73 - Threshold of Hypothermia to Detect Serious Bacterial Infection in Young Infants

Poster Number: 73
Publication Number: 73.105

Yu Hsiang J. Lo, University of Michigan, Ann Arbor, MI, United States; Christopher Graves, WakeMed Children's Hospital WakeMed Health and Hospitals, Raleigh, NC, United States; Jamie Holland, University of Utah School of Medicine, Salt Lake City, UT, United States; Alexander J. Rogers, University of Michigan Medical School, Ann Arbor, MI, United States; Nathan Money, University of Utah School of Medicine, Salt Lake City, UT, United States; Andrew N. Hashikawa, University of Michigan Medical School, Dexter, MI, United States; Sriram Ramgopal, Ann & Robert H. Lurie Children's Hospital of Chicago, Chicago, IL, United States

Background: Hypothermia is a risk factor for serious bacterial infections (SBI) among young infants presenting to the emergency department (ED). There is no universally accepted temperature threshold for hypothermia to detect SBI in this population.

Objective: To describe the prevalence of SBI, and optimal temperature to define hypothermia among young infants presenting to the ED.

Design/Methods: We performed a retrospective cross-sectional study of infants ≤90 days old presenting to 4 academic pediatric EDs from 1/1/2015 to 12/31/2019. We included infants with any measured rectal temperature of ≤36.5°C during their ED stay (suggested World Health Organization criteria). Our primary outcomes were SBI (bacteremia, bacterial meningitis, and/or UTI) and invasive bacterial infection (IBI; bacteremia and/or bacterial meningitis). We constructed receiver operating characteristic (ROC) for each outcome, calculated areas under the ROC curves (AUROC), and determined an optimal cutoff value for hypothermic temperature.

Results: Of 3,743 patients included in our study, 54.6% were younger than 30 days old (median age 25 days, IQR 6-52), with a median minimum temperature of 36.3°C (IQR 36.1-36.4°C). Blood cultures were obtained in 25.8% of patients, urine culture in 23.2%, and CSF culture in 13.0%. SBI was found in 78 (2.1%) and IBI in 16 (0.4%) patients. The most common organism associated with SBI was Escherichia coli [Table 1]. Among patients with SBI, the median temperature was 36.1°C (IQR 36.0-36.5°C), which was similar to the median temperature among patients without SBI (36.2°C; IQR 36.1-36.4 °C; p=0.25 by the Wilcoxon rank-sum test). A similar finding was noted with respect to IBI (median temperature 36.0°C, IQR 35.8-36.4°C, among patients with IBI compared to median temperature 36.2°C, IQR 36.1-36.4°C, among patients without IBI; p=0.27). For SBI, the AUROC was 0.56. Using a cutoff of 36.3°C resulted in a sensitivity of 55% and specificity of 54%. For IBI, the AUROC was 0.54. A cutoff temperature of 36.1°C resulted in a sensitivity of 50% and specificity of 70% [Figure]. Analysis of sensitivity and specificity in 0.2°C intervals from 36.3 to 34.9°C showed compromises in the detection of SBI and IBI [Table 2]. AUROC and metrics of diagnostic accuracy remained poor when inclusion was limited to the subset with blood culture performed (n=967).Conclusion(s): A low proportion of infants with hypothermia in the ED have SBIs (2.1%). No cutoff temperature could be identified with satisfactory diagnostic accuracy for SBI or IBI. Further investigations are needed to identify clinical risk factors for this vulnerable population.
Yu Hsiang Johnny Lo CVPAS Award CV.pdf
SBI ProfileSerious bacterial infections (SBI) identified in study cohort, organized by infection type.