,315 individuals developed T2D. With adjustment for demographics, the odds of T2D increased fourfold for each percentage-unit increase in HbA_1c. The odds ratio (OR) was 4.00 (95% CI 3.14, 5.10) for blacks and 4.73 (3.10, 7.21) for whites, resulting in a combined OR of 4.50 (3.35, 6.03). After adjustment for fasting laboratory tests and clinic data, the combined OR was 2.68 (2.15, 3.34) over 20 years, 5.79 (2.51, 13.36) within 8 years, and 2.23 (1.94, 2.57) after 8 years.

Conclusions: HbA_1c predicts T2D in different common scenarios and is useful for identifying individuals with elevated T2D risk in both the short- and long-term.

Figure 1

Association of HbA_1c with incident T2D over two decades adjusted for other clinical predictors by cohort and race. In all four models, each 1%-unit increase in HbA_1c was associated with a 2.7- to 4.5-fold higher incidence of T2D in FHS whites, ARIC whites, and ARIC blacks, as well as in the meta-analysis of these three groups. “HbA_1c-only” model: adjusted for age and sex; “HbA_1c + fasting laboratory tests” model: adjusted for age, sex, FG, TG, and HDLs; “HbA_1c + clinic visit” model: adjusted for age, sex, SBP, family history of T2D, and BMI; “HbA_1c + fasting laboratory tests + clinic visit” model: adjusted for age, sex, FG, TG, HDLs, SBP, family history of T2D, and BMI. As we observed heterogeneity in the effect estimates, we performed the meta-analysis using random effects. OR, OR per 1%-unit increase in HbA_1c; I-squared, Higgins I² test for heterogeneity.

Figure 2

Incidence of T2D over a two-decade follow-up period in FHS and ARIC by HbA_1c (≥5.7% [39 mmol/mol] vs. <5.7% [39 mmol/mol]) and race (blacks and whites). Kaplan-Meier curves showed clear separation of the curves for HbA_1c ≥5.7% (39 mmol/mol) vs. <5.7% (39 mmol/mol) in blacks (ARIC) and whites (pooled across ARIC and FHS). While the incidence of T2D among those with HbA_1c <5.7% (39 mmol/mol) was higher in blacks (14.5 events per 1,000 person-years [95% CI 12.9, 16.2]) than whites (10.6 events per 1,000 person-years [95% CI 10.1, 11.1]; log-rank test P < 0.0001), incidence of T2D among those with HbA_1c ≥5.7% was similar between blacks (30.5 events per 1,000 person-years [95% CI 28.6, 32.5]) and whites (29.4 events per 1,000 person-years [95% CI 26.8, 32.1]; log-rank test P = 0.25).

Figure 3

Absolute risks of incident T2D over two decades, within 8 years, and after 8 years by HbA_1c levels and HbA_1c-FG categories. A: HbA_1c (%) levels and T2D risk over two decades. B: HbA_1c (%) levels and T2D risk within 8 years and after 8 years. C: HbA_1c (%)-FG (mg/dL) categories and T2D risk over two decades. D: HbA_1c (%)-FG (mg/dL) categories and T2D risk within 8 years and after 8 years. A: Compared with lower HbA_1c levels, higher HbA_1c had higher predicted absolute risk for incident T2D over the 20-year follow-up period. B: Likewise, higher HbA_1c had higher predicted absolute risk for incident T2D over both short-term (within 8 years) and long-term (only after 8 years) follow-up. C: The predicted absolute risk over the 20-year follow-up period was higher with each successive HbA_1c level (<5.4, 5.4–5.7, 5.7–6.0, and ≥6.0% [<36, 36–39, 39–42, and ≥42 mmol/mol]) within each FG level (<100, 100–100, and ≥110 mg/dL). D: The predicted absolute risk period was higher with each successive HbA_1c level within each FG level over both short-term (within 8 years) and long-term (only after 8 years) follow-up. Regression models in A and B included HbA_1c, age, sex, race, and cohort. Regression models in C and D included HbA_1c, FG, age, sex, race, and cohort. Box plots are represented by the first quartile (lower hinge) and median and third quartile (upper hinge) and whiskers indicating 1.5 times the interquartile range.