Abstract 17098: Evaluation of a Non-laboratory-based Cardiovascular Disease Risk Score in Predicting Cardiovascular Mortality in the Southern Community Cohort Study
Introduction: Cardiovascular disease (CVD) risk prediction guides clinical recommendations regarding primary prevention strategies. Recently, a non-laboratory-based CVD risk score was derived and validated in several populations. We evaluated the performance of the non-laboratory CVD risk score among individuals of low socioeconomic status in the southeastern US, with high CVD incidence rates.
Methods: The Southern Community Cohort Study (SCCS) is a prospective cohort study of black and white adults enrolled primarily from community health centers in the southeastern US between 2002-2009. We calculated c-statistics and 95% confidence intervals (CI) to assess the performance of the non-laboratory risk score in discriminating 5- year risk of CVD death (ICD-10 I00-I99) in 14,243 participants (median enrollment age 51 yrs; 74% black, 71% female, 61% income < $15,000 yr) with no history of MI/CABG or stroke at baseline and with complete data on all input variables (age, sex, smoking, diabetes mellitus, reported hypertension (HTN), measured SBP, and BMI). Reported HTN was substituted for anti-HTN therapy, used in the original risk score, because the latter was available only for a subset of the SCCS population. However, among this subset, 82% of individuals who reported HTN also reported use of anti-HTN drugs.
Results: CVD risk factors were common: HTN (66%), smoking (59%), diabetes (29%), median SBP 130 mmHg, and median BMI 31 kg/m2. CVD death occurred in 263 participants (1.8%). The non-laboratory CVD risk score demonstrated poor discrimination for CVD mortality, with c-statistics of 0.65 (95% CI: 0.61-0.68) overall and 0.64 (0.59-0.69), 0.65 (0.6-0.7), 0.67 (0.63-0.71) and 0.58 (0.51-0.65) for males, females, blacks and whites, respectively. In an additional logistic regression model fit to the SCCS data, the inclusion of education and income did not substantially improve model performance; overall c-statistic 0.70 (0.67-0.73).
Conclusion: In this population of predominantly low-income blacks and whites, a non-laboratory CVD risk score did not perform well in predicting CVD death in any of the race and gender groups. Future studies are warranted to develop algorithms for identifying individuals at increased CVD risk in low resource populations.
Author Disclosures: J. Singh: None. E.K. Kabagambe: None. M. Xu: None. T.A. Gaziano: None. T.J. Wang: Other Research Support; Modest; Diasorin. Consultant/Advisory Board; Modest; Takeda. W.J. Blot: None. D.K. Gupta: None. L. Lipworth: None.
- © 2015 by American Heart Association, Inc.