Continued Relevance of Initial Temperature Measurement in Trauma Patients

Key Takeaways

Hypothermia is an independent predictor of mortality when controlling for injury severity.

Introduction

Hypothermia occurs in 30-50% of severely injured trauma patients. ¹ Causes of hypothermia in trauma patients include exposure to cold temperature at the time of injury or transport, exposure during primary and secondary surveys, and dysfunctional thermoregulation. Along with acidosis and coagulopathy, hypothermia is part of the classic “lethal triad” and has traditionally demanded prompt recognition and management. Hypothermia is associated with various clinical presentations and metabolic derangements such as left shift of the oxygen dissociation curve, impaired coagulation, reduced cardiac output, and cardiac arrhythmias.^2,3 The increased affinity by hemoglobin for oxygen decreases oxygenation of tissues at the capillary level, which increases anaerobic metabolism and lactate buildup. There are also deleterious effects on enzymes in the renal tubule that affects ability for bicarbonate reabsorption and hydrogen ion excretion, worsening acidosis, and, subsequently, coagulopathy. ³ Advanced Trauma Life Support^® (ATLS^®) continues to recommend accurate core temperatures in injured trauma patients using rectal, esophageal, or bladder temperatures to quickly and accurately identify hypothermia. ⁴

Prior studies have demonstrated conflicting results on whether temperature is an independent predictor of morbidity and mortality when controlling for comorbidities and clinical presentation.^5-9 The goal of this study is to determine if hypothermia is an independent predictor of mortality in the trauma population. Secondary outcomes include hypothermia’s impact on intensive care unit (ICU) length of stay and transfusion requirement.

Methods

Following institutional review board approval, we retrospectively reviewed data at our American College of Surgeons verified Level I Trauma Center on all trauma alert patients from October 2015 through May 2020. Our criteria for full and partial trauma team activation are included in Table 1. Our team collected data on patient demographics, initial core temperature in trauma bay (measured by rectal or temperature-sensing Foley catheter), Glasgow Coma Scale (GCS) on presentation, injury severity score (ISS), blood products given, ICU length of stay, and mortality rate. Patients were stratified into groups based on GCS ≥12 vs < 12 and temperature <35°C, 35-39°C, and >39°C). Our criteria for full and partial trauma team activation are included in Table 2.

OPEN IN VIEWER

Table 1.

Comparison of Hypothermic and Normothermic Patients. Statistically Significant Differences Between Groups Denoted by Asterisk (*).

Variable	Hypothermic (n = 252)	Normothermic (N = 15,293)	P-value	Test
Age, years, median (IQR)	50 (29, 66)	44 (26.5, 63)	.0156*	Mann-Whitney U
Female, n (%)	77 (34.1)	5505 (36.0)	.5574	Chi-square
Prehospital systolic blood pressure, mmHg, mean (SD) (n = 137, n = 8824, n = 13)	125.6 (32.7)	137.2 (28.0)	<.0001*	Satterthwaite t-test
Emergency department systolic blood pressure, mmHg, mean (SD) (n = 208, n = 15,240, n = 31)	120.6 (35.3)	137.2 (27.4)	<.0001*	Satterthwaite t-test
Emergency department Glasgow Coma Score, median (IQR) (n = 225, n = 15,285, n =31)	3 (3, 15)	15 (15, 15)	<.0001*	Mann-Whitney U
Injury severity score, median (IQR) (n = 224, n = 14,912, n = 30)	16.5 (9, 26)	9 (5, 14)	<.0001*	Mann-Whitney U
pH, median (IQR) (n = 105, n = 1641, n = 13)	7.28 (7.16, 7.36)	7.31 (7.25, 7.38)	.0047*	Mann-Whitney U
Intensive care unit length of stay, days, median (IQR)	2 (0, 4)	0 (0, 2)	<.0001*	Mann-Whitney U
Received massive transfusion protocol, n (%) (n = 220, n = 15053, n = 31)	15 (6.8)	149 (1.0)	<.0001*	Fisher’s exact

OPEN IN VIEWER

Table 2.

On-Scene/In-Transport Criteria for “Modified” and “Full” Trauma Alters at Our Institution. All Trauma Alerts Must Have Objective Evidence of Trauma.

Trauma Alert Level	Clinical Criteria
“Modified”	Long bone fractures, flail chest, pelvic fractures, paralysis, penetrating extremity injuries due to gunshot wounds, burns greater than 15% body surface area, obvious vascular injuries, traumatic pneumothorax, traumatic amputation, GCS <12, sustained pulse rate greater than 120
“Full”	Confirmed blood pressure <90 mmHg, respirations <10 or >29, penetrating trauma to head, neck, abdomen, or chest, Glasgow Coma Scale less than or equal to 8, or bilateral femur fractures

Outcomes measured were blood products given, ICU length of stay, and mortality rate. Cohort outcomes were compared using correlations and logistic regression. Patients were also separately stratified by ISS. All data were analyzed using SAS^® version 9.4.

Results

We reviewed data from 18,188 patients, 2621 of which were excluded due to incomplete survival data or lack of core temperature or GCS documented on presentation. Survival and temperature data were reviewed on the remaining 15,567 patients (Figure 1). Overall, there were 252 hypothermic patients (presenting temperature less than 35°C), 15,293 normothermic patients (35-39°C), and 22 hyperthermic patients (greater than 39°C). Hyperthermic patients were excluded from demographics comparison due to low number of patients (n = 22). Hypothermic patients tended to be older with increased injury severity scores and decreased pH. They were also noted to have decreased systolic blood pressure on presentation and decreased GCS. Insignificant factors between the 2 groups that were analyzed but removed from table for clarity include prehospital heart rate, oxygen saturation, and red blood cell units received (Table 2). Comorbidities compared between the 2 groups were statistically insignificant, including rates of substance use disorder, anticoagulant use, tobacco use, chronic obstructive pulmonary disease, chronic kidney disease, congestive heart failure, diabetes mellitus, and hypertension.

OPEN IN VIEWER

Figure 1.

Entire patient cohort as demonstrated by distribution of presenting temperature in degree Celsius.

The initial temperature in our trauma patient cohort was not significantly associated with ICU length of stay or red blood cell units given (P = .21 and P = .084, respectively). Mortality risk was charted against temperature (Figure 2). Hypothermic patients have an overall mortality rate of 38.5%, compared to mortality rate of 4.3% in normothermic patients and 13.6% in hyperthermic patients. Odds ratio of mortality in hypothermic patients (<35°C) compared to normothermic patients (35-39°C) was 3.95 (95% CI 2.90-5.41).

OPEN IN VIEWER

Figure 2.

Mortality rate as a function of temperature on presentation (rounded to nearest degree Celsius).

In the cohort of patients stratified by GCS, temperature was a significant independent predictor of mortality. In patients with GCS >12, patients with temperatures less than 35°C have increased mortality rate compared to patients with temperature greater than 35°C (7.6% compared to 1.6%). In patients with GCS less than 12, mortality of hypothermic patients was 56.6% compared to 25.6% in normothermic patients. In the entire cohort, when controlling for GCS, temperature was a significant predictor of morality (P < .001). As temperature increased per 5/9^th degree Celsius (1°F), odds of death decreased 14.3% (95% CI 10.5% - 18.0%).

When patients were stratified by injury severity score, the data continue to support hypothermia as an independent increased risk of mortality. When controlling for ISS, the odds of mortality decreased 30.2% (95% CI 26.4% - 33.8%) for every 5/9^th degree Celsius (1°F) increase in temperature.

Discussion

Our data support that hypothermia is an independent predictor of mortality when controlling for injury severity and GCS. However, hypothermia on presentation was not significantly associated with ICU length of stay or blood product requirements. These data support continued early identification and treatment of hypothermic trauma patients. While previous studies have focused on national databases or single institution experiences with limited patient population, this research focuses on a large patient cohort from a single institution with standardized central temperature measurement as part of the secondary survey. Standard practice in our institution is to utilize passive warming strategies such as removing wet clothing and placing warm blankets on all trauma patients on arrival. If patients are found to be hypothermic, treatment is initiated with warmed fluids and forced-air warming blankets prior to leaving the trauma bay. Invasive rewarming strategies such as peritoneal lavage are rarely utilized or indicated in our patient population but are available adjuncts.

This study supports prior observations that hypothermia on arrival in trauma patients is a negative predictor of survival; however, our research further stratifies patients based on clinical presentation.^6-9 Prior studies have demonstrated conflicting results on whether hypothermia is an independent predictor of mortality and other outcomes such as ICU length of stay and blood products given. Notably, Steinemann et al found in their patient cohort hypothermia was a predictor for mortality, but when stratified for injury type and severity, there was no significant difference. ⁵ However, more recent studies have demonstrated increased mortality rate in hypothermic patients even when controlling for injury severity.^6,7 Rösli et al demonstrated that hypothermic patients tend to have increased injury severity but did not stratify outcomes based on injury severity in hypothermic patients. ⁹

Prehospital management and prevention of hypothermia, such as warmed fluid use, removal of wet clothing, and use of blankets, is now becoming standard practice in many communities. ¹⁰ Goal-directed treatment algorithms aimed at targeting hypothermia in the prehospital setting have demonstrated improved morbidity and mortality; however, these have been limited to small patient cohorts with confounding factors such as decreased injury severity.^11,12 Though this study focused on hypothermia on arrival, Gregory et al, demonstrated that although many patients arrive to the hospital normothermic, many become hypothermic throughout their initial resuscitation, particularly when undergoing operative intervention. ¹³ Upon arrival to the ICU, trauma patients are hypothermic in nearly 37-58% of cases, and hypothermia on arrival to the ICU is a significant predictor of 24-hour and 30-day mortality.^14,15 Thus, hypothermia may be poorly recognized or inconsistently managed during the initial evaluation and resuscitation period.

One of the major strengths of this study is our institution’s uniform measurement of core body temperatures. Obtaining accurate core temperatures to rapidly diagnose hypothermia is of importance as accuracy of non-invasive temperature measurement can widely vary in hypothermic patients. ¹⁶ Newer non-invasive insulated thermometers that utilize probes placed over the temporal artery have demonstrated satisfactory accuracy for intraoperative monitoring but have not been studied in trauma patients. ¹⁷ All patients in our trauma database had their temperatures measured either rectally or with a temperature-sensing Foley catheter, both of which have been demonstrated to provide more accurate core body temperatures than current forehead or oral temperatures. ¹⁸ Another major strength of this study is the large patient database and varied nature of patients. This allows the results to be more generalizable to the public as there are both penetrating and blunt mechanisms of injury in addition to a wide age range.

Limitations of this study include that it only represents a single institution’s experience and only measured the initial temperature on arrival to the trauma bay. This research demonstrates a correlation between hypothermia on presentation with outcome. However, patients may become hypothermic after admission which may have an impact on mortality rates. However, whether hypothermia was prolonged or rapidly corrected was not evaluated and applied to outcomes. While prior studies have demonstrated that hypothermia on arrival to the trauma center or ICU is predictive of outcomes, further research should focus on whether aggressively treating hypothermia or time to return to normothermia affects trauma patient outcomes. Further prospective research should also evaluate whether statistical models based on initial temperature and injury severity score are predictive of mortality.

In summary, our study demonstrates that hypothermia is an independent risk factor for mortality in trauma patients. Even when evaluating patients with similar clinical presentation, presenting hypothermia is a key clinical indicator of outcomes. Therefore, obtaining accurate core temperatures and rapidly treating hypothermia continues to be a vital part of the initial evaluation of trauma patients.