Turning 18 with congenital heart disease: prediction of infective endocarditis based on a large population

Aims The risk of infective endocarditis (IE) in adults with congenital heart disease is known to be increased, yet empirical risk estimates are lacking. We sought to predict the occurrence of IE in patients with congenital heart disease at the transition from childhood into adulthood. Methods and results We identified patients from the CONCOR national registry for adults with congenital heart disease. Potential predictors included patient characteristics, and complications and interventions in childhood. The outcome measure was the occurrence of IE up to the age of 40 and 60. A prediction model was derived using the Cox proportional hazards model and bootstrapping techniques. The model was transformed into a clinically applicable risk score. Of 10 210 patients, 233 (2.3%) developed adult-onset IE during 220 688 patient-years. Predictors of IE were gender, main congenital heart defect, multiple heart defects, and three types of complications in childhood. Up to the age of 40, patients with a low predicted risk (<3%) had an observed incidence of less than 1%; those with a high predicted risk (>= 3%) had an observed incidence of 6%. The model also yielded accurate predictions up to the age of 60. Conclusion Among young adult patients with congenital heart disease, the use of six simple clinical parameters can accurately predict patients at relatively low or high risk of IE. After confirmation in other cohorts, application of the prediction model may lead to individually tailored medical surveillance and educational counselling, thus averting IE or enabling timely detection in adult patients with congenital heart disease.