We will develop a multi-level risk estimation of outcome states from pediatric intensive care incorporating poor functional/adaptive survival state(s) as well as death. This severity of illness method will be an important advance over survival/death predictors and will enhance quantitative quality assurance studies, risk adjustment, cost evaluations, and individuals' decision-making. This study will involve pediatric ICU to estimate the multi-state outcome predictor using poly-logistic regression analysis. Predictor variables will include physiologic variables, pre-admission location, operative status, principle diagnosis categorizations, disease stage, age, computed tomography (CT) severity classification and baseline functional status. Survival categories will be determined using the Vineland Adaptive Behavior Scale (VABS). The Phase I primary aim is to develop a scale and evaluate quicker and more convenient measures of functional status than the VABS. Such a simplified tool will facilitate performance of the Phase II study and the acceptance of the resulting model. The appropriate indications and/or screeningtests will be chosen by pediatricians, intensivists, and developmentalist. Statistical comparison will bedone to evaluate inter-rater reliability and consistency. Phase I will also develop a manual for the collectors of this information in Phase II, and design Phase II, including the statistical methodology.
