783 IDENTIFICATION OF SERIOUS ILLNESS IN FEBRILE INFANTS ONDER 60 DAYS BY PEDIATRIC RESIDENTS: COMPARATIVE EVALUATION OF THE “McCARTHY OBSERVATION SCALE” AND A SIMPLE QUESTIONNAIRE

Abstract

A logistic regression based Observation Scale (O.S.) was shown by McCarthy et al to be a sensitive predictor of serious illness in children, when used by experienced pediatricians. A prospective study was begun in January 1984 to compare the predictive efficacy of the O.S. in comparison with a simple yes/maybe/no questionnaire (Q) when used by pediatric residents. Of 109 infants<60 days hospitalized during this period, 52 were found to have serious illness. Linear regression and correlation co-efficient matrices were obtained using outcome as the dependant variable. High correlation coefficients were seen with the Q, band count, temperature, WBC count, whereas the O.S. showed very low correlation, particularly when used by junior residents (r=.06). The sensitivities of the O.S. and Q were 46% and 67% respectively, while the specificities were 73% and 58% respectively. When combined with an abnormal WBC count (< 5000 or >15000/mm), the sensitivity of both the O.S. and Q increased. The simple Q was found to be better than the O.S. for identifying serious illness in febrile infants 60 days. The sophisticated O.S. may well be a good predictive tool only when used by experienced clinicians.