A Novel Non-Invasive Thermometer for Continuous Core Body Temperature: Comparison with Tympanic Temperature in an Acute Stroke Clinical Setting

Among the physiological parameters commonly monitored in clinical practice, body temperature is one of the major clinical signs and is used to influence clinical management decisions and the diagnosis of certain diseases [1]. Several sites and devices are commonly used to measure body temperature [2,3,4], and due to local metabolism effects, anatomical characteristics, tissue conduction, and blood flow, measurements taken by peripheral thermometers can be highly variable and underestimate the core temperature [5,6,7,8,9]. The inter- and intraindividual variability in normal body temperature further complicate the definition of fever conditions [10], and the most important factors are the site of measurement and the patient’s age [11]. Fever has been defined as a core temperature above 38.3 °C [12]; nevertheless, differences are present between core and peripheral body temperatures [13], and core body temperature is rarely measured in non-surgical or intensive care settings due to the invasiveness of the devices and techniques—e.g., the insertion of vascular, oesophageal, rectal, or bladder catheters [1,14,15]. Fever can be found in about half of patients admitted to intensive care units (ICU) and can be associated with clinical worsening and increased mortality [16,17], and a similar prevalence has been found in patients admitted with stroke [18]. In stroke patients, the continuous monitoring of body temperature is recommended, but it can be often impractical, and, in most cases, it relies on peripheral temperature measurements [19,20].

Due to the great importance of measuring body temperature with non-invasive and accurate methods, novel devices have been developed to measure or estimate core temperature from the skin surface, including the zero-heat-flux thermometers that have been mostly evaluated in surgery and ICU [21]. Other devices have been developed, deriving core body temperature noninvasively from heat flux gradients [22]. Such devices—particularly those that are small and adherent to the patients’ body and do not require wired connections—could be applicable in a wide range of clinical settings, including people with acute neurological diseases. In addition, especially during the COVID-19 pandemic, there was a growing use of forehead temperature measurements in clinical settings as well [23,24,25].

The main aim of the present study was to evaluate the use of a novel wireless non-invasive heat flux-based thermometer in acute stroke patients admitted to a stroke unit and to compare the measurements with the currently used infrared (IR) tympanic measurements. Moreover, the secondary aim was to compare IR forehead temperature readings with tympanic measurements performed at the same time.

An ecologically valid design was used to compare the measurements from the novel device and the practices currently used in the stroke unit. All the patients that were consecutively admitted to the stroke unit from 01.02.2021 to 31.07.2021 with a diagnosis of acute stroke were considered eligible to enter the study. Patients were excluded if they were expected to stay in the stroke unit for less than 3 days, if they were found to be positive for SARS-CoV-2 based on a nasopharyngeal swab at admission, or if they suffered from ear inflammation, forehead swelling, head injuries, or skin or peripheral vascular disease which might have impacted the temperature readings. Participants were also excluded if it was not possible to apply the novel device more than 4 h from admission. Demographics and clinical data were also registered. The device was applied to the left side of the chest according to the manufacturer’s instructions and using the provided adhesive patch. The Bluetooth connection was activated and connected with a tablet for clinical use. The recordings for each participant started from 1 h after the device application and lasted for a maximum of 72 h. All the procedures were performed according to the Declaration of Helsinki, and the local institutional review board and ethics committee (CEUR-FVG) approved the study with an approval number of 115/2018.

The data were collected from a sample of 30 patients (6 females, 71.9 ± 18.9 y, age range 22–97 y) who were admitted with a diagnosis of acute ischemic stroke within 6 h of the onset of the symptoms. The body mass index was higher than 25.0 in 14 (46.6%) patients in the sample. The NIHSS was 7 ± 6, and all the subjects had an anamnestic mRS ≤ 2. Demographics and stroke characteristics are reported in Table 1. The novel CORE device was applied between 30 min and 2 h and 35 min from stroke unit admission, and recorded data were included starting from the second hour of measurement to allow for proper stabilization. Routine body measurements were taken, on average, 3 times/day, with a maximum of 5 times/day. A total of 305 measurements for the Tcore and Ttym recordings were included in the final analyses, while 246 Tfor measurements were collected due to technical issues.

The mean error of the novel wireless non-invasive heat flux-based device compared to the standard tympanic temperature measurements was 0.11 °C. In particular, 85% of all the Tcor measurements were within ±0.5 °C of the tympanic measurements.

In Figure 1, the estimated core temperatures (Tcore) and the forehead temperatures (Tfor) were plotted against the tympanic temperatures (Ttym) for each measurement. A significant linear correlation was observed between Tcore and Ttym (r = 0.89, p < 0.001). Similarly, a moderate correlation was also observed between Tfor and Tymp (r = 0.74, p < 0.001).

The comparison of the Tcore and Ttym measurements by Bland–Altman analysis (Figure 2; left panel) showed a good agreement between them, with a very low mean difference of 0.11 ± 0.34 °C (95%CI −0.55–0.77 °C). No proportional bias was detected (B = −0.003, p = 0.923).

The Bland–Altman comparison between the Tfor and Ttym (Figure 2; right panel) showed a mean difference of −0.14 ± 0.44 °C (95%CI −0.72–1.00 °C). However, a proportional bias was detected (B = −0.388, p < 0.001), with a trend of an overestimation of lower temperatures and an underestimation of higher temperatures. Due to the aforementioned trend, an error up to 1.0 °C was possible in the range between 36.0 °C and 38.5 °C.

The strong desire to monitor human health data is driving the rapid growth of wearable biosensing technologies. Among the different wearable biosensing devices, temperature measurement has proven to be an important physiological parameter [29]. The accurate and continuous measurement of the human core body temperature by a wearable device is of great significance for human healthcare and disease monitoring [30,31]. However, accurate monitoring of body temperature often represents a challenge in clinical practice due to the different devices available, the site of measurement, and the possibility of the healthcare staff obtaining frequent measurements when continuous devices are not available.

This study provides early evidence of the usability of a novel non-invasive wireless heat flux-based thermometer that could provide continuous monitoring and estimation of core body temperature. The estimated core temperatures showed a good agreement (0.11 ± 0.34 °C) and no proportional bias with the tympanic measures, which is a commonly used method in everyday clinical practice in post-stroke patients. The observed mean error compared to the standard tympanic temperature measurements was comparable with that observed in previous clinical studies [6,9], while the calculated limits of agreement were slightly higher than the ±0.5 °C limits that have been used in previous studies in the clinical setting [6,9,32,33,34] and which correspond to the usual magnitude of the human circadian temperature variation [35,36].

Indeed, although such technology might have some limitations in providing an assessment of core body temperature during exercise and its precision should be further evaluated [37,38,39], it might be useful for detecting hyperthermic events in the clinical setting, particularly when compared to other non-invasive methods which might not provide continuous temperature readings [40,41,42] or when patients have to be remotely monitored [1,43,44,45,46]. Our results showed that a novel non-invasive wireless heat flux-based thermometer was able to detect body temperature changes measured by the tympanic thermometer as the presence or absence of hyperthermia (94.1%) and the presence or absence of fever (97.4%) in the post-stroke clinical setting.

In stroke patients, the American Heart Association (AHA)/American Stroke Association (ASA) guidelines recommend obtaining body temperature measurements every 30 min while in the emergency department and every 4 h in the acute setting (stroke unit) [47,48]. However, these guidelines do not define the most appropriate site of measurement, despite the variety of practices ranging from oral, axillary, tympanic, and core body temperature, but recommend caution, as all these methods might differ from each other and might underestimate brain temperature [48]. These differences should be considered when defining the body temperature limits to start cooling and antipyretic treatments, as 37.5 °C is often reported as a cut-off value for worsening stroke symptoms or other declines in neurological status [47].

In this ecologically valid study, the standard clinical practice was maintained by the healthcare staff, meaning they used the standard devices and approaches to measure body temperature. The first finding is represented by the frequency of routinely taken measurements, suggesting that most of the patients were monitored three times per day (usually at 7:00, 14:00, and 22:00), and only in case of fever did they receive additional measurements up to five times per day. This observation suggests that, despite the guidelines recommending more frequent monitoring of body temperature (six times per day), it might be impractical and unfeasible in some clinical settings. The main device used to measure body temperature was the infrared ear thermometer, which represents one of the most common methods to non-invasively assess body temperature by measuring tympanic temperature [49]. Our results also showed the importance of continuous monitoring. Indeed, in fifteen out of thirty patients, at least one possible hyperthermic or fever episode was not detected by non-continuous measurements by the standard tympanic method. The higher measurement frequency allowed by the wearable device detected nineteen hyperthermic and two fever episodes which were not identified by tympanic measurements. Due to the absence of a continuous gold-standard core body temperature measurement, it is not possible to define if those missing events were effective hyperthermic/fever events or false positives; most of the missed events (76%) occurred during the night, when usually 8–10 h passed between the two tympanic temperature measurements.

Among the new devices that have been applied to non-invasively measure body temperature and which became particularly popular during the recent COVID-19 pandemic, non-contact infrared thermometers have been used to measure the surface temperature at the forehead. However, conflicting results have been found in determining its accuracy to predict fever, particularly when non-standardized protocols are adopted (e.g., distance, position, etc.) and the devices are not properly calibrated [8,50,51,52,53,54,55].

Recently, to eliminate the environmental effects and thus improve the accuracy in estimating core body temperature, methods based on direct and correcting temperature data for measurements with an infrared thermal camera [56] and methods based on a dynamic model for measurements with a passive heat flow sensor [57] were proposed. In addition, it has been shown that personalized models can improve measurement accuracy [30].

The forehead is also used as a site of measurements for the zero-heat-flux method and the double sensors method—non-invasive solutions to estimate core body temperature that were shown to be reliable in a wide set of conditions [58,59,60]. Indeed, several active or passive sensing techniques for the non-invasive estimation of core body temperature based on heat flow have been proposed. The zero-heat-flux method [61] is based on active sensors which typically contain a self-heating layer to prevent heat transfer between the skin and the environment. Usually, an active sensing device consists of an insulator patch covered by an electric heater, which creates an isothermal tunnel from the body to the skin surface, allowing for the estimation of the temperature of deeper tissues from the skin temperature. These sensors provide clinicians with the ability to non-invasively monitor body temperature, present a good agreement with core body temperature [62] and sufficient accuracy during stable temperature periods [63,64,65], and are increasingly being used to substitute for more invasive core temperature measurements during surgery and in critical care [21,62]. However, the main disadvantage of the zero-heat-flux method is represented by the presence of a servo-controlled heater, necessary for the heat loss to the ambient atmosphere and requiring considerable alternating current power, therefore making it necessary to have it connected to a socket [59]. If this characteristic minimally influences its application in severe patients requiring prolonged bed rest, in confused or agitated patients, or in moderately severe patients who might be able to stand up and walk (e.g., to go to the bathroom), this might represent a major limitation and prevent its wide use in semi-intensive care units and stroke units. In contrast to the zero-heat-flux method, a novel non-invasive wireless heat flux-based device does not require active heating and is a low-power wireless mobile solution allowing for remote monitoring. Compared to the double sensor method, where the temperature sensors need to be stacked on top of each other, the heat flux and temperature sensor of a novel device are placed on the same level, allowing for a thinner and more compact design and integration into a small non-invasive wearable tool.

This study presents some relevant limitations, which arise from the ecologically valid setting of the study and the maintenance of standard clinical practice. The results were obtained from a relatively small population of only 30 post-stroke patients in a single stroke unit and should be confirmed in a larger study. However, the study encompassed different stroke severity patients, and, on average, every patient was monitored for more than 2.8 days. We also did not observe many temperature readings above 39 °C in order to be able to evaluate the device’s accuracy in a higher temperature range. In addition, the absence of a comparison with a well-validated gold-standard continuous core body temperature measurement, along with the inability to confirm hyperthermic/fever events with other parameters (e.g., blood analyses), suggest considering the results presented in this study as a comparison between devices, and further clinical validation studies, including gold-standard and continuous body temperature measurements, are therefore encouraged.

In conclusion, temperature monitoring with a novel wireless non-invasive heat flux-based thermometer could be a reliable alternative method for assessing core temperature in acute ischemic stroke patients compared to standard tympanic temperature measurements. The estimated core temperatures showed a good agreement and no proportional bias compared to commonly used tympanic measurements in everyday clinical practice in post-stroke patients, and they provided wireless continuous monitoring in the stroke unit.

Conceptualization, A.B.S., M.A., G.F. and P.M.; Methodology, A.B.S. and M.A.; Software, A.B.S. and M.A.; Validation, G.F., P.C., M.N. and P.M.; Formal Analysis, A.B.S., M.A. and A.A.; Investigation, A.B.S., M.A., G.F., P.C. and M.N.; Resources, A.A. and P.M.; Data Curation, M.A. and A.A.; Writing—Original Draft Preparation, A.B.S. and M.A.; Writing—Review & Editing, all of the authors; Visualization, M.A.; Supervision, A.A. and P.M. All authors have read and agreed to the published version of the manuscript.

The authors declare no conflict of interest for the present study. The research was conducted according to the principles of the Declaration of Helsinki. This study was approved by the Local Ethics Committee CEUR (Comitato Etico Unico Regionale, FVG, Italy).

All participants released their informed consent for the treatment of clinical data after all procedures had been fully explained, as for a standard institutional procedure.

Anonymized data are available upon reasonable request to the corresponding author.

The authors declare no conflict of interest.

This research received no external funding.