Abstract 2957: Mortality Rates Predicted by ADHERE CART Model Overestimate Mortality Rates in Clinical Trials: Results from REVIVE I & II
Background: Classification and regression tree (CART) analysis of ADHERE data identified blood urea nitrogen (BUN), systolic blood pressure (SBP), and serum creatinine (Cr) at admission as significant determinants of in-hospital mortality. REVIVE I & II (REVIVE) were pilot and confirmatory trials, respectively, that compared levosimendan to placebo, in addition to standard-of-care, in patients with acute decompensated heart failure (ADHF). We hypothesized that mortality from REVIVE would be similar to ADHERE CART mortality. Because of the high variability in length of hospital stay, 14-day mortality data may also be valuable.
Methods: REVIVE enrolled a total of 700 patients with ADHF and dyspnea at rest. Both in-hospital and 14-day mortality data from REVIVE were mapped using the same variables and cut-points obtained by the ADHERE CART analysis.
Results: Compared to ADHERE, proportionately more patients in REVIVE had SBP < 115 mmHg (56.4% vs. 18.6%; p < 0.001), and more (3.0% vs. 1.9%; p < 0.05) were in the subgroup with the highest predicted mortality (SBP < 115 mmHg, BUN ≥43 mg/dL, & Cr ≥2.75 mg/dL). For the total population and for every CART defined subgroup, REVIVE in-hospital mortality rates were lower than ADHERE in-hospital mortality rates. REVIVE 14-day mortality rates were identical, or slightly lower, than REVIVE in-hospital mortality rates for each CART defined subgroup.
Conclusion: This analysis suggests that clinical trials, such as REVIVE, may enroll proportionately more patients with risk factors for high mortality than are found in the general population (ADHERE). Despite the predicted higher risk of mortality, mortality rates in REVIVE appeared to be lower than those resulting from the ADHERE CART model for the total population and for every CART specified subgroup, suggesting additional factors such as older age and comorbid conditions not identified in the ADHERE CART analysis may be responsible for the poorer prognosis in non-trial populations.