Emergency Medicine

The association between the degree of fever as measured in the emergency department and clinical outcomes of hospitalized adult patients

a b s t r a c t

Background: Fever is a Physiologic response to a wide range of pathologies and one of the most common com- plaints and clinical signs in the emergency medicine department (ED). The association between fever magnitude and clinical outcomes has been evaluated in specific populations with inconsistent results.

Objectives: In this study we aimed to investigate the association between the degree of fever in the ED and clinical outcomes of hospitalized febrile adult patients.

Methods: This was a retrospective single-center cohort study of all the patients with maximal body temperature (BT) >= 38.0 ?C, as recorded during the ED evaluation, who were hospitalized between January 2015 and Decem- ber 2020. Patients with heatstroke were excluded. The primary outcome was 30-day all-cause mortality and sec- ondary outcomes were intensive care unit admission and development of Acute kidney injury .

Results: Fever was recorded among 8.1% of patients evaluated in the ED. Elevated BT was associated with in- creased risk of hospital admission (70.3% vs. 49.4%, p < 0.001), 30-day mortality (12.3% vs. 2.6%, p < 0.001),

ICU admission (5.7% vs. 2.8%, p < 0.001), and AKI 11.7% vs. 3.8%, p < 0.001).

After exclusion of nine patients with heatstroke, 21,252 hospitalized febrile patients were included in the final analysis. BT > 39.7 ?C was progressively associated with increased mortality (OR 1.64-2.22, 95% CI 1.16-2.81, p < 0.005) as compared to BT 38.0-38.1 ?C. More AKI events were observed in patients with BT > 39.5 ?C (OR 1.48-2.91, 95% CI 1.11-3.66, p < 0.007). Temperature between 39.2 and 39.5 ?C was associated with lower mor- tality (OR 0.62-0.71, 95% CI 0.51-0.87, p < 0.001). In a multiple logistic regression analysis BT > 39.9 ?C was in- dependently associated with increased mortality and AKI. BT > 39.7 ?C was progressively associated with an increased risk of ICU admission.

Conclusion: Among febrile patients admitted to the hospital, BT > 39.5 ?C was associated with adverse clinical course, as compared to patients with lower-grade fever (38.0-38.1 ?C). These patients should be flagged on ar- rival to the ED and likely warrant more aggressive evaluation and treatment.

(C) 2021

  1. Introduction

Fever is a physiologic response to a wide range of infective and in- flammatory pathologies and may have a positive effect on the function of the immune system. [1] On the other hand, fever increases metabolic rate and has deleterious effects on cardiac, respiratory, and nervous sys- tems and is associated with increased mortality in patients with stroke, head injury, and cardiac arrest. [2].

* Corresponding author.

E-mail address: [email protected] (E. Marcusohn).

Fever is one of the most common complaints and clinical signs among adult patients in the emergency department (ED), accounting for approximately 5% of visits and up to 15% of visits among the elderly. [3].

Body temperature (BT) is a vital sign routinely measured upon ar- rival to the ED and a feverish feeling is commonly reported by patients.

[4] Elevated BT is an important marker of a patient’s clinical status and its presence and magnitude often prompt changes in patients’ manage- ment and evaluation. Although fever does not necessarily reflect infec- tious pathology, as the fever rises so does the suspicion for underlying infection. [5] The association between fever magnitude and clinical


0735-6757/(C) 2021

outcomes has been evaluated in specific populations. In infants, fever above 40 ?C was associated with an increased risk of severe bacterial in- fection. [6,7] In adult patients admitted to intensive care units with an infectious disease, higher peak temperature during the first 24-h from admission was associated with lower in-hospital mortality. However, among patients who were not diagnosed with an infection, peak tem- perature above 39.0 ?C was associated with progressively increasing in-hospital mortality. [1] A meta-analysis of studies including only non-neurological critically ill patients (with or without infection) found that high-grade fever (39.3 ?C to 39.5 ?C) was associated with in- creased mortality. [8].

There is a relatively limited body of literature investigating the asso- ciation between fever magnitude as measured in the ED and clinical outcomes among febrile adult patients. In this study, we assessed the prognostic significance of fever magnitude in the ED among hospital-

ized adult patients who had a recorded temperature >= 38.0 ?C in the ED.

  1. Materials and methods
    1. Study design and setting

This retrospective cohort study included all patients with maximal BT of 38.0 ?C or higher in the ED who were hospitalized in Rambam Health Care Campus (RHCC) between January 2015 and December 2020. RHCC is a 1000-bed tertiary academic medical center serving more than two million residents. Over 60,000 patients are treated in the adult ED annually, and approximately 30% of them are admitted for further treatment and investigation.

Data analyzed in this study were retrieved from Prometheus, an in- tegrated electronic medical records (EMR) system used in our medical center. Data were extracted from the patients’ files using the MDClone system (Beer-Sheba, Israel). MDClone extracts data from the electronic medical records, including patients’ hospitalizations, coded diagnoses, medications, laboratory tests, demographics, and administrative infor- mation. The data are presented in an anonymous and standardized for- mat. The system allows the retrieval of a wide range of variables, in a defined time frame, around an index event. [9].

The study was approved by the Institutional Review Board of our medical center. The need for written informed consent was waived due to the retrospective study design.

    1. Participants

The study population included all consecutive patients with fever (de- fined as BT >= 38.0 ?C), as recorded during ED evaluation, who were admit- ted for further workup or treatment. If any patient had more than one visit that met the inclusion criteria during the study period, only the

First visit was included. Admitted patients who did not have fever (max- imal ED BT < 38.0) were assessed to provide context for our main out- comes of interest but were excluded for the remainder of the study.

In the absence of a single number that can unequivocally distinguish patients with abnormal BT, we chose 38 ?C as a cutoff. [10-12] Due to its high specificity (over 99%) for fever detection, the oral temperature measurement is used as the primary modality in our ED. [10] In RHCC, when oral temperature is below 38 ?C but the clinician suspects that the patient may be febrile, rectal temperature is measured upon re- quest. Additionally, patients with suspected heatstroke, hypothermia, and those unable to cooperate with oral temperature measurement have rectal temperature measured routinely. The highest temperature recorded during the ED evaluation was included in the analysis. Patients with heatstroke (ICD10 code T67) were excluded from the study.

    1. Parameters and outcome measures

The following parameters were extracted from the EMR. If more than one vital sign or laboratory measurement was recorded, the most

extreme value (in the direction indicated in parentheses below) was used:

  1. Demographic characteristics: age and gender.
  2. Comorbidities (based on ICD10 coding): diabetes mellitus, hyperten- sion, cerebrovascular diseases, ischemic heart disease, congestive heart failure, chronic kidney disease, chronic obstructive lung dis- ease, Chronic liver disease, and malignancy (solid and hematologic).
  3. Vital signs in the ED: body temperature (highest recorded), heart rate (highest recorded), systolic blood pressure (lowest recorded), and peripheral oxygen saturation while breathing ambient air (low- est recorded). We defined tachycardia as a heart rate above 100 bpm; hypotension as systolic blood pressure below 90 mmHg; and desaturation as oxygen saturation below 90% while breathing room air.
  4. Laboratory values at admission: white blood cell count (highest re- corded), hemoglobin (lowest recorded), platelet count (lowest re- corded), and serum creatinine (highest recorded).
  5. Hospital discharge Diagnosis codes (ICD10). The following ICD10 codes were defined as infectious diagnoses: A00-B99, G00-G09, H66, H67, H70, I38, I39, J00-J22, J85, J86, L00-L08, M00-M02, N10.

All other diagnoses were categorized as non-infectious.

The primary outcome of this study was 30-day all-cause mortality as recorded in the Ministry of Health database which is synchronized with the MDClone system. Secondary outcomes were [1] admission to an in- tensive care or intermediate care unit (abbreviated together as ICU) and

[2] Acute Kidney Injury , defined as an increase in serum creatinine to above 200% baseline, which corresponds to KDIGO (Kidney Disease Improving Global Outcomes) stage 2 or higher. [13].

    1. Statistical analysis

Patients’ characteristics were summarized with descriptive statis- tics. Continuous variables were described using medians with inter- quartile ranges (IQR). Categorical variables were described using percentages. Logistic regression was used to evaluate an association be- tween temperature and predefined outcomes. A multiple logistic re- gression model was created for each of the predefined outcomes using a backward selection procedure with a p-value threshold of 0.1 for the likelihood ratio test. The model associations were expressed as odds ra- tios (OR) with 95% confidence intervals (95% CI). Missing data were handled using list-wise deletion. We considered temperature as a cate- gorical variable and report OR relative to the lowest category (38.0-38.1 ?C). All available data were used for graphs generation. A p-value of 0.05 was considered statistically significant; corrections for multiple testing were not made in this exploratory study.

Data analysis was conducted with Statistical Package for the Social Sciences, version 25.0 (IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp) and MedCalc for Windows, version 15.0 (MedCalc Software, Ostend, Belgium).

  1. Results

Over the study period, 373,181 patients were evaluated in the ED of RHCC. BT was not documented among 361 patients and 1193 additional patients had BT < 36.0 ?C. Among the remaining 371,627 patients, 341,372 (91.9%) had normal BT of whom 168,742 (49.4%) were admit-

ted. Fever (BT >= 38.0 ?C) was recorded in 30,255 patients (8.1%), of

whom 21,261 (70.3%) were hospitalized (see Fig. 1). Among all 190,003 admitted patients (with any BT), the 30-day all-cause mortality rate was 3.7%. Two thousand six hundred eleven patients (12.3%) in the febrile cohort died within 30 days of the admission, compared to 2.6% of those hospitalized with normal BT (p < 0.001). Those with normal BT also had lower rates of ICU admission and AKI (2.8% vs. 5.7%, p < 0.001 and 3.8% vs. 11.7%, p < 0.001, respectively).

Image of Fig. 1

Fig. 1. Study flow chart.

    1. Hospitalized febrile patients

Among the patients hospitalized with elevated BT, nine had a pri- mary diagnosis of heatstroke and were excluded. Among the 21,252 pa- tients included in the final analysis, a high rate of comorbidities was apparent, with 69.5% having at least one documented comorbidity; 16,742 (78.8%) were admitted to medical departments and 4510 (21.2%) to surgical wards. One thousand fifty-five patients (5%) were transferred to ICU from the ED. The baseline demographic, clinical, and laboratory characteristics of the study cohort, as well as their clini- cal outcomes, are presented in Table 1. Rectal temperature measure- ment was used in the analysis of 9369 (44.1%) patients.

Patients’ characteristics by the maximal temperature category in the ED are presented in Table 2. Patients with BT >=40 ?C had higher rates of tachycardia, hypotension, and desaturation. The odds ratios for mortal-

ity, ICU admission, and AKI, by ED BT (using 38.0-38.1 ?C as the compar- ator), are presented in Fig. 2 and Table 1 in the supplementary material. Among the febrile patients, BT above 39.7 ?C was associated with in- creased 30-day all-cause mortality (39.8-39.9 ?C: OR 1.64, 95% CI 1.16-2.3, p = 0.005 and >= 40 ?C: OR 2.22, 95% CI 1.75-2.81, p < 0.001)

and increased ICU admissions (39.8-39.9 ?C: OR 2.28, 95% CI

1.50-3.46, p < 0.001 and >= 40 ?C: OR 2.43, 95% CI 1.79-3.31,

p < 0.001) as compared to BT 38.0-38.1 ?C. More AKI events were ob-

served in patients with fever above 39.5 ?C (39.6-39.7 ?C: OR 1.48, 95% CI 1.11-1.98, p = 0.007, 39.8-39.9 ?C: OR 1.68, 95% CI 1.18-2.40,

p = 0.004, and >= 40 ?C: OR 2.91, 95% CI 2.31-3.66, p < 0.001) than

those with BT 38.0-38.1 ?C. Temperature between 39.2 and 39.5 ?C

was associated with lower mortality rates (39.2-39.3 ?C: OR 0.71, 95% CI 0.59-0.87, p < 0.001 and 39.4-39.5 ?C: OR 0.62, 95% CI 0.51-0.74,

p < 0.001) but not with a decrease in other adverse outcomes. Fig. 3 and Tables 2 and 3 in the supplementary material present the ORs for mortality, ICU admission, and AKI according to age category (older

and younger than 65 years). Younger patients (<=65 years) with temper- atures above 39.8 ?C had higher ORs for all 3 of the adverse outcomes of

interest as compared to older patients (>65 years).

In a multiple logistic regression analysis that included demographic data, comorbidities, vital signs, and laboratory tests, BT >=40 ?C was inde- pendently associated with increased mortality (adjusted OR 1.63, 95% CI 1.25-2.12, p < 0.001) and increased risk of AKI (OR 2.29, 95% CI 1.81-2.9, p < 0.001). Temperature above 39.8 ?C was independently as-

sociated with increased risk of ICU admission (39.8-39.9 ?C: adjusted OR 1.90, 95% CI 1.22-2.96, p = 0.004 and >= 40 ?C: adjusted OR 2.11,

95% CI 1.55-2.89, p < 0.001). Patients with temperatures between

39.2 and 39.5 ?C had significantly lower mortality rates (39.2-39.3 ?C: adjusted OR of 0.76, 95% CI 0.62-0.93, p = 0.006 and 39.4-39.5 ?C: OR

Table 1 Demographic, clinical, and laboratory characteristics of 21,252 febrile patients included in the final analysis.

Age, years (IQR)

68.3 (60-80.3)

Male gender, n (%)

11,503 (54.1%)

Elderly (>65 years), n (%)

11,975 (56.3%)

Medical patients, n (%)

16,742 (78.8%)

Vital signs

HR, bpm (IQR)

100 (87-114)

Tachycardia (HR > 100 bpm), n (%)

10,134 (47.7%)

SBP, mmHg (IQR)

132 (117-150)

Hypotension (SBP < 90 mmHg), n (%)

598 (2.8%)

SO2, % (IQR)

96 (93-98)

Desaturation (SO2 < 90%), n (%)

2306 (11.0%)

Laboratory parameters

Leukocytes, Ku/L (IQR) 11.5 (7.9-15.8)

Hemoglobin, g/dL (IQR) 12.2 (10.6-13.6)

Platelets, 103 u/L (IQR) 207 (155-275)

Creatinine, mg/dL (IQR) 1.03 (0.8-1.4)


Diabetes Mellitus, n (%) 6552 (30.8%)

Hypertension, n (%) 11,532 (54.3%)

Cerebrovascular disease, n (%) 3525 (16.6%)

IHD, n (%) 3407 (16.0%)

CHF, n (%) 2592 (12.2%)

COPD, n (%) 1614 (7.6%)

Chronic liver dis., n (%) 840 (4.0%)

Chronic kidney disease, n (%) 1063 (5.0%)

Malignancy, n (%) 1956 (9.2%)

One comorbidity, n (%) 4816 (22.7%)

Two comorbidities, n (%) 4749 (22.3%)

Three or more comorbidities, n (%) 5136 (24.2%) Discharge diagnosis

Infectious disease, n (%) 10,543 (49.6%)


30-day mortality, n (%) 2611 (12.3%)

ICU admission, n (%) 1216 (5.7%)

Acute kidney injury, n (%) 2485 (11.7%)

IQR- interquartile range; HR- heart rate; bpm- beats per minute; SBP- systolic blood pres- sure; SO2– oxygen saturation while breathing ambient air; IHD- ischemic heart disease; CHF- congestive heart failure; COPD- chronic obstructive lung disease; CKD- chronic kidney disease; ICU- intensive or intermediate care unit.

0.72, 95% CI 0.60-0.87, p < 0.001). A complete list of independent risk factors for the predefined outcomes is presented in Table 3. The rate of missing data was low, and 20,773 (97.7%) patients were included in the multivariate analysis.

At discharge, the diagnosis of 10,543 febrile patients (49.6%) was categorized as infectious. The proportion of patients with an infectious diagnosis increased from 49.6% in those with a maximum BT of 38.0-38.4 ?C recorded in the ED to 56.5% in those with BT > 40 ?C. Ninety-five patients (0.4%) included in the analysis had a primary diag- nosis of stroke (ICD10 codes I60-I63 and I67-I69) or intracranial trau- matic injury (ICD10 code S06). Fig. 4 presents the ORs for mortality according to the discharge diagnosis categorized as infectious or not. Among patients with non-infectious diagnoses, an increase in BT was not associated with increased mortality until the recorded ED BT rose above 39.7 ?C. Patients with an infectious disease had higher OR for mortality across the whole range of BTs (compared to those with non- infectious disease). In both groups, the lowest ORs were observed among those with BT between 39.4 ?C to 39.5 ?C.

  1. Discussion

In this large retrospective cohort study conducted in a tertiary uni- versity hospital, fever (BT >= 38.0 ?C) was recorded among 8.1% of pa- tients evaluated in the ED. Elevated BT was associated with increased

risk of admission, 30-day all-cause mortality, ICU admission, and AKI. Among febrile ED patients who were hospitalized, we found an in- creased risk of mortality and ICU admission in patients with a fever above 39.7 ?C. Increased risk of AKI was observed when BT reached

39.5 ?C. BT between 39.2 and 39.5 ?C was associated with the lowest

Table 2

Demographic, clinical, and laboratory characteristics of 21,252 febrile patients included in the final analysis by maximal body temperature recorded during evaluation in the emergency department.

Maximal temperature in the Emergency Department (?C)


(n = 10,034)


(n = 4919)


(n = 4271)


(n = 1598)


(n = 430)

Age, years (IQR)

68.8 (51.7-80.8)

69.9 (52.4-81.0)

66.2 (47.5-78.5)

67.7 (49.2-79.2)

71.1 (56.4-81.2)

Male gender, n (%)

5252 (52.3%)

2660 (54.1%)

2441 (57.2%)

900 (56.3%)

250 (58.1%)

Elderly (>65 years), n (%)

5714 (56.9%)

2878 (58.5%)

2233 (52.3%)

878 (54.9%)

272 (63.3%)

Medical patients, n (%)

7753 (77.3%)

3899 (79.3%)

3363 (78.7%)

1355 (84.8%)

372 (86.5%)

Vital signs

HR, bpm (IQR)

98 (85-111)

100 (87-114)

103 (90-116)

106 (91-120)

113 (96-130)

Tachycardia (HR > 100 bpm), n (%)

4242 (42.3%)

2360 (48.0%)

2301 (53.9%)

943 (59.2%)

288 (67.1%)

SBP, mmHg (IQR)

132 (117-150)

133 (118-151)

130 (116-149)

132 (116-151)

130 (112-146)

Hypotension (SBP < 90 mmHg), n (%)

262 (2.6%)

138 (2.8%)

121 (2.8%)

48 (3.0%)

29 (6.8%)

SO2, % (IQR)

96 (94-98)

96 (93-98)

96 (94-98)

96 (93-98)

95 (92-98)


1111 (11.2%)

549 (11.3%)

405 (9.6%)

184 (11.7%)

57 (13.7%)

(SO2 < 90%), n (%)

Laboratory parameters

Leukocytes, Ku/L (IQR)

11.7 (8.1-15.8)

11.7 (8.1-16.1)

11.1 (7.3-15.6)

11.1 (7.3-15.3)

11.3 (7.6-16.4)

Hemoglobin, g/dL (IQR)

12.1 (10.5-13.6)

12.1 (10.6-13.6)

12.3 (10.8-13.6)

12.4 (10.9-13.7)

12.3 (10.8-13.7)

Platelets, 103 u/L (IQR)

215 (161-285)

209 (156-276)

198 (147-259)

189 (139-249)

198 (142-271)

Creatinine, mg/dL (IQR)

1.0 (0.8-1.4)

1.0 (0.8-1.4)

1.0 (0.8-1.4)

1.1 (0.9-1.5)

1.2 (0.9-1.7)


Diabetes Mellitus, n (%)

3061 (30.5%)

1581 (32.1%)

1280 (30.0%)

503 (31.5%)

127 (29.5%)

Hypertension, n (%)

5512 (54.9%)

2743 (55.8%)

2182 (51.1%)

848 (53.1%)

247 (57.4%)

Cerebrovascular disease, n (%)

1741 (17.4%)

841 (17.1%)

626 (14.7%)

231 (14.5%)

86 (20.0%)

IHD, n (%)

1670 (16.6%)

805 (16.4%)

623 (14.6%)

234 (14.6%)

75 (17.4%)

CHF, n (%)

1302 (13.0%)

618 (12.6%)

445 (10.4%)

171 (10.7%)

56 (13.0%)

COPD, n (%)

818 (8.2%)

350 (7.1%)

311 (7.3%)

101 (6.3%)

34 (7.9%)

Chronic liver dis., n (%)

402 (4.0%)

188 (3.8%)

167 (3.9%)

62 (3.9%)

21 (4.9%)

Chronic kidney disease, n (%)

524 (5.2%)

269 (5.5%)

176 (4.1%)

68 (4.3%)

26 (6.0%)

Malignancy, n (%)

890 (8.9%)

454 (9.3%)

420 (9.9%)

146 (9.2%)

44 (10.3%)

Discharge diagnosis

Infectious disease, n (%)

4980 (49.6%)

2282 (46.4%)

2134 (50.0%)

904 (56.6%)

243 (56.5%)


30-day mortality, n (%)

1282 (12.8%)

631 (12.8%)

420 (9.8%)

175 (11.0%)

103 (24.0%)

ICU, n (%)

579 (5.8%)

244 (5.1%)

232 (5.4%)

107 (6.7%)

54 (12.6%)

Acute kidney injury, n (%)

1103 (11.0%)

567 (11.5%)

488 (11.4%)

214 (13.4%)

113 (26.3%)

IQR- interquartile range; HR- heart rate; bpm- beats per minute; SBP- systolic blood pressure; SO2– oxygen saturation while breathing ambient air; IHD- ischemic heart disease; CHF- congestive heart failure; COPD- chronic obstructive lung disease; CKD- chronic kidney disease; ICU- Intensive or intermediate care unit.

Image of Fig. 2

Fig. 2. Odds ratios with 95% confidence intervals for (a) 30-day all-cause mortality, (b) intensive or intermediate care unit admission, and (c) acute kidney injury, relative to body temperature of 38.0-38.1 ?C. By definition, patients with body temperature 38.0-38.1 ?C have an odds ratio of 1.0.

Image of Fig. 3

Fig. 3. Odds ratio with 95% confidence intervals for (a) 30-day all-cause mortality, (b) intensive or intermediate care unit admission, and (c) acute kidney injury among patients <=65 years old and those >65 years old, relative to body temperature of 38.0-38.1 ?C. By definition, patients with body temperature 38.0-38.1 ?C have an odds ratio of 1.0.

mortality rates. To the best of our knowledge, this is the first study to focus on the prognostic value of fever magnitude elevation in the ED among a non-selected patients’ cohort.

Numerous studies evaluated the association between pyrexia and clinical outcomes in highly selected groups of patients. The negative ef- fect of high fever on outcomes of patients with stroke, neurological in- jury, and following cardiopulmonary resuscitation is well established. [2,14] Infants with BT above 40 ?C have a higher risk of serious bacterial infections when compared to pediatric patients with fever between 38 ?C to 39.9 ?C. [6] Among 926 adult patients with bacterial infection, high fever (BT >= 40.0 ?C) was not associated with increased mortality,

compared to febrile patients with a lower degree of fever. [15] In a large-scale study of ICU patients, there was no significant change in in-hospital mortality among patients with a wide range of BTs above

38.0 ?C and an infectious disease diagnosis. Among patients with a non-infectious diagnosis, increasing peak temperature above 39.0 ?C was associated with a progressively increasing risk of in-hospital mor- tality. [1] In another prospective observational study of critically ill pa- tients, a maximal BT during ICU stay above 39.5 ?C was associated with increased mortality in both septic and non-septic groups. [16] A meta-analysis of observational studies dealing with an association of fever with mortality in non-neurological critically ill patients found

Table 3

Independent predictors of 30-day all-cause mortality, intensive or intermediate care unit admission, and acute kidney injury, using backward multiple logistic regression with demograph- ics factors, vital signs, laboratory tests, and comorbidities.

Mortality Intensive or intermediate care unit admission

Acute kidney injury

Adjusted OR (95% CI)


Adjusted OR (95% CI)


Adjusted OR (95% CI)


Age, years

1.05 (1.04-1.05)


0.99 (0.99-1)


1.01 (1.00-1.02)


Male gender

0.87 (0.79-0.95)



1.67 (1.52-1.83)


1.40 (1.23-1.57)


1.56 (1.42-1.71)



3.14 (2.56-3.85)


3.3 (2.6-4.18)


2.67 (2.19-3.25)



2.33 (2.08-2.61)


2.04 (1.75-2.39)


1.46 (1.30-1.65)


Leukocytes, Ku/L

1.01 (1.00-1.01)


1.01 (1.00-1.01)


1.01 (1.00-1.01)


Platelets, 103 u/L

1 (1.00-1.00)


Creatinine, mg/dL

1.23 (1.18-1.27)


1.18 (1.13-1.23)


1.37 (1.33-1.42)



0.79 (0.71-0.87)


Diabetes Mellitus

1.21 (1.10-1.32)


Cerebrovascular disease

1.72 (1.54-1.91)


1.57 (1.36-1.83)


1.40 (1.25-1.56)



1.33 (1.13-1.57)



1.98 (1.66-2.37)


Chronic liver disease

1.30 (1.03-1.64)



2.17 (1.88-2.51)


0.70 (0.55-0.89)


1.39 (1.20-1.60)


Temperature 39.2-39.3 ?C

0.76 (0.62-0.93)


Temperature 39.4-39.5 ?C

0.72 (0.60-0.87)


Temperature 39.8-39.9 ?C

1.90 (1.22-2.96)


Temperature >= 40 ?C

1.63 (1.25-2.12)


2.11 (1.55-2.89)


2.29 (1.81-2.90)


CHF- congestive heart failure; COPD- chronic obstructive lung disease.

Image of Fig. 4

Fig. 4. Odds ratio with 95% confidence intervals for 30-day all-cause mortality among patients with an infectious and non-infectious discharge diagnosis, relative to body temperature of 38.0-38.1 ?C. By definition, patients with body temperature 38.0-38.1 ?C have an odds ratio of 1.0.

that BT >= 39.3 ?C was significantly associated with an increased mortal- ity. [8] This finding is consistent with our results, although with a slightly higher cutoff of 39.6 ?C, among a more generalized cohort; only 5.7% of our patients required ICU admission and 0.4% had a stroke

or a traumatic brain injury.

Fever, in the setting of infectious disease, may confer a survival ben- efit. Elevated BT impairs microbial replication and viability, leading to an antimicrobial effect. [17] It is noteworthy that at the beginning of the 20th century fever induction was routinely used for syphilis and gonor- rhea treatment. [18] Patients unable to mount high fever may be more susceptible to a more severe disease course. Some studies found an as- sociation between fever reduction (using antipyretics and active mea- sures) with increased mortality in patients with infectious diseases. [16,19,20] However, this finding has been challenged by more recent studies. [21-23] An additional explanation for the positive effect of high fever may be related to the fact that patients with higher BT are treated in a more timely manner by physicians in the ED. Patients with higher BTs are more likely to get early treatment, compared with those with lower BT. [5].

The negative effect of high BT is also prominent. It causes discomfort, increases metabolic rate and oxygen consumption, causes vasodilata- tion and myocardial depression, and may harm some vital body sys- tems, especially the nervous system where the temperature is up to

2.5 ?C higher than the core temperature. [2,17,23-25] In our study, pa- tients with BT > 40 ?C had higher rates of tachycardia, hypotension, and desaturation compared to those with lower BT. These may reflect the negative Physiologic effects of such extremely high BT.

The U-shape of the mortality-temperature curve, seen in our study, may occur when the protective effects of fever are overwhelmed by its

negative physiologic effects as BT rises above 39.5-39.7 ?C. The rela- tively protective effect of BT between 39.0 and 39.5 ?C is more promi- nent among patients with infection. In our cohort, among patients hospitalized with fever in the ED, roughly half had a non-infectious di- agnosis at hospital discharge. This finding is consistent with that previ- ously described in ED patients admitted with fever. [26] Noninfectious etiologies of fever include Surgical emergencies (although some of these, e.g. cholecystitis, cholangitis, appendicitis, may be categorized as infectious), malignancy, myocardial infarction, pulmonary embolism, cerebrovascular insults, autoimmune and rheumatologic diseases, drug reactions, hyperthyroidism, pancreatitis, and acute adrenal insuffi- ciency, among others. [3] In this population, increasing BT was not asso- ciated with increased mortality (compared to BT of 38.0-38.1 ?C) up to

39.7 ?C. Interestingly, in patients without an infectious disease, the low- est mortality was also observed among those with BT of 39.0 to 39.5 ?C, possibly suggesting immunity-independent protective effects of BT in this range. These findings should be interpreted cautiously as using ICD10 discharge diagnoses as Surrogate markers for infection may have only moderate validity. [27].

From a practical point of view, our findings suggest that adult ED pa- tients who are admitted to the hospital with BT above 39.5-39.7 ?C should be considered at higher risk for adverse outcomes. Although the causes of death were not included in our analysis, in the absence of a clear alternative diagnosis, sepsis workup and empiric antibiotic treatment should be strongly considered in these patients. Our study was not designed to evaluate the effect of antipyretics and active cooling measures among patients with fever, however it may support the strat- egy of avoiding such measures among patients with a fever below

39.5 ?C. Exceptions include subgroups of patients where fever has been shown to be deleterious (e.g., after cardiac arrest or brain, stroke, or traumatic brain injury).

  1. Limitations

Our study has several limitations. First, the current work was a single-center study conducted in the ED of a tertiary, university hospital and it included only patients who were hospitalized after evaluation in the ED. The results may not necessarily be generalizable to other set- tings. Second, the method of measuring BT was not standardized as BT could have been measured either orally or rectally. Although rectal mea- surement may be more accurate, this is a pragmatic “real life” study, as it is not currently practical to measure core BT in all ED patients. Third, we did not include respiratory rate in our analysis. The documentation level of this important vital sign is poor in our institution, as well as in other health care systems. [28] Fourth, no adjustments were made using se- verity scores, such as Sequential Organ Failure Assessment Score, quick SOFA score, Rapid Emergency Medicine score (REMS), or Acute Physiology and Chronic Health Evaluation II (APACHE-II) score. These would have allowed more accurate adjustment for disease sever- ity than vital signs, laboratory values, and comorbidities. Practically, however, these scores are not readily available for physicians working in the ED and are not routinely documented in the ED. Retrospective cal- culation of these scores is also problematic due to the paucity of infor- mation regarding respiratory rate and arterial PaO2. Fifth, we were unable to determine the usage of antipyretics before and during the ED evaluation. These data, as well as the time course of the febrile dis- ease, could not be efficiently extracted from the EMR. However, the ben- efits of antipyretics in treating fever in ambulatory and ED settings are unclear. In children, recent antipyretic use was not associated with lower maximal BT measured during ED evaluation, though it is unclear if the results would be differ in adult patients. [29] Sixth, the usage of immune-modifying drugs, antibiotics, and the etiology of fever were not considered. As mentioned above, the etiology of fever may not be clear during the initial evaluation in the ED. Finally, as this was a retro- spective study, some data collected from the EMR may be incomplete. However, the number of missing values was small (2.2%), and we

used mortality as a primary outcome due to its robust nature. As the mortality events in our study were extracted from the Ministry of Health database, it is very unlikely that there are any missing primary outcomes.

  1. Conclusion

Among a non-selected population of adult, febrile patients admitted to the hospital, BT above 39.5 ?C in the ED was associated with increased 30-day all-cause mortality, ICU admission, and AKI as compared to pa- tients with lower-grade fever (38.0-38.1 ?C). These patients should be flagged on arrival to the ED and likely require more aggressive evalua- tion and treatment. The protective effect of BT between 39.0 and

39.5 ?C warrants further research.


No funding was required for the conduction of the study.

Conflicts of interest

The authors have no conflicts of interest to declare.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2021.11.045.


  1. Young P, Saxena M, Beasley R, Bellomo R, Bailey M, Pilcher D, et al. Early peak tem- perature and mortality in critically ill patients with or without infection. Intensive Care Med. 2012 Mar;38(3):437-44.
  2. Greer D, Funk S, Reaven N, Ouzounelli M, Uman G. Impact of fever on outcome in pa- tients with stroke and neurologic injury: a comprehensive meta-analysis. Stroke. 2008 Nov 1;39(11):3029-35.
  3. DeWitt S, Chavez SA, Perkins J, Long B, Koyfman A. Evaluation of fever in the emer- gency department. Am J Emerg Med. 2017 Nov 1;35(11):1755-8.
  4. Rueegg M, Nickel CH, Bingisser R. Disagreements between emergency patients and physicians regarding chief complaint – Patient factors and prognostic implications. Int J Clin Pract. 2021 May 1;75(5):e14070.
  5. Sunden-Cullberg J, Rylance R, Svefors J, Norrby-Teglund A, Bjork J, Inghammar M. Fever in the emergency department predicts survival of patients with severe sepsis and septic shock admitted to the ICU*. Crit Care Med. 2017 Apr 1;45(4):591-9.
  6. Rosenfeld-Yehoshua N, Barkan S, Abu-Kishk I, Booch M, Suhami R, Kozer E. Hyperpy-

rexia and high fever as a predictor for serious bacterial infection in children–a systematic review. Eur J Pediatr. 2018 Mar 1;177(3):337-44.

  1. Michelson KA, Neuman MI, Pruitt CM, Desai S, Wang ME, DePorre AG, et al. Height of fever and invasive bacterial infection. Arch Dis Child. 2021 Jun 1;106(6):594-6.
  2. Egi M, Morita K. Fever in non-neurological critically ill patients: a systematic review of observational studies. J Crit Care. 2012 Oct;27(5):428-33.
  3. Reiner Benaim A, Almog R, Gorelik Y, Hochberg I, Nassar L, Mashiach T, et al. Analyz- ing medical research results based on synthetic data and their relation to real data results: systematic comparison from five observational studies. JMIR Med Inform. 2020;8(2):e16492.
  4. Bijur P, Shah P, Esses D. Temperature measurement in the adult emergency depart- ment: oral, tympanic membrane and temporal artery temperatures versus rectal temperature. Emerg Med J. 2016 Dec 1;33(12):843-7.
  5. Obermeyer Z, Samra JK, Mullainathan S. Individual differences in normal body tem- perature: longitudinal big data analysis of patient records. BMJ. 2017 Dec 13;359: 5468.
  6. Varney S, Manthey D, Culpepper V, Creedon J. A comparison of oral, tympanic, and rectal temperature measurement in the elderly. J Emerg Med. 2002;22(2):153-7.
  7. Khwaja A. KDIGO clinical practice guidelines for acute kidney injury. Nephron. 2012 Aug 7;120(4):c179-84.
  8. Diringer MN, Reaven NL, Funk SEUG. Elevated body temperature independently contributes to increased length of stay in neurologic intensive care unit patients. Crit Care Med. 2004 Jul;32(7):1489-95.
  9. Yamamoto S, Yamazaki S, Shimizu T, Takeshima T, Fukuma S, Yamamoto Y, et al. Body temperature at the emergency department as a predictor of mortality in pa- tients with bacterial infection. Med. 2016;95(21).
  10. Lee BH, Inui D, Suh GY, Kim JY, Kwon JY, Park J, et al. Association of body tempera- ture and antipyretic treatments with mortality of critically ill patients with and without sepsis: multi-centered prospective observational study. Crit Care. 2012; 161:1-13 2012 Feb 28;16(1):.
  11. Casadevall A. Thermal restriction as an antimicrobial function of fever. PLoS Pathog. 2016 May 1;12(5).
  12. Simpson W. Artificial fever therapy of syphilis and gonococcic infections. Br J Vener Dis. 1936 Jul 1;12(3):133-66.
  13. Ye S, Xu D, Zhang C, Li M, Zhang Y. Effect of Antipyretic therapy on mortality in crit- ically ill patients with sepsis receiving mechanical ventilation treatment. Can Respir

J. 2017;2017.

  1. Schulman CI, Namias N, Doherty J, Manning RJ, Li P, Elhaddad A, et al. The effect of antipyretic therapy upon outcomes in critically ill patients: a randomized, prospec- tive study. Surg Infect (Larchmt). 2005;6(4):369-75.
  2. Jefferies S, Weatherall M, Young P, Eyers S, Perrin KG, Beasley CRW. The effect of an- tipyretic medications on mortality in critically ill patients with infection: a system- atic review and meta-analysis. Crit Care Resusc J Australas Acad Crit Care Med. 2011 Jun;13(2):125-31.
  3. Schortgen F, Clabault K, Katsahian S, Devaquet J, Mercat A, Deye N, et al. Fever con- trol using external cooling in septic shock doi:101164/rccm201110-1820OC , 185 (10); 2012 Dec 14; 1088-95.
  4. Mohr NM, Doerschug KC. Point: should antipyretic therapy be given routinely to fe- brile patients in septic shock? Yes Chest. 2013 Oct 1;144(4):1096-8.
  5. Haupt M, Rackow E. Adverse effects of febrile state on cardiac performance. Am Heart J. 1983;105(5):763-8.
  6. Mcilvoy L. Comparison of brain temperature to core temperature: a review of the lit- erature. J Neurosci Nurs. 2004;36(1):23-31.
  7. Knott JC, Tan S-L, Street AC, Bailey M, Cameron P. febrile adults presenting to the emergency department: outcomes and markers of serious illness. Emerg Med J. 2004 Mar 1;21(2):170-4.
  8. DP H, SL N, CB L, J H, C P, AT L. How well do discharge diagnoses identify hospitalised patients with Community-acquired infections?-a validation study. PLoS One. 2014 Mar 24;9(3).
  9. Cretikos MA, Bellomo R, Hillman K, Chen J, Finfer S, Flabouris A. Respiratory rate: the neglected vital sign. Med J Aust. 2008 Jun 2;188(11):657-9.
  10. Huang SY, Greenes DS. Effect of recent antipyretic use on measured fever in the pe- diatric emergency department. Arch Pediatr Adolesc Med. 2004 Oct 1;158(10): 972-6.

Leave a Reply

Your email address will not be published. Required fields are marked *