Association between clinical risk scores and mortality in atrial fibrillation: Systematic review and network meta-regression of 669,000 patients.

Aims Many clinical scores for risk stratification in patients with atrial fibrillation have been proposed, and some have been useful in predicting all-cause mortality. We aim to analyse the relationship between clinical risk score and all-cause death occurrence in atrial fibrillation patients. Methods We performed a systematic search in PubMed and Scopus from inception to 22 July 2017. We considered the following scores: ATRIA-Stroke, ATRIA-Bleeding, CHADS(2), CHA(2)DS(2)-VASc, HAS-BLED, HATCH and ORBIT. Papers reporting data about scores and all-cause death rates were considered. Results Fifty studies and 71 scores groups were included in the analysis, with 669,217 patients. Data on ATRIA-Bleeding, CHADS(2), CHA(2)DS(2)-VASc and HAS-BLED were available. All the scores were significantly associated with an increased risk for all-cause death. All the scores showed modest predictive ability at five years (c-indexes (95% confidence interval) CHADS(2): 0.64 (0.63-0.65), CHA(2)DS(2)-VASc: 0.62 (0.61-0.64), HAS-BLED: 0.62 (0.58-0.66)). Network meta-regression found no significant differences in predictive ability. CHA(2)DS(2)-VASc score had consistently high negative predictive value (>= 94%) at one, three and five years of follow-up; conversely it showed the highest probability of being the best performing score (63% at one year, 60% at three years, 68% at five years). Conclusion In atrial fibrillation patients, contemporary clinical risk scores are associated with an increased risk of all-cause death. Use of these scores for death prediction in atrial fibrillation patients could be considered as part of holistic clinical assessment. The CHA(2)DS(2)-VASc score had consistently high negative predictive value during follow-up and the highest probability of being the best performing clinical score.