Risk scores and surgery for infective endocarditis: in search of a good predictive score

Objectives: To evaluate scoring systems that have been created to predict the risk of death post-surgery in infective endocarditis (IE). Design: Eight scores - (1) The Society of Thoracic Surgery (STS) risk score for IE, (2) De Feo score, (3) PALSUSE score (prosthetic valve, age ≥70, large intracardiac destruction, Staphylococcus spp, urgent surgery, sex [female], EuroSCORE ≥10), (4) ANCLA score (anemia, New York Heart Association class IV, critical state, large intracardiac destruction, surgery of thoracic aorta), (5) Risk-Endocarditis Score (RISK-E), (6) score for heart valve or prosthesis IE (EndoSCORE), and (7,8) Association pour l'Étude et la Prévention de l'Endocadite Infectieuse (AEPEI) score I and II - were evaluated in 324 (mean age, 61.8 ± 14.6 years) consecutive patients having IE and undergoing cardiac operation (1999-2018, Regione Autonoma Friuli-Venezia Giulia, Italy). Results: There were 45 (13.9%) in-hospital deaths. Despite many differences on the number and the type of variables, all the investigated scores showed good goodness-of-fit (Hosmer-Lemeshow test, p ≥.28). For five scores, accuracy of prediction (receiver-operating characteristic curve analysis) was good (ANCLA score) or fair (STS risk score for IE, PALSUSE score, AEPEI score I and II). When compared one-to-one (Hanley-McNeil method), accuracy of prediction of ANCLA score was higher than all of other risk scores except for AEPEI score I (p = .077). Conclusions: Five of eight scores that were evaluated in this study showed satisfactory performance in predicting in-hospital mortality following surgery for IE. The ANCLA score should be preferred.