Electrocardiographic abnormalities and reclassification of cardiovascular risk: insights from NHANES-III.

BACKGROUND: We aimed to assess the additive value of electrocardiogram (ECG) findings to risk prediction models for cardiovascular disease.

METHODS: Our dataset consisted of 6025 individuals with ECG data available from the National Health and Nutrition Examination Survey-III. This is a self-weighting sample with a follow-up of 79,046.84 person-years. The primary outcomes were cardiovascular mortality and all-cause mortality. We compared 2 models: Framingham Risk Score (FRS) covariates (Model A) and ECG abnormalities added to Model A (Model B), and calculated the net reclassification improvement index (NRI).

RESULTS: Mean age of our study population was 58.7 years; 45.6% were male and 91.7% were white. At baseline, 54.6% of individuals had ECG abnormalities, of which 545 (9%) died secondary to a cardiovascular event, compared with 194 individuals (3.2%) (P <.01) without ECG abnormalities. ECG abnormalities were significant predictors of cardiovascular mortality after adjusting for traditional cardiovascular risk factors (hazard ratio 1.44; 95% confidence interval, 1.13-1.83). Addition of ECG abnormalities led to an overall NRI of 3.6% subjects (P <.001) and 13.24% in the intermediate risk category. The absolute integrated discrimination index was 0.0001 (P <.001).

CONCLUSION: Electrocardiographic abnormalities are independent predictors of cardiovascular mortality, and their addition to the FRS improves model discrimination and calibration. Further studies are needed to assess the prospective application of ECG abnormalities in cardiovascular risk prediction in individual subjects.