Clinical implications of a multivariate stratification model for the estimation of prognosis in ventricular septal defect

Objectives To derive and validate a multivariate stratification model for prediction of survival free from intervention (SFFI) in ventricular septal defect (VSD). A secondary aim is for this model to serve as proof of concept for derivation of a more general congenital heart disease prognostic model, of which the VSD model will be the first component. Study design For 12 years, 2334 subjects with congenital heart disease were prospectively and consecutively enrolled. Of these, 675 had VSD and form the derivation cohort. One hundred seven other subjects with VSD followed in another practice formed the validation cohort. The derivation cohort was serially stratified based on clinical and demographic features correlating with SFFI. Results Six strata were defined, the most favorable predicting nearly 100% SFFI at 10 years, and the least favorable, a high likelihood of event within weeks. Strata with best SFFI had many subjects with nearly normal physiology, muscular VSD location, or prior intervention. In the validation cohort, the relation between predicted and actual SFFI at 6 months, 1 year, 2 years, and 5 years follow-up had areas under the receiver operating characteristic curves 0.800 or greater. Conclusions A prediction model for SFFI in VSD has been derived and validated. It has potential for clinical application to the benefit of patients and families, medical trainees, and practicing physicians.