Table 2.

Included studies performing in-ear monitoring of temperature.

Study	Device for in-ear T measurement	Comparator/reference	Experimental setting and population	Main results
Bagley et al. (2011)	IRTT (Thermoscan ExacTemp IRT 4520, Braun).	Rectal probe.	Rest and exercise in cold environment (7 ± 1°C; 48 ± 5% relative humidity, 150 min) in 25 healthy subjects.	Rectal T significantly exceeded tympanic T (p < 0.0005) in all measurements during rest and exercise conditions in cold, with heteroschedasticity between the two measurements (r2 = 0.181; bias = −1.4°C).
Basak et al. (2013)	IRTT (Genius First Temp M3000A, Tyco Healthcare Group).	Oral probe, forehead non-contact IR thermometer.	Rest condition in 452 healthy subjects at constant room temperature	Strong correlation was observed between tympanic and oral T (r2# = 0.75, P < 0.01) and tympanic and forehead T (r2# = 0.64, P < 0.01).
Basset et al. (2011)	Thermocouple thermometer (Mon-a-therm 400 series thermistor, model 90058, Mallinckrodt Medical Inc).	Rectal and esophageal probes, skin thermistors.	T measurement before, during, and after immersion of lower body surface in cold (5 ± 0.1°C) water in a thermoneutral air environment (~21.5 ± 0.5°C) in 12 healthy subjects.	Rectal T significantly dropped from baseline values (−1.0 ± 0.4°C, P < 0.05) by the end of the cooling phase, while smaller changes were observed in tympanic (−0.3 ± 0.3°C) and esophageal T (−0.1 ± 0.3°C).
Bhangu and Parmar (2010)	IRTT (Thermoscan IRT 4520, Braun).	Oral probe.	“Tough Guy” outdoor endurance event in 64 athletes (environmental T = 2.8°C; indoor T ~21°C).	Tympanic T was significantly higher than oral T at discharge [35.0°C (95% CI, 34.6–35.3°C) vs. 33.8°C (95% CI, 33.2–34.5), P = 0.003]
Chaglla et al. (2018)	Prototype thermometer with graphene-inked MLX90614-DCA IR thermopile and 3D printed ear hook enclosure. Bluetooth module for wireless connectivity to a smartphone application.	Original sensor, Cosinuss One ear thermometer, IRTT (ThermoScan 7AgePrecision IRT652, Braun).	Rest condition in 10 healthy subjects and exercise (total 25 min) in 1 subject, at room temperature (21°C).	The graphene-inked prototype demonstrated higher accuracy and was less influenced by ambient T than commercial sensors. At rest, the mean bias* was −0.36°C for the graphene-inked vs. −0.51°C for the original IRTT. During exercise, the graphene-inked prototype was less affected by airflow and ambient T.
Flouris and Cheung (2010a)	Flexible thermistor thermometer (MA-100, Thermometrics, Edison).	Rectal probe, mean body T (skin heat flow with 6 pen-shaped humidity probes), and mean skin T (12 thermistors)	Rest (1 h) and exercise (to volitional exhaustion on a cycle ergometer at 20 W with step increments of 20 W h−1) in environmental chamber at 42°C in 10 healthy males.	Tympanic T linearly increased with rectal T, mean body and mean skin T, albeit with a 100 min delay.
Flouris and Cheung (2010b)	Flexible thermistor (MA-100, Thermometrics, Edison).	Rectal probe and exhaled breath T (chip thermistor inside a valve connected to a mask).	Immersion in water tanks at 42°C/12° C until increase/decrease of rectal T with respect to baseline T, in 15 healthy individuals.	Tympanic T was significantly different (P < 0.05) from rectal T and exhaled breath T, albeit it showed good correlation with rectal T (r2# ranging from 0.20 to 0.96). Tympanic T displayed moderate delays in repetitive changes of body T in the water tanks.
Fogt et al. (2017)	IRTTs (T1: Welch Allyn Braun Pro 4000, Braun; T2: Covidien Genius2, Covidien).	Gastrointestinal pills, oral and temporal probes.	Graded exercise until voluntary exhaustion in an environmental chamber (35.5 ± 0.6°C, 53.9 ± 5.8% relative humidity) 14 young, active and healthy adults.	No differences were observed between mean T1 and pill T (LoA = ±1.90°C), while mean T2 was significantly higher (P = 0.008) than pill T (LoA = ±2.15°C).
Gagnon et al. (2010)	Thermocouple probe thermometer (Mon-a-therm, model 503-0021, Covidien-Nellcor).	Rectal and esophageal probes.	Exercise in the heat (42°C, 30% relative humidity) till hyperthermia in 24 healthy subjects. Subsequent immersion in water (26°) until rectal T reached 37.5°C (12 subjects) or recovery in a temperate environment (30°C, 30% relative humidity) for 60 min (12 subjects).	At the end of the exercise session in the heat, tympanic T was significantly lower than esophageal T (P < 0.001), but was not significantly different from rectal T. In the subsequent cold immersion or recovery stage, tympanic T was significantly lower than rectal T (P < 0.001), but was not significantly different from esophageal T.
Harmanci et al. (2018)	IRTT (Thermoscan ExacTemp IRT 4520; Braun).	Gastrointestinal pills.	Rest and two treadmill exercise sessions in a climate- controlled exercise laboratory (45–50% relative humidity and 22–25°C room T) in 15 healthy female futsal players.	No significant difference was observed between gastrointestinal and tympanic T at rest (P > 0.05). Tympanic T was significantly lower than gastrointestinal T after 10, 20, and 30 min of exercise (P < 0.05).
Kallmünzer et al. (2011)	IRTT (Genius 2, Tyco Healthcare Group).	Rectal probe.	Gel-based neck cooling for 190 min in 10 healthy subjects.	After neck cooling, a significant drop from baseline (−1.7°C, P = 0.001) was observed in tympanic T vs. a smaller decrease (−0.65°C, P = 0.019) in rectal T, concurrent with a decrease in HR of 15 bpm.
Keene et al. (2015)	IRTT (ThermoScan IRT 4520, Braun).	Gastrointestinal probe.	Two 20 min work in heat chamber set at 100°C in 37 professional firefighters.	Tympanic T underestimated gastrointestinal T by 1.3 ± 0.5°C before entering the chamber, and by 1.0 ± 0.8°C following exercise.
Lee et al. (2011)	IRTT (CE Thermo, Nipro Corporation) equipped with a silicon mold to fit into the ear. Outer ear tightly sealed using surgical tape. Wireless transmitter and receiver for telemetric system.	Rectal probe.	12 conditions: 2 activities × 3 clothing levels × 2 air T (25°C and 32°C with 50% relative humidity). WBGT: 19.4 and 25.4°C for 25 and 32°C, respectively, in 8 healthy males.	Although changes in tympanic T were significantly (P < 0.05) greater than those in rectal T for different clothing levels at rest, an overall agreement between the two T measurements was found. During exercise, tympanic T was lower than rectal T for control and HDPE conditions, but finally reached to/exceeded it for PVC conditions at both 25 and 32°C.
Morán-Navarro et al. (2019)	IRTTs (T1: Braun Thermoscan IRT6520, Braun; T2: JPD-FR100, Etekcity).	Oral and skin probes, ingestible sensor.	Outdoor 60 min submaximal exercise in the heat (40.1 ± 0.5°C, 39.5 ± 3.4% relative humidity) on cycle ergometer in 12 well-trained athletes.	T1 registered temperatures similar to the ingestible core T sensor at rest (bias* = 0.1°C; P = NS) and during exercise in the heat with (bias* = −0.1°C; P = NS) and without wind (bias* < −0.1°C; P = NS), but it registered colder T during recovery (bias* = −0.7°C; P < 0.001). T2 registered colder T values at rest (bias* = −1.4°C; P < 0.001) and during recovery (bias* = −1.1°C; P < 0.00), but warmer T during exercise in the heat without wind (bias* = 0.9°C; P < 0.001).
Muth et al. (2010)	IRTT (First Temp Genius, Sherwood Medical).	Sublingual probe.	Swimming (45 min water temperature 28°C) with immersed ears or not, in 25 healthy subjects.	At baseline, sublingual T was significantly lower than tympanic T in immersed (P < 0.001) and control (P = 0.002) groups. In the head-immersed group, sublingual T was significantly (P < 0.001) higher than tympanic T. In the head-not-immersed group, sublingual T was significantly lower than tympanic T (P = 0.002).
Nagano et al. (2010)	Tympanic thermocouple probe in earplug (3M 1110, 3M Health Care).	Esophageal, skin, and rectal probes.	Work-rest cycle simulation in warm environment (climatic chamber) in 6 healthy subjects.	Tympanic T successfully tracked rectal T during the work-rest-work simulated cycles, with slight T underestimation (bias* = −0.45 ± 0.08, −0.36 ± 0.11, and −0.3 ± 0.12 at ambient T of 25, 30, and 35°C, respectively).
Nakada et al. (2017)	Customized IRTT embedded into an earplug, measuring T at three different locations in the external auditory canal.	Esophageal and rectal probes.	Exercise (at 80 W for 45 min) in a climatic chamber (24, 32, and 40°C, at 50% relative humidity) in 11 healthy male volunteers.	Tympanic T showed progressively lower values from proximal to distal positions from the tympanic membrane (mean T values decreased from 36.86 ± 0.78°C to 36.40 ± 1.24°C to 35.12 ± 2.82°C at the three locations). Tympanic T was markedly affected by ambient T and radiation. The combination of T at proximal and distal location from the tympanic membrane can help to better estimate esophageal T.
Ota et al. (2017)	Customized 3-D printed IRTT. Bluetooth module for wireless connectivity to a smartphone application.	Skin IR thermometer, commercial IRTT.	Exercise in environmental chamber at T ranging from 6 to 40°C in 1 healthy subject.	IRTT not affected by external perturbations.
Pryor et al. (2012)	IRTT (T1: Thermoscan, model IRT 3520, Braun) and thermocouple probes (T2s: Mon-a-therm Tympanic T probe, thermistor YSI 400 Series, and Mon-a-therm Model 4070, Mallinckrodt Medical Inc).	Gastrointestinal pills, temporal artery probe, skin forehead probe, skin thermistors.	Treadmill walking protocol (<50 min) with thermal protective clothing in warm room (38.2 ± 0.9°C, 22.2 ± 3.9% relative humidity) in 50 firefighters. Subsequent forearm immersion in refrigerated water (5°C) in 25 subjects.	Both IRTT and thermocouple probes greatly underestimated deep gastric T (T1: bias = −1.31°C; LoA = 2.75°C; T2: bias = −3.28°C, LoA = 5.00°C).
Skaiaa et al. (2015)	Tympanic thermistor-based probe (Métraux®).	Rectal probe.	Exposition to four simulated in-field conditions with local cooling of head/ears (ambient air/wind without insulation, ambient air/wind with insulation, snow in ear canal with insulation, cold water in ear canal with insulation) in 13 healthy subjects.	At baseline, tympanic T significantly lower (P < 0.001) than rectal T (bias = 0.8°C). After 5 and 10 min of cold exposure, tympanic T was significantly lower (P < 0.006) than baseline T, and its significantly underestimated rectal T (bias* = −1.5/−3.2°C at 5 min and = −1.2/−2.0°C at 10 min). After surface cooling of head and neck, tympanic T did not accurately reflect rectal T within the first 10 min of measurement, with risk of over-triage.
Strapazzon et al. (2015)	Tympanic thermistor-based probe (M1024233, GE Healthcare).	Esophageal probe.	Exposition to change in ambient T from normal (23.2 ± 0.4°C) to very low (−18.7 ± 1.0°C), with and without insulation by ear protectors, in 31 healthy subjects.	Tympanic T was influenced by ambient T with increasing deviation from esophageal T at low ambient T (P = 0.007 and P < 0.001 without and with ear insulation, respectively). Ear insulation reduced T bias by 82% (from 7.2 to 1.3°C) at low T.
Suzuki et al. (2010)	IRTT (EM-30CPLB, Terumo Corporation).	Axillary thermometer, facial thermography.	Rest in cold (12.6°C, 31% relative humidity) and warm (20°C, 36% relative humidity) environments in 50 healthy subjects.	Tympanic T was significantly lower than axillary T in cold environment, and higher than facial T in cold and warm environments (P < 0.01). The ROC curve for tympanic T with axillary T as reference had AUC of 0.62 and 0.74 in cold and warm condition, respectively.
Teunissen et al. (2011)	Tympanic thermistor (P-8432, ICBT) mounted bilaterally inside a customized silicon mold, with or without protection from the environment with a cotton patch covering the complete auricle.	Esophageal and rectal probes.	Rest at 21, 10, and 30°C (50% humidity), followed by cyclo-ergometer exercise and recovery at 30°C in 7 subjects. Protocol repetition with and without face-wind during the rest period at 30°C. Extra auricle insulation at one side.	Ambient temperature affected significantly tympanic T, while rectal and esophageal T remained stable. Insulating the auricle mitigated but did not abolish the effect, nor protected from the wind. Acceptable agreement with rectal T was observed during recovery from exercise without wind (bias = −0.66 ± 0.21°C covered vs. −1.20 ± 0.15°C uncovered), while differences increased significantly with wind (bias = −1.73 ± 0.11°C covered vs. −2.49 ± 0.04°C uncovered).
Yamakoshi et al. (2013)	Microminiature IR thermo-pile sensor (10TP583T, Ishizuka Electronics Corp) and thermistor (SXK-67 & SZL-64, Technol Seven Co), molded with a micro-speaker in soft silicon material. Connectable with race face helmet.	Gastrointestinal pills.	Simulated racing environment and after a rest period of 10 min underwent a bathing period of 30 min, with the water preset to 42°C and then a 40 min period, with body T gradually reduced by natural cooling in the laboratory in 10 healthy volunteers. Real racing conditions in 2 professional drivers.	Good agreement between the tympanic IR thermopile sensor and both the thermistor (r2# = 0.97, P < 0.001; bias* = −0.01°C) and the gastrointestinal T (r2# = 0.86, P < 0.001; bias* = −0.27°C), respectively. The system functioned well during real competitive racing conditions.

Agreement between measurements is given as bias [mean, mean ± standard deviation, or median (95% limits of agreement, LoA)], calculated as (tympanic T-reference/comparator T), area under the curve (AUC) of the receiver-operating characteristic curve (ROC), and coefficient of determination (r²), as available.

Bpm, beats per minute; HDPE, high-density polyethylene coverall; HR, heart rate; IRTT, infrared tympanic thermometer; min, minutes; NIRS, near infrared-red spectroscopy; NS, non-significant; P, p-value; PVC, polyvinyl chloride coverall; T, temperature; WBGT, wet bulb globe temperature. ^#Calculated from r values in the studies; *bias sign reversed with respect to the original article to indicate (tympanic T—reference/comparator T).