Clinician Management Practices for Infants With Hypothermia in the Emergency Department

We conducted a survey of ED clinicians from 32 US pediatric hospitals participating in a multicenter study evaluating risk factors for SBI among infants with hypothermia presenting to the ED conducted by the Pediatric Emergency Medicine Clinical Research Committee. We identified sites through outreach after the initiation of this study to sites that had previously worked on a multicenter study within the Pediatric Emergency Medicine Clinical Research Committee or had expressed an interest in participating. Each site had 1 principal investigator (PI), who identified and distributed the survey to eligible participants at their respective sites. Eligible clinicians were those who self-reported that they worked “primarily in the ED.” This study was approved by the Ann & Robert H. Lurie Children’s Hospital of Chicago Institutional Review Board.

We iteratively developed a survey instrument to characterize clinician management approaches to infants with hypothermia. Four authors (SR, CG, PA, and AR) developed key questions related to the management of infants with hypothermia with a focus on the evaluation and treatment of SBI. Questions pertaining to diagnostic testing and management were queried within the age groups of 0 to 7, 8 to 21, 22 to 28, and 29 to 60 days, following the age categorization provided in the recent American Academy of Pediatrics febrile infant clinical practice guideline.¹¹ 

After the initial survey draft, 9 experts (1 expert in survey methodology, 4 clinicians in pediatric emergency medicine, 1 in emergency medicine, 1 in pediatric hospital medicine, and 2 in pediatric critical care) reviewed the survey materials to ensure content and response validity. Additionally, we conducted 4 cognitive interviews (2 with pediatric emergency medicine physicians and 2 with pediatric hospitalists) to obtain additional feedback on the content and clarity of questions, with subsequent revisions performed on the basis of these responses. Survey content included: (1) the presence of a formal guideline or pathway for the management of infants with hypothermia at the respondent’s institution, (2) confidence in managing an infant with hypothermia (relative to an infant with fever) using a 5-point Likert scale, (3) defining a cutoff temperature for hypothermia, (4) clinical testing and management for infants with hypothermia, (5) alternative diagnoses other than SBI and HSV in this population, and (6) whether management practices might change if there is documented hypothermia before ED arrival, but none during the ED encounter. The survey was developed and distributed on the Qualtrics platform (Provo, UT). The full survey instrument is described in the Supplemental Information.

Eligible participants included attending physicians, fellows in pediatric emergency medicine, nurse practitioners, and physician assistants/associates. The first question of the survey identified ineligible participants (those who did not work primarily in the ED). As an incentive, participants had the option to participate in an anonymous gift card raffle.

Individual e-mail notices were sent to site PIs at all 32 institutions with solicitation suggestions for all eligible participants at their respective sites. Site PIs were responsible for contacting eligible participants at their sites through individual emails, with 1 follow-up e-mail sent ∼2 weeks after the initial request. We collected data on the number of eligible participants contacted by each site PI to determine the overall response rate. Surveys were distributed from December 1, 2022 to March 15, 2023.

We calculated the completed survey response rate across all sites. We compared the relative confidence in the management of infants with hypothermia versus fever using the Wilcoxon rank test. Considering each hospital as an independent cluster, we explored whether within-hospital response rates for testing and treatment preferences were related to the number of respondents from each hospital.¹²  We presented raw percentages for all demographic variables about clinicians and cluster-weighted percentages for all other variables.¹³  We described differences in the testing and treatment of SBI, HSV, and respiratory viral infections, sub-analyzing by age group. To evaluate age-based associations in testing and treatment decisions, we constructed a univariable generalized linear mixed model for each decision, treating age group as a categorical variable and random intercepts per respondent. We a priori categorized free-text responses on alternative diagnoses besides SBIs and HSV. We further stratified results into groups by attending physician, fellow, nurse practitioner, and physician assistant/associate. We then grouped responses by institution and evaluated institutional differences in testing for the 4 age groups via boxplots. Analyses were performed by using the lme4 (1.1-34)¹⁴  and htestClust (v0.2.2)^12,13  packages in R, version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

A total of 1935 surveys were distributed to 1001 pediatric emergency medicine physicians, 109 emergency medicine physicians, 187 general pediatricians, 266 fellows, 214 nurse practitioners, and 158 physician assistants/associates. A response was initiated in 1508 (77.9%). Of these, 16 respondents were considered ineligible based on primary work location, and 261 (17.3%) surveys were incomplete. The remaining 1231 (63.6% of all distributed surveys) surveys were included for analysis. The survey respondents included 711 pediatric emergency medicine attending physicians, 32 general emergency medicine attending physicians, 99 general pediatricians, 199 fellows, 105 nurse practitioners, 76 physician assistants/associates, and 9 attending physicians with a different specialty (6 with a pediatric subspeciality, 3 not stated). Demographics of the participants are provided in Table 1. Most endorsed that their primary ED site was at a freestanding children’s hospital (82.9%). The median number of respondents per site was 55 (interquartile range [IQR] 38–74). There was no significant association between the number of respondents per site (cluster size) and respondent answers for testing or treatment decisions.

The median threshold to define hypothermia was ≤36.0°C (IQR 35.9–36.0), which constituted most responses. Fewer respondents selected different thresholds, such as ≤35.5°C and ≤36.5°C (Fig 1). A total of 160 respondents (12.8%) stated that they do not have a specific threshold for hypothermia.

A plurality of respondents (47.1%) stated that their ED did not have a formal pathway, and 25.0% were unsure. Among those who did have an institutional guideline for the management of infants with hypothermia (27.9%), 93.2% stated that it was included within their guidelines for infants with fever.

The median point scoring of confidence for infants with hypothermia was 4 (IQR 3–4) on a 1 to 5 scale, with 5 representing the most confidence. The median point scoring of confidence of febrile infants was higher (median score 5, IQR 4–5; P < .01). A total of 834 (67.7%) respondents reported caring for at least 1 infant with hypothermia over the past 6 months.

Testing preferences are provided in Table 2. When choosing to obtain blood testing for SBI, nearly all respondents endorsed that they would obtain a blood culture (98.9%) and a complete blood count with differential (98.7%). When testing for HSV, 74.9% of clinicians stated that they would obtain a cerebrospinal fluid (CSF) polymerase chain reaction (PCR), with a lower proportion stating that they would obtain liver function tests (57.3%), whole blood PCR (40.8%), or surface swabs (39.9%).

Clinical testing preferences varied by patient age (Fig 2). The proportion of clinicians who would obtain blood and urine tests from an infant with hypothermia was high among younger age groups, then subsequently declined. In generalized linear mixed models, blood, urine, and CSF testing decisions for SBI were similar between the 0- to 7-day and 8- to 21-day age groups but were significantly lower in the 22- to 28-day and 29- to 60-day age groups relative to the 0- to 7-day age group. HSV and respiratory viral testing were significantly lower in all age groups relative to the 0- to 7-day age group.

Most respondents (68.0%) stated that they would obtain testing if hypothermia was only noted before the encounter. Among those who would obtain testing, 89.7% stated that they would obtain this testing if the hypothermia were documented at a physician’s office, whereas 23.3% stated that they would test if it was documented only at home.

The proportion of clinicians who would change their testing plan if the infant was no longer hypothermic after bundling and warming in the ED ranged from 31.7% for infants 0 to 7 days of age to 79.9% for infants 29 to 60 days of age. When determining a change in plan for these infants, 66.7% endorsed that they would do fewer diagnostic tests, with the remainder endorsing that they would perform no diagnostic tests.

Clinicians endorsed that they would use empirical antibiotics differently based on age, ranging from 91.3% of infants 0 to 7 days of age to 19.3% of infants 29 to 60 days of age (Fig 3). Similarly, clinicians more frequently endorsed that they would use acyclovir among infants 0 to 7 days of age (76.3%) and least frequently among infants 29 to 60 days of age (4.1%). Patterns of hospitalization similarly revealed a downward trend with age (81.5% among infants 0–7 days of age, 6.4% among infants 29–60 days of age). In generalized linear mixed models, antibiotic, acyclovir, and admission dispositions were significantly lower in all age groups relative to the 0- to 7-day age group.

Alternative etiologies for hypothermia were provided by 1113 respondents, including environmental exposure (27.2%), infectious processes unrelated to SBI or HSV (11.5%), and metabolic disorders (9.6%; Table 3).

Testing preferences for SBI for blood and urine revealed little variation among the included hospitals except for older infants (29–60 days; Fig 4). Testing preferences for CSF revealed greater variability in the 22- to 28-day age group. There was greater variability in antibiotic and acyclovir preferences for infants 22 to 28 days of age relative to older age groups and for admission decisions relative to other testing and treatment metrics.

Demographics stratified by background are provided in Supplemental Table 4. Overall comfort with infants with hypothermia was lower among nonattending clinicians (Supplemental Table 5). Nonattending clinicians reported that they would perform more testing for SBI relative to attending physicians (Supplemental Figs 5 and 6).

We performed a multicenter survey of ED clinicians working in children’s hospitals to evaluate testing patterns for the evaluation of infants with hypothermia. Most clinicians defined hypothermia using a cutoff of ≤36.0°C. Confidence in managing infants with hypothermia was high, although respondents were less confident managing infants with hypothermia compared with those with fever. Our findings reveal the variability in the definition of hypothermia in the ED and testing and treatment among various age groups, reflecting the need for improved evidence to optimize the management of these patients.

The preferred hypothermia cutoff of ≤36.0°C in our study may stem from the Pediatric Sepsis Consortium Criteria, which defines temperatures of <36°C and >38.5°C as a component of the systemic inflammatory response syndrome.¹⁵  This contrasts with broader definitions of hypothermia, such as ≤36.4°C used in World Health Organization criteria,¹⁶  and narrower definitions, such as ≤35.5°C or lower reported in some older literature.^17,18  Our findings correspond with previous retrospective studies,^7,19  with one revealing that blood culture was performed more frequently among infants with lower temperatures (median temperature of 36.0°C for those who had a blood culture performed, vs 36.3°C for those who did not have a blood culture performed).¹⁹  Previous work has also revealed that SBIs are more frequent with a lower minimum temperature, potentially supporting 36.0°C as a target cutoff for future studies.²⁰ 

We demonstrate clinician preferences for the diagnostic evaluation of infants with hypothermia that carry both similarities and differences from clinical practice guidelines for febrile infants.¹¹  Differing from clinical practice guidelines for febrile infants,¹¹  which recommend urinalysis and blood culture acquisition in all infants ≤60 days of age, a high proportion of respondents reported that they might not perform urinalysis or blood cultures in infants >28 days of age. In one multicenter study, younger infants with hypothermia were more likely to get blood testing compared with those who did not (median age 13 days vs 26 days).¹⁹  One previous retrospective study of 360 infants with a temperature <36°C in the ED revealed that all cases of SBI occurred in those ≤28 days of age. Among those included, a higher proportion of SBIs occurred in infants 15 to 28 days of age (4/43; 9.3%) compared with those 0 to 14 days of age.⁴  This finding supports clinician preferences for diagnostic testing for younger infants with hypothermia. Other practice preferences, such as greater use of empirical antibiotics and hospitalization, generally corroborate practice patterns for febrile infants.

We additionally identified differences in practice patterns for other characteristics of hypothermic infants that warrant further investigation. A high proportion of clinicians would defer testing a euthermic infant who only had hypothermia measured at home, but not when measured at a physician’s office, which is likely reflective of the greater faith of survey respondents in the validity of temperature measurements reported in physician offices than by parents at home. One recent single-center study did not identify any cases of SBI among 212 well-appearing infants referred from primary care clinics for a temperature <36.5°C over an 8.5-year period.²¹ 

Our study reveals a strong interest on the use of inflammatory markers for managing hypothermic infants. Although some components of the complete blood count with differential, including the white blood cell count and absolute neutrophil count, have revealed an association with SBI in 1 study,¹⁹  their role has not been more broadly investigated. One case series of infants with hypothermia revealed that nearly one-half of those with invasive bacterial infections (defined as meningitis and/or bacteremia) had a normal white blood cell count, revealing the limited role of this blood test. No study has evaluated the utility of other inflammatory markers, such as procalcitonin and C-reactive protein, which are being increasingly used in this population.¹⁰ 

We identify the important aspects and needs of future research on the management of infants with hypothermia. The American Academy of Pediatrics practice guideline for the management of febrile infants, guided by large-scale prospective studies of febrile infants,^22,23  provides little guidance for the management of infants with hypothermia and only suggests that HSV testing should be considered for hypothermic infants.¹¹  Notably, when identifying preferences for HSV testing, many clinicians elected to only get HSV CSF studies, which differs from recommended practices for the diagnosis of this infection.²⁴ 

Our findings are subject to limitations. The survey base for this study was composed of clinicians from children’s hospitals. These findings may, therefore, lack generalizability to other settings. Most respondents were pediatric emergency medicine attending physicians. Because we had fewer responses from general emergency medicine physicians, we were unable to compare practice differences between these 2 specialty types. We used targeted emails to improve our response rate; although high (>60%), nonresponses may represent a source of bias. The survey did not include any internal measures of validity, and survey responses do not necessarily equate to actual clinical practice. Because we wished to keep survey prompts brief, we focused on relatively simple case presentations that may have lacked sufficient specificity for respondents. Finally, because we focused on the ED evaluation of infants with hypothermia, our findings are not reflective of the opinions of clinicians who may care for these infants in other settings, including in the primary care office, neonatal intensive care unit, or hospital ward. Despite these limitations, the findings from this survey reveal key gaps in knowledge provided to infants with hypothermia and identify potential discrepancies in practice in need of further research.

In this survey of clinician preferences for the diagnosis and management of infants with hypothermia, we identified age-based patterns in testing and treatment that largely mirrored those of infants with fever. A temperature cutoff of ≤36.0°C was most frequently used to define hypothermia among respondents. Clinicians generally felt more confident in caring for infants with fever than those with hypothermia. There was variation in testing and treatment preferences by age, particularly for older infants, among the studied institutions. More research is required to safely risk-stratify these infants to avoid under- and overutilization of diagnostic testing and treatment modalities. Future efforts targeted toward the implementation of study findings may benefit by providing real-time guidance on the management of these infants within electronic health record systems to ensure appropriate and targeted management practices for these children to avoid missing potentially serious infections while preventing overutilization of resources.

Yale School of Medicine: Sofia Grigoria Athanasopoulou, MD (New Haven, CT); Children’s Hospital of Los Angeles: Marsha Elkhunovich, MD (Los Angeles, CA); University Hospitals Cleveland Medical Center/Rainbow Babies and Children’s Hospital: Megan Evers DO, MS, MSc (Cleveland, OH); Boston Children’s Hospital: Alexandra Geanacopoulos, MD (Boston, MA); Children’s Hospital of Orange County: Theodore Heyming, MD (Orange, CA); University of Colorado School of Medicine, Children’s Hospital Colorado: Jonathan R. Higgins, MD (Aurora, CO); Medical College of Wisconsin: Jamie Lynn Holland, MD (Milwaukee, WI); Warren Alpert Medical School of Brown University: Sara E. Holmstrom, MD (Providence, RI); Johns Hopkins Children’s Center: Mary Beth Howard, MD, MSc (Baltimore, MD); Norton Children’s Hospital, University of Louisville School of Medicine: Adam Isacoff, MD, MSc, MBA (Louisville, KY); Children’s Hospital of Michigan: Nirupama Kannikeswaran, MBBS (Detroit, MI); Emory University School of Medicine, Children’s Healthcare of Atlanta: Naghma S. Khan, MD (Atlanta, GA); Weill Cornell Medicine: Yu Hsiang Lo, MD (New York, NY); Columbia University Vagelos College of Physicians and Surgeons: Son H. McLaren, MD, MS (New York, NY); University of Utah School of Medicine: Nathan M. Money, DO (Salt Lake City, UT); Massachusetts General Hospital: Nicole Nadeau, MD (Boston, MA); UPMC Children’s Hospital of Pittsburgh: AG Nuwan Perera, MD (Pittsburgh, PA); Medical University of South Carolina: Christopher M. Pruitt, MD (Charleston, SC); St Louis Children’s Hospital, Washington University in St Louis: James Rudloff, MD (St Louis, MO); Children’s Hospital of Philadelphia: Laura F. Sartori, MD, MPH (Philadelphia, PA); Harbor-UCLA Medical Center: John Stockton, MD, MPH (Los Angeles, CA); Nemours Children’s Hospital Delaware: Amy D. Thompson, MD (Wilmington, DE); University of Washington: Indi Trehan, MD, MPH, DTM&H (Seattle, WA); Rady Children’s Hospital: Stacey Ulrich, MD (San Diego, CA); University of Alabama at Birmingham: Alicia Webb, MD (Birmingham, AL); University of Cincinnati College of Medicine: Paria M. Wilson, MD, MEd (Cincinnati, OH); Monroe Carell Jr Children's Hospital at Vanderbilt: Amelia Wong, MD (Nashville, TN); UT Southwestern Medical Center, Children’s Medical Center Dallas: Amy Z. Zhou, MD, PhD (Dallas, TX).

We thank Marie Heffernan, PhD, Michelle L. Macy, MD, MS, Todd A. Florin, MD, MSCE, Jennifer A. Hoffmann, MD, MS, Elizabeth C. Powell, MD, MPH, Megan M. Attridge, MD, MS (Northwestern University Feinberg School of Medicine), Jillian Cotter, MD, MSCS (University of Colorado School of Medicine), Andrea V. Rivera-Sepulveda, MD (Nemours Children’s Health), Gayatri Boddupalli Madduri, MD (Stanford University School of Medicine), and Jonathan Pelletier, MD (Northeast Ohio Medical University) for assistance in development of the survey instrument.

CSF

cerebrospinal fluid

ED

emergency department

HSV

herpes simplex virus

IQR

interquartile range

PI

principal investigator

SBI

serious bacterial infection