| ||||||||||||||
|
|
|||||||||||||



From the *Department of Anesthesia and Intensive Care, Medical University, Vienna, Austria;
Department of Anesthesiology, Inselspital, Bern, Switzerland;
Department of Anesthesiology, Wagner-Jauregg Nervenklinik, Lint, Austria; and
Outcomes Research Institute and Department of Anesthesiology, University of Louisville, Kentucky.
Address correspondence to Oliver Kimberger, MD, Department of Anesthesiology, Medical University Vienna, Vienna 1090, Austria. Address e-mail to study{at}kimberger.at or www.or.org.
| Abstract |
|---|
|
|
|---|
METHODS: In 35 patients undergoing neurosurgical interventions and 35 patients in the neurosurgical intensive care unit, measurements from the temporal artery thermometer were compared with those from a bladder thermometer. Four measurements were obtained from each patient.
RESULTS: Overall 280 measurement pairs were obtained. The mean bias between the methods was 0.07°C ± 0.79°C; the limits of agreement were
3 times greater than the a priori defined limit of ±0.5°C (–1.48 to 1.62). The sensitivity for detecting fever (core temperature >37.8°C) using the temporal artery thermometer was 0.72, and the specificity was 0.97. The positive predictive value for fever was 0.89; the negative predictive value was 0.94. The sensitivity for detecting hypothermia (core temperature <35.5°C) was 0.29, and the specificity was 0.95. The positive predictive value for hypothermia was 0.31, and the negative predictive value was 0.95.
CONCLUSIONS: The results of this study do not support the use of temporal artery thermometry for perioperative core temperature monitoring; the temporal artery thermometer does not provide information that is an adequate substitute for core temperature measurement by a bladder thermometer.
| Introduction |
|---|
|
|
|---|
Accurate core temperature measurements can be obtained from several locations including the distal esophagus, rectum, bladder, and pulmonary artery, among others. Obtaining these core temperature measurements in the anesthetized and tracheally intubated patient is fairly easy, yet for unanesthetized patients, many core temperature measurement methods are either too invasive or simply not feasible. In the neurosurgical intensive care unit (ICU), patients are often unanesthetized to enable close neurological monitoring. Core temperature monitoring can also be difficult in patients with neuraxial anesthesia. However, these patients are as likely as patients with general anesthesia to become hypothermic, and so proper temperature monitoring is no less important (9). There is a similar need for a fast and accurate yet noninvasive core temperature monitoring system in pediatric care, where rectal thermometry is still considered the "gold standard" of temperature measurement.
There are several noninvasive methods for core temperature measurements and they have proven to be of limited value: Axillary methods (10–13), aural canal infrared thermometers (14–17), and liquid crystal forehead thermometers (18) are not accurate enough for clinical purposes; both axillary and aural canal thermometers have a low sensitivity for detecting fever in comparison to standard core temperature measurement methods (16). Likewise, in pediatric patients, supralingual thermometry with pacifier thermometers cannot replace rectal thermometry (19).
A new type of infrared temporal artery thermometer has recently been developed (TemporalScannerTM TAT-5000, Exergen, Boston, MA) with the goal of providing a more accurate noninvasive measure of core temperature. This new device records skin temperature on the area over the temporal artery at a rate up to 1000 measurements per second. The highest temperature is registered, and core temperature is calculated using a proprietary algorithm that compensates for ambient temperature. Previous versions of this thermometer have been tested in several, predominantly pediatric, studies. These studies have shown the temporal thermometer to be more accurate than aural infrared thermometers (20), and useful as a screening method for clinically important fever in children (21–23), yet it was not recommended for routine clinical use and deemed too inaccurate to replace standard invasive temperature measurements (19,22–24). However, the new version of the thermometer, TAT-5000, claims to be as accurate and precise as invasive core temperature thermometry because of improvements in the hardware and the user guidelines.
We tested the TAT-5000 temporal artery thermometer in a neurosurgical patient population, which has a high risk for fever (25). Specifically, we compared the TAT-5000 temporal artery thermometer with a bladder thermometer (26) and assessed the sensitivity and specificity of the temporal artery thermometer for detecting hypothermia and fever. The goal of the study was to determine if the temporal artery thermometer agreed closely enough with bladder temperature measurements to be a useful substitute in the perioperative and the intensive care setting.
| METHODS |
|---|
|
|
|---|
Morphometric and demographic patient characteristics were initially recorded. Patients were managed according to the discretion of the responsible anesthesiologist. All temporal artery thermometer measurements were conducted by one experienced anesthesiologist trained to use the device.
A bladder temperature sensor (accuracy ±0.2°C) incorporated into a urinary drainage catheter (Smiths-Medical, London, United Kingdom) (26,27) was inserted after arrival to the neurosurgical operating room (OR) or to the neurosurgical ICU, respectively. Temporal temperatures were recorded in the OR at four times: 1) at arrival in the OR; 2) 30 minutes after anesthetic induction; 3) at the end of surgery; and 4) 30 minutes after arrival in the postanesthesia care unit. In the neurosurgical ICU, temperature was recorded four times at 1-h intervals. From each OR- and ICU-patient, four measurements were acquired (280 measurements overall).
Temporal artery measurements were performed according to the TAT-5000 manual: The patients hair was brushed aside if necessary, and the probe was positioned on the center of the patients forehead. The measurement button was depressed, and the measurement was performed sliding from the center midline across the forehead to the hairline. With the button still depressed, the probe was lifted from the forehead and the region behind the earlobe was scanned, as recommended, to exclude falsely low temperature readings caused by evaporative cooling due to forehead sweating. Then, the button was released and the temperature recorded. The duration of this procedure was
5 seconds. This procedure was repeated immediately to assess repeatability. For method comparison, the two measurements were averaged. During the first temporal artery measurement, bladder temperature was recorded. Other recorded variables were forehead sweating and the use of a warming/cooling forced air device.
Thermal management in the neurosurgical OR consisted of forced air warming (Bair HuggerTM, Arizant, MN). In the neurosurgical ICU, treatment for fever was performed according to a routine protocol, which included the administration of metamizol and forced air cooling (PolarAirTM, Arizant, MN).
Our primary outcome was the agreement in temperature reading between the established method (bladder temperature) and the temporal artery thermometer. Our secondary outcome was sensitivity and specificity for fever and hypothermia. Fever was defined as core temperature >37.8°C, and hypothermia was defined as core temperature <35.5°C. A priori the acceptable limits of agreement were chosen to be ±0.5°C. These limits can be considered clinically relevant, as a change of >0.5°C exceeds the usual temperature cycle variations in humans (28) and has been used in similar, previous studies (24,29).
Statistical Analysis
Bland-Altman analysis (30) was used to assess the difference between the average of two subsequent temporal artery measurements and the bladder thermometer measurement. Limits of agreement <±0.5°C were defined a priori as clinically acceptable. If the observed limits of agreement (
two standard deviations around the mean difference) were clinically acceptable as a priori defined (<±0.5°C), the two methods were considered equivalent. Ninety-five percent confidence intervals were calculated for mean bias, upper and lower limits of agreement, sensitivity, specificity, positive predictive values, and negative predictive values for fever and hypothermia. For assessment of repeatability, the intraclass correlation coefficient and the repeatability coefficient (31) were calculated. To exclude a relationship between the magnitude of measurement and method—discrepancy or repeatability error, a rank correlation coefficient was calculated. Receiver operating characteristic curves (ROC) were plotted for the detection of fever >37.8°C and hypothermia <35.5°C and the areas under the curves were calculated.
Data were analyzed using MedCalc (MedCalc Corporation, Belgium) for Bland-Altman and ROC-curve analysis; for other analysis, SPSS (SPSS, Chicago, IL) was used.
| RESULTS |
|---|
|
|
|---|
|
Method Comparison
Bladder temperatures ranged from 34.8°C to 41.3°C, temporal artery measurements ranged from 32.6°C to 41.1°C. There was no significant correlation between method discrepancy and magnitude of measurement (P = 0.8). As the discrepancy between the thermometers did not differ between groups or time points (results not shown), data were pooled for Bland-Altman analysis. The mean bias between the methods was low, yet the limits of agreement were
3 times more than the 0.5°C limit considered a priori to be clinically acceptable (Table 2, Fig. 1). Mean bias in mildly febrile or hypothermic patients was significantly greater than overall bias (P = 0.005, P < 0.001, respectively; Mann-Whitney U Test, Table 2). Sweating or the use of a forced air cooling/warming device had no influence on the temporal artery measurements (results not shown).
|
|
Repeatability
There was no significant correlation between repeatability error and magnitude of measurement (P = 0.6). The repeatability coefficient for 2 subsequent temporal artery measurements was 1.25°C. The intraclass correlation coefficient was 0.73 (95% confidence interval [CI]: 0.67–0.78).
The TAT-5000 thermometers sensitivity for the detection of fever and for the detection of hypothermia is displayed in Table 3. The area under the ROC curve for detection of fever >37.8°C was 0.95 (95% CI: 0.92–0.98; Fig. 2); the area under the ROC curve for detection of hypothermia <35.5°C was 0.71 (95% CI: 0.66–0.77; Fig. 3).
|
|
|
| DISCUSSION |
|---|
|
|
|---|
This is the first report comparing the new TAT-5000 temporal artery thermometer to bladder thermometry in a perioperative and intensive care setting. Several studies have evaluated consumer and professional models of temporal artery thermometers manufactured by Exergen, primarily in a pediatric setting. In these studies, the temporal artery thermometer was compared to standard core temperature measurement sites: bladder temperature, pulmonary artery temperature, rectal temperature, axillary temperature, and aural infrared temperature. The temporal artery thermometer proved to surpass axillary and aural infrared thermometers, and was found to be sufficiently accurate to serve as a screening method for high rectal fever in a pediatric population. All of the following studies tested previous versions of the TAT-5000: Suleman et al. (24) compared the temporal thermometer in a fairly small sample size with pulmonary artery catheters and bladder catheters; the performance of the thermometer was very poor and a sensitivity of 0% for fever was measured in adults. Greenes and Fleisher (20) found an adequate sensitivity for the temporal artery thermometer only for patients with high rectal fever and recommended it solely as a better alternative to the aural infrared thermometer. Schuh et al. (22) found the professional version of the medical-grade temporal artery thermometer to be an effective screening method, yet it was not able to replace rectal thermometry. In a recent study, Hebbar et al. (23) does not recommend the temporal artery thermometer as a replacement for invasive core temperature measurements, yet judges it to be more reliable and applicable than axillary temperature measurements. In accordance with these studies, we are not able to recommend the use of the new temporal artery thermometer as a substitute for standard core temperature measurements because of its inadequate agreement with bladder temperature, its poor performance for detecting mild hyper- and hypothermia, and its wide range of repeatability error.
A limitation of our study is the use of bladder temperature to measure core temperature. It is not an undisputed core temperature measurement site, and pulmonary artery temperature measurements would certainly have been preferable, yet they were generally considered too invasive in our patients. The inaccuracies caused by this core temperature measurement site may have contributed somewhat to the study results. Furthermore, Greenes and Fleisher (35) found that rectal temperature measurements lag behind those measured at the temporal artery. It can therefore not be excluded that a similar temperature lag further contributed to a deterioration of our temporal artery thermometer results. Another limitation is that although four measurements were obtained in one patient, these measurements were treated as independent measurements. For method comparison, two subsequent temporal artery measurements were averaged; discrepancies between the thermometry methods might have been even greater if only a single temporal artery measurement had been performed.
The range of bladder temperature in our patients was narrow. It included only a few patients with high fever or moderate-to-deep hypothermic temperatures. Thus, statements about the performance of the temporal artery thermometer under febrile conditions or during moderate or deep hypothermia are not possible. Finally, 74% of the study patients had some kind of cranial surgery. It is not known to what extent a surgical cranial intervention influences temporal artery flow and the temperature of the skin over the temporal artery.
In conclusion, screening for fever and measurement of core temperature, when other more invasive measurement methods are not feasible, may be applications for the temporal artery thermometer. However, the new version of the temporal artery thermometer, Temporal Scanner TAT-5000, cannot replace more invasive temperature monitoring systems when highly reproducible core temperature measurements are required.
| ACKNOWLEDGMENTS |
|---|
| Footnotes |
|---|
Presented, in part, at the Euroanesthesia-Congress, Vienna, 2005.
None of the authors has a personal financial interest related to this work.
Reprints will not be available from the author.
| REFERENCES |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
E. Bridges and K. Thomas Noninvasive Measurement of Body Temperature in Critically Ill Patients Crit. Care Nurse, June 1, 2009; 29(3): 94 - 97. [Full Text] [PDF] |
||||
![]() |
O. Kimberger, R. Thell, M. Schuh, J. Koch, D. I. Sessler, and A. Kurz Accuracy and precision of a novel non-invasive core thermometer Br. J. Anaesth., May 29, 2009; (2009) aep134v1. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|