JOURNAL OF TROPICAL PEDIATRICS, VOL. 57, NO.

6, 2011

A Comparison of Different Methods of Temperature

Measurements in Sick Newborns
by Sinan Uslu,1,2 Hamus Ozdemir,2 Ali Bulbul,1 Serdar Comert,1 Fatih Bolat,1 Emrah Can,1 and Asiye Nuhoglu1
1
Department of Pediatrics, Division of Neonatology, Sisli Etfal Children Hospital, Istanbul, Turkey
2
Department of Pediatrics, Division of Neonatology, Diyarbakir Children Hospital, Diyarbakir, Turkey

Correspondence: Sinan Uslu, Department of Pediatrics, Division of Neonatology, Darusafaka mah. Acelya sok. Yonca sit.
1B blok D:9 Istinye, 34460, Istanbul, Turkey; Sisli Etfal Children Hospital, Halaskargazi Cad. Sisli, 34360, Istanbul, Turkey.
Tel: þ90 532 737 00 15, Fax: þ90 212 234 11 21. E-mail: <[email protected]>.

Summary
We aimed to compare the accuracy of digital axillary thermometer (DAT), rectal glass mercury therm-
ometer (RGMT), infrared tympanic thermometer (ITT) and infrared forehead skin thermometer (IFST)
measurements with traditional axillary glass mercury thermometer (AGMT) for intermittent tempera-
ture measurement in sick newborns. A prospective, descriptive and comparative study in which five
different types of thermometer readings were performed sequentially for 3 days. A total of 1989
measurements were collected from 663 newborns. DAT and ITT measurements correlated most closely
to AGMT (r ¼ 0.94). The correlation coefficent for IFST and RGMT were 0.74 and 0.87, respectively.
The mean differences for DAT, ITT, RGMT and IFST were þ0.02 C, þ0.03 C, þ0.25 C and
þ0.55 C, respectively. There were not any clinical differences (defined as a mean difference of
0.2 C) between both mean AGMT&DAT and AGMT&ITT measurements. Our study suggests that
tympanic thermometer measurement could be used as an acceptable and practical method for sick
newborn in neonatal units.

Key words: Temperature, newborn, thermometers, axillary, rectal, tympanic, digital, infrared.

Introduction potential harm and toxic vapor effects to health

Maintainance and monitoring of the thermoregula- workers and the patients. Therefore, glass mercury
tion of newborn infants is a basic requirement of thermometers are now used rarely in developed coun-
good neonatal management and play a key role for tries and studies are conducted to understand the ef-
neonatal nursing care in the neonatal intensive care ficacy of the various types of equipment available to
unit (NICU). Determining accurate measurement of measure temperature [6].
temperature is very important because abnormal The aim of this study was to compare the accuracy
temperature is strongly associated with a serious con- of digital axillary thermometer (DAT), rectal glass
dition [1, 2]. Temperature of newborns may be mea- mercury thermometer (RGMT), infrared tympanic
sured in a variety of sites, including rectum, tympanic thermometer (ITT) and infrared forehead skin
membrane, forehead and axilla and using a number thermometer (IFST) measurements with traditional
of different tools, including glass mercury, digital, axillary glass mercury thermometer (AGMT) for
electronic and infrared thermometers. intermittent temperature measurement in the NICU.
Temperature measurement by the axillary method
has become the accepted neonatal nursing care [3, 4]. Methods
Because of the greater surface vasculature, increased This study was performed between 1 December 2008
body fat and thermal uniformity, temperature mea- and 1 April 2009 in the NICU of the Diyarbakir
sured in the axillary area is considered reliable and is Children Hospital, in Turkey. During the study
used as a standard measurement site in newborns. No period, the newborns who were hospitalized in
significant differences have been noted between left NICU of the Diyarbakir Children Hospital, were
and right axillae [5]. included in the study. The patients with unstable con-
Glass mercury thermometers are historically the ditions such as dysmorphic appearance, congenital
most acceptable standard methods of temperature anomaly, septic shock and circulatory problem that
measurement. However, glass mercury thermometers would effect measurement were excluded from
has some disadvantages such as danger of breakage, the trial. This was a prospective, descriptive and

 The Author [2011]. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected] 418
doi:10.1093/tropej/fmq120 Advance Access published on 18 January 2011

Downloaded from https://academic.oup.com/tropej/article-abstract/57/6/418/1735550

by guest
on 27 March 2018
 S. USLU ET AL.

comparative study in which three temperature read- 3 min was needed for axillary and 2 min for rectal
ings were performed sequentially 3 days from each thermometer, approximately 1–3 min with digital
newborn in incubator by the same neonatal nurse. thermometer, 3 s with tympanic thermometer and
A health worker helped the nurse for stabilization 1 s with infrared skin thermometer. All measurements
of the patients. Prior to the onset of data collection, were terminated nearly at 8–10 min.
the data collector nurse was instructed for correct Five temperature measurements were obtained by
temperature measurements procedures utilizing the data collector for each sample. Data analysis
five methods. An oral informed and written consent included Pearson’s r coefficients (to determine the
were obtained from the parents of the infants. strength of the correlation), paired t-tests (to deter-
Temperature was measured using the following: mine statistically significant difference), standard de-
glass mercury thermomether (for axillary and rectal viation, mean and range using SPSS statistical
tools separately), digital thermometer (Microlife package. Scatter plot method was used in order to
digital thermometer, model MT 3001, Microlife AG compare axillary glass thermometer with every other
Swiss Corp., Widnay, Switz), tympanic thermometer devices. Clinical signifance was defined as a mean
(First Temp Genius, Tyco Healthcare Kendall, difference of 0.2 C between axillary glass tempera-
Mansfield, Massachusetts) with disposable probe ture and other four measurements [9].
covers and infrared skin thermometer (Thermoflash
LX-26, Visiomed France, Mountreuil, France). Results
The glass mercury thermometers and the tympanic
thermometer were supplied by hospital’s central ser- During the study period, 742 patients were hospita-
vice. The infrared skin thermometers and digital lized in the NICU. Seventy nine neonates who met
thermometers were provided by the authors of the the exclusion criteria were not included in the trial.
study. The glass mercury thermometer (GMT) and Six hundred sixty-three newborn infants were
digital thermometer were used separately for each included in this study and 1989 measurements of
patients. data were used during statistical analysis. Three hun-
Bilateral axillae were utilized for glass and digital dred and forty-one of them were males (51.4%) and
thermometers, and right ear was chosen for tympanic the average gestational age, birth weights and post-
thermometer. Infrared skin temperature was ob- natal age during the study period were 36  3.6
tained on the central part of forehead. Axillary and weeks, 2468  11 g and 11  7.4 days, respectively.
forehead region were dried using a towel before the Among the babies in the study group, 305 (46%)
measurement. The GMT was shaken before using in were preterm whereas 358 (54%) term. The study
order to decrease the reading below 35 C. Infrared group did not include any postmature baby.
skin thermometer was put in incubator or it was kept Among the babies, 319 (48.1%) were found to be
at room temperature for 15 min before using. First, low birth weight (LBW) (<2500 g) whereas 344
digital and glass mercury thermometers were ran- (51.9%) had a birth weight 2500 g. The temperature
domly placed in bilateral axillas separately in values of the patients and mean differences according
supine position at the same time. Patients were sta- to gestational age (preterm and term) and birth
bilized by a health worker. Simultaneously tympanic weight (<2500 g and 2500 g) were summarized in
thermometer was inserted into the right external Table 1. When all methods of temperature measure-
auditory canal by pulling the pinna straight back ment were taken into consideration, any clinical dif-
and the probe was directed toward the eye. The ference which was defined as a mean difference of
prob was held in the ear canal until a beep was 0.2 C was not observed between patients regarding
heard. After this procedure according to principles gestational age and birth weight. The hospitalization
and precautions of manufacturer, infrared skin etiologies of patients in the study group were shown
thermometer readings were recorded three times by in Table 2.
placing the device approximately 5–15 cm above fore- The correlation between five methods were given
head skin (mean value was accepted). Immediately in Table 3. DAT and ITT measurements correlated
after IFST measurement, a rectal glass thermometer most closely to AGMT (r ¼ 0.94). The correlation
was inserted upwards to a depth of 3 cm in a term coefficent for IFST and RGMT were 0.74 and 0.87,
and 2 cm in a preterm baby. respectively. The comparison of measurement of
All the temperature measurements were recorded DAT, IFST, RGMT and ITT with AGMT were
in the morning between hours of 08:00 till 12:00 shown in Figs 1–4, respectively. The comparison of
during the study period. The room air temperature the temperature readings done by AGMT and DAT,
and relative humidity were kept constantly at and AGMT and ITT is seen in Figs 1 and 2, respect-
22–26 C and a relative humidity of 30–60%, respect- ively. A significant correlation was found between
ively [7]. The temperatures of the incubators were comparable methods (Fig. 1, Rsq 0.886, Fig. 2, Rsq
adjusted according to the standard temperature 0.885). The comparison of AGMT and RGMT tem-
reccommendations based on gestational age [8]. All perature measurements were shown in Fig. 3. There
readings were done by celcius ( C) scale. A time of is good correlation between these methods (Fig. 3,

Journal of Tropical Pediatrics Vol. 57, No. 6 419

Downloaded from https://academic.oup.com/tropej/article-abstract/57/6/418/1735550

by guest
on 27 March 2018
 S. USLU ET AL.

TABLE 1
The temperature measurement of the patients according to gestational age (preterm and term) and
birth weight (<2500 g and 2500 g)

Types of Gestational age Birthweight

Thermometera
Preterm Term babies Mean <2500 g 2500 g Mean
babies (n ¼ 305) (n ¼ 358) differenceb (n ¼ 319) (n ¼ 344) differenceb

AGMT 36.71  0.41 36.76  0.41 0.05 36.70  0.40 36.77  0.41 0.07
DAT 36.72  0.41 36.77  0.40 0.05 36.72  0.40 36.78  0.41 0.06
ITT 36.73  0.41 36.78  0.41 0.05 36.72  0.41 36.80  0.41 0.08
RGMT 36.94  0.42 36.98  0.41 0.04 36.94  0.41 36.99  0.42 0.05
IFST 37.21  0.47 37.24  0.48 0.03 37.21  0.48 37.23  0.48 0.02
a
Values were given as mean  SD by celcius ( C) scale.
b
There were no clinical differences (defined as a mean difference of 0.2 C) between patients according to gestational age
(preterm and term) and birth weight (<2500 g and 2500 g).

TABLE 2 TABLE 3
The hospitalization etiologies of patients in Correlation coefficents for temperature
the study group measurements between axillary glass thermometer
and other four methods
Etiologies Infants (n ¼ 663) n (%)
Types of Pearson’s correlation
Respiratory diseases 273 (41.2) Thermometer coefficents
Perinatal asphyxia 109 (16.4)
Sepsis 91 (13.7) AGMT 1.0000
Prematurity 78 (11.8) DAT 0.94
Hyperbilirubinemia 41 (6.2) ITT 0.94
Surgical diseases 23 (3.5) RGMT 0.87
Other diseases 48 (7.2) IFST 0.74

FIG. 1. Scatter plot of the difference between tem- FIG. 2. Scatter plot of the difference between tem-
peratures measured by AGMT and DAT. peratures measured by AGMT and ITT.

Rsq 0.758). Poor correlation between AGMT and

IFST temperature measurements is seen Fig. 4 (35.4–38.6 C). The IFST had the widest range
(Rsq 0.596). (35.9–39.6 C). Mean AGMT measurement (36.74 
The mean temperature measurements were 0.41 C) was compared with DAT (36.75  0.40 C),
given in Table 4. The DAT had the smallest range ITT (36.76  0.42 C), RGMT (36.97  0.42 C) and

420 Journal of Tropical Pediatrics Vol. 57, No. 6

Downloaded from https://academic.oup.com/tropej/article-abstract/57/6/418/1735550

by guest
on 27 March 2018
 S. USLU ET AL.

FIG. 3. Scatter plot of the difference between tem- FIG. 4. Scatter plot of the difference between tem-
peratures measured by AGMT and RGMT. peratures measured by AGMT and IFST.

TABLE 4 acquired infections [2, 6]. Therefore, several studies

Measurement ranges of five methods have been conducted to determine the most accurate
and practical device.
Types of Mean ( C) Standard Range ( C) Digital thermometers have been shown to be a
Thermometer deviation valuable method for neonatal temperature measure-
ments. Sganga [6], Smith [11] and Leick-Rude et al.
AGMT 36.74 0.41 35.5–38.7 [12] found that the digital thermometer had a high
DAT 36.75 0.40 35.4–38.6
ITT 36.76 0.42 35.4–38.7
correlation with the AGMT. In our study, the mean
RGMT 36.97 0.42 35.5–39.1 difference between AGMT and DAT was statistically
IFST 37.22 0.47 35.9–39.6 significant, but not clinically significant. We demon-
strated a good correlation between AGMT and
DAT.
ITT is useful for clinical temperature measurement
as long as moderately variability between patients is
IFST (37.22  0.47 C). When we evaluated the cor-
acceptable [13]. There are studies supporting the use
relation between AGMT and the other methods, the
of ITT among various age groups including new-
mean differences for DAT, ITT, RGMT and IFST
borns [14–16]. Weiss et al. [17] and Weiss [18]
were þ0.02 C, þ0.03 C, þ0.25 C and þ0.55 C,
found that mean ear and axillary temperatures were
respectively. The mean differences were significant
highly correlated in newborns. However, its accuracy
(p ¼ 0.001, p ¼ 0.001, p < 0.001, p < 0.001, respective-
has been questioned for newborns in some studies
ly). But there were no clinical differences (defined as
[6, 19]. In our study, we found a high correlation
a mean difference of 0.2 C) between mean AGMT
between AGMT and ITT measurements. Although
and DAT measurement, and mean AGMT and ITT
the mean difference of temperature measurements
measurement.
was statistically significant, a clinical significance
was not found.
Discussion Rectal temperature measurement with glass mer-
Accurate and practical temperature monitoring is cury thermometer was reported to be the accurate
essential for clinical neonatal care and time manage- method compared with the axillary measurements
ment in neonatology units. Glass mercury thermom- in newborns [20, 21]. Fulbrook et al. [22] stated
eters and axillary region have been considered the that rectal temperatures to be consistently higher
gold standard method of temperature measurement than temperatures taken at other sites possibly due
in newborns [6, 10]. Although glass mercury thermo- to increased metabolic activity or bacteria. Hissink
meters have been used for a long period of time, this et al. [23] showed a wide variation of the mean dif-
procedure had also some important disadvantages ference between axillary and rectal temperature in
such as danger of breakage, potential harm and newborns. In a systematic review, Craig et al. [24]
toxic vapor effects; difficulties in reading the values concluded that the difference between temperature
on the device; and possible role in spread of hospital reading at the axilla and rectum using glass mercury

Journal of Tropical Pediatrics Vol. 57, No. 6 421

Downloaded from https://academic.oup.com/tropej/article-abstract/57/6/418/1735550

by guest
on 27 March 2018
 S. USLU ET AL.

thermometer showed wide variation across studies. 2. Rosenthal HM, Leslie A. Measuring temperature
Our study had same result also. There was a statis- of NICU patients-A comparison of three devices.
tical and clinical significance between AGMT and J Neonatal Nurs 2006;12:125–9.
3. National Association of Neonatal Nurses (NANN).
RGMT.
Neonatal thermoregulation guidelines for practice.
Kemp et al. [25] compared AGMT and infrared Glenview, IL: NANN, 1997.
axillary skin temperature and found good correlation 4. Haddock BJ, Merrow DL, Swanson MS. The falling
between two devices but the measurements were from grace of axillary temperatures. Pediatr Nurs 1996;22:
the different regions. Can et al. [26] concluded that 121–5.
the noncontact infrared thermometer could not be 5. Shenep JL, Adair JR, Hughes WT, et al. Infrared, ther-
recommended for assessment of body temperature mistor, and glass-mercury thermometry for measure-
in newborns admitted to NICU. Similarly, we ment of body temperature in children with cancer.
Clin Pediatr 1991;30(Suppl. 4):36–41.
found a less correlation between AGMT and IFST 6. Sganga A, Wallace R, Kiehl E, et al. A comparison
results. The mean difference of measurements was of four methods of normal newborn temperature
significant both statistically and clinically. measurement. MCN Am J Matern Child Nurs 2000;
According to our study results, IFST measurement 25:76–9.
does not seem to be suitable and accurate for sick 7. White RD. Recommended standards for the newborn
newborns. ICU. J Perinatol 2007;27:S4–19.
This study has some limitations. Measurements 8. Rutter N. Temperature control and disorders.
were performed from the different body region by In: Rennie JM (ed.). Roberton’s Textbook of
Neonatology, 4th edn. Philadelphia: Elsevier Churchill
the same neonatal nurse. Different body regions
Livingstone, 2005; pp. 267–79.
have separate surface vasculature, metabolic activity 9. Keeling EB. Thermoregulation and axillary tempera-
and body fat composition. This condition might ture measurements in neonates: a review of the litera-
affect the measurements of the temperature. ture. Matern Child Nurs J 1992;20:124–40.
However, Kunnel et al. [20] found no difference in 10. Devrim I, Kara A, Ceyhan M, et al. Measurement
temperatures taken rectally, femorally, from the accuracy of fever by tympanic and axillary thermom-
axilla or skin. Another limitation of our study was etry. Pediatr Emerg Care 2007;23:16–9.
that we did not have a true measure of core body 11. Smith J. Are electronic thermometry techniques suitable
alternatives to traditional mercury in glass thermometry
temperature to use as a criterion standard. In the techniques in the paediatric setting? J Adv Nurs 1998;
literature, esophageal or pulmonary artery tempera- 28:1030–9.
tures are generally considered to be true measures of 12. Leick-Rude MK, Bloom LF. A comparison of
core body temperature [27]. But both methods are temperature-taking methods in neonates. Neonatal
invasive procedures. Netw 1998;17:21–37.
The ideal measuring device should be a noninva- 13. Ç|nar ND, Filiz TM. Neonatal thermoregulation.
sive, with a rapid result, accurate and practical to use. J Neonatal Nurs 2006;12:69–74.
We measured temperature using five different devices 14. Kocoglu H, Goksu S, Isik M, et al. Infrared tympanic
thermometer can accurately measure the body tempera-
and different sites in the same baby three times to ture in children in an emergency room setting. Int J
find an ideal tool. We found statistically significant Pediatr Otorhinolaryngol 2002;65:39–43.
differences between measurements, but concluded 15. Erickson RS, Woo TM. Accuracy of infrared ear
that the differences between AGMT and DAT or thermometry and traditional temperature methods in
ITT seen (0.01–0.02 C) were not clinically significant. young children. Heart Lung 1994;23:181–95.
Both of them were noninvasive, but ITT had more 16. Fulbrook P. Core body temperature measurement: a
rapid results. comparison of axilla, tympanic membrane and pulmon-
ary artery blood temperature. Intensive Crit Care Nurs
1997;13:266–72.
Conclusion 17. Weiss ME, Poeltler D, Gocka I. Infrared tympanic
Good correlation with glass mercury thermometer, thermometry for neonatal temperature assessment.
rapid result delivery, improved patient comfort, J Obstet Gynecol Neonatal Nurs 1994;23:798–804.
being an easy and noninvasive procedure and lacking 18. Weiss ME. Tympanic infrared thermometry for full-
term and preterm neonates. Clin Pediatr 1991;
the disadvantages of glass mercury thermometer are 30(Suppl. 4):42–5.
the advantages of tympanic thermometer. Our study 19. Yetman RJ, Coody DK, West MS, et al. Comparison
suggests that tympanic thermometer measurement of temperature measurements by an aural infrared
could be used as an acceptable method for sick new- thermometer with measurements by traditional rectal
borns in the neonatal units. and axillary techniques. J Pediatr 1993;122:769–73.
20. Kunnel MT, O’Brien C, Munro BH, Medoff-Cooper B.
Comparisons of rectal, femoral, axillary, and skin-to-
References mattress temperatures in stable neonates. Nurs Res
1. World Health Organization. Thermal protection 1988;37:162–4.
of the newborn: a practical guide. World Health 21. Jirapaet V, Jirapaet K. Comparisons of tympanic mem-
Organization. WHO/RHT/MSM/97.2, 1997. brane, abdominal skin, axillary, and rectal temperature

422 Journal of Tropical Pediatrics Vol. 57, No. 6

Downloaded from https://academic.oup.com/tropej/article-abstract/57/6/418/1735550

by guest
on 27 March 2018
 S. USLU ET AL.

measurements in term and preterm neonates. Nurs 25. Kemp C. Infrared skin thermometer good alternative
Health Sci 2000;2:1–8. for newborns. AAP News 2008;29:2.
22. Fulbrook P. Core temperature measurements in adults: 26. Can E, Bulbul A, Uslu S, et al. Comparison of
a literature review. J Adv Nurs 1993;18:1451–60. non-contact infrared forehead thermometer to standard
23. Hissink Muller PC, van Berkel LH, de Beaufort AJ. temperature measurement in neonatal intensive care
Axillary and rectal temperature measurements poorly unit patients. Turk Arch Ped 2010;45:257–63.
agree in newborn infants. Neonatology 2008;94:31–4. 27. Harasawa K, Kemmotsu O, Mayumi T, et al. Compar-
24. Craig JV, Lancaster GA, Williamson PR, et al. ison of tympanic, esophageal and blood temperatures
Temperature measured at the axilla compared with during mild hypothermic cardiopulmonary bypass: a
rectum in children and young people: systematic study using an infrared emission detection tympanic
review. Br Med J 2000;320:1174–8. thermometer. J Clin Monit 1997;13:19–24.

Journal of Tropical Pediatrics Vol. 57, No. 6 423

Downloaded from https://academic.oup.com/tropej/article-abstract/57/6/418/1735550

by guest
on 27 March 2018