Improving risk prediction in heart failure: MAGGIC plus natriuretic peptides
Michaels AT, Radjef R, She R, Liu B, Peterson EL, Pinto Y, Williams KL, Sabbah H, Lanfear D. Improving risk prediction in heart failure: MAGGIC plus natriuretic peptides. J Card Fail 2016; 22(8 Suppl):S99.
J Card Fail
Background: Risk stratification of patients with heart failure (HF) remains challenging but is a critical need. The MAGGIC score is a clinical risk model derived from meta-analysis of nearly 40k patients. Natriuretic peptides (NP) have consistently shown powerful risk prediction in HF patients, but the incremental value in addition to MAGGIC score is not known. Methods: In this single center study 4264 patients were analyzed from two cohorts; a prospective ambulatory registry of HF patients (n = 1314) who had baseline NTproBNP levels measured, and a retrospective cohort collected utilizing administrative data from hospital discharges for HF (January 1 st , 2014 through July 30 th , 2015; n = 2503) with clinical BNP levels measured at or near discharge. The hospital discharge cohort were all assigned NYHA class IV. The primary end-point was all cause mortality. Performance of the MAGGIC score and NP levels was assessed within each cohort utilizing Cox regression and receiver operating curves (ROC) analysis (MAGGIC alone vs. MAGGIC+NP) with the net reclassification improvement (NRI) also calculated. Results: The overall cohort had an average age of 71.2 years, was 47.8% females, and 41% self-identified African Americans. Median follow up was 1.52 years during which there were 1139 deaths (27%). The MAGGIC score was a strong predictor of outcome in both cohorts ( P < .001). In ROC analysis of the ambulatory registry, NP significantly improved area under the curve (AUC) compared to MAGGIC alone from 0.74 to 0.79 ( P = .002) and had a NRI of 0.354 (Figure). In contrast, within the hospital discharge cohort NP levels did not significantly add to MAGGIC score (AUC 0.681 vs. 0.676, NRI = 0.033, P = .284) (Figure). Conclusion: In our study, NP levels in the ambulatory setting significantly improved risk stratification provided by the MAGGIC score, but discharge NP levels did not improve MAGGIC prediction of post-hospital survival. Overall risk stratification and particularly NP utility is much better in the ambulatory setting.