Circumstances in which a patient's temperature requires careful monitoring include the following:

Oral temperature measurement continues to be used in practice as a quick and easy method of obtaining an estimate of the core body temperature (Bijur et al. [19]). When taking an oral temperature, the thermometer is placed in the posterior sublingual pocket of tissue at the base of the tongue (Sund‐Levander and Grodzinsky [192]). The temperature difference between the anterior and posterior sublingual regions can be up to 1.6°C in febrile patients (Sund‐Levander and Grodzinsky [192]). This area is in close proximity to the thermoreceptors, which respond rapidly to changes in the core temperature, hence changes in core temperature are reflected quickly here (Sund‐Levander and Grodzinsky [192]). Oral temperatures are thought to be affected by ingested foods and fluids, smoking, the muscular activity of chewing, and rapid breathing (Sund‐Levander and Grodzinsky [192]).

Rectal temperatures have been demonstrated in clinical trials to be more accurate than oral or axillary; however, due to its invasive nature, assessing temperature via this route is not advocated, and aspects relating to privacy, dignity and patient choice need to be considered (Bijur et al. [19], Hernandez and Upadhye [79]).

The rectal temperature is often higher than the peripheral one and offers greater precision in terms of obtaining the core temperature as it is more sheltered from the external environment (Sund‐Levander and Grodzinsky [192]); however, false readings can occur in the presence of stool (Tait et al. [194]). In adults, the rectal thermometer should be inserted at least 4 cm into the rectum for an accurate reading (Adam et al. [2]); however, use of a rectal thermometer is contraindicated in immunosuppressed patients and in children under 5 years old as it carries a risk of rectal ulceration or perforation (NICE [138]).

The axilla is considered less desirable than the other sites because of the difficulty in achieving accurate and reliable readings (Asadian et al. [8]) as it is not close to major vessels, and skin surface temperatures vary more with changes in the temperature of the environment (Marui et al. [113]). However, axillary temperature measurement, using an electronic or chemical dot thermometer, is recommended in children aged 4 weeks to 5 years to detect fevers (NICE [138]). In the clinical setting, it is mainly used in neonatal or paediatric units, where ambient temperatures are stable and patients are unsuitable for, or cannot tolerate, rectal thermometers (El‐Radhi [59], NICE [138]).

To take an axillary temperature reading, the thermometer should be placed in the centre of the armpit, with the patient's arm firmly against the side of the chest (Barry et al. [13]). It is important that the same arm is used for each measurement, as there is often variation in temperature between left and right (Barry et al. [13]).

Measuring temperature using the tympanic membrane has become increasingly popular as it is less invasive than other methods, reduces the risk of cross‐contamination and provides rapid results (<1 minute) (El‐Radhi [59]). A tympanic thermometer uses infra‐red light to detect thermal radiation; it is designed for intermittent use and provides a digital reading (Barry et al. [13]). It has been suggested by some that tympanic membrane thermometers give an accurate representation of actual body temperature (Kiekkas et al. [95]). This is because the tympanic membrane lies close to the temperature regulation centre in the hypothalamus, shares the same artery and is therefore considered to directly reflect core temperature in adults (Basak et al. [14]).

However, some limitations to this method of temperature measurement have been identified. In 2003, the Medicines and Healthcare products Regulatory Agency (MHRA) published a medical device alert following reports of tympanic thermometers providing low temperature readings, which were attributed to dirty probes and probe covers and user error (MHRA [116]). A soiled probe or probe cover would record a low temperature because the infra‐red emissions from the tympanic membrane would be affected (Basak et al. [14]). In addition, this method is not suitable for patients who have undergone ear surgery or who use a hearing aid. Occlusion and otitis media resulting from ear wax can also lead to inaccurate values (Basak et al. [14]).

The accuracy of these devices is still subject to much debate within the literature. For example, their use is not yet recommended by the National Institute for Health and Care Excellence (NICE) for measuring temperature perioperatively as they do not provide a direct estimate of core body temperature (devices used in this setting should be accurate to within 0.5°C) (NICE [133]). However, their use is recommended in children aged 4 weeks to 5 years when screening for fever (NICE [138]).

Temperature from the temporal region can be measured using an infra‐red temporal artery thermometer (El‐Radhi [59]). However, readings can be inaccurate due to the presence of perspiration, make‐up and hair (Geijer et al. [70]). Thickness of tissue and bone, local blood flow, device placement and ambient temperature can also affect accuracy (Sund‐Levander and Grodzinsky [192]). This is especially the case during rapid changes in body temperature, for example during episodes of pyrexia (Sund‐Levander and Grodzinsky [192]).

Use of non‐touch infra‐red devices on the temporal artery site produces an indirect estimate of the true temperature using a correction factor to account for the difference in temperature between the outer surface of the head and the arterial blood flow (NICE [133]). Temporal artery thermometers produce rapid readings and do not require any skin contact (Basak et al. [14], Gasim et al. [69]), which reduces the risk of infection and discomfort. They are also cost‐effective and easy to use (Basak et al. [14]). Measurements from this area are considered accurate because it shares the same blood flow as the hypothalamus, originating from the carotid artery, and is therefore considered to reflect the body's core temperature (Gasim et al. [69]). However, studies disagree on the precision and accuracy of this temperature measurement method: some say that the existing evidence does not support its use and that it cannot replace ordinary invasive and non‐invasive methods in adult patients (Geijer et al. [70], Kiekkas et al. [95]), while others support the use of these devices as an alternative to invasive temperature measurement (Barry et al. [13], Reynolds [175], Smith et al. [188]) but on paediatric patients only (Morgensen et al. [120]).