Head-to-head comparison of MAGGIC-HF, SHFM and BCN-Bio HF scores. (25th November 2020)

• Record Type: Journal Article
• Authors: Codina, P; De Antonio, M; Domingo, M; Santiago-Vacas, E; Zamora, E; Santesmases, J; Subirana, I; Buchaca, D; Diez-Quevedo, C; Boldo, M; Rivas, C; Velayos, P; Nunez, J; Lupon, J; Bayes-Genis, A
• Is Part Of: European heart journal. Volume 41:(2020)Supplement 2
• Publication Date: 2020-11-25
• Subjects: Chronic Heart Failure - Epidemiology, Prognosis, Outcome; Cardiology -- Periodicals; Heart -- Diseases -- Periodicals; 616.12005
• DOI: 10.1093/ehjci/ehaa946.0958
• Languages: English
• ISSNs: 0195-668X

Abstract

Abstract: Background: Three heart failure (HF) web-based risk scores are currently used in clinical practice. Two only include clinical variables; one adds biomarkers, known to refine pathophysiological pathways in HF and to improve mortality prediction. Objective: To assess inter-score agreement and to compare mortality prediction discrimination between MAGGIC-HF risk score, Seattle HF Model (SHFM) and Barcelona Bio-HF Risk Calculator (BCN Bio-HF), which includes both clinical and biomarker variables (NTproBNP). Methods: Out of 1745 consecutive patients with HF from different etiologies admitted at our Unit from May 2006 to November 2018, 1689 with NTproBNP measurement at first visit were included. Absolute and consistency intraclass correlation coefficient (ICC) at individual level among the three risk estimation tools was assessed. Bland Altman graphics were used to illustrate the differences between scores across the broad spectrum of mortality risk. Discrimination of the three prediction tools was compared by AUC of the ROC curves for 1-, 3- and 5-year all-cause mortality. Patients used previously to derivate the BCN Bio-HF were excluded. Results: Patients age 66.3±13.3 years, 70.4% men, LVEF 36.4%±14.4, ischemic etiology 43.7%. Absolute ICC was poor for the three tools (from 0.18 [–0.006 to 0.35] to 0.53 [0.42 to 0.62], while consistency ICC was slightly better (from 0.28 [0.24 to 0.33] to 0.57 [0.53 to 0.60], being highest the ICC of MAGGIC-HF and BCN Bio-HF. Correlation was better among scores in low-mortality risk profile patients but clinical scores tended to infraestimate the risk in comparison with the BCN Bio-HF in high-risk patients (Figure). Discrimination was numerically better with the BCN Bio-HF at every time risk (Table), significantly better when compared with SHFM. Conclusions: In comparison with other clinical scores, the BCN Bio-HF calculator predicted better higher mortality risk. Correlation was globally poor for the three tools at individual level, although improved in the low risk patients. Discrimination tended to be numerically better with the BCN Bio-HF calculator, reaching statistical significance when compared with the SHFM. Funding Acknowledgement: Type of funding source: None