Abstract
OBJECTIVE: Severe asthma exacerbations account for a large share of asthma morbidity, mortality, and costs. Here, we aim to identify early predictive factors associated with pediatric intensive care unit (PICU) admission. METHODS: We performed a retrospective observational study of 5,185 emergency department (ED) encounters at a large children's hospital, including 86 (1.7%) resulting in PICU admission between 10/1/2015 and 8/7/2018 with ICD9/ICD10 codes for "asthma," "bronchospasm," or "wheezing." Vital signs and demographic information were obtained from electronic health record data and analyzed for each encounter. Predictive factors were identified using adjusted regression models, and our primary outcome was PICU admission. RESULTS: Higher mean heart rates (HRs) and respiratory rates (RRs), and lower SpO2 within the first hour of ED presentation were independently associated with PICU admission. Odds of PICU admission increased 70% for each 10 beats/min higher HR, 125% for each 10 breaths/min higher RR, and 34% for each 5% lower SpO2. A binary predictive index using 1-h vitals yielded OR 13.4 (95% CI 8.1-22.1) for PICU admission, area under receiver operator characteristic (AUROC) curve 0.84 and overall accuracy of 80.1%. Results were largely unchanged (AUROC 0.84-0.88) after adjusting for surrogates of asthma severity and initial ED management. In combination with a secondary standardized clinical asthma distress score, positive predictive value increased by sevenfold (6.1%-46%). CONCLUSIONS: A predictive index using HR, RR, and SpO2 within the first hour of ED presentation accurately predicted PICU admission in this cohort. Automated vital signs trend analysis may help identify vulnerable patients quickly upon presentation.