Comparing SF-36® scores versus biomarkers to predict mortality in primary cardiac prevention patients

Abstract

BackgroundRisk stratification plays an important role in evaluating patients with no known cardiovascular disease (CVD). Few studies have investigated health-related quality of life questionnaires such as the Medical Outcomes Study Short Form-36 (SF-36®) as predictive tools for mortality, particularly in direct comparison with biomarkers. Our objective is to measure the relative effectiveness of SF-36® scores in predicting mortality when compared to traditional and novel biomarkers in a primary prevention population. Methods7056 patients evaluated for primary cardiac prevention between January 1996 and April 2011 were included in this study. Patient characteristics included medical history, SF-36® questionnaire and a laboratory panel (total cholesterol, triglycerides, HDL, LDL, ApoA, ApoB, ApoA1/ApoB ratio, homocysteine, lipoprotein (a), fibrinogen, hsCRP, uric acid and urine ACR). The primary outcome was all-cause mortality. ResultsA low SF-36® physical score independently predicted a 6-fold increase in death at 8years (above vs. below median Hazard Ratio [95% confidence interval] 5.99 [3.86–9.35], p<0.001). In a univariate analysis, SF-36® physical score had a c-index of 0.75, which was superior to that of all the biomarkers. It also carried incremental predictive ability when added to non-laboratory risk factors (Net Reclassification Index=59.9%), as well as Framingham risk score components (Net Reclassification Index=61.1%). Biomarkers added no incremental predictive value to a non-laboratory risk factor model when combined to SF-36 physical score. ConclusionThe SF-36® physical score is a reliable predictor of mortality in patients without CVD, and outperformed most studied traditional and novel biomarkers. In an era of rising healthcare costs, the SF-36® questionnaire could be used as an adjunct simple and cost-effective predictor of mortality to current predictors.