Noninvasive zero-heat-flux thermometer was first invented by Fox and Solman in 1971 [15]. Safety and usefulness of this method has been already reported. Boisson et al. compared esophageal temperature with that of BHTMS at forehead showing bias and limit of agreement (LOA) of 0.1℃ ± 0.5℃ [9]. Iden et al. compared this method with nasopharyngeal temperature showing almost identical results (0.07℃, ± 0.42℃) [10]. Eshraghi et al. showed the difference between the temperature obtained from BHTMS and that of the pulmonary artery was − 0.23 ℃, LOA of ± 0.82℃ [12]. Also, Tachibana et al. compared esophageal temperature and BHTMS at neck showing mean bias of 0.05℃, LOA ± 0.35℃ [13]. Considering Tachibana et al.’s study, the mean value of the difference (bias) < 0.4℃ and 2 standard deviations (SD) < ± 1.0℃ was set as the standard for comparing reliability in this study [13].
In this study, we compared \({\text{T}}_{\text{e}\text{s}\text{o}}\) and \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\) for the primary outcome. Wilcoxon signed rank test showed that \({\text{T}}_{\text{e}\text{s}\text{o}}\) and \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\) had no significant difference. Spearman correlation showed moderate positive correlation. However, the result of Bland-Altman plot of \({\text{T}}_{\text{e}\text{s}\text{o}}\) and \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\) was mean bias (\({\text{T}}_{\text{e}\text{s}\text{o}}\) - \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\)) of 0.14℃ with 2SD of ± 1.44 ℃ which did not meet the standard for reliability as we stated earlier.
For secondary outcomes, the data within 0 to 40 minutes (T0, T10, T20, T30, T40) were analyzed. First, we ran a paired t test to compare accuracy between \({\text{T}}_{\text{e}\text{s}\text{o}}\) and \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\). T0 and T10 showed significant difference. And T20, T30, T40 showed no significant difference. There was positive correlation between \({\text{T}}_{\text{e}\text{s}\text{o}}\) and \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\) at T0 - T40. T0 showed weak correlation. However, T10, T20, T30, T40 showed moderate to strong correlation. Bland-Altman analysis was performed. From the result, we noticed that only T30 and T40 can be accepted as reliable which met the earlier LOA standard. For further analysis, we also performed Bland-Altman analysis with the 198 sets of data collected from 30 minutes (T30). The mean bias of \({\text{T}}_{\text{e}\text{s}\text{o}}\) was 0.08℃ under \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\) with 2SD of ± 0.96℃, which meet the standard for reliability. Considering the results above, the BHTMS at wrist is relatively accurate and equivalent to esophageal temperature at least 30 minutes after the first temperature measurement.
As other zero-heat-flux thermometer sensors, BHTMS sensor is also consisted of two thermistors, a thermal insulator and an electrical heater. BHTMS has some advantages such as its accuracy, easy to use with disposable sensor, noninvasiveness. However, BHTMS has some unfavorable situations. We cannot apply BHTMS sensor to forehead during the surgeries of brain or head and neck surgeries. Also, it may interfere with the attachment of other monitoring on the forehead such as electroencephalogram or cerebral oximetry. Considering the principle of zero-heat-flux thermometer, we could make a hypothesis that core body temperature can be obtained from wrist area by using BHTMS. Wrist area we considered has radial artery running about 5 mm below the skin, has little influence by fatty tissues, and is easy to approach [16].
In our study, at the beginning of the study (T0), 10 out of 20 patients showed immediate close approximation between \({\text{T}}_{\text{e}\text{s}\text{o}}\) and \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\) ≤ ± 1.0℃. And the other half of patients showed wide difference between \({\text{T}}_{\text{e}\text{s}\text{o}}\) and \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\) up to 3.1℃ and the wide difference decreased progressively, by times. The exact mechanism of the close approximation of \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\) to \({\text{T}}_{\text{e}\text{s}\text{o}}\) after 30 minute needs to be explicit. One reason we assume is the redistribution of body heat. Before the induction of general anesthesia, peripheral compartment distant from heart such as wrist region typically shows 2–4℃ lower temperature than core body temperature. This is normal core-to-peripheral temperature gradient. As known, general anesthesia reduces vasoconstriction and occurs peripheral vasodilation which leads to perioperative hypothermia. And this redistribution increases temperature of peripheral region [17]. Also, the other reason we assumed is the direct heating effect of upper body warm blanket. When we turned off the warming system due to concern of hyperthermia of patient, we could observe wider difference between \({\text{T}}_{\text{e}\text{s}\text{o}}\) and \({\text{T}}_{\text{w}\text{r}\text{i}\text{s}\text{t}}\). Therefore, we are considering the upper body warming blanket has direct heating effect. The time interval until the blankets were applied is another considering factor of direct heating effect. We applied BHTMS sensor before intubation, and warm blanket was applied after intubation or mostly after positioning the patients for the operation. The time interval was not correctly measured during study, but it may have taken up to 10 minutes.
Our study has several limitations. First, we monitored and analyzed esophageal temperature range of 35.2 to 37.2℃ in orthopedic surgeries. We cannot assure about whether BHTMS is correct measuring hyperthermia or hypothermia out of that range and other surgeries. Second, for the secondary outcome, we did not compare the temperature by times after 40 minutes because the length of surgery time differed by every patients. We cannot assure about comparison by times after 40 minutes. Third, we did not compare with group of patients that did not apply warming system because of potential ethical problem. Fourth, peripheral circulation of each patient was not evaluated but it may have influence on BHTMS at wrist, since it is targeting the radial artery. Also, we did not measure peripheral skin temperature to compare. Finally, the influence of the intravenous fluid and the temperature of fluid was not evaluated. However, considering the result of secondary outcome, under the Bair Hugger™ upper body warming blankets, the influence of fluid after 30 minutes could be considered as minimal.
In conclusion, BHTMS at wrist area under upper body warm blanket may be another possible alternative method for monitoring core body temperature after 30 minutes compared to esophageal temperature.