Prognostic utility of a risk prediction model for predischarge major residual lesions or unplanned reinterventions following congenital mitral valve repair

Abstract

ObjectiveWe sought to develop a risk prediction model for predischarge major mitral valve (MV) residual lesions or unplanned MV reinterventions following congenital MV repair. MethodsPatients who underwent congenital MV repair (excluding primary repair, but including secondary repair, of canal-type defects) at a single institution from January 2000 to December 2020 and survived to discharge were retrospectively reviewed. The primary outcome was major MV residua (mean gradient >6 mm Hg or moderate or greater regurgitation on the discharge echocardiogram) or predischarge unplanned MV reintervention. Risk factors of interest included age, single-ventricle physiology, preoperative and intraoperative postrepair MV stenosis and regurgitation severity, MV annular diameter z score, systemic ventricle ejection fraction, unfavorable anatomy, concomitant left-heart procedure, and various technique-related categories. Logistic regression was used to develop a weighted risk score for the primary outcome. Internal validation using bootstrap-resampling was performed. ResultsOf 866 patients who underwent congenital MV repair at a median age of 2.7 years (interquartile range, 0.7-9.1 years), 202 (23.3%) patients developed the primary outcome. The final risk prediction model had a C-statistic of 0.82 (95% confidence interval, 0.78-0.85). A weighted risk score was formulated per the variables in this model. The median risk score was 8 (interquartile range, 6-11) points. Patients were categorized as low (score 0-5), medium (score 6-10), high (score 11-15), or very high (score ≥16) risk. The probability of the primary outcome was 5.0 ± 1.7%, 15.2 ± 6.7%, 45.9 ± 12.6%, and 76.7 ± 8.8% for low-, medium-, high-, and very-high-risk patients, respectively. ConclusionsOur risk prediction model may guide prognostication of patients following congenital MV repair.