The World Health Organization defines normal core body temperature as roughly 36.5–37.5 °C and classifies fever in adults as a temperature at or above 38 °C, with variations depending on the measurement site and the clinical context. These thresholds anchor a broader set of WHO guidelines on temperature monitoring that span neonatal thermal care, fever management in children through the Integrated Management of Childhood Illness (IMCI) framework, and vital-signs surveillance during outbreaks. Yet applying a single number to every patient, every measurement device, and every clinical setting is harder than it sounds, and the gaps between guideline ideals and real-world practice are surprisingly wide.
What the WHO Framework Actually Covers
WHO guidance on temperature monitoring is not a single document but a family of recommendations woven into disease-specific and population-specific guidelines. For neonates, the WHO thermal protection guidelines classify body temperature into four zones: normal (36.5–37.5 °C), mild hypothermia or “cold stress” (36.0–36.4 °C), moderate hypothermia (32.0–35.9 °C), and severe hypothermia (below 32 °C). These thresholds drive immediate clinical actions in delivery rooms and neonatal units worldwide. For children under five, the IMCI guidelines use fever as a branching point for deciding whether to administer antibiotics, test for malaria, or refer to a hospital. An evaluation of the IMCI fever module in Bangladesh found that if the guidelines had been followed, about 78% of children with confirmed bacterial infections would have received antibiotics, though coverage was uneven across infection types, with nearly all pneumonia and meningitis cases captured but only about half of bacteremia cases.1Bulletin of the World Health Organization. Diagnosis and management of febrile children using the WHO/UNICEF guidelines for IMCI in Dhaka, Bangladesh
For adults, WHO guidance folds temperature monitoring into broader vital-signs protocols for conditions like sepsis, surgical care, and infectious disease outbreaks. The 38 °C fever threshold is commonly used as a screening cutoff, but WHO materials consistently emphasize that no single reading should be interpreted in isolation. Context matters: the patient’s baseline temperature, the measurement site, the device used, and the time of day all influence what any given number means.
Where You Measure Matters More Than You Think
One of the most practical questions in temperature monitoring is which body site to use. Core temperature, measured via the pulmonary artery, esophagus, or bladder, is the gold standard but is obviously invasive and limited to intensive care. Everything else is a peripheral estimate of that core value, and the estimates vary. A systematic review and meta-analysis of non-invasive methods in ICU patients found that axillary, tympanic infrared, and zero-heat-flux thermometers all tended to underestimate true intravascular temperature, with only esophageal measurements showing clinically acceptable accuracy.2PubMed Central. Accuracy of non-invasive body temperature measurement methods in adult patients admitted to the intensive care unit: a systematic review and meta-analysis
In emergency departments, where rectal temperature serves as the practical reference, a study of nearly 1,000 patients found that tympanic membrane readings were the most precise among non-invasive options, with a standard deviation of just 0.4 °C from rectal values, compared with 0.6 °C for both oral and temporal artery readings. When the goal was detecting a fever of 38 °C or higher, tympanic and temporal artery thermometers caught about two-thirds to 70% of true fevers, while oral thermometers caught only about a third. Lowering the tympanic cutoff to 37.5 °C pushed sensitivity above 90% while keeping specificity around 90%.3PubMed. Temperature measurement in the adult emergency department: oral, tympanic membrane and temporal artery temperatures versus rectal temperature That tradeoff, lowering the threshold to catch more fevers at the cost of more false positives, is exactly the kind of calibration clinicians face daily.
In intensive care settings, tympanic thermometry has also performed well. A study in ICU patients found that both left and right tympanic readings showed very good agreement with the standard method, achieving 83% sensitivity and 100% specificity, while forehead readings were the least accurate of the peripheral methods tested.4PubMed Central. Accuracy and precision of four common peripheral temperature measurement methods in intensive care patients
The Problem With Non-Contact Infrared Thermometers
Non-contact infrared thermometers became ubiquitous during the COVID-19 pandemic, used at building entrances, airports, and clinics worldwide. WHO guidance acknowledged their role in outbreak screening but cautioned about their limitations. The evidence backs that caution up strongly. A clinical evaluation of several commercially available models found that depending on the model, between 48% and 88% of individual temperature readings fell outside the manufacturer’s stated accuracy. Sensitivity for detecting a temperature above 38 °C ranged from essentially zero to 0.69, meaning some devices missed the majority of febrile individuals.5PubMed Central. Clinical evaluation of non-contact infrared thermometers
Part of the problem is environmental. A study among healthy adults found that raising the ambient temperature from 18 °C to 28 °C significantly increased forehead, neck, and temple readings, while wrist readings stayed relatively stable. Most neck readings (about 95%) and forehead readings (about 75%) fell within normal range classifications, whereas wrist and temple readings were frequently classified as below normal. Agreement in temperature classification between the two ambient conditions was only fair for the neck and lower for other sites.6PubMed Central. Cross-sectional analysis of site- and ambient temperature–related variability in noncontact infrared thermometry among healthy Black African adults In other words, the reading you get depends heavily on how warm or cool the room is and where on the body you aim the device. For outbreak screening, where the goal is catching as many febrile individuals as possible, this level of unreliability is a real problem.
Older Adults and the Blunted Fever Problem
Standard fever thresholds can be dangerously misleading in older adults. WHO and other clinical guidelines generally acknowledge this, but the degree to which fever responses are blunted with age is underappreciated. A systematic review of human and animal studies confirmed that older adults are less able to mount an effective febrile response than younger people, with age-related changes in immune, vascular, and metabolic activity all contributing to the dampened signal.7PubMed Central. Altered Febrile Responses in Older Adults: A Systematic Review
The clinical consequences are concrete. A review of infections in elderly patients estimated that fever may be absent or blunted in roughly 20% to 30% of cases, potentially delaying diagnosis in a population already at elevated risk of serious complications from infection.8PubMed. Fever in the elderly A study in nursing home residents showed the same pattern in practice: the mean baseline oral temperature among randomly selected residents was about 97.4 °F (36.3 °C), already below the textbook 98.6 °F. During documented infections, nearly half of fever episodes were “blunted,” never reaching 101 °F (38.3 °C). About a quarter of those blunted readings actually reflected an adequate rise from the individual’s low baseline but simply never crossed the standard threshold.9PubMed. Fever response in elderly nursing home residents: are the older truly colder?
This is why WHO-aligned clinical guidance increasingly encourages tracking change from an individual’s baseline rather than relying on a fixed cutoff. For an older person whose usual temperature sits around 36.0 °C, a reading of 37.5 °C could signal a serious infection even though it falls below the conventional fever line.
Neonatal Thermal Care
Temperature monitoring in newborns is one area where WHO guidelines are especially prescriptive, because thermal instability kills. Hypothermia in low-birth-weight and preterm infants remains a leading contributor to neonatal morbidity in low- and middle-income countries, and WHO recommends immediate skin-to-skin contact, delayed bathing, and continuous temperature surveillance in the hours after birth. Kangaroo mother care, which WHO promotes as a cornerstone intervention, has been shown to keep neonatal temperatures relatively stable and improve physiological indices including heart rate, respiratory rate, and oxygen saturation in low-birth-weight babies.10PubMed Central. Kangaroo Method Care: Benefits, Proponents and Barriers: A Narrative Review
The WHO neonatal thermal classification, with its four zones from normal down to severe hypothermia, is designed to trigger specific interventions at each stage: additional warming measures for cold stress, active rewarming and monitoring for moderate hypothermia, and urgent resuscitation-level care for severe hypothermia below 32 °C. Axillary measurement is the standard for neonates in most low-resource settings because it is non-invasive and requires only a basic digital thermometer.
Circadian Rhythm and What It Means for Readings
Body temperature is not static over the course of a day, and this natural fluctuation has real implications for how temperature readings should be interpreted. The circadian system drives a predictable pattern: temperature rises during the day, peaks in the early evening, and falls overnight to a minimum around the time of waking.11PubMed. The circadian rhythm of human body temperature – Clinical implications and review of the literature In healthy people, the difference between the daily low and high can range from about 0.3 °C to over 1 °C.12PubMed Central. Circadian rhythmicity of body temperature and metabolism
This matters because a temperature taken at 6 a.m. and one taken at 6 p.m. on the same person can look meaningfully different even when nothing is wrong. It also means that illness can disrupt the rhythm itself, not just raise the overall level. A study of hospitalized patients found that the circadian temperature swing was dramatically flattened, with only a 0.2 °F difference from trough to peak compared to 0.5–1.9 °F in healthy volunteers. Fewer than 20% of hospitalized patients showed a detectable daily rhythm at all, and among those who did, it was present only on a minority of hospital days. The timing of the trough also shifted: in healthy people it typically occurs around 6 a.m., but in elderly hospitalized patients it drifted as late as noon.13PubMed. Disruption of the Body Temperature Circadian Rhythm in Hospitalized Patients
The disruption of the circadian temperature rhythm appears to carry diagnostic and prognostic value in conditions like infections, autoimmune diseases, and cancer, though research in this area is still developing.11PubMed. The circadian rhythm of human body temperature – Clinical implications and review of the literature
Temperature Trajectories in Sepsis
Sepsis is where temperature monitoring becomes genuinely life-or-death, and where single-point readings are least useful. WHO sepsis guidance emphasizes frequent vital-signs assessment, but emerging research shows that the pattern of temperature over time carries far more information than any isolated reading. A study of over 1,800 septic patients with hypothermia identified three distinct 48-hour temperature trajectory classes. Compared with the intermediate pattern, the trajectory that stayed persistently low was associated with nearly double the 7-day mortality risk, while the trajectory that warmed most aggressively was linked to better 28-day survival.14PubMed Central. Temperature trajectory in septic patients with hypothermia: When the ice begins to melt—A retrospective cohort study
A separate analysis that modeled temperature as a time-varying variable in sepsis patients identified five distinct trajectory classes and found a U-shaped relationship between temperature and mortality: the lowest risk clustered near the upper range of normal, while both hypothermia and high fever were associated with worse outcomes. The prognostic power of temperature patterns peaked around four days after ICU admission. Dynamic prediction models that incorporated temperature trajectories outperformed those relying on a single measurement.15PubMed Central. Body temperature as a dynamic host-response biomarker in sepsis: trajectories and time-varying mortality risk The practical takeaway is that in critically ill patients, trend monitoring is essential. A normal temperature at one moment does not rule out danger if the trend is heading in the wrong direction.
Perioperative Monitoring Gaps
Inadvertent hypothermia during surgery is a well-known risk: it increases bleeding, wound infections, and cardiac events. WHO surgical safety guidelines and national equivalents recommend continuous temperature monitoring during operations and active warming when needed. The reality falls short. A period prevalence study across five hospitals found that perioperative temperature monitoring from the preoperative baseline through to discharge from the post-anesthesia care unit was “sparse and intermittent.” Over half of patients had two or fewer temperature readings recorded during their entire perioperative care, and a third had no temperature data at all before arriving in recovery. Even among patients who received active warming during surgery, more than two-thirds had no temperature monitoring recorded during the procedure.16International Journal of Nursing Studies. Perioperative temperature monitoring for patient safety: A period prevalence study of five hospitals
This is a striking gap between what guidelines say and what actually happens in operating rooms. If you are warming a patient but not measuring their temperature, you have no way to know whether the intervention is working or whether the patient has crossed into hypothermia.
Continuous Monitoring and Wearable Devices
Traditional spot-check thermometry, where a nurse takes a reading every few hours, misses a lot. Wearable sensors that measure temperature continuously are closing that gap. A clinical evaluation of a wearable temperature-monitoring patch found that it measured body temperature with a mean error of about 0.29 °F compared to clinic-assessed oral readings. Using a confidence threshold, the patch detected clinic-recorded fevers with 90% sensitivity and 88% specificity, outperforming patient self-assessed oral readings taken around the same time, which achieved only 62% sensitivity. The patch detected fevers a median of 4.3 hours before they were caught by routine clinical measurements, with some fevers flagged more than 18 hours early.17PubMed Central. A Novel Wearable Device for Continuous Temperature Monitoring & Fever Detection
That kind of lead time could change clinical decision-making, especially for patients at high risk of sepsis or neutropenic fever. Continuous monitoring also makes it possible to track the circadian and trajectory patterns discussed earlier, giving clinicians a richer picture than any series of spot checks.
Monitoring in Low-Resource Settings
WHO guidelines are written to be implementable globally, but temperature monitoring in low-resource settings faces practical barriers that rarely appear in guideline documents. Equipment shortages, unreliable power, and limited nursing staff all compromise how often and how accurately vital signs are checked. A study in Malawi evaluating a continuous pediatric monitoring system found that healthcare providers reported the monitors improved care by measuring multiple vital signs reliably and by providing backup power lasting about four hours along with alarm functions. After the monitors were introduced, staff actually spent more total time with patient monitoring because the system flagged problems that might otherwise have gone unnoticed.18PubMed Central. Implementing the IMPALA continuous monitoring system for paediatric critical care in Malawi: A mixed methods study of barriers and facilitators
Low-cost device development is an active area of research. Efforts to build non-invasive vital-signs monitoring devices specifically for pregnant women in low-resource settings aim to address the high paucity of such equipment in facilities where maternal mortality is greatest.19PubMed Central. Low cost, non-invasive, and continuous vital signs monitoring device for pregnant women in low resource settings (Lvital device) The challenge is not just building a cheap thermometer. It is building one that stays calibrated, withstands heat and humidity, runs on minimal power, and integrates into clinical workflows that are already stretched thin.
Device Hygiene and Calibration
Even when the right device is available, how it is maintained affects accuracy and safety. WHO infection prevention guidelines recommend disinfecting shared patient-care devices between uses, and thermometers are no exception. A study among physicians in Ethiopia found that only about 20% reported disinfecting non-infrared thermometers after every use, despite 87% agreeing that disinfection after each use was necessary. About 40% of physicians reported never having disinfected any thermometer at all.20PubMed Central. Disinfection of Stethoscope and Non-Infrared Thermometer: Practices of Physicians in Ethiopia in the Era of COVID-19
Calibration is the other often-overlooked piece. Clinical thermometers, like all measuring instruments, drift over time. International standards such as ISO/IEC 17025 provide a framework for ensuring that calibration results are traceable and reliable, and the principle applies as much to a ward thermometer as to a laboratory analyzer.21International Journal of Research and Innovation in Applied Science. ISO/IEC 17025-Compliant Calibration of Biomedical Equipment: Frameworks, Methods, and Measurement Uncertainty Assessment A thermometer that is off by half a degree is clinically meaningless in some contexts and potentially dangerous in others, particularly in neonates or when tracking the subtle temperature changes that precede sepsis.
Medications That Shift the Thermostat
A factor that rarely gets mentioned in temperature monitoring guidelines but affects real patients daily is the influence of medications on body temperature. A systematic review and meta-analysis found that during heat stress, drugs with high anticholinergic properties raised core temperature by an average of about 0.4 °C at ambient temperatures above 30 °C, primarily by suppressing sweating. Non-selective beta-blockers, adrenaline, and anti-Parkinson’s agents also elevated core temperature, though by smaller margins. Antidepressants, diuretics, and drugs with weak anticholinergic effects did not significantly alter core temperature responses.22PubMed Central. The effect of prescription and over-the-counter medications on core temperature in adults during heat stress: a systematic review and meta-analysis
This has practical implications for monitoring. A patient on high-dose anticholinergic medication during a heat wave may show an elevated temperature that reflects drug-induced impairment of sweating rather than infection. Conversely, some medications that suppress the immune response can blunt fever, creating the same false-reassurance problem seen in older adults. Clinicians interpreting temperature readings need to know what their patient is taking.
Targeted Temperature Management After Cardiac Arrest
Temperature monitoring becomes a therapeutic tool in its own right when clinicians deliberately control a patient’s body temperature after cardiac arrest. For years, guidelines recommended cooling comatose survivors to 33 °C. That practice was upended by large trials showing that outcomes at 33 °C were no better than at 36 °C, and a subsequent trial found that strict normothermia (actively preventing fever without inducing hypothermia) performed just as well. Based on that evidence, the International Liaison Committee on Resuscitation and the American Heart Association updated their recommendations, moving away from deep cooling and toward a broader emphasis on preventing hyperthermia.23Circulation. Temperature Management for Comatose Adult Survivors of Cardiac Arrest: A Science Advisory From the American Heart Association
Precise, continuous temperature monitoring is essential for this approach. Drifting even a degree or two above target can worsen neurological outcomes, and the monitoring hardware, typically an esophageal or bladder probe connected to a feedback-controlled cooling system, is far more demanding than the equipment used on a general ward. It is also a vivid illustration of how the purpose of temperature monitoring shifts with clinical context: the same measurement that serves as a simple screening tool at the hospital entrance becomes a tightly regulated therapeutic variable in the ICU.