Development and Internal Validation of a Predictive Model Including Pulse Oximetry for Hospitalization of Under-Five Children in Bangladesh

• DOI: 10.1371/journal.pone.0143213
• Publication Date: 2015-11-18T14:01:06Z
• Authors: Shahreen Raihana; Dustin Dunsmuir; Tanvir Huda; Guohai Zhou; Qazi Sadeq-ur Rahman; Ainara Garde; Md Moinuddin; Walter Karlen; Guy A. Dumont; Niranjan Kissoon; Shams El Arifeen; Charles Larson; J. Mark Ansermino
• Keywords: Lethargy; Pulse Oximetry; Interquartile range; Tachypnea; Irritability; Respiratory Rate

Background The reduction in the deaths of millions children who die from infectious diseases requires early initiation treatment and improved access to care available health facilities. A major challenge is lack objective evidence guide front line workers community recognize critical illness earlier their course. Methods We undertook a prospective observational study less than 5 years age presenting at outpatient or emergency department rural tertiary hospital between October 2012 April 2013. Study physicians collected clinical signs symptoms facility records, with mobile application performed recordings oxygen saturation, heart rate respiratory rate. Facility decided need for admission without knowledge saturation. Multiple logistic predictive models were tested. Findings Twenty-five percent 3374 assessed children, median (interquartile range) 1.02 (0.42–2.24), admitted hospital. unable contact 20% subjects after visit. regression model using continuous rate, temperature combined dichotomous chest indrawing, lethargy, irritability cough, diarrhea fast difficult breathing predicted an area under receiver operating characteristic curve 0.89 (95% confidence interval -CI: 0.87 0.90). At risk threshold 25% admission, sensitivity was 77% CI: 74% 80%), specificity 87% 86% 88%), positive value 70% 67% 73%) negative 91% 90% 92%). Conclusion combination readily obtained hospitalization critically ill children. External validation this setting will be required before adoption into practice.