Prediction of mortality and mode of death by clinical risk score systems in 2.6 million patients with atrial fibrillation: a nationwide analysis

Abstract

Abstract Atrial fibrillation (AF) is associated with a higher mortality, but modes of death may vary and their respective predictors have been insufficiently defined. Charlson comorbidity index (CCI) is a tool to quantify multimorbidity and a strong estimator of mortality. The quantifiable frailty phenotype is also predictive of mortality and disability and claims data can be used to classify individuals as frail and non-frail using the Claims-based Frailty Index (CFI). We evaluated whether these tools may help to predict mortality and the different modes of death in AF. Methods Based on the France nationwide administrative hospital-discharge database, we collected information for all AF patients treated between 2010 and 2019 in France. Adverse outcomes were investigated during follow-up. CHA2DS2VASc score, CCI and CFI were calculated for each patient. Results Among 2,641,626 patients with AF, 670,541 patients died during a follow-up of 2.0±2.3 years (median 1.1) (yearly rate 12.6%, 30.3% cardiovascular and 69.7% non-cardiovascular deaths). Death occurred more often in patients with higher CHA2DS2VASc, CCI and CFI scores. CCI was a better predictor of total mortality than CFI and CHA2DS2VASc score (see C-statistics in table); however, the CHA2DS2VASc score was a better predictor of cardiovascular mortality than CCI and CFI. By contrast, CCI was a better predictor of non-cardiovascular mortality than CFI and CHA2DS2VASc score. The optimal predictive performances were better for non-cardiovascular death than for cardiovascular death. Conclusion Multimorbidity assessed with CCI demonstrated better performances in predicting total mortality and non-cardiovascular mortality than CHA2DS2VASc score and Frailty assessed with CFI in AF patients. By contrast, CHA2DS2VASc score was a better predictor of cardiovascular mortality than CCI and CFI in these patients. Funding Acknowledgement Type of funding source: None