Abstract 874: Education Independently Predicts Risk of Cardiovascular Disease but Only Modestly Improves Performance of Framingham Risk Functions
Background: Socio-economic status has been shown to be an important predictor of cardiovascular disease (CVD) risk in non-US cohorts. Whether it improves the performance of Framingham risk equations is unknown.
Methods: Binary education level (at least college vs. not) was used as a proxy for socio-economic status. Self-reported information was available on 7650 (4116 women) participants of the Framingham Heart Study from the larger pool of 8491 individuals whose data have been used to the develop the most recent Framingham CVD risk prediction models. Ten-year, sex-specific Cox proportional hazard models adjusted for standard CVD risk factors (age, treated or untreated systolic blood pressure, total and HDL cholesterol, smoking and diabetes status) and stratified on the presence or absence of education level were fit to make full use of the data. Change in c statistic, net reclassification improvement (NRI; with categories <6%, 6 –20%, >20%) and integrated discrimination improvement (IDI) were calculated to assess gains in model performance and risk reclassification after adding education to Framingham risk algorithm.
Results: As seen in table⇓ below education level was statistically significant for both women and men. However, it offered very modest improvement in model performance. Different definitions of education status did not change the results.
Conclusions: Socio-economic status defined by education level is an independent predictor of CVD risk but offers very modest improvement in model performance.