The Association Between Fever and Subsequent Deterioration Among Hospitalized Children With Elevated PEWS

Abstract

A B S T R A C TOBJECTIVES: To evaluate the association between fever and subsequent deterioration among patients with Pediatric Early Warning Score (PEWS) elevations to $4 to inform improvements to care escalation processes at our institution. METHODS:We performed a cohort study of hospitalized children at a single quaternary children's hospital with PEWS elevations to $4 between January 1, 2014 and March 31, 2014. Bivariable analysis was used to compare characteristics between patients with and without unplanned ICU transfers and critical deterioration events (CDEs) (ie, unplanned ICU transfers with life-sustaining interventions initiated in the first 12 ICU hours). A multivariable Poisson regression was used to assess the relative risk of unplanned ICU transfers and CDEs. RESULTS:The study population included 220 PEWS elevations from 176 unique patients. Of those, 33% had fever (n 5 73), 40% experienced an unplanned ICU transfer (n 5 88), and 19% experienced CDEs (n 5 42). Bivariable analysis revealed that febrile patients were less likely to experience an unplanned ICU transfer than those without fever. The same association was found in multivariable analysis with only marginal significance (adjusted relative risk 0.68; 95% confidence interval 0.45-1.01; P 5 .058). There was no difference in the CDE risk for febrile versus afebrile patients (adjusted relative risk 0.79; 95% confidence interval 0.43-1.44; P 5 .44). CONCLUSIONS:At our institution, patients with an elevated PEWS appeared less likely to experience an unplanned ICU transfer if they were febrile. We were underpowered to evaluate the effect on CDEs. These findings contributed to our recognition that (1) PEWS may not include all relevant clinical factors used for clinical decision-making regarding care escalation and (2) further study is needed in this area. the section head of pediatric hospital medicine who assisted with the study design and provided mentorship; Mr Thomas is a biostatistician who performed the statistical analysis; Ms Wathen is a clinical practice specialist in the PICU at Children' s Hospital Colorado who assisted in study design and data collection; Dr Brittan is a pediatric hospitalist who assisted with the study design and data analysis; Ms Iwanowski is a quality improvement specialist who assisted in the study design and data collection; Dr McLeod is a pediatric hospitalist who assisted with study design, assisted with data analysis, and provided mentorship; and all authors approved the final manuscript as submitted. Total PEWS, median (IQR) 4.0 (4.0-5.0) PEWS subset scores, median (IQR) Cardiovascular 1.0 (1.0-1.0) Respiratory 2.0 (1.0-3.0) Behavior 1.0 (1.0-2.0) Change from previous PEWS, median (IQR) 2.0 (1.0-3.0)Acute HHFNC, n (%) 61 (28) Continuous albuterol, n (%) 19 (9) Nighttime event, n (%) 99 (45) IQR, interquartile range.

Development and External Validation of a Machine Learning Model for Prediction of Potential Transfer to the PICU

“…Our finding that the F io 2 inhaled by a child is important to predicting ICU transfer is consistent with established literature (36, 37). Other variables deemed important by pCART include temperature, blood pressure, WBC count, and platelets, thereby indicating the presence of infection or inflammatory conditions could lead to deterioration and illness (38–40). We also demonstrate through case analysis of model explanations for an individual patient that pCART is appropriately incorporating information to make prediction, thereby aligning with clinical intuition.…”