Abstract 15272: Estimating Life Expectancy of Systolic Blood Pressure Intervention Trial Participants Using Pooled Epidemiologic Cohort Data

Abstract

Introduction: Intensive systolic blood pressure (SBP) treatment (<120 mm Hg) in the Systolic Blood Pressure Intervention Trial (SPRINT) improved overall survival compared to standard treatment (<140 mm Hg). Economic analyses of SPRINT require extrapolation of treatment effects beyond the trial data. Methods: We projected life expectancy after SPRINT using six US cohort studies in the National Heart, Lung, and Blood Institute Pooled Cohorts Study (NHLBI-PCS). We included SPRINT-eligible NHLBI-PCS participants as those aged >=50 years with SBP 130-180 mm Hg and increased cardiovascular disease (CVD) risk without diabetes or history of stroke. We used propensity scores to weight NHLBI-PCS participants to resemble SPRINT participants. In SPRINT participants, we estimated in-trial survival (<4 years) using a time-based flexible parametric survival model (FPSM). In SPRINT-eligible NHLBI-PCS participants, we estimated post-trial survival (>=4 years) using an age-based FPSM and applied the formula to SPRINT participants to predict post-trial survival. We combined in- and post-trial survival to project overall life expectancy for each SPRINT participant and compared it to commonly used Gompertz methods. Results: We included 8,584 SPRINT and 10,610 SPRINT-eligible NHLBI-PCS participants. After propensity weighting, mean (SD) age was 67.9 (9.4) and 68.7 (8.8) years, 35.5% and 38.3% were female in SPRINT and NHLBI-PCS, respectively. Predicted in-trial survival was similar to that observed in SPRINT with both FPSM and Gompertz models (Figure). Assuming constant treatment effects, projected mean life expectancy using the NHLBI-PCS method was 21.1 (7.4) years with intensive and 19.3 (7.2) years with standard treatment; compared to 11.2 (2.3) and 10.5 (2.2) years, respectively, using the Gompertz method. Conclusions: Combining SPRINT and NHLBI-PCS observed data may offer a more realistic estimate of life expectancy than by parametrically extrapolating SPRINT data.