Abstract 2999: Single Versus Paired Blood Pressure Components and Risk for Cardiovascular Disease: The Framingham Heart Study
The utility of any single versus paired blood pressure (BP) components in predicting cardiovascular disease (CVD) risk is not established. We compared the importance of paired systolic BP (SBP) and diastolic BP (DBP) (measured variables), paired pulse pressure (PP) and mean arterial pressure (MAP) (derived variables that indicate stiffness and resistance, respectively), and each of the four single BP components in predicting CVD risk. We studied participants free of CVD and without BP therapy from combined original (n = 4,760) and offspring (n = 4,897) cohorts of the Framingham Heart Study from which 1,439 CVD events (myocardial infarctions, stroke, heart failure, CVD deaths) occurred over serial 4-year intervals from 1952 to 2001.
Results: In pooled logistic regression that uses BP categories, the interaction term of SBP and DBP improved model fit for predicting CVD events when added to SBP and DBP, after adjusting for covariates (differences in Chi-sq = 35.9, p = 0.02). Similar findings were noted for interaction of PP and MAP when added to PP and MAP (differences in Chi-sq = 43.1, p = 0.01). Also, in models with continuous BP measures, adjusted for covariates, there were increased odds of CVD events with increasing SBP [odds ratio (OR) per sd = 1.39, p = 0.0001] or increasing DBP (OR per sd = 1.25, p = 0.0001). Model fit was not improved when DBP was added to SBP (differences in Chi-Sq = 0.49, p = 0.48), but adding a DBP*DBP term (but not SBP*SBP) to DBP and SBP improved the fit (differences in Chi-Sq = 9.97, p = 0.0016), indicating that there was a quadratic relation with DBP. Similarly, in a separate model adjusted for covariates, there was increased odds of CVD with increasing PP (OR per sd = 1.34, p < 0.0001) or increasing MAP (OR per sd = 1.35, p < 0.0001). When MAP was added to PP, the fit improved (differences in Chi-Sq = 33.7, p < 0.0001), adding MAP*MAP or PP*PP to MAP and PP did not improve the fit, consistent with a linear relation. Paired PP and MAP and paired SBP and DBP were superior to any single BP component for predicting CVD events and extent of CVD risk varied with the level of each BP component. In contrast to paired SBP and DBP, paired PP and MAP has a monotonic relation with risk that may provide direct physiologic correlates to each BP component.
This research has received full or partial funding support from the American Heart Association, AHA National Center.