Diagnostic accuracy of the Embolic Risk French Calculator for symptomatic embolism with infective endocarditis among Japanese population

Background Recently, the Embolic Risk French Calculator (ER-Calculator) was designed to predict symptomatic embolism (SE) associated with infective endocarditis (IE), but external validation has not been reported. This study aimed to determine predictors of SE and the diagnostic accuracy of the ER-Calculator in left-sided active IE among a Japanese population. Methods This retrospective cohort study included 166 consecutive patients with a definite diagnosis of left-sided IE from 1994 to 2015 in our institution. SE during the period after initiation of antibiotic therapy was defined as new SE and embolism during the period before initiation of antibiotic therapy was defined as previous embolism. The primary endpoint was new SE. Results The mean age of patients was 63 ± 17 years. New SE occurred in 23 (14%) patients at a median of 6 days (interquartile range: 2.5–12.5 days) after initiation of antibiotic therapy. The cumulative incidence of new SE at 12 weeks was 18.2%. The 2-week probability by the ER-Calculator as well as previously reported predictors, such as previous embolism, vegetation length (>10 mm), and their combination, were associated with a high risk of new SE. By receiver operating characteristic analysis, the area under the curve of the 2-week probability by the ER-Calculator for prediction of new SE was 0.75 and the optimal cut-off value was 8%. A 2-week probability >8% by the ER-Calculator was the most useful predictor of new SE (hazard ratio 3.63, 95% confidence interval 1.50–8.37; p = 0.006), which was more remarkable for fatal embolic events (hazard ratio 13.9, 95% confidence interval 3.19–95.4; p = 0.004). Conclusions The ER-Calculator is a useful predictor of new SE. Predictive ability is more remarkable for critical embolic events.