Expert Review Measurement of Body Temperature
Timothy Wait* and Paul Dennis†
……………………………………………………………………………………………………………………………………..
The Journal of Clinical Examination 2013 (07): 29-41
Abstract
This article describes an evidence-based method for measurement of body temperature consistent with The
Principles of Clinical Examination [1]. Temperature measurement at oral, aural, rectal and axillary sites is
addressed, and measurement technique, accuracy and precision, and the normal ranges are discussed. We
recommend a method for the measurement of body temperature based on the best evidence available at the time
of publication. We welcome responses to this article which provide evidence for improvements to the method
described here. Word count: 1,789 (excluding abstract, figures, tables and references).
Key words: temperature, thermometers, examination, fever, hypothermia
Address for correspondence: [email protected]
Author affiliations: * FY1 General Surgery, Pilgrim Hospital, Boston † Department of Pharmacology, University of
Oxford
……………………………………………………………………………………………………………………………………
Introduction
Core body temperature is the average temperature
of the core thermal compartment, most accurately
approximated using a pulmonary artery catheter. In
hospital and general practice settings, the core
temperature is often approximated by measurement
of temperature at peripheral sites. Abnormalities in
body temperature are an important clinical sign and
frequently influence decision making in clinical
practice.
Defining normal body temperature
The normal range for core temperature is generally
considered to be between 36.5 °C and 37.5 °C.
However, a number of studies comparing core
temperature (measured via a pulmonary artery
catheter) with simultaneous measurement of
peripheral temperatures suggest that there may be
discrepancies between the two. In general terms,
measurements obtained at oral, aural and axillary
sites tend to underestimate the true core
temperature. In contrast, measurements obtained
by rectal thermometry may overestimate the true
core temperature.
There is significant uncertainty in the available data.
The precision of peripheral temperature
measurement for estimating pulmonary artery
temperature is poor (evidence boxes 1-4). The
authors recommend that temperatures measured at
peripheral sites are compared with the (relatively
well established) normal ranges for temperature at
that site (table 1), rather than extrapolating to an
estimated “true” core temperature.
Elevated body temperature is most commonly
caused by infection but the differential diagnosis is
broad and includes systemic inflammatory response
syndrome (SIRS) (of which infection is the most
common
cause),
malignant
disease,
hyperthyroidism, environmental and exertional
hyperthermia, drug effects and drug reactions
(importantly including neuroleptic malignant
syndrome and malignant hyperthermia). The
differential diagnosis for hypothermia includes
sepsis,
SIRS,
environmental
exposure,
hypothyroidism,
adrenal
insufficiency,
hypoglycaemia, and drug or alcohol intoxication.
The most important use of measurements of body
temperature is in diagnosis but extreme
derangements in body temperature require urgent
treatment to bring the core temperature into a safe
range.
29
Site
Oral
Aural
Axillary
Rectal
Age group
studied
18-40
>70
11 - 85
>85
>70
>70
Lower limit (°C)
36.0
36.2
35.5
36.7
Upper limit
(°C)
37.7
37.0
37.6
37.5
37.0
37.5
Reference
[26]
[28]
[21]
[21]
[28]
[28]
Table 1 The limits of normal body temperature measured at different sites
Literature search
Abnormalities of temperature
Anatomy and
regulation
physiology
of
temperature
Central temperature-sensitive neurons are found in
the preoptic and anterior hypothalamic nuclei of the
hypothalamus, and peripheral temperature
receptors are found in the skin, abdominal viscera,
spinal cord and great veins of the thorax. Afferent
signals from these receptors are integrated in the
posterior hypothalamus, which tightly regulates
body temperature under normal conditions. When
body temperature increases, heat loss is achieved
by peripheral vasodilation and by sweating. When
body temperature falls, heat production is increased
by shivering and by sympathetic stimulation of
metabolic thermogenesis, and heat is also
conserved by peripheral vasoconstriction. In the
longer term, the production of thyrotropin-releasing
hormone is upregulated by the hypothalamus in
response to cold exposure, increasing metabolic
heat production via release of thyroid-stimulating
hormone and ultimately through the action of
thyroxine. The temperature set-point of the
hypothalamus is increased by exposure to
pyrogens, which include components of the
inflammatory cascade as well as products of
bacterial cell lysis and tissue breakdown. Animal, in
vitro, and some limited human data suggest that the
resulting fever may have adaptive value in the
setting of infection [2], and the widespread use of
antipyretics in this context is increasingly being
questioned. Results of the HEAT trial, a randomised
double-blinded placebo controlled trial of
paracetamol for febrile intensive care patients with
known or suspected infection, are eagerly awaited
[3].
Pubmed was searched for (Thermometers[MeSH
Terms] OR thermography[MeSH Terms]) plus any
of the search terms oral, tympanic, ear, rectal,
oesophageal, esophageal, axilla, chemical dot,
forehead, liquid crystal, bladder, temporal artery or
pulmonary artery, with filters humans and english.
The Cochrane Database of Systematic Reviews
was searched using the terms thermometers and
body temperature
The following textbooks were also used in the
preparation of this article:
McGee’s Evidence-Based Physical Diagnosis [4]
Sapira’s Art and Science of Clinical Diagnosis [5]
Preparation
Clinical examination should ideally take place in a
warm, private environment with good illumination.
The examiner should be appropriately dressed [6]. It
may be appropriate to offer a chaperone to the
patient especially for rectal temperature
measurement. Wash your hands and take
appropriate infection control measures [7]. Introduce
yourself to the patient and confirm their identity.
Obtain informed consent. Check that the patient is
comfortable, then position the patient appropriately
and ensure adequate exposure depending on the
method of measurement you intend to use. Ensure
you have the appropriate equipment to hand,
including a thermometer and any necessary probe
covers.
30
The use of an infrared emission detection aural
thermometer is the most common method for
measuring temperature in the NHS and is usually a
suitable method. See Figures 1-3 for an example of
a typical device. However other methods may be
equally suitable and in some situations infrared
emission detection aural thermometry may not be
suitable. So, the most appropriate thermometer, site
and method should be chosen for each case and
the results interpreted accordingly.
Figure 3 Aural thermometer probe covers
Thermometer Selection
Figure 1 Electronic
thermometer
oral
and
axillary
Figure 2 Infrared emission detection aural
thermometer
Infrared emission detection ‘tympanic’ thermometers
are the most commonly used device in routine
clinical practice in the NHS. They measure the
temperature of the aural canal rather than
specifically that of the tympanic membrane, and
some devices apply correction factors to estimate
“core”, “oral” or “rectal” temperature when these
modes are selected. The estimate of aural
temperature and the effect of these correction
factors depend on the manufacturer’s calibration of
the particular model of thermometer. As a very
general rule, these devices are more likely to
overestimate core temperature in “core” or “rectal”
equivalence modes, and more likely to
underestimate core temperature in “ear/equal” or
“oral” equivalence modes (evidence box 2).
Electronic
contact
thermometers
measure
temperature using a thermistor (a resistor designed
so that resistance varies with temperature to a far
greater extent than it usually would), and may
operate in predictive or monitoring modes.
Monitoring modes continuously display temperature
and may provide an audible tone when the rate of
increase falls below a preset level. In contrast,
predictive modes use the rate of temperature
increase to extrapolate a final temperature, resulting
in a faster but potentially less accurate reading.
Chemical dot thermometers use an array of
coloured dots formulated to change colour at
specific temperature increments. Either an
electronic contact thermometer in a monitoring
mode (properly calibrated) or chemical dot
thermometer is a reasonable choice for routine
clinical practice.
31
Traditional mercury-in-glass thermometers require
at least 3 minutes (and preferably 8-10 minutes) for
an accurate reading, and carry a small but
significant risk of harm both from broken glass and
from the toxicity of mercury [8]. The use of mercury
thermometers for oral, rectal and axillary
temperature measurement has largely been
replaced by electronic contact thermometers and
chemical dot thermometers.
Choice of measurement site
No peripheral temperature measurement site has
been shown to reflect pulmonary artery temperature
with good precision. Aural temperature
measurement is a reasonable choice for the
majority of situations. Aural measurement appears a
little less precise than measurement at the oral site,
but with relatively good repeatability within the same
patient, fewer confounding factors, and a significant
advantage in terms of patient comfort and speed of
measurement, all of which facilitate serial
measurement. For a single temperature
measurement in a fully cooperative patient with no
confounding factors (see evidence box 6), the oral
site may be more reliable. Axillary temperature
measurement is probably less reliable in unselected
patients compared to oral and aural sites and
carries theoretical concerns about underestimation
of core temperature. Rectal thermometry has
greater sensitivity for detection of fever and of
severe hyperthermia than oral or aural thermometry.
These issues are discussed in evidence boxes 1-4,
6 and 7.
form a seal. Depending on the model of
thermometer, press or hold the measurement
button. The end of the measurement period is
usually signalled by an audible tone.
Oral temperature measurement
Use a properly calibrated electronic contact
thermometer in a monitoring mode, or chemical dot
thermometer. You need to be familiar with the
model of thermometer used locally. Use a
disposable probe cover. If you feel it is appropriate,
pass the thermometer to the patient rather than
putting it in the mouth yourself. Ask them to put the
tip of the thermometer under the tongue, close the
mouth and breathe normally through the nose.
Remove the thermometer once it has equilibrated
with oral temperature.
Axillary temperature measurement
Axillary temperature measurement is an option
when other measurement sites are inappropriate,
although the potential to underestimate core
temperature should be borne in mind. Use an
electronic contact thermometer in a monitoring
mode, with disposable probe cover, or a chemical
dot thermometer. Position the thermometer high in
the axilla, with the arm tight against the chest wall,
until the temperature equilibrates.
Method for measurement of temperature
Aural/tympanic temperature measurement
Use an infrared emission detection aural
thermometer in “ear/equal” or “core” mode
depending on the model of thermometer. You need
to be familiar with the model of thermometer used
locally. Ensure that the thermometer has been
calibrated according to manufacturer’s instructions
and that the lens is clean. Use a disposable probe
cover, applied by pressing the probe tip into the
probe cover until it engages. Apply gentle upward
and backward traction to the pinna as for routine
otoscopy (see figure 1). Align the probe tip with the
ear canal, which runs forwards at about a 20° angle
(the direction of the earpieces on a stethoscope),
and insert the probe tip gently but firmly enough to
Figure 1 Direction of gentle traction on pinna
32
Rectal temperature measurement
Rectal temperature measurement is relatively
contraindicated by local pathology (haemorrhoids,
colitis, recent surgery etc.), and may cause an
increase in vagal outflow to the heart, potentially
resulting in bradycardia. Explanation and informed
consent is especially important if rectal temperature
measurement is considered.
A chaperone should be offered and disposable
gloves should be worn. Ask the patient to lie in the
left lateral position with hips and knees flexed.
Underwear should be removed and a towel or sheet
used to preserve patient dignity. Use an electronic
rectal thermometer and apply the probe cover.
Lubricate the tip of the thermometer using waterbased lubricant at room temperature. Holding the
thermometer in your right hand and, if necessary,
using your left hand to part the patient’s buttocks,
gently insert the tip of the probe 2-3 inches into the
rectum. Leave the probe in position until the
temperature equilibrates. Once the temperature
reading is obtained, a tissue should be used to
remove excess lubricant.
Concluding the examination
Most electronic thermometers signal that
equilibration has been achieved with an audible
tone; manufacturer’s instructions should be followed
in the case of chemical dot thermometers.
Equilibration typically takes about one minute but
may be quicker. Read off the temperature and
dispose of the probe cover in an appropriate clinical
waste bin. Most thermometers have a release
button for the probe cover, which should be used if
available, rather than removing the probe cover
manually. Finally, wash your hands, thank the
patient, allow him or her to redress in privacy, and
make a record of the temperature measured.
Acknowledgements:
Thanks to Associate Editor, Dr Sam Lynn for help
and advice.
Conflicts of interest
None declared
33
References
[1] Jopling, H. The principles of clinical examination.
The Journal of Clinical Examination 2006; 1:1-6
[2] Young, PJ et al. Fever and antipyresis in
infection. Med J Aust 2011; 195:458-459
[3]
http://www.anzics.com.au/ctg/currentresearch/255-heat
[4] McGee, S Evidence based physical diagnosis.
Second edition, 2007. Publisher: Saunders.
[5] Orient, JM Sapira’s art and science of bedside
diagnosis. Fourth edition, 2010. Publisher:
Lippincoat
[6] Casey, D, Pottle J and Webster, DP Clothing in
Clinical Examination The Journal of Clinical
Examination 2013; 13: 1-12
[7] Pottle, L and Webster DP Infection Control in
Clinical Examination The Journal of Clinical
Examination 2013; 13: 13-28
[8] Blumenthal, I Should we ban the mercury
thermometer? Discussion paper. J R Soc
Med 1992; 85:553-555
[9] Lawson, L et al. Accuracy and precision of
noninvasive temperature measurement in adult
intensive care patients. Am J Crit Care 2007
16;485-496
[10] Erickson, RS & Kirklin, SK. Comparison of earbased, bladder, oral and axillary methods for core
temperature measurement Crit Care Med 1993;
21:1528-1534
[11] Giuliano, KK et al. Temperature measurement
in critically ill adults, a comparison of tympanic and
oral methods. Am J Crit Care 2000; 9:254-261
[14] Fisk, J & Arcona, S. Comparing tympanic
membrane and pulmonary artery temperatures.
Dimens Crit Care Nurs 2001; 20:44-9
[15] Milewski et al. Comparison of pulmonary artery,
rectal and tympanic membrane temperatures in
adult intensive care unit patients. Clin Paediatrics
1991; 30(4 Suppl):13-16
[16] Stavem, K et al. Accuracy of infrared ear
thermometry in adult patients Intensive Care
Med 1997; 23:100-105
[17] White et al. A comparison of tympanic
thermometer readings to pulmonary artery catheter
core temperature recordings. Appl Nurs Res 1994;
7:165-169
[18] Moran, JL et al. Tympanic temperature
measurements: are they reliable in the critically ill?
A clinical study of measures of agreement. Crit Care
Med 2007; 35:155-164
[19] Hasper, D et al. Tympanic temperature during
therapeutic hypothermia. Emerg Med J 2011;
28:483-485
[20] Farnell, S et al. Temperature measurement:
comparison of non-invasive methods used in adult
critical care. J Clin Nurs 2005; 14:632-9
[21] Chamberlain, JM et al. Determination of normal
ear temperature with an infrared emission detection
thermometer. Ann Emereg Med 1995; 25:15-20
[22] Lefrant, J-Y et al. Temperature measurement in
intensive care patients: comparison of urinary
bladder, oesophageal, rectal, axillary, and inguinal
methods versus pulmonary artery core
method. Intensive Care Med 2003; 29:414-418
[12] Mangat, J et al. A comparison of technologies
used for estimation of body temperature. Physiol
Meas 2010; 31:1105-1108
[23] Fulbrook, P. Core temperature measurement: a
comparison of rectal, axillary and pulmonary artery
blood temperature. Intens and Crit Care Nurs 1993;
9:217-225
[13] Rotello, LC et al. Comparison of infrared ear
thermometer derived and equilibrated rectal
temperatures in estimating pulmonary artery
temperatures. Crit Care Med 1996; 24:1501-1506
[24] Weingart, S et al. Rectal probe temperature lag
during rapid saline induction of hypothermia after
resuscitation
from
cardiac
arrest. Resuscitation 2009; 80:837-838
34
[25] Giuffre, M et al. The relationship between
axillary and core body temperature measurements.
Appl Nurs Res 1990; 3:52-5
[26] Mackowiak, PA et al. A critical appraisal of
98.6°F, the upper limit of the normal body
temperature, and other legacies of Carl Reinhold
August Wunderlich. JAMA 1992; 268:1578-1580
[27] Purssell et al. Tympanic thermometry – normal
temperature and reliability. Paediatr Nurs 2009;
21:40-3
[36] Terndrup, TE & Rajk, J. Impact of operator
technique and device on infrared emission detection
tympanic thermometry. J Emerg Med 1992; 10:683687
[37] Daanen, H. Infrared tympanic temperature and
ear canal morphology. J Med Eng Tech 2006;
30:224-234
[38] Rajee, M & Sultana, RV. NexTemp
thermometer can be used interchangeably with
tympanic or mercury thermometers for emergency
department use. Emerg Med Aus 2006 18:245-251
[28] Darowski, A et al. Normal rectal, auditory canal,
sublingual and axillary temperatures in elderly
afebrile patients in a warm environment. Age and
Ageing 1991; 20:113-119
[39] Doezema, D et al. Cerumen occlusion lowers
infrared tympanic temperature measurement. Acad
Emerg Med 1995; 2:17-19
[29] Tandberg, D & Sklar, D. Effect of tachypnea on
the estimation of body temperature by an oral
thermometer. NEJM 1983; 308:945-946
[40] Varney, SM et al. A comparison of oral,
tympanic, and rectal temperature measurement in
the elderly. J Emerg Med 2002; 22:153-157
[30] Rabinowitz, RP et al. Effects of anatomic site,
oral stimulation, and body position on estimates of
body temperature. Arch Intern Med 1996; 156:777780
[41] Barnett, BJ et al. Oral and tympanic membrane
temperatures are inaccurate to identify fever in
emergency department adults. West J Emerg Med
2011; 12(4):505-511
[31] Newman, BH & Martin, CA. The effect of hot
beverages, cold beverages and chewing gum on
oral temperature. Transfusion 2001; 41:1241-1243
[42] El-Radhi, AS & Patel, S. An evaluation of
tympanic thermometry in a paediatric emergency
department. J Emerg Med 2006; 23:40-1
[32] Terndrup, TE et al. A comparison of oral, rectal,
and tympanic membrane-derived temperature
changes after ingestion of liquids and smoking. Am
J Emerg Med 1989; 7:150-4
[43] Zengeya, ST & Blumenthal, I. Modern
electronic and chemical thermometers used in the
axilla are inaccurate. Eur J Pediatics : 2006;
1551005-8
[33] Quatrara, B. et al. The effect of respiratory rate
and ingestion of hot and cold beverages on the
accuracy of oral temperatures measured by
electronic thermometers. Medsurg Nurs 2007;
16:105-8
[44] Duce, SJ. A systematic review of the literature
to determine optimal methods of temperature
measurement in neonates, infants and children.
DARE Review 1996; 4:1-124
[34] Rampen, AJ et al. Tympanic measurement of
body temperature in stroke patients “turned on its
ear”. J Neurol Neurosurg Psychiatry 2005; 76:10411042
[35] Muth, CM et al. Infrared ear thermometry in
water-related accidents-not a good choice. J Emerg
Med 2010; 38:417-421
[45] Dodd, SR et al. In a systematic review, infrared
ear thermometry for fever diagnosis in children finds
poor sensitivity. J Clin Epidemiol 2006; 59:354-357
[46] Briner, WW. Tympanic membrane vs. Rectal
temperature measurement in marathon runners.
JAMA 1996; 276:194
35
Evidence box 1
Question: What is the accuracy and precision of oral temperature measurement for estimation of pulmonary
artery temperature in adults?
References: Lawson et al. 2007 [9], Erickson & Kirklin 1993 [10], Giuliano et al. 2000 11], Mangat et al. 2010 [12]
Discussion: Studies comparing electronic contact oral thermometry with simultaneous pulmonary artery
temperature measurement in adults have found mean offsets of oral temperature measurements from pulmonary
artery temperature (reflecting accuracy)* of -0.09 °C, 0.05 °C, and 0.04 °C, and standard deviations for these
offsets (reflecting precision) of 0.43 °C, 0.26 °C, and 0.5 °C respectively [9-11]. The factors contributing to this low
precision are unclear. The study populations included both intubated patients and non-intubated patients requiring
intensive care. Erickson & Kirklin [10] note a mean offset from pulmonary artery temperature of 0.12 °C in intubated
patients and -0.29 °C in non-intubated patients.
Mangat et al. [12] report a repeatability coefficient (defined as the value below which 95% of the differences
between two readings by the same operator are expected to lie) of 0.60°C for an electronic contact oral
thermometer in a predictive mode. Their study population included 30 febrile patients and 70 normal volunteers.
Repeatability coefficients were 0.33°C in normal volunteers and 0.98°C in the subgroup of patients with fever.
Key points: On average, oral temperature most likely underestimates pulmonary artery temperature in nonintubated patients. There is significant variability of oral temperature readings relative to pulmonary artery
temperature in these studies, the contributing factors to which are unclear.
*In this review, mean offsets are reported as negative if the test thermometer underestimated pulmonary artery
temperature, positive if the test thermometer overestimated.
36
Evidence box 2
Question: What is the accuracy and precision of infrared emission detection aural temperature measurement for
estimation of pulmonary artery temperature in adults?
References: Lawson et al. 2007 [9], Erickson & Kirklin 1993 [10], Giuliano et al. 2000 [11], Mangat et al. 2010 [12],
Rotello et al. 1996 [13], Fisk et al. [14], Milewski et al. 1991 [15], Stavem et al. 1997 [16], White et al. 1994 [17], Moran et
al. 2007 [18], Hasper et al. 2011 [19], Farnell et al. 2005 [20]
Discussion: Rotello et al. [13] found mean offsets of measured aural temperature from pulmonary artery
temperature of 0.06 °C, 0.07 °C, and -0.22 °C for three models of aural thermometer in 21 intensive care patients.
One of these (mean offset 0.07 °C) was the model used in Chamberlain et al.’s study on the normal range for
aural temperature [21], set in “ear/equal” mode in both studies. Giuliano et al .[11] reported mean offsets of -0.17 °C
and -0.05 °C for two thermometer models using a “core” equivalence mode in a population of 72 patients.
Erickson et al. [10] found a mean offset of 0.07 °C in 38 patients using a “core” equivalence mode. Fisk et al. [14]
found mean offsets between -0.18°C and 0.00°C in 56 patients. Milewski et al. [15] found mean offsets between 0.01 and 0.35°C in a study with nine participants. Several studies have reported aural thermometers
systematically overestimating pulmonary artery temperature: Lawson et al. [9] reported a mean offset of 0.36°C
using a “core” equivalence mode in 60 patients, Stavem et al. [16] 0.45°C in 16 patients using a “rectal” equivalence
mode, and White et al. [17] 0.17-0.36°C in 19 patients. Moran et al. [18] reported a systematic underestimate of
0.36°C in 110 patients using a “core” mode. Hasper et al. [19] have found good accuracy relative to oesophageal
temperature in patients undergoing mild therapeutic hypothermia (mean offset +0.02°C).
A number of studies report the standard deviation for the mean offset from pulmonary artery or oesophageal
temperature at around 0.4-0.5 °C [10, 13, 16, 17, 19]. Giuliano et al .[11] found standard deviations as high as 0.65 °C,
Lawson et al. 0.56 °C [9], Fisk et al. [14] 0.51-0.68°C, Farnell et al. [20] 0.6°C, and White et al. [17] 0.32-0.61°C.
Mangat et al. [12] report tighter limits of agreement relative to nasopharyngeal temperature, corresponding to
standard deviations of 0.26-0.28 °C, while Milewski et al. [15] found standard deviations as low as 0.1-0.23 °C.
Repeatability of aural temperature measurements is relatively good, with the standard deviation for triplicate
readings reported in the range of 0.13-0.2 °C [10, 13], and repeatability coefficients in the range of 0.29-0.44 °C
(0.26-0.49°C in the subgroup of patients with fever) [12].
Key points: As with oral temperature measurement, there is significant variability in aural temperature
measurements relative to pulmonary artery temperature, the contributing factors to which are unclear.
Repeatability of aural measurements within the same patient appears relatively good.
37
Evidence box 3
Question: What is the accuracy and precision of rectal temperature measurement for estimation of pulmonary
artery temperature in adults?
References: Rotello et al. 1996 [13], Milewski et al. 1991
Fulbrook 1993 [23], Weingart et al. 2009 [24]
[14],
Stavem et al. 1997
[16],
Lefrant et al. 2003
[22],
Discussion: The mean offset of rectal temperature from pulmonary artery temperature is reported in the range of
-0.16 - 0.3°C [13, 16, 22, 23]. Milewski et al. report much greater mean offsets of 0.46-1.10°C in their nine participants
[14]. Standard deviation for the offset from pulmonary artery temperature is reported in the range of 0.05-0.5 °C. An
important aspect of rectal temperature measurement is that the rectal temperature may lag significantly as core
temperature changes. This phenomenon of “probe lag” has been elegantly demonstrated by Weingart et al. [24] in
the setting of rapid saline induction of therapeutic hypothermia (the authors recommend oesophageal temperature
monitoring in this setting).
Key points: Rectal temperature may overestimate pulmonary artery temperature. Rectal temperature exhibits
substantial “probe lag” and is unsuitable for monitoring active patient warming or cooling.
Evidence box 4
Question: What is the accuracy and precision of axillary temperature measurement for estimation of pulmonary
artery temperature in adults?
References: Lawson et al. 2007 [9], Erickson & Kirklin 1993
Fulbrook 1993 [23], Giuffre et al. 1990 [25]
[10],
Mangat et al. 2010
[12],
Lefrant et al. 2003
[22],
Discussion: Studies in the intensive care population comparing axillary temperature to simultaneous pulmonary
artery temperature measurement have found mean offsets of axillary from pulmonary artery temperature of -0.23
°C, -0.27 °C, -0.33°C, -0.19°C, and -0.68°C [9, 10, 22, 23, 25]. Reported standard deviations were 0.44°C, 0.45°C,
0.16°C, and 0.57°C [9, 10, 22, 23, 25]; the precision in this setting is similar to other measurement sites.
Mangat et al. report a repeatability coefficient of 0.92°C for axillary temperature measurement using an electronic
axillary thermometer in a predictive mode. Repeatability coefficients were 0.5°C in normal volunteers, and 1.49°C
in the subgroup of patients with fever [12].
Key points: Axillary temperature underestimates pulmonary artery temperature. Some of the variability in axillary
temperature measurements observed in the general population is not reproduced in the intensive care
environment.
38
Evidence box 5
Question: What is the normal range for body temperature measured at oral, aural, rectal and axillary sites?
References: Mackowiack et al. 1992 [26], Chamberlain et al. 1995 [21], Purssell et al. 2009 [27], Darowski et al. 1991
[28]
Discussion: Mackowiack et al. [26] analysed 700 oral temperature measurements obtained from 148 healthy
volunteers aged 18-40 over a 2½ day period. Diurnal variation in temperature was observed. The upper limit of
normal (99th centile) was 37.2 °C at its 6am nadir and 37.7 °C at its 4pm zenith, with lower limits of 36.0°C and
36.3°C respectively.
Chamberlain et al. [21] measured aural temperatures using an infrared emission detection thermometer
(Thermoscan Pro-1® set in “equal” mode, to display the measured temperature in the ear canal) in 2447 subjects
across eight age brackets from 0-2days to >85 years, without potentially febrile illness or use of medication known
to affect body temperature. The upper limit of normal (99th centile) was 37.9 °C for children under 11 years, 37.6
°C for all subjects over 11 years, and 37.5 °C for adults over 85 years. In a longitudinal study of 5 patients, a
pattern of diurnal temperature variation was observed similar to that reported by Mackowiack et al.
Purssell et al. [27] recorded aural temperatures using and infrared emission detection thermometer (no “core” or
other correction factor specified) in 244 children between one and six years old and without suspicion for febrile
illness, in an outpatient setting. Mean temperature was 36.65°C with 95% confidence intervals of 35.8°C -37.5°C.
Darowski et al. [28] measured oral, rectal, ear canal and axillary temperatures in 50 hospital inpatients aged 70
years or over and with no suspicion for febrile illness. Aural temperatures were measured with an electronic
thermocouple probe and with a zero gradient device, rather than with an infrared emission detection thermometer.
Measurements were made between 10am and 6pm. Temperature ranges were lower than those from studies with
a younger sample; 36.2-37.0 °C for oral, 36.4-37.2 °C for ear canal, 36.7-37.5 °C for rectal and 35.5-37.0 °C for
axillary temperatures.
Key points: In the general population, oral or infrared emission detection aural temperature outside the range
36.0-37.7 is abnormal. In the elderly population, oral temperature greater than 37.0 or aural temperature greater
than 37.5 should be suspicious for fever. The upper limit of normal body temperature is lower in the early morning.
39
Evidence box 6
Question: What factors are known to affect the accuracy of temperature measurement at peripheral sites?
References: Darowski et al. 1991 [28], Lawson et al. 2007 [9], Erickson & Kirklin 1993 [10], Mangat et al. 2010
[12],Tandberg & Sklar 1983 [29], Rabinowitz et al. 1996 [30], Newman & Martin 2001 [31], Terndrup et al. 1989 [32],
Quatrara et al. 2007 [33], Rampen et al. 2005 [34], Muth et al. 2010 [35], Terndrup & Rajik 1992 [36], Daanen 2006 [37],
Rajee & Sultana 2006 [38], Doezema et al. 1995 [39]
Discussion: Oral temperatures are a little higher in orally intubated patients and may be misleadingly low in
tachypnoeic patients [9, 10, 29], including in the setting of exertional hyperthermia. Chewing gum for 5 minutes
appears to increase oral temperature by 0.3-0.5°C, an effect which can persist for as long as 30 minutes [30,31]. The
effect of hot and cold drinks on oral temperature is maximal in the first few minutes but may persist for 10-30
minutes [29-32]. Smoking has been found to cause a small delayed increase in oral temperature (mean 0.2°C),
starting at around 10 minutes and persisting until about 30 minutes [30]. The repeatability of oral temperature
measurements worsens significantly in febrile patients [12], although this may in part be due to difficulties in
maintaining nasal breathing during temperature measurement.
While oral temperatures increase for up to 30 minutes following 5 minutes of chewing gum, mean aural
temperatures appear to fall by around 0.3°C on average over this period [30]. Mean aural temperature can be up to
0.4 °C higher in the downward ear when a patient has been lying on one side, but resolves within ten minutes of
upright positioning [34]. Mean aural temperatures may fall by several degrees Celsius following immersion in water
[12], and infrared aural thermometry should not be used in this setting.
The mean aural temperature reading by infrared thermometry is increased by traction on the pinna to straighten
the ear canal [36], and varies with ear canal morphology, in particular the circumference of the ear canal at its distal
bend [37]. Repeat measurements in the same ear appear to have a small bias of 0.1°C towards a lower second
reading [38]. A randomised single-blind trial found occlusion of the ear canal by cerumen to lower the mean
temperature reading by 0.3 °C [39].
Axillary measurements will theoretically underestimate core temperature when the core-peripheral temperature
gradient is high (e.g. vasoconstriction, hypoperfusion). Darowski et al.’s study on the normal range of temperature
at different sites in elderly patients found the lower limit for axillary temperature to be substantially lower than for
other measurement sites, at 35.5 °C vs. 36.2 °C, 36.4 °C and 36.7 °C for oral, aural and rectal temperatures
respectively [28].
Key points: Aural temperature measurement should be preferred over the oral site in patients with tachypnoea,
those who cannot maintain nasal breathing, and following eating, drinking or smoking (although chewing appears
to have some effect on aural temperature). Aural temperature measurement should not be used following
immersion in water, and should not be used in the downward ear when a patient has been lying on one side.
Axillary temperature measurement is unreliable in unselected patients.
40
Evidence box 7
Question: What is the sensitivity of oral, aural and axillary thermometry for fever compared with rectal
thermometry?
References: Varney et al. 2002 [40], Barnett et al. 2011 [41], El-Radhi & Patel 2006
2006 [43], Duce 1996 [44], Dodd et al. 2006 [45], Briner 1996 [46]
[42],
Zengeyha & Blumenthal
Discussion: Varney et al. [40] measured the temperature of 95 patients over the age of 60 and presenting with
symptoms suggestive of infection using oral, aural and rectal thermometry. Allowing for rectal temperature to be
on average 0.5 °C higher than oral temperature, they identified fever >38 °C by rectal but not oral thermometry in
14.7% of patients overall, and by rectal but not aural thermometry in 12.2%. Of those patients tested, 5.6% were
febrile by rectal thermometry but afebrile by both oral and aural thermometry. Of note, no patients were febrile on
oral or aural measurement but afebrile on rectal measurement. Barnett et al. [41] measured oral, aural and rectal
temperatures in 457 adults presenting to the emergency department with a mean age of 65 years old. In this
population, a binary cut-off value of 37.2 °C for oral temperature had a sensitivity of 84.7% and specificity of
83.5% for rectal temperatures >38 °C. Using an aural thermometer with a “rectal” mode manufacturer’s correction
factor, a cut-off value of 37.9 °C had a sensitivity of 80.0% and specificity of 80.8% for rectal temperature >38 °C.
Two recent studies in the paediatric population have reported very poor sensitivity of axillary temperature
measurement for identifying fever on rectal temperature measurement [42, 43]. A 1996 Cochrane review [44]
concluded that in the paediatric population, rectal temperature measurement should be used until the child is old
enough to cooperate with oral temperature measurement. Aural temperature measurement may be a reasonable
alternative provided imperfect sensitivity is acknowledged and the results taken in their clinical context, with one
study demonstrating that temperatures >37.6 °C on infrared aural thermometry has a 76% sensitivity for rectal
temperatures >38 °C [42]. A systematic review in 2006 based on heterogeneous sensitivity estimates found a
pooled sensitivity of 63.7% (with a specificity of 95.2%) [45].
In the specific setting of suspected severe hyperthermia (>41°C), where reliable detection of elevated temperature
is important for prompt therapeutic cooling, rectal temperature measurement should be strongly considered [46].
Key points: In the general population, aural and oral thermometry have a sensitivity in the region of 80-85% for
fever at the rectal site. In the paediatric population, aural thermometry has a sensitivity in the region of 60-75% for
fever at the rectal site.
41