Online Program

WITHDRAWN: Derivation of Clinical Prediction Rules for Sever Dehydration of Children Living in Resource-Limited Settings: A Dhaka Study

Chieh Chu, Brown University 
Adam C. Levine, Brown University 
Christopher Schmid, Brown University  

Keywords: Dehydration, Diagnostic, Prediction, Children

Each year, diarrhea-induced dehydration caused 700,000 child deaths worldwide. Current assessment tools like the WHO scale and CDC scale have never been prospectively validated, or were developed in high income areas. This study aimed to establish an accurate clinical assessment of severe dehydration which can be applied to low income countries. The research setting took place in Dhaka Hospital, where a sample of children under age 5 with a history of diarrhea were prospectively enrolled. 10 symptoms of dehydration (eyes, general appearance, heart rate, mucous membranes, radial pulse, respiration, skin pinch, tears, thirst, and urine) were selected from the existing scales. Dehydration was measured by weight change upon rehydration. The logistic model selected eyes, general appearance, heart rate, skin pinch and tears as predictors. The adjusted c-statistic was 0.774. The ordinal model included eyes, general appearance, respiration, skin pinch and tears. The adjusted c-statistic was 0.786. The classification tree split on general appearance, eyes, skin pinch and respiration, and the adjusted c-statistic was 0.739. A strong interaction was found between eyes and general appearance.