Validation of a model to predict hospitalization due to RSV of infants born at 33–35 weeks' gestation

Abstract

A model to predict hospitalization due to respiratory syncytial virus (RSV) of infants born at 33- 35 weeks' gestation was developed using seven risk factors from the Spanish FLIP study "birth +/-10 weeks from the beginning of the RSV season", "birth weight", "breast fed <or=2 months", "number of siblings >or=2 years", "number of family members with atopy", "number of family members with wheezing", and "gender". The aim of this study was to validate the model using French data. The FLIP model [predictive accuracy 71%, receiver operating characteristic (ROC) 0.791] was tested against the French data (77 hospitalized infants with RSV born at 33-35 weeks and 154 age-matched controls) using discriminatory function analysis by applying the FLIP coefficients to the French data and by generating the seven variable model from the French data. Applying the FLIP coefficients to the French dataset, the model correctly classified 69% of cases (ROC 0.627). The predictive power increased to 73% (ROC 0.654) when "number of siblings >or=2 years" was substituted for "number of children at school". The number needed to treat (NNT) in order to prevent 70% of hospitalizations was 18. The model derived using French data could correctly classify 62% of cases in the French data (ROC 0.658). The model was successfully validated and may potentially optimize immunoprophylaxis in French infants born at 33-35 week's gestation.