Lower respiratory tract illness (LRI) is the most common reason for hospitalization of infants beyond the neonatal period, and the US cost is over $700 million annually. The goal of this research is to develop severity measures that will be useful in developing guidelines in decision making for management for LRI in infants. The focus is on hospitalization because this is a high stakes decision arena for emergency department and office-based providers. Consequences of false positives (unnecessary hospitalizations) are expensive, and potential consequences of false negatives (infant death in a child unsafely sent home) are disastrous. Specific objectives follow. (1a) to develop and (1b) to validate a physiologically-based measure of respiratory illness severity termed the Acute Infant Respiratory Dysfunction index (AIR-DI). Development of the AIR-DI will be based on physiologic variables recognized as indicators of respiratory dysfunction. Validation will be based on overall severity estimates made by physicians, measures of severity based on nursing resource allocation and on therapy, and risk indicators. (2) To develop a prediction rule for peak AIR-DI (worst dysfunction during the illness) based on observations available to physicians when hospitalization decisions are made. The prediction rule, termed the Risk for Respiratory Dysfuntion (RRD) Score, will predict severe respiratory dysfunction, a dichotomous variable based on the AIR-DI. The RRD, developed empirically from analysis of LRI episodes in a derivation data set and assessed through a validation data set, will represent the ability of physicians to predict outcome on the basis of risk factors and careful clinical evaluation. (3) Using decision modeling techniques, to test the cost-effectiveness of basing hospitalization decisions on the prediction rule. The decision model will be structured to address decision alternatives (hospitalize or not) considering the proximate clinical outcome (severe respiratory dysfunction) and final outcomes (e.g., recovery, death, vulnerable child syndrome) identified by a sample of physicians and a sample of parents. Utilities for decision analysis will be based on preferences of physicians and parents for outcome states in the decision model. This research involves the fields of psychobiology, neuroscience, nutrition, and gastrointestinal physiology. It will advance our basic knowledge of the psychobiology of carbohydrate and fat appetite and the development of food preferences. The findings may provide practical benefits for current attempts to alter dietary fat and carbohydrate intake and control adiposity in humans.
