Western Journal of Emergency Medicine: Integrating Emergency Care with Population Health

Introduction: Fever as an indicator of infection is frequently used as an aid in triggering concern for sepsis in the emergency department (ED). Adults with sepsis presenting to the ED with a normal temperature have been shown to have delays in treatment and greater mortality. The association between temperature and timeliness of sepsis-related care in the ED remains poorly characterized in children. Our objective in this study was to measure the association between body temperature at the physiologic onset of sepsis and the time to initiation of antibiotic treatment and fluid bolus among children with clinically defined sepsis.

Methods: We conducted a retrospective, cohort study of pediatric patients with sepsis presenting to the ED. Data collected from an existing quality improvement database were supplemented via chart extraction. We assessed body temperature at physiologic onset of sepsis (PO-S), the date and time when a patient first met clinical criteria for sepsis as defined by Goldstein et al.1 Our primary outcomes were time from PO-S and administration of antibiotics and fluid bolus. Secondary outcomes included maximum vasoactive-inotropic scores, need for extracorporeal membrane oxygenation (ECMO) within 30 days of presentation, presence and type of organ dysfunction, 30-day hospital- and intensive care unit (ICU)-free days, and mortality. We summarized and compared data  by temperature group. Multivariable quantile regression was used to evaluate adjusted associations between body temperature and time to initiation of antibiotic treatment and fluid bolus.

Results: Of 928 patients screened, 385 (41%) met inclusion criteria. Median time to antibiotic treatment did not differ between temperature groups at PO-S—≤ 36.0 °C: median (IQR) 48.5, (41.3-104.8); 36.1-37.9 oC: median, 95.5, (41.3-104.8;), and ≥ 38.0 oC: median 84, 45-151; (P = .24). Median time to fluid bolus administration also did not differ between temperature groups at PO-S—≤ 36.0 °C: median 39, (20.8-65.8); 36.1-37.9 oC: median, 42.5 (21.3-86.3); and ≥ 38.0 oC: median, 54 (29-84); (P =.07). In addition, mortality differed by temperature at PO-S (≤ 36.0 °C: 1/22 (4.5%); 36.1-37.9 oC: 4/80 (5.0%); and ≥. 38.0 oC: 3/283 (1.1%), (P = .04); as did organ dysfunction at 72 hours: ≤. 36.0 °C: 15/22 (68.2%); 36.1-37.9 oC: 43/80 (53.8%), ≥ 38.0 oC: 74/283 (26.1%); (P < .001) and median (IQR) 30-day ICU- and hospital-free days—≤ 36.0 °C: median, 24, (20,-26.8); 36.1-37.9 oC: median, 28 (24.8-30), ≥ 38.0 oC: median, 30 (27-30), (P < .001); and at ≤. 36.0°C: median, 22, (17-25); 36.1-37.9 oC: median, 24 (17.8-27); ≥ 38.0 oC: median, 25 (20, 27), (P = .04), respectively. We did not observe an association between temperature and median time to antibiotic administration (β: 2.5, 95% CI, -4.2 to 9.1, P = .50) or first fluid bolus administration (β: 1.7, 95% CI, -1.4 to 4.8, P = .30).

Conclusion: Time to fluid bolus administration and time to antibiotic administration did not differ statistically by temperature from physiological onset of sepsis. Children presenting with hypothermia (≤ 36.0 °C) had worse outcomes.