(1305) Hyperthermia and Hypothermia Detection Inconsistencies Across Racial and Ethnic Groups

Introduction: Accurate measurement of body temperature is a critical aspect in diagnosis, but previous studies have shown discrepancies in temperature reading depending on measurement location. This study explores this variability further by looking at patients of different self-identified racial groups. Specifically, the study analyzes temperature inconsistencies for patients with darker skin pigmentation, as proxied from self-identified race/ethnicity, from missed fever and missed hypothermia from temporal thermometers.

Methods: The study harmonized three EHR databases (MIMIC-IV, eICU-CRD-1, eICU-CRD-2) of US ICU patients (2008-2022). Temporal and contact (rectal, core, oral) readings were aligned within 1 hour and combined with sociodemographic and clinical parameters. Patients were excluded if they had no pairs or temperature measurements of > 45°C and < 30°C. The outcomes were defined as: hidden fever (temporal ≤ 38°C, contact > 38°C), hidden hypothermia (temporal ≥ 36°C, contact < 36°C), and the primary outcome hidden hyperthermia or hypothermia (HHH). Generalized estimating equations assessed outcome likelihoods across races and ethnicities, adjusting for sociodemographic and clinical variables.

Results: The created dataset contained 13,251 paired temperature measurements. A total of 6,783 first temperature pairs were studied across 6,725 hospitalizations, with the cohort consisting of 0.5% American Indian/Alaska Native (11.1% HHH), 1.9% Asian (12.6%), 10.3% Black (10.1%), 2.8% Hispanic/Latino (9.9%), and 77.2% White (6.6%). Patients had significantly higher adjusted odds of HHH compared to White patients for the following groups: Asian [OR (95% CI): 2.01 (1.18, 3.44), p = 0.010] and Black [OR (95% CI): 1.71 (1.29, 2.26), p < 0.001]. Higher adjusted odds were also observed for American Indian/Alaska Native [OR (95% CI): 1.92 (0.67, 5.47), p = 0.225] and Hispanic/Latino [OR (95% CI): 1.56 (0.95, 2.54), p = 0.076].

Conclusions: We observe an association between racial/ethnic groups and HHH. With significantly higher adjusted odds of having HHH, patients of Asian and Black backgrounds may require alternative methods to record temperature to mitigate the risk of misdiagnosis and delayed treatment.