Hemoglobin A1c levels predicted cardiovascular disease risk better than diabetes diagnosis alone in a recent analysis of two large cohort studies — a finding that could influence current CVD risk models.
“These results may be particularly helpful in light of current discussion around treatment choices for diabetic patients for prevention of CVD, including use of statins and aspirin,” Nina P. Paynter, PhD, of Brigham and Women’s Hospital in Boston, and colleagues wrote in Archives of Internal Medicine.
Diabetes is considered an independent risk factor for CVD, but including HbA1c levels as part of the overall CVD risk score may help predict which patients with diabetes are at greater risk, while enabling clinicians to use one standard risk assessment model for patients with and without the disease.
To explore these questions, the researchers analyzed data from 24,674 women who participated in the Women’s Health Study (WHS) and 11,280 men who participated in the Physician’s Health Study (PHS). They identified 685 women and 563 men with diabetes. Median follow-up in the WHS study was 11.8 years and 10.2 years in the PHS study.
In the WHS study, women with diabetes experienced 125 CVD events vs. 666 events among those without the disease. In the PHS, there were a total of 170 CVD events among men with diabetes and 1,382 among those without.
The researchers determined that adding HbA1c to two risk assessment models – the National Cholesterol Education Program for Adults (ATP III) and the Reynolds Risk Score (RRS), which include factors such as C-reactive protein and parental history of premature myocardial infarction – resulted in statistically significant improvements.
Among women with diabetes that participated in the WHS, models that included HbA1c improved discriminatory power (C statistic 0.177, P<0.001) and increased net reclassification improvement (NRI) by 26.7% on the ATP III (P=0.001) and 23.6% on the RRS (P=0.003).
The improvements were more modest for men with diabetes who participated in PHS, but adding HbA1c still significantly improved the discriminatory power of the ATP III (C statistic 0.039, P=0.02) and increased NRI 9.2% for ATP III (P=0.04) and 12.4% for RRS (P=0.004).
Compared with using a dichotomous term for diabetes, HbA1c levels improved risk assessment predictions in women (NRI 11.8%, P=0.03), but not men.
“We found in these large population-based cohorts of both men and women, presence of diabetes alone did not confer a 10-year risk of CVD higher than 20%, and the measurement of HbA1c levels in diabetic subjects improved risk prediction compared with classification as cardiovascular risk equivalent,” the researchers wrote.
However, Mark J. Pletcher, MD, MPH, of the University of California in San Francisco, pointed out that although the study included almost 36,000 men and women, the number of patients with diabetes included in the analyses was relatively small number (n=1,248) and most had well-controlled diabetes, as indicated by HbA1c of 6.5% to 7.0%. Both of these factors limit the accuracy of the modeling.
He added that “it is unclear if including HbA1c in the risk equation is truly better than simply including indicator variable for diabetes,” such as 1 or -1.
Despite these limitations, Pletcher wrote that “allowing downward reclassification” for some patients with diabetes using a risk model that includes HbA1c will “likely improve the overall accuracy of 10-year CVD risk prediction.”
He suggested that guidelines committees develop better strategies to incorporate new developments in CVD risk prediction into clinical guidelines more quickly and efficiently, as this “could have very large public health benefits.”