|Home | About | Journals | Submit | Contact Us | Français|
Most clinically available thermometers accurately report the temperature of whatever tissue is being measured. The difficulty is that no reliably core-temperature measuring sites are completely non-invasive and easy to use — especially in patients not having general anesthesia. Nonetheless, temperature can be reliably measured in most patients. Body temperature should be measured in patients having general anesthesia exceeding 30 minutes in duration, and in patients having major operations under neuraxial anesthesia.
Core body temperature is normally tightly regulated. All general anesthetics produce a profound dose-dependent reduction in the core temperature triggering cold defenses including arterio-venous shunt vasoconstriction and shivering. Anesthetic-induced impairment of normal thermoregulatory control, and the resulting core-to-peripheral redistribution of body heat, is the primary cause of hypothermia in most patients. Neuraxial anesthesia also impairs thermoregulatory control, although to a lesser extant than general anesthesia. Prolonged epidural analgesia is associated with hyperthermia whose cause remains unknown.
In previous articles, I've reviewed heat balance in surgical patients,1 complications associated with perioperative thermal perturbations,2 and the etiology and treatments of postoperative shivering.3 Heier and Caldwell have reviewed the effects of hypothermia on the response to neuromuscular blocking drugs.4 Furthermore, an entire book is devoted to the emerging field of therapeutic hypothermia.5 In this article, I will belatedly review temperature monitoring and the effects of general and regional anesthesia on thermoregulatory control.
Surgery typically involves exposure to a cold environment, administration of unwarmed intravenous fluids, and evaporation from within surgical incisions. However, these factors alone would not usually cause hypothermia; instead, thermoregulatory defenses would normally maintain core temperature in the face of comparable environmental stress. That hypothermia is typical in unwarmed surgical patients reflects a failure of effective thermoregulatory defenses. Understanding the effects of anesthetics on normal thermoregulatory control is thus the key to perioperative thermal perturbations because ineffective thermoregulation — much more than cold exposure — underlies most temperature changes observed in surgical patients.
I will first briefly review temperature monitoring and normal thermoregulation, and then discuss the effects of general and neuraxial anesthesia on temperature control.
Body temperature is not homogeneous: deep thoracic, abdominal, and central nervous system (i.e., core) temperatures usually range from 2 to 4°C cooler than the arms and legs — and much of the skin surface is cooler yet. Unlike core temperature, which is tightly regulated, skin temperature varies markedly as a function of environmental exposure; temperature of peripheral tissues (mostly the arms and legs) depends on current exposure, exposure history, core temperature, and thermoregulatory vasomotion. Core temperature, while by no means completely characterizing body heat content and distribution, is the best single indicator of thermal status in humans.
Core temperature monitoring (e.g., tympanic membrane, pulmonary artery, distal esophagus, and nasopharynx) is used to monitor intraoperative hypothermia, prevent overheating, and facilitate detection of malignant hyperthermia. Because these sites are not necessarily available or convenient, a variety of “near-core” sites are also used clinically. These include the mouth, axilla, bladder, rectum, and skin surface. Each has distinct limitations but can be used clinically in appropriate circumstances.
What level of accuracy is clinically necessary has yet to be established. But a good rule-of-thumb, one that has been used in many studies, is that the combined inaccuracy of a site/thermometer combination should not exceed 0.5°C. One basis for this choice is that it is the smallest difference that has been shown to be associated with hypothermia-induced complications.6
Muscle or skin-surface temperatures may be used to evaluate vasomotion7 and assure validity of peripheral neuromuscular monitoring.4 Muscle temperatures are also used to determine peripheral compartment temperatures and regional distribution of body heat.8-10 Both core and mean skin-surface temperature measurements are required to determine the thermoregulatory effects of different anesthetic drugs11 and estimate mean-body temperature.12
Mercury-in-glass thermometers are slow and cumbersome, and spilt mercury is a biohazard; they have thus all but disappeared from clinical use — although they remain useful for laboratory calibration of other systems. The most common electronic thermometers are thermistors and thermocouples. Thermistors are temperature-sensitive semi-conductors, whereas thermocouples depend on the tiny current generated when dissimilar metals are joined. Both devices are sufficiently accurate for clinical use and inexpensive enough to be disposable. However, the signals from each are inherently non-linear and thus need to be linearized by calibrated compensating units.
Infrared sensors are another type of thermometer that has become popular in the last decade. They work by evaluating infrared energy that is emitted by all surfaces above absolute zero degrees. They can consequently be used without actually touching the surface in question (which is useful for measuring the temperature of molten lava or metals, for example). These thermometers are accurate and relatively inexpensive. Clinical models can measure temperature of the skin surface to within a tenth of a degree or so. When infrared signals are actually obtained from the tympanic membrane, the result is core temperature.13,14 However, nearly all available systems are intentionally too large to even fit more than a few mm into the aural canal and do not “see” anywhere near the tympanic membrane. As normally used, that is directed into the aural canal15 or near the temporal artery,16 infrared systems are insufficiently accurate for clinical use (fig. 1). In light of their poor performance, it seems unfortunate that they have become so popular.
An interesting method of measuring core temperature from the surface of the skin is to use a system originally proposed by Fox17,18 and refined by Togawa.19 The technique is to combine a heater with a thermal flux transducer (which is, effectively, two thermometers separated by a known thermal insulator). The heater is then servo-controlled until flux is zero. At this point, heat and skin temperature are, by definition, equal since there would otherwise be a flow of heat. However, the same logic suggests that there is no flow of heat from skin to deeper tissues; otherwise, heat would accumulate, which would violate the Second Law of Thermodynamics. This logic is not quite accurate since it ignores blood-borne lateral convection of heat. But in practice, these thermometers accurately determine the temperature of tissues to about a centimeter below the skin surface. In many parts of the body, notably the chest and forehead, a centimeter is sufficient to approximate core temperature (fig. 2).20 Unfortunately, these otherwise excellent monitors are not currently available in Europe or the United States.
Core temperature monitoring is appropriate during most general anesthetics both to facilitate detection of malignant hyperthermia and to quantify hyperthermia and hypothermia. Malignant hyperthermia is best detected by tachycardia and an increase in end-tidal PCO2 out of proportion to minute ventilation.21 Although increasing core temperature is not the first sign of acute malignant hyperthermia, it certainly helps confirm the diagnosis. More common than malignant hyperthermia is intraoperative hyperthermia having other etiologies including excessive warming, infectious fever, blood in the fourth cerebral ventricle, and mismatched blood transfusions. Because hyperthermia has so many serious etiologies, any perioperative hyperthermia requires diagnostic attention.
By far the most common perioperative thermal disturbance is inadvertent hypothermia. Prospective, randomized trials have shown that even mild hypothermia causes numerous adverse outcomes in a variety of patient populations. Hypothermia-induced complications include morbid myocardial outcomes22 secondary to sympathetic nervous system activation,23 surgical wound infection,24,25 coagulopathy6,26-33 increased allogeneic transfusions,6,24,26,27,31,33-37 negative nitrogen balance,38 delayed wound healing,24 delayed post-anesthetic recovery,39 prolonged hospitalization,24 shivering,40 and patient discomfort.41
The major cause of hypothermia in most patients given general anesthesia is an internal core-to-peripheral redistribution of body heat that usually reduces core temperature by 0.5 to 1.5°C in the first 30 minutes following induction of anesthesia. Hypothermia results from internal redistribution of heat and a variety of other factors whose importance in individual patients is hard to predict.9 Core temperature perturbations during the first 30 minutes of anesthesia thus are difficult to interpret and measurements not usually required. Body temperature should, however, be monitored in most patients undergoing general anesthesia exceeding 30 minutes in duration and in all patients whose surgery lasts longer than one hour. Measuring body temperature (and maintaining normothermia) is now essentially the standard-of-care during prolonged general anesthesia, especially for large operations where the risk of hypothermia is substantial.
Hypothermia, resulting largely from core-to-peripheral redistribution of body heat,8 is as common during epidural and spinal anesthesia as it is during general anesthesia, and can be nearly as severe.42 Because neuraxial anesthesia impairs behavioral thermoregulatory responses (i.e., patient sensation of cold),43 patients and physicians are both frequently unaware that hypothermia has developed (fig. 3).42 Core temperature should therefore be measured during regional anesthesia in patients likely to become hypothermic, including those undergoing body cavity surgery — although temperature monitoring during neuraxial anesthesia remains relatively uncommon.44,45
The core thermal compartment is composed of highly perfused tissues whose temperature is uniform and high compared with the rest of the body. Temperature in this compartment can be evaluated in the pulmonary artery, distal esophagus, tympanic membrane, or nasopharynx.46,47 Even during rapid thermal perturbations (e.g., cardiopulmonary bypass), these temperature-monitoring sites remain reliable — although there may be transient real differences among them.
Temperature probes incorporated into esophageal stethoscopes must be positioned at the point of maximal heart sounds, or even more distally, to provide accurate readings.48 Modern tympanic thermocouples are soft and pliable. There is thus little if any risk of perforating the membrane, although it is possible to push a bolus of wax onto the tympanic membrane. Inserting tympanic probes is somewhat more difficult than it sounds, especially in conscious subjects, because the aural canal is several cm long and is not straight. The difficulty is that subjects and people inserting the probes often mistake the bend in the canal for the tympanic membrane and thus do not position the probes on the membrane itself. Once properly positioned, it is helpful to occlude the aural canal with wool to prevent air currents from cooling the thermocouple. Nasopharyngeal probes should be inserted at least a few cm past the nares to obtain core temperature; nasopharyngeal temperature are probably only accurate in patients who are not breathing through their nostrils.
Core temperature can be estimated with reasonable accuracy using oral, axillary, and bladder temperatures except during extreme thermal perturbations.46,47 Each of these sites is subject to artifact so clinicians should use reasonable judgment in selecting a monitoring site (and type of thermometer) for a given patient. For example, oral temperatures can be inaccurate in patients who breathe through their mouths or have recently ingested hot or cold liquids. Axillary temperatures are reasonably accurate,49 but work best when the probe is positioned over the axillary artery and the arm is kept at the patient's side. Differences in technique may explain reported differences in accuracy.50
Skin-surface temperatures are considerably lower than core temperature51; forehead skin temperature, for example, is typically 2°C cooler than core. Perhaps surprisingly, even the intense vasodilation associated with sweating and the intense vasoconstriction associated with shivering only slightly alter the core-to-forehead temperature gradient (fig. 4).52 Skin temperature is determined by the balance of heat provided by subcutaneous tissues and heat lost to the environment. Dissipation of heat from the skin surface, mostly by radiation and convection, depends on ambient temperature. While each type of heat loss is controlled by different equations, most of which are highly non-linear, cutaneous heat loss is approximately linear over small ranges of ambient temperature. The 1−2°C ambient temperature differences usually observed during surgery thus have little effect on the core-to-forehead temperature gradient (fig. 5).52 Forehead skin temperature is thus a surprisingly accurate measure of core temperature so long as a +2°C compensation is included.
A special case of skin-temperature monitoring is temporal artery thermometers. These are infrared skin-surface thermometers that record skin temperature at approximately 10 Hz and detect the highest temperature as the device is scanned across the forehead, including the region of the temporal artery. The theory is that the blood in the temporal artery is near core temperature and, therefore, that supervening skin temperature will also approximate core temperature. While the theory is attractive, the devices are much too inaccurate for clinical use.16,53
A distinct limitation of skin temperatures is that they fail to reliably confirm the clinical signs of malignant hyperthermia (tachycardia and hypercarbia) in swine (fig. 6)54 and have not been evaluated for this purpose in humans. Rectal temperature also normally correlates well with core temperature,46,47 but fails to increase appropriately during malignant hyperthermia crises54 and under other documented situations including heat stroke.55,56 Consequently, rectal and skin-surface temperatures must be used with considerable caution.
The four core temperature monitoring sites (e.g., tympanic membrane, nasopharynx, pulmonary artery, and esophagus) remain useful even during cardiopulmonary bypass. In contrast, rectal temperatures lag behind those measured in core sites. Consequently, rectal temperature is considered an “intermediate” temperature in deliberately cooled patients. During cardiac surgery, bladder temperature is equal to rectal temperature (and therefore intermediate) when urine flow is low, but equal to pulmonary artery temperature (and thus core) when flow is high.57 Because bladder temperature is strongly influenced by urine flow, it may be difficult to interpret in these patients. The adequacy of rewarming is best evaluated by considering both “core” and “intermediate” temperatures.
Mean-skin temperature is the area-weighted average temperature of the skin surface. Mean-skin temperature, while less important than core temperature, is nonetheless important for at least three reasons: 1) cutaneous heat loss is a function of mean-skin and ambient temperatures; 2) central thermoregulatory control is determined by a combination of core and mean-skin temperatures; and 3) the combination of core and mean-skin temperatures can be used to estimate mean-body temperature and, therefore, body heat content.
Unsurprisingly, the accuracy of mean-skin temperature measurements increases with the number of measurement sites. Thus, 15 or more sites are usually used in thermoregulatory studies. For example, the following sites and regional weightings have been used in a hundred or more studies: head—6%, upper arms—9%, forearms—6%, hands—2.5%, fingers—2%, back— 19%, chest—9.5%, abdomen—9.5%, medial thigh—6%, lateral thigh—6%, posterior thigh—7%, anterior calves—7.5%, posterior calves—4%, feet—4%, and toes—2%.58 This large number of measurement sites results in accurate measurements even in the context of regional thermal manipulations (active heating or cooling) and when different amounts of insulation are used in various areas.
When thermal management (insulation or active heating or cooling) is uniformly distributed over the entire body, simpler formulae can be used without great loss of accuracy. A formula with only four sites was developed by Ramanathan in 1964 and remains in common use 59: Mean-skin temperature = 0.3(chest + upper arm) + 0.2(thigh + calf)].
Changes in mean-body temperature over time can be determined by integrating the difference between metabolic heat production (oxygen consumption) and cutaneous heat loss (measured with thermal flux transducers). Mean-body temperature can also be approximated as the mass-weighted sum of regional temperature distributions, which can be determined by integration of radial temperature distributions.60 However, the technique is invasive and the computations tedious. Its use is consequently restricted to controlled studies in laboratories possessing the necessary equipment.61
In 1935 Burton62 cleverly proposed that mean-body temperature (MBT) could be calculated from a formula: MBT = a·TCore + (1–a)·TSkin. The general form of the equation was based on the logic that core tissues are relatively homogeneous, whereas tissue temperature in the peripheral decreases parabolically from core temperature to skin temperature. The value of a, the coefficient describing the contribution of core temperature to mean-body temperature, was then estimated by simultaneously measuring the change in body heat content (in a calorimeter), core temperature, and mean-skin temperature. The resulting value of the coefficient alpha was 0.64, thus giving the formula: MBT = 0.64·TCore + 0.36·TSkin.
A similar approach has been used by others, including Hardy and DuBois,63 who proposed a coefficient, a, of 0.7 for a neutral environment; Stolwijk and Hardy,64 who proposed a coefficient of 0.7 for a hot environment; and Snellen,65 who found the coefficient to be ≈0.8 during muscular work in a hot environment. Subsequently, Colin et al.66 showed in an elegant study that Burton's coefficient was correct for a neutral environment, but that the coefficient increased to 0.79 in an extremely warm environment.
Given all the assumptions about distribution of heat within the body that are necessary to estimate mean-body temperature from core and skin temperatures, it would be surprising if a simple formula based on core and mean-skin temperatures were sufficient. But remarkably, it is. Even during cardiopulmonary bypass, the formula of Colin et al.66 estimates mean-body temperature reasonably well (fig. 7).
Body temperature is normally tightly regulated, more so even than blood pressure or heart rate. The control system is complex and involves parallel positive- and negative-feedback systems that are so widely distributed that nearly every part of the autonomic nervous system participates to some extent.
As early as 1912, physiologists recognized that the hypothalamus is the dominant thermoregulatory site in mammals because control was markedly compromised by injury or destruction of the hypothalamus. (The spinal cord serves this function in birds.) Interestingly, it took nearly another half-century before the importance of thermal input from the skin was appreciated. It is now known that thermal signals from a variety of tissues and structures contribute thermal signals to the hypothalamus, and that there is considerable pre-processing of thermal information on the way from peripheral to central tissues.67 Thus, thermoregulation is based on multiple, redundant signals from nearly every type of tissue. The processing of thermoregulatory information occurs in three phases: afferent thermal sensing, central regulation, and efferent responses.
While all physiologic processes are, to some extent, temperature dependent, specific cells are markedly activated or inhibited by thermal perturbations. The assumption is that these cells are temperature sensors, and they are referred to as warm- or cold-sensing cells. Cold receptors, for example, increase their activity as tissue cools, whereas the reverse is true for heat-sensors.
Because of its accessibility, cutaneous thermoreception is relatively well understood. (See monograph by Hensel for details).68 Human skin is phenomenally sensitive to temperature: An increase in forehead temperature of as little as 0.003°C can be detected! Apparent skin temperature and, more importantly, the ability to influence thermoregulatory responses is not uniform across the skin surface. The face is approximately five times as sensitive as other areas. Furthermore, sensitivity at differing sites depends somewhat on whether the skin is being warmed or cooled. The skin is far more sensitive to rapid thermal perturbations than to those occurring slowly. Similarly, the static skin temperature contributes less to thermoregulatory responses than even small changes.
Cold signals from the skin travel primarily via A nerve fibers whereas warm signals are transduced by unmyelinated C fibers.69 Until recently, little was know about how A and C fibers actually detect cutaneous temperature. However, it now appears that Transient Receptor Potential (TRP) vanilloid (V) and menthol (M) receptors may be the fundamental temperature sensing elements both in skin and the dorsal root ganglia. These receptors, which have only been well characterized in recent years, are a family notable for having unusually high temperature sensitivity. Most change their activity by more than a factor-of-ten over a 10°C range (Q10 > 10). TRPV1−4 receptors are heat activated, whereas TRPM8 and TRPA1 are activated by cold.70,71
Most ascending thermal information traverses the spino-thalamic tracts in the anterior spinal cord, but no single spinal tract is critical for conveying thermal information. Recently, for example, an afferent somatosensory pathway via lateral parabrachial neurons has been shown to transmit signals directly to the preoptic thermoregulatory control center.72 Consequently, the entire anterior cord must be destroyed to ablate thermoregulatory responses. The hypothalamus, other parts of the brain, the spinal cord, deep abdominal and thoracic tissues, and the skin surface each contribute roughly 20 percent of the total thermal input to the central regulatory system.73,74 Hence although the hypothalamus is the dominant and most precise thermoregulatory controller, its temperature per se is not especially important.
The simplest thermoregulatory model is the “set-point” system in which all thermoregulatory responses are simultaneously turned on or off in response to hypothalamic temperature. This model is known to be an inadequate representation of the thermoregulatory system because: 1) responses are determined by thermal input from nearly every portion of the body; 2) responses do not occur simultaneously or at similar temperatures; 3) the model does not incorporate a “null zone” in which no thermoregulatory responses occur; and 4) this model cannot explain thermal adaptation and a host of other observed phenomena.
Consequently, I will review here a model which is somewhat more complicated, but considerably more useful. As with all models, this should be considered a framework from which to analyze thermoregulatory responses, not an actual mechanism by which the body produces those responses. In this model, thermal input from tissues throughout the body are integrated at a variety of centers (including the spinal cord and brain stem), but most importantly the hypothalamus. Individual responses are coordinated on the basis of weighted averages of the diverse inputs.
Temperature is regulated by central structures that compare integrated thermal inputs from the skin surface, neuraxis, and deep tissues with thresholds (triggering core temperatures) for each thermoregulatory response. Control is distributed in the sense that thermal input is integrated at various levels within the neuraxis, but the dominant controller in mammals is the hypothalamus, with autonomic control being centered in the anterior hypothalamus and behavioral control being centered in the posterior hypothalamus. This hierarchical arrangement presumably developed when the evolving thermoregulatory control system co-opted previously existing mechanisms.67 For example, muscles used for shivering were probably developed for posture and locomotion; similarly, thermoregulatory vasomotion is probably an offshoot of systems originally developed for hemodynamic control. It is likely that some thermoregulatory responses can be mounted by the spinal cord alone.74 For example, animals and patients with high spinal-cord transections regulate temperature much worse than normal — but are not poikilothermic.
The slope of response intensity versus core temperature defines the gain of a thermoregulatory response. The maximum intensity of the response is defined as when response intensity no longer increases with further deviation in core temperature. Figure 8, for example, shows the normal sweating response as a function of distal esophageal core temperature during surface warming. There is only background insensible water loss from the skin without anesthesia until the threshold is reached at a core temperature of 36.5°C. The sweating rate then increases quickly as core temperature increases an additional 0.5°C (gain), but remains essentially constant with further hypothermia (maximum response intensity). Although the threshold increases as a function of isoflurane concentration, the gain and maximum intensity remain similar during anesthesia.75
Control of autonomic responses is approximately 80 percent determined by thermal input from core structures76,77 and remains similar during anesthesia (fig. 9). In contrast, fully half of the input controlling behavioral responses is derived from the skin surface.78
Humans apparently measure temperature to great precision, but nonetheless tolerate an interthreshold range over which autonomic responses are not activated. This range of temperatures thus defines normal core temperature under given circumstances (i.e., time of day, menstrual phase). Normal core temperatures in humans typically range from 36.5°C to 37.5°C; values <36°C or >38°C usually indicate loss of control or a thermal environment so extreme that it overcomes thermoregulatory defenses.
Thermoregulatory modeling is thus complicated by interactions with other regulatory responses (i.e., vascular volume control) and time-dependent effects. An area of continuing interest to physiologists is how humans handle environmental stress that would normally provoke opposing compensations. Heat stroke, for example, often results from dehydration in an excessively hot environment. Dehydration would normally activate water-retention mechanisms whereas hyperthermia normally provokes sweating. Heat stroke, in fact, usually develops because the body cannot simultaneously compensate effectively for both perturbations.
Most thermoregulatory models (including the one described above) do not adequately account for the rate at which central and peripheral temperatures change. Consequently, they should be applied to vigorously dynamic situations with caution. Similarly, at least under some circumstances thermoregulatory responses are not determined only by instantaneous thermal inputs, but instead reflect the recent history of thermal perturbations. The extent to which time- and temperature-dependent factors contribute to human thermoregulatory responses remains unclear.
How the body determines absolute threshold temperatures is incompletely understood, but appears to involve inhibitory postsynaptic potentials in hypothalamic neurons79 that are modulated by norepinephrine, dopamine, 5-hydroxytryptamine, acetylcholine, prostaglandin E1, and neuropeptides. The thresholds vary daily by 0.5−1°C in both sexes (circadian rhythm)80 and by ≈ 0.5°C with menstrual cycles in women81. Exercise, nutrition, infection, hypo- and hyperthyroidism, drugs (including alcohol, sedatives, and nicotine), and cold- and warm-adaptation all alter threshold temperatures. But each of these effects is small compared to the profound impairment induced by general anesthesia.
The interthreshold range (core temperatures not triggering autonomic thermoregulatory responses) is bounded by the sweating threshold at its upper end and by the vasoconstriction threshold at the lower end. Within this range, temperatures are presumably sensed accurately but do not trigger regulatory responses. Teleologically, sacrificing a small degree of temperature regulation is prudent because energy and nutrients are not wasted aggressively combating small environmental changes. Some animals such as camels and desert rats use this strategy extensively, allowing body temperature to change up to 10 C during a 24-hour period.
The interthreshold range is usually only 0.2−0.4°C in humans,82 and that range defines normal body temperature. For unclear reasons, control is only half as tight at the circadian nadir near 3:00 AM (fig. 10).80 Because energy cost and nutrients are conserved without excessive autonomic control or evaporative water loss within the interthreshold range, some animals such as camels and desert rats maintain a wide interthreshold range, allowing core temperature changes up to 10°C each day. However, this is very much the exception and most mammals tightly control core temperature.
Both sweating and vasoconstriction thresholds are 0.3−0.5°C higher in women than men, even during the follicular phase of the menstrual cycle (i.e., first ten days).75 Differences are even greater during the luteal phase.83 Central thermoregulatory control is apparently intact even in slightly pre-mature infants,84 but is presumably immature in less-developed infants such as those weighing less than a kilogram. The shivering threshold is well maintained in some elderly subjects well into their 9th decade, whereas others that age regulate poorly; regulation though appears consistently normal in people aged less than 80 years.85
Some thermoregulatory responses are rarely, if ever, activated except by thermal perturbations. Such responses include sweating, peripheral cutaneous vasoconstriction, and brown fat metabolism. In other cases, the thermoregulatory system has co-opted effector mechanisms developed for other purposes including shivering (postural and locomotive muscular activity) and vasomotion (blood pressure and osmotic control). Adaptation of preexisting systems for thermoregulatory control is consistent with the hierarchical thermoregulatory model proposed by Satinoff,67 and may explain why thermoregulatory control is so widely disbursed.
Thermal perturbations, (defined by body temperature difference from a specific threshold) triggers effector responses that actually mediate appropriate increases in environmental heat loss or increases in metabolic heat production. Each response has its own threshold and gain. The control system is thus able to activate responses in an efficient order (i.e., vasoconstriction before shivering which is metabolically costly) and only to the extent actually necessary to maintain core temperature.
Behavioral regulation (intentional manipulation of heat exchange with the environment) is the most powerful thermoregulatory effector. It is such modification that allows humans to live in the warmest and coldest climates on earth. Animals also use behavioral modification to alter heat balance with the environment. Behavioral regulation is most dramatic in reptiles and amphibians. These animals, often referred to as “cold-blooded,” actually regulate their temperatures remarkably well and even develop behavioral “fever.”86 Given access to a reasonable range of environmental temperatures, they will position themselves to maintain a central temperature within a few degrees of “normal.” Interestingly, the temperatures maintained as optimal by most reptiles is similar to that in mammals, near 37 C. Similarly, fish provided with a thermal gradient will position themselves to maintain a nearly constant central temperature.87 One investigator was even able to train a goldfish to maintain his water (and therefore body) temperature nearly constant by pushing a button!88 Even bacteria, given an opportunity, will position themselves to maintain optimal temperature.
Aggressive behavioral modification of environmental heat loss is not necessary in mammals exposed to reasonable environments. This has the evolutionary advantage of maintaining a nearly constant central temperature (presumably necessary for optimal enzyme function) without requiring behavioral modifications that might compromise survival. Nonetheless, when autonomic thermoregulatory responses are insufficient for maintaining central temperature, behavioral responses become critical for survival. Behavioral adaptations take many forms, but most commonly involve simple maneuvers such as moving from direct sun into shade, dressing more warmly, or altering ambient temperature using a heating/air conditioning system. Behavioral responses require a conscious perception of body temperature. Intriguingly, humans appear to poorly sense changes in central temperature; in contrast, minute changes in skin-surface temperature are easily perceived. Thus, behavioral thermoregulation is about half mediated by skin temperature78 whereas mean-skin temperature contributes only 10−20% to the control of autonomic thermoregulatory defenses.76,89
Most metabolic heat is lost from the skin surface and cutaneous and vasoconstriction, the most consistently used autonomic effector mechanism, reduces this loss. Total digital skin blood flow is divided into nutritional (mostly capillary) and thermoregulatory (mostly arterio-venous shunt) components.90 Shunts are typically 100 μm in diameter, which means that one shunt can convey 10,000-fold as much blood as a comparable length of capillary 10-μm in diameter. Arterio-venous shunt flow tends to be “on” or “off” which is simply a way of saying that the gain of this response is high. Roughly 10 percent of cardiac output traverses arterio-venous shunts; consequently, shunt vasoconstriction increases mean arterial pressure ≈15 mmHg.91
Arterio-venous shunts are located only in acral regions (fingers, toes, nose, etc.). These specialized thermoregulatory vessels are under alpha adrenergic control and are constricted by norepinephrine released from sympathetic nerves. Circulating factors appear to have little direct influence on arterio-venus shunts, although hormones such as angiotensin are known to facilitate the response to a given sympathetic stimulus. Most blood vessels constrict in response to local hypothermia, but arterio-venus shunts are relatively resistant regional temperature perturbations and appear to be almost exclusively controlled by central thermoregulatory status. In a thermoneutral environment (e.g., body temperature within the interthreshold range) or in a denervated extremity, arterio-venus shunts are fully dilated. However, at typical ambient temperatures tonic sympathetic stimulation maintains minimal shunt flow.
Non-shivering thermogenesis is defined as an increase in metabolic heat production not associated with muscular activity. This increase occurs largely in specialized fat called brown adipose tissue located largely in the intrascapular and perirenal areas. Brown fat has a dark hue because it is loaded with mitochondria. When stimulated, this tissue has by far the highest metabolic rate of any organ (up to 0.5 W/g). Ordinarily, mitochondrial metabolism produces a proton which is secreted outside the sarcoplasmic reticulum. The proton gradient across this membrane subsequently activates the sodium-potassium ATPase, producing ATP from ADP. When stimulated by norepinephrine released from sympathetic nerves, mitochondrial respiration in brown ATPase tissue proceeds normally. However, production of ATP is prevented by an “uncoupling protein” which allow protons to reenter the sarcoplasmic reticulum without driving the sodium-potassium ATPase.92
Nonshivering thermogenesis is the primary defense against cold in small mammalian species such as mice and rats, and can easily double or triple metabolic heat production (measured as whole-body oxygen consumption) without producing mechanical work. Nonshivering thermogenesis also doubles heat production in infants.93 The intensity of nonshivering thermogenesis is a linear function of the difference between mean body temperature and its threshold.
But despite its importance in small animals and human infants, non-shivering thermogenesis is relatively unimportant or non existent in species having a relatively large body size (i.e., greater than fifty kg). In adult humans, non-shivering thermogenesis is poorly developed94 and contributes little to thermal balance in adult humans.
Sustained shivering augments metabolic heat production 50 to 100 percent in adults. This increase is small compared with that produced by exercise (which can, at least briefly, increase metabolism five-fold) and is, thus, surprisingly ineffective. Shivering is manifested as an irregular tremor which on electromyographic analysis consists of randomly overlapping myofibril depolarization spikes. Superimposed on this rapid and apparently disorganized local activity, is a 4 − 10 cycles/minute waxing-and-waning activity. Notably, this slow amplitude modulation is synchronous and occurs simultaneously in all muscles throughout the body.95 Shivering does not occur in newborn infants and probably is not fully effective until children are several years old. Because the shivering threshold is a full degree less than the vasoconstriction threshold,82 shivering appears to be a “last resort” response to extreme cold.
Sweating is mediated by post-ganglionic, cholinergic nerves.96 It thus is an active process that is prevented by nerve block or atropine administration.97 Even untrained individuals can sweat up to one liter/hour, and athletes can sweat at twice that rate. Sweating is the only mechanism by which the body can dissipate heat in an environment exceeding core temperature. Fortunately, the process is remarkably effective: each gram of evaporated sweat dissipates 0.58 kcal. In a dry, convective environment, individuals can thus easily dissipate many times their basal metabolic rate which is very roughly a kcal·kg−1·h−1. Of course sweat which drips off the skin without evaporating contributes nothing to heat balance, but does promote dehydration.
During exercise, muscle blood flow increases enormously and blood pressure can only be maintained by vigorous vasoconstriction. Furthermore, exercise produces considerable heat which in most environments must be dissipated by increased capillary blood flow and sweating (a liter/hour or more). Both these thermoregulatory compensations compete with the needs of muscle for increased blood flow. Consequently, it is unsurprising that maximum capillary blood flow and sweating rate are impaired by insufficient vascular volume and cardiovascular compromise. In light of the huge cardiovascular stresses imposed by exercise and the thermoregulatory compensation for the attendant increase in metabolic heat production, it is remarkable that humans can perform vigorously in a warm environment and maintain a reasonable blood pressure.
In contrast to shunt flow, capillary blood flow is minimal both at typical ambient temperatures and at thermoneutral temperatures. During heat stress, active dilation of pre-capillary arterials increases capillary blood flow enormously. This dilation certainly involves withdrawal of tonic sympathetic stimulation but also likely involves release of the yet-to-be identified factor from sweat glands; the mediator may be nitric oxide or neuropeptide Y.98 Because active vasodilation requires intact sweat gland function, it also is largely inhibited by nerve block. During extreme heat stress, blood flow through the top millimeter of skin can reach 7.5 liters/minute — equaling the entire resting cardiac output.99 The threshold for active vasodilation usually is similar to the sweating threshold, but maximum cutaneous vasodilation usually is delayed until sweating intensity is at its maximum.
All potential thermoregulatory responses are ideally available and used in a specific order depending on their respective thresholds and response gains. However, one or more effectors may be disabled by circumstances. For example, social convention may restrict voluntary movement or the ability to seek a warmer or cooler environment. Or a muscle relaxant may prevent shivering or a vasodilator may restrict vasoconstriction. In such circumstances, remaining effectors compensate to the limit of their abilities. The result is that core temperature is usually nonetheless maintained, although the range of tolerated environments decreases.
Hyperthermia is a generic term simply indicating a core body temperature exceeding normal values. In contrast, fever is a regulated increase in the core temperature targeted by the thermoregulatory system. Hyperthermia can result from a variety of causes and, unlike perioperative hypothermia, usually requires diagnosis and often intervention.
Passive intraoperative hyperthermia results from excessive patient heating and is most common in infants and children. Hyperthermia was common in the tropics, before air conditioning became routine, and was aggravated by the frequent use of atropine.100 Passive hyperthermia, by definition, does not result from thermoregulatory intervention. Consequently, it can easily be treated by discontinuing active warming and removing excessive insulation.
The increase in body temperature during malignant hyperthermia results from an enormous increase in metabolic heat produced by both internal organs and skeletal muscles. Central thermoregulation presumably remains intact during acute crises, but efferent heat loss mechanisms may be compromised by intense peripheral vasoconstriction resulting from circulating catecholamine concentrations 20 times normal.101
Body temperature is minimally influenced by circulating factors such as thyroid hormones; instead it is normally maintained by neuronal systems. In contrast, fever is mediated by endogenous pyrogens which increase the thermoregulatory target temperature (“setpoint”). Endogenous pyrogens include interleukin-1, tumor necrosis factor, interferon alpha, endothelin-1, and macrophage inflammatory protein-1.102,103 There is increasing evidence that vagal afferents mediate between systemic pyrogens and the hypothalamus104, although several systems probably contribute.105 Most endogenous pyrogens have peripheral actions (e.g., immune system activation) in addition to their central generating capabilities. The relative contributions of fever per se and the systemic action of endogenous pyrogens remains unclear; however, it appears that fever itself is an important immune defense.106
Fever is relatively rare during general anesthesia, considering how many patients presumably experience febrile stimuli, including surgical tissue injury. The reason intraoperative fever is rare is that volatile anesthetics per se inhibit expression of fever,107 as do opioids.108,109 Infection is by far the most common cause of fever. Such fevers may reflect pre-existing infection or result, for example, from urological manipulations. However, perioperative fever also occurs in response to mis-matched blood transfusions, blood in the fourth cerebral ventricle, drug toxicity, and allergic reactions.110,111 Some degree of fever is also typical after surgery, and presumably results from the inflammatory response to surgery.112 There is no evidence to support the common attribution of postoperative fever to atelectasis. Instead, the causes of fever are sufficiently diverse — and potentially serious — that physicians caring for febrile patients should consider potential etiologies.
Treatment of hyperthermia depends on the etiology; the critical distinction is between actively maintained fever and hyperthermia that results from excessive heating, inadequate dissipation of metabolic heat, or excessive heat production. A simple way to distinguish the etiologies is that patients with fever and increasing core temperature will have constricted, cold fingertips whereas those with other types of hyperthermia will be vasodilated and have warm fingertips. It is always appropriate to treat underlying causes, but non-febrile hyperthermia will also improve with cooling.
The first- and second-line treatments for fever are amelioration of the underlying cause and administration of anti-pyretic medications.113 The first treatment strategy often fails because the etiology of fever remains either unknown or unresponsive. The second strategy also often fails or is only partially effective, perhaps because some fever is mediated by mechanisms that bypass conventional anti-pyretics.102 It is in these patients that third-line treatment is most likely to be implemented: active cooling. Active cooling of febrile patients is a natural response. However, it often fails to reduce core temperature — while simultaneously worsening the situation by triggering thermoregulatory defenses including intense discomfort, shivering, and autonomic nervous system activation.114,115
Active cooling should thus be used with considerable caution in febrile patients, with great attention to the metabolic and vasomotor consequences — to say nothing of the resulting thermal discomfort. Systems that directly cool the core116-118 provoke less thermoregulatory stress than surface-based systems,115 especially when intense core cooling is combined with gentle surface warming. A general clinical guideline is that cooling which maintains or decreases oxygen consumption is likely to be helpful119, whereas an increasing metabolic rate indicates a potentially harmful activation of thermoregulatory responses.
Anesthetized patients cannot activate behavioral responses, leaving them to rely on autonomic defenses and external thermal management. All general anesthetics so far tested markedly impair normal autonomic thermoregulatory control. Anesthetic-induced impairment has a specific form: warm-response thresholds are elevated slightly, if at all, whereas cold-response thresholds are markedly reduced. Consequently, the interthreshold range increases ten-fold to approximately 2−4°C.109,120-123 The gain and maximum intensity of some responses remain normal,75 whereas general anesthesia reduces others.124,125
Propofol,120 alfentanil,109 dexmedetomidine,121 isoflurane,123 and desflurane122 all increase the sweating threshold only slightly, if at all. Warm defenses are thus well preserved even during general anesthesia. A consequence is that inadvertent hyperthermia during forced-air warming is relatively rare because patients are usually able to dissipate excess heat into their dry, convective micro-environment. They are less protected against hyperthermia with the newer circulating-water garments that not only transfer more heat,61 but are impervious to moisture, thus preventing evaporative heat loss.
Propofol,120 alfentanil,109 and dexmedetomidine,121 produce a marked and linear decrease in the vasoconstriction and shivering thresholds. In contrast, isoflurane123 and desflurane122 decrease the cold-response thresholds non-linearly. Consequently, the volatile anesthetics inhibit vasoconstriction and shivering less than propofol at low concentrations, but more than propofol at typical anesthetic doses.
Interestingly, the normal approximately 1°C difference between the vasoconstriction and shivering thresholds is maintained even when patients are given sedatives or general anesthesia. That the relationship between these two thresholds is so precisely maintained under a large variety of circumstances suggests that both major autonomic cold defenses are similarly controlled, perhaps by an identical central regulator. The only exceptions to comparable control identified to date are nefopam126 and meperidine, which reduces the shivering threshold twice as much as the vasoconstriction threshold127 — explaining the drug's potent anti-shivering action.128,129
The dose-dependent response thresholds for four anesthetic drugs are shown in figure 11. These responses are characteristic of the drugs and drug combinations that have so far been tested. The combination of increased sweating thresholds and reduced vasoconstriction thresholds increases the interthreshold range ten-fold, from its normal value near 0.2−0.4°C to approximately 2−4°C. Temperatures within this range do not trigger thermoregulatory defenses; by definition, patients are thus poikilothermic within this temperature range.
Halothane130, enflurane,131 and the combination of nitrous oxide and fentanyl132 decrease the vasoconstriction threshold 2 − 4°C from its normal value near 37°C. The effects of these drugs on sweating or shivering remain unknown, but experience with other drugs suggests that they are unlikely to have much effect on sweating, but have a profound effect on shivering. Clonidine synchronously decreases cold-response thresholds,133 while slightly increasing the sweating threshold.134 Nitrous oxide decreases the vasoconstriction135 and shivering136 thresholds less than equi-potent concentrations of volatile anesthetics.
Midazolam, in typical clinical doses, minimally influences thermoregulatory control.137,138 Painful stimulation slightly increases vasoconstriction thresholds131 just as pain has an anti-anesthetic effect139 and regional anesthesia has a pro-anesthetic action.140 Consequently, thresholds will be somewhat lower when surgical pain is prevented by simultaneous local or regional anesthesia. Both amino acid141 and fructose142 infusions increase the vasoconstriction threshold by ≈0.5°C.
The effects of vascular volume on thermoregulatory vasoconstriction have not been evaluated during anesthesia. But, positive end-expiratory pressure increases the vasoconstriction threshold while increasing central blood volume by leg raising reduces the threshold.143 Baroreceptor unloading augments the peripheral vasoconstrictor and catecholamine response to core hypothermia while simultaneously reducing thermogenesis — which consequently aggravates hypothermia in the upright position. Upright posture attenuates the thermogenic response to core hypothermia but augments peripheral vasoconstriction. This divergent result suggests that input from the baroreceptor modifies the individual thermoregulatory efferent pathway at a site distal to the common thermoregulatory center or neural pathway.144
Both the gain and maximum intensity of sweating remain normal during isoflurane (Fig. 8)75 and enflurane anesthesia.145 However, the gain of arterio-venous shunt vasoconstriction is reduced three-fold during desflurane anesthesia (fig. 12),124 even though the maximum vasoconstriction intensity remains normal.146 Volatile anesthetics thus not only markedly decrease the vasoconstriction threshold,122,123 but once triggered, three times as much additional hypothermia as normal is required to reach maximum vasoconstriction. Fortunately, maximum intensity is finally reached and once reached, is effective, usually preventing further core hypothermia.10
Shivering is rare with surgical doses of general anesthesia, which is consistent with its threshold being roughly 1°C less than the vasoconstriction threshold.109,120-123 The reason is that vasoconstriction is effective, constraining metabolic heat to the core thermal compartment, thus usually preventing additional hypothermia.10 Consequently, it is rare even for unwarmed patients to become cold enough to induce shivering. Nonetheless, sufficient active cooling can induce shivering.
Gain and maximum shivering intensity remain normal during both meperidine and alfentanil administration.147 Gain also remains nearly intact during nitrous oxide administration, although maximum intensity is reduced.148 Isoflurane changes the macroscopic pattern of shivering to such an extent that it is no longer possible to easily determine gain. The drug does, however, reduce maximum shivering intensity.125
To sum up, sweating is the thermoregulatory defense that is best preserved during anesthesia. Not only is the threshold only slightly increased, but also the gain and maximum intensity are well preserved. In contrast, the thresholds for vasoconstriction and shivering are markedly reduced, and furthermore, these responses are less effective than normal even after being activated.
It would be intuitive to conclude that surgical patients become hypothermic because they are minimally covered, exposed to a cold environment, washed with cold fluids that are allowed to evaporate, because surgery per se increases heat loss from within incisions, and because general anesthesia reduces metabolic rate. However, even the combination of all these factors would rarely produce hypothermia in subjects with intact thermoregulatory defenses. Anesthetic-induced thermoregulatory impairment is thus by far the most important cause of perioperative hypothermia.
As we have seen, thermoregulatory control is profoundly impaired by most any type of general anesthesia in adults, resulting in a large interthreshold range (i.e. 2−4°C) over which core temperature perturbations fail to trigger regulatory defenses. Thermoregulatory control is equally bad in anesthetized infants and children, but does not appear to be worse. For example, thermoregulatory vasoconstriction is comparably impaired in infants, children, and adults given isoflurane149 or halothane150 (fig. 13). In contrast, the vasoconstriction threshold is about 1°C less in patients aged 60−80 years than in those between 30 and 50 years old (fig. 14).151,152 Infants are nonetheless at special risk of hypothermia because their large surface area-to-mass ratio increases the relative difference between heat loss to heat production.
Nonshivering thermogenesis does not occur in anesthetized adults,153 which is hardly surprising since this response is not particularly important in unanesthetized adults.94 In contrast to adult humans, nonshivering thermogenesis is an important thermoregulatory response in animals and human infants. However, nonshivering thermogenesis in animals is inhibited by volatile anesthetics,154 and it fails to increase the metabolic rate in infants anesthetized with propofol.155 It thus appears that nonshivering thermogenesis is relatively unimportant in perioperative patients and certainly has a small effect compared with the approximately 30% reduction in metabolic rate associated with general anesthesia.
Central thermoregulatory control is slightly impaired by neuraxial anesthesia, but this is combined with reduced gain and maximum response intensity of shivering. Autonomic impairment is compounded by an impairment of behavioral regulation so that patients do not recognize that they are hypothermic. And finally, core temperature is not usually monitored during neuraxial anesthesia.
The result is that patients undergoing neuraxial anesthesia typically become hypothermic and do not sense the hypothermia. In addition, the anesthesiologist does not detect the hypothermia. This is problematic because there is little reason to believe that patients having neuraxial anesthesia are protected from the well-established complications of hypothermia.
Epidural43,156 and spinal156,157 anesthesia each decrease the thresholds triggering vasoconstriction and shivering (above the level of the block) about 0.6°C (fig. 15). Although the magnitude is less, the pattern of impairment is thus similar to that observed with general anesthetics and opioids, suggesting an alteration in central, rather than peripheral control seems most likely. The mechanism by which peripheral administration of local anesthesia impairs centrally mediated thermoregulation remains unknown, but is proportional to the number of spinal segments blocked (fig. 16).158
Reduced thresholds during neuraxial anesthesia does not result from recirculation of neuraxially administered local anesthetic because impairment is similar during epidural and spinal anesthesia,43,156,157 although the amount and location of administered local anesthetic differs substantially. Furthermore, lidocaine administered intravenously in doses producing plasma concentrations similar to those occurring during epidural anesthesia has no thermoregulatory effect.159 Finally, neuraxial administration of 2-chloroprocaine, a local anesthetic which has a plasma half life well under a minute, also impairs thermoregulatory control.160
Since neuraxial anesthesia prevents vasoconstriction and shivering in blocked regions, it is unsurprising that epidural anesthesia decreases the maximum intensity of shivering. However, epidural anesthesia also reduces the gain of shivering which suggests that the regulatory system is unable to compensate for lower body paralysis (fig. 17).125 Thermoregulatory defenses, once triggered, are thus less effective than usual during regional anesthesia.
Sedative and analgesic medications all impair thermoregulatory control to some extent.109,127,137,161 Such inhibition may be severe when combined with the intrinsic impairment produced by regional anesthesia and other factors, including advanced age or pre-existing illness (fig. 18).85
Interestingly, core hypothermia during regional anesthesia may not trigger a perception of cold.43,162 The reason is that thermal perception (behavioral regulation) is largely determined by skin rather than core temperature.78 During regional anesthesia, core hypothermia is accompanied by a real increase in skin temperature. The paradoxical result is often a perception of continued or increased warmth, accompanied by autonomic thermoregulatory responses including shivering (fig. 19).43,162
Taken together, these data indicate that neuraxial anesthesia inhibits numerous aspects of thermoregulatory control. The vasoconstriction and shivering thresholds are reduced by regional anesthesia,43,156-158,163 and further reduced by adjuvant drugs109,137 and advanced age.85 Even once triggered, the gain and maximum response intensity of shivering are about half normal.164 Finally, behavioral thermoregulation is impaired.162 The result is that cold-defenses are triggered at a lower temperature than normal during regional anesthesia, defenses are less effective once triggered, and patients frequently do not recognize that they are hypothermic. Because core-temperature monitoring remains rare during regional anesthesia,44 substantial hypothermia often goes undetected in these patients.42
Shivering-like tremor is common during neuraxial anesthesia and has at least four potential etiologies: 1) normal thermoregulatory shivering in response to core hypothermia; 2) normal shivering in normothermic or even hyperthermic patients who are developing a fever; 3) direct stimulation of cold receptors in the neuraxis by injected local anesthetic; and, 4) non-thermoregulatory muscular activity that resembles thermoregulatory shivering. However, other etiologies remain possible. For example, a convincing cause has yet to be identified for the intense shivering that so often occurs immediately after induction of spinal or epidural anesthesia for cesarean delivery — well before core temperature has had time to decrease.
Most shivering associated with neuraxial anesthesia appears to be normal shivering, the expected response to hypothermia. And at least in volunteers given neuraxial anesthesia, shivering is always preceded by core hypothermia and vasoconstriction (above the level of the block).43 Furthermore, electromyographic analysis indicates that the tremor has the 4−8 cycles/minute waxing-and-waning pattern that characterizes normal shivering.160 Fever is defined by a regulated increase in thermoregulatory response thresholds and can thus provoke shivering even in normothermic individuals. Nonetheless, perioperative fever is probably a relatively rare cause of shivering.
All mammals and birds have spinal thermoreceptors. There is thus the theoretical possibility that injection of relatively cool (i.e., ambient temperature) local anesthetic into the epidural space might provoke shivering by stimulating local temperature sensors. Consistent with this possibility, the incidence of shivering in pregnant women was reported to be greater when they are given refrigerated epidural anesthetic than when the anesthetic is warmed before injection.165 However, epidural administration of large amounts of ice-cold saline does not trigger shivering in non-pregnant volunteers.166 Furthermore, the incidence of shivering is comparable in volunteers43 and non-pregnant patients167 given warm or cold epidural anesthetic injections. These data indicate that temperature of injected local anesthetic rarely provokes shivering during major conduction anesthesia.
Not all shivering-like tremor is thermoregulatory. It is possible to detect low-intensity shivering-like muscular activity in both surgical patients168 and during labor.169 The cause of this muscular activity remains unknown, but it is associated with pain and may thus result from sympathetic nervous system activation.170
Since skin temperature contributes to control of thermoregulatory responses, shivering of any type can be treated by warming the skin surface.171 This is why shivering so often stops in a matter of seconds after entering a warm room even though core temperature hasn't had time to change at all. However, the entire skin surface contributes 20% to thermoregulatory control76,89 and the lower body contributes about 10%,163 sentient skin warming is likely to only compensate for small reductions in core temperature. As might thus be expected, skin warming is only effective in a fraction of patients.
Most often, pharmacologic treatments will be required for moderate or severe shivering. The same drugs that are effective for shivering after general anesthesia can be used to treat shivering during neuraxial anesthesia: these include meperidine (25 mg, IV or epidurally),172 clonidine (75 μg, IV),173 ketanserin (10 mg, IV),173 and magnesium sulfate (30 mg/kg, IV).174
Prolonged epidural analgesia for labor and delivery is occasionally associated with hyperthermia, typically to 38.5−39.5°C. Hyperthermia develops only in a sub-set of women.175 Hyperthermia typically develops after at least five hours of labor, and then increases over time.176-179 The clinical consequence of this hyperthermia is that women given epidural analgesia for labor are more often given antibiotics than in those treated conventionally, and their offspring are more commonly treated for sepsis.177,180,181
Although best studied and most concerning in the context of labor, the association between epidural analgesia and hyperthermia is by no means restricted to labor; it also occurs in non-pregnant post-operative patients.182 It is thus apparent that this hyperthermia is not restricted to pregnancy and must have a more general etiology.
There are several potential explanations for hyperthermia during labor analgesia. For example, it could simply be passive hyperthermia resulting from excessive heat production and inadequate heat dissipation to the environment. Labor certainly involves muscular effort that increases metabolic rate; furthermore, maternal metabolism is already increased by the fetus. Nonetheless, maternal metabolic rate remains small compared with even gentle exercise which perhaps doubles metabolic rate, and does not provoke hyperthermia in any but the most extreme environments. There is not reason to believe that epidural analgesia per se alters whatever increase in metabolic rate might normally accompany labor. And of course metabolic rate is near-normal in postoperative patients who also develop hyperthermia with epidural analgesia.
A dense epidural block would inhibit sweating, which is sympathetically mediated, in the blocked region; but epidural analgesia for labor does not normally produce a sufficiently dense block. Furthermore, in a relatively dry and cool hospital environment, patients could easily dissipate many times their basal metabolic rates just from the upper body. It thus seems unlikely that an imbalance between heat production and loss is the explanation for hyperthermia during labor analgesia. A corollary is that hyperthermia during labor analgesia is a regulated fever rather than simple passive hyperthermia.
Hyperthermia during labor could just be the normal febrile response to infection. “Fever work-ups” and antibiotic treatments are common responses to maternal hyperthermia, and some hyperthermia surely is infectious fever.183 Nonetheless, typical epidural-associated hyperthermia seems unlikely to result from infection and the current consensus is that infection is rarely the cause.
Inflammation is a different matter, though. There are many potential sources of non-infectious inflammation in laboring patients, to say nothing of postoperative patients who obviously have injured tissues. For example, Dashe et al concluded: “Epidural analgesia is associated with intrapartum fever, but only in the presence of placental inflammation.”184 It seems likely that inflammation provokes a regulated febrile response during labor (and in postoperative patients). Consistent with this theory, high-dose steroids — powerful antiinflammatory drugs — nearly eliminate fever during labor.185 In contrast, acetaminophen did not prevent hyperthermia, although the drug is usually an effective antipyretic.186 That prolonged labor is associated with a greater risk of hyperthermia is consistent with a longer period in which to develop inflammation, especially placental inflammation which is likely to release a variety of pyrogenic cytokines. And of course longer labor is associated with factors that promote inflammation.187
The difficulty is that epidural analgesia surely does not augment the general inflammatory response to labor or surgery. Nor does it increase the risk of fetal malposition or need for cesarean delivery.188 It thus remains unclear why epidural analgesia augments the risk of hyperthermia during labor and in postoperative patients. The conventional assumption is that hyperthermia is somehow caused by the technique; although no even slightly convincing mechanism has been proposed.
It is worth remembering, though, that when hyperthermia during labor is studied, pain in the “control” patients is usually treated with opioids — which themselves blunts thermoregulatory defenses109,127 and specifically attenuates fever.108 Fever associated with infection or tissue injury might then be suppressed by low doses of opioids that are usually given to the “control” patients while being expressed normally in patients given epidural analgesia.189 The extent to which this mechanism contributes remains to be determined, and the theory is controversial.190 However, no convincing alternative explanation has been advanced.
Core temperature, while by no means completely characterizing body heat content and distribution, is the best single indicator of thermal status in humans. Core temperature can be accurately monitored at the tympanic membrane, pulmonary artery, distal esophagus, and nasopharynx. Under appropriate circumstances, core temperature can also be reliably estimated from the mouth, axilla, and bladder. In contrast, infrared aural canal (“tympanic”) and temporal artery systems are insufficiently accurate for clinical use.
Body temperature should be monitored in most patients undergoing general anesthesia exceeding 30 minutes in duration and in all patients whose surgery lasts longer than one hour. Measuring body temperature (and maintaining normothermia) is now the standard-of-care during prolonged general anesthesia, especially for large operations where the risk of hypothermia is substantial. Core temperature should also be measured during regional anesthesia in patients likely to become hypothermic, including those undergoing body cavity surgery.
The processing of thermoregulatory information occurs in three phases: afferent thermal sensing, central regulation, and efferent responses. Transient Receptor Potential (TRP) vanilloid (V) and menthol (M) receptors may be the fundamental temperature sensing elements. Most ascending thermal information traverses the spino-thalamic tracts in the anterior spinal cord, but no single spinal tract is critical for conveying thermal information. The hypothalamus, other parts of the brain, the spinal cord, deep abdominal and thoracic tissues, and the skin surface each contribute roughly a fifth of the total thermal input to the central regulatory system.
Temperature is regulated by central structures that compare integrated thermal inputs from the skin surface, neuraxis, and deep tissues with thresholds (triggering core temperatures) for each thermoregulatory response. The slope of response intensity versus core temperature defines the gain of a thermoregulatory response. The maximum intensity of the response is defined as when response intensity no longer increases with further deviation in core temperature. The interthreshold range (core temperatures not triggering autonomic thermoregulatory responses) is bounded by the sweating threshold at its upper end and by the vasoconstriction threshold at the lower end. The interthreshold range is usually only 0.2−0.4°C in humans, and that range defines normal body temperature.
Behavioral regulation is the most powerful thermoregulatory effector, and it is behavioral regulation that allows humans to tolerate extreme environments. However, surgical patients much largely depend on autonomic responses including sweating, vasoconstriction, and shivering. Among these defenses, vasoconstriction is the most important and accounts for most perioperative thermal perturbations.
Hyperthermia is any increase in core temperature; in contrast, fever is a regulated increase in the core temperature targeted by the thermoregulatory system. Fever is mediated by circulating endogenous pyrogens and is an active process. Hyperthermia can result from a variety of causes, many of which are serious including infection, mis-matched blood transfusion, allergic reactions, and malignant hyperthermia. Perioperative hyperthermia thus deserves a serious diagnostic effort, and often intervention.
General anesthetics and opioids have little influence on sweating, but profoundly reduce the vasoconstriction and shivering thresholds. The results is a 10−20-fold increase in the interthreshold range. In contrast, general anesthetics have relatively little effect on the gain and maximum intensity of thermoregulatory responses. It is thermoregulatory impairment not — as one might assume — exposure to a cool operating room environment that causes most perioperative thermal perturbations. Thermoregulatory defenses are reasonably well maintained in infants and children, but somewhat impaired in the elderly.
Central thermoregulatory control is slightly impaired by neuraxial anesthesia, but this is combined with reduced gain and maximum response intensity of shivering. Autonomic impairment is compounded by an impairment of behavioral regulation so that patients do not recognize that they are hypothermic. The result is that patients undergoing neuraxial anesthesia typically become hypothermic and do not sense the hypothermia. Temperature should thus be measured in patients having major surgery under regional anesthesia, and they should be actively warmed as necessary to maintain normothermia.
Many of the studies described in this review were supported by National Institutes of Health Grants GM 39723, 27345, 58273, and 061655 (Bethesda, Maryland) and the Joseph Drown Foundation (Los Angeles, California).
Summary statement: This article reviews perioperative temperature monitoring and the effects of anesthetic drugs on body temperature control.