Best Practice & Research Clinical Anaesthesiology

Vol. 22, No. 4, pp. 627–644, 2008

doi:10.1016/j.bpa.2008.06.004
available online at http://www.sciencedirect.com

Physiology of Thermoregulation

Andrea Kurz * M.D.

Professor and Vice Chair
Department of Outcomes Research, The Cleveland Clinic, 9500 Euclid Avenue, P77 Cleveland, Ohio 44195, USA

Core body temperature is one of the most tightly regulated parameters of human physiology. At
any given time, body temperature differs from the expected value by no more than a few tenths
of a degree. However, slight daily variations are due to circadian rhythm, and, in women, monthly
variations are due to their menstrual cycle. Importantly, both anesthesia and surgery dramati-
cally alter this delicate control, and as a result intraoperative core temperatures 1 to 3  C below
normal are not uncommon.
Consequently, perioperative hypothermia leads to a number of complications including post-
operative shivering (which unacceptably increases patients’ metabolic rates), impaired coagula-
tion, prolonged drug action, and negative postoperative nitrogen balance. In this review I will
describe how anesthesia and surgery impair thermoregulation, the resulting changes in heat bal-
ance, and the physiological responses provoked by perioperative alterations in body
temperature.

Key words: anesthesia; core temperature; heat balance; shivering; surgery; sweating;
vasoconstriction.

NORMAL THERMOREGULATION

Like many other physiological control systems, thermoregulation uses negative-feed-

back to minimize perturbations from preset, ‘‘normal’’ values. Early research in
1912 showed that if the hypothalamus was destroyed, animals were unable to ade-
quately regulate body temperature. In the late 1950’s experiments showed that
mice placed in a cold environment shivered without decreasing their hypothalamic tem-
peratures and thus was recognized the importance of thermal input from the skin sur-
face. Then, in the early 1960’s, physiologists reported active thermoregulation in
response to isolated warming and cooling at sites other than the hypothalamus or
skin surface; these included extra-hypothalamic portions of the brain, deep abdominal
tissues, and the spinal cord. Collectively, this research shows that normal

* Tel.: þ1 216 445 9924; Fax: þ1 216 444 6135.

E-mail address: ak@or.org
1521-6896/$ - see front matter ª 2008 Published by Elsevier Ltd.
628 A. Kurz

thermoregulation is based on multiple, redundant signals from nearly every tissue type.
It is now understood that the processing of thermoregulatory information occurs in
three phases: afferent thermal sensing, central regulation, and efferent responses (Figure 1).

Afferent input and central control

Receptors for cold and warm are distributed throughout the body. Cold signals tra-
verse A—delta fibers whereas signals from warmth receptors are conveyed by C fi-
bers. Thermal inputs are integrated at numerous levels within the spinal cord and
central nervous system, finally arriving at the hypothalamus which is the dominant
thermoregulatory controller in mammals.1 The skin surface, deep abdominal and tho-
racic tissues, spinal cord, hypothalamus, and other portions of the brain each contrib-
ute very roughly 20 percent to autonomic thermoregulatory control.2,3 Behavioral
responses, in contrast, may depend more on skin temperature.4
Threshold, gain, and maximum response intensity describe each thermoregulatory
response. Thresholds are defined by the core temperature triggering each thermoreg-
ulatory defense (at a given mean skin temperature); this is analogous to the tempera-
ture setting on a home thermostat. Gain characterizes the extent to which response
intensity increases with further deviation from the triggering threshold; it is analogous
to a proportional controller which augments heater output as ambient temperature

Anterior Hypothalamus
Sweating

37 Vasodilation

Skin
Vasoconstriction
Deep tissues
36 Non-Shivering
Thermogenesis
Spinal cord
Shivering
Brain
(non-hypothalamic)
35

Figure 1. The hypothalamus is the dominant thermoregulatory controller in mammals, and is shown here as
a large square. The skin surface, deep abdominal and thoracic tissues, spinal cord, and non-hypothalamic por-
tions of the brain each contribute very roughly 20 percent to control of autonomic thermoregulatory de-
fenses. These inputs are shown entering the hypothalamus from the left side of the figure. Hypothalamic
temperature, per se, also contributes roughly 20 percent to thermoregulatory control. In the hypothalamus,
integrated body temperature is compared to thresholds which are the temperatures triggering specific ther-
moregulatory responses. Temperatures exceeding warm-response thresholds (i.e., sweating) or less than
cold-response thresholds (i.e., vasoconstriction and shivering) initiate the corresponding thermoregulatory
defense. Temperatures between the sweating and vasoconstriction thresholds define the interthreshold
range — temperatures not triggering thermoregulatory defenses. The interthreshold range is normally
only 0.2  C. Because thermoregulatory defenses are effective in most environments, body temperature
rarely deviate more than a few tenths of a degree from its time-adjusted target value. The sweating, vaso-
constriction, and shivering thresholds are from Lopez et al.,5 and are shown as means and standard devia-
tions; the nonshivering thermogenesis threshold is estimated.
 Physiology of Thermoregulation 629

progressively deviates from the thermostat setting. In this context, the highest possible
heat output from the furnace would constitute the maximum response intensity. An
approximately 1  C circadian cycle and an approximately 0.5  C menstrual cycle are
superimposed on the normal human core temperature of 37  C.
Sweating and active vasodilation, the thresholds for warm responses, normally ex-
ceed vasoconstriction, the threshold for the first cold defense, by only 0.2  C. Temper-
atures between the initial warm- and cold-response thresholds define the
interthreshold range. Temperatures within this range do not trigger autonomic ther-
moregulatory defenses. Higher or lower temperatures, though, do trigger effective
thermoregulatory defenses; consequently, the thermoregulatory system usually main-
tains core temperature within approximately 0.2  C of the time-adjusted target value.
The precision of thermoregulatory control is similar in men and women5, but is dimin-
ished in the elderly.6
Mean body temperature is used to assess thermoregulatory responses and is defined
as a physiologically-weighted average reflecting the thermoregulatory importance of
various tissues. In unanesthetized subjects, mean body temperature is z0.85
tcentral þ 0.15 tskin, where average skin temperature can be determined using the for-
mula tskin ¼ 0.3 (tchest þ tarm) þ 0.2 (tthigh þ tleg). The difference between the lowest
warm and highest cold thresholds indicates the sensitivity of the system. The interthres-
hold range (temperature range over which no regulatory responses occur) is typically
z0.2  C. Central body temperature is frequently substituted for mean body temper-
ature in clinical studies.
Given similar warm and cold threshold temperatures and relatively high response
gains, then the regulatory system can also be modeled as a thermostat or ‘‘setpoint’’
(i.e., responses fully activated or completely inactivated at the same temperature). De-
spite being a simplification of a complex system, the setpoint model often describes
thermoregulation remarkably well. However, it is inadequate in anesthetized or se-
dated patients and cannot explain the consistent order in which effectors are initiated.
Although unknown, the mechanism, which determines absolute threshold temper-
atures, appears to be mediated by norepinephrine, dopamine, 5-hydroxytryptamine,
acetylcholine, prostaglandin E1, and neuropeptides. The thresholds vary daily in both
sexes (circadian rhythm) and monthly in women by z0.5  C. Exercise, food intake,
infection, hypo- and hyperthyroidism, anesthetic and other drugs (including alcohol,
sedatives, and nicotine), and cold- and warm-adaptation alter threshold temperatures.
Central regulation is intact in infants, but it may be impaired in the elderly or ex-
tremely ill patients.

Efferent responses
Changes in human behavior are the most effective response to changes in body tem-
perature. It is primarily behavioral defenses that allow humans to live and work in ex-
treme environments. Behavioral strategies are mediated by thermal discomfort, which
provokes responses such as dressing warmly or adjusting ambient temperature.
Sweating and active cutaneous vasodilation are the major autonomic defenses
against heat. Sweating is mediated by post-ganglionic, cholinergic nerves that terminate
on widely, but unevenly, distributed glands. Sweat is an ultrafiltrate of plasma whose
composition depends on the rate of sweating, hydration status, and a number of other
factors. The maximum sweating rate exceeds 0.5 liters/hour in most adults, and is
two- or three-fold greater in trained athletes.7 Each gram of evaporated sweat absorbs
584 cal. Consequently, sweating can easily dissipate many times the basal metabolic
630 A. Kurz

rate in a dry environment. The efficacy of sweating is augmented by pre-capillary ther-

moregulatory vasodilation. Active thermoregulatory vasodilation is a uniquely human
response that is mediated by a yet-to-be-identified factor released from sweat glands.
It increases cutaneous blood flow enormously8, thereby facilitating transfer of heat
from the core to the skin for eventual dissipation to the environment.
Cutaneous vasoconstriction, the most consistently used effector mechanism, reduces
metabolic heat loss from convection and skin surface radiation. Total digital skin blood
flow is divided into nutritional (capillary) and thermoregulatory (arterio-venous shunt)
components. The arterio-venous shunts are anatomically and functionally distinct from
the capillaries supplying nutritional blood to the skin (thus vasoconstriction does not
compromise the needs of peripheral tissues). Shunts are typically 100 mm in diameter,
which means that one can convey 10,000-fold as much blood as a comparable length of
10 mm-diameter capillary. Blood flow through the arterio-venous shunts tends to be
‘‘on’’ or ‘‘off.’’ During heat stress, flow may be 100-fold greater than necessary to sup-
ply the nutritional needs of the skin, but thermoregulatory constriction can decrease
flow more than 10-fold.
Local alpha-adrenergic sympathetic nerves mediate constriction in the thermoreg-
ulatory arterio-venous shunts; flow is minimally affected by circulating catecholamines.
Systemic hemodynamic changes are not observed during thermoregulatory vasocon-
striction because, except during heat exposure, 10% of cardiac output traverses
these vessels; larger arterioles that control blood pressure are not influenced.
Since thermoregulatory shunts are concentrated in the fingers and toes vasocon-
striction reduces flow to a greater extent to the distal extremities than to the central
skin. Skin-surface temperature gradients (forearm temperature — finger tip tempera-
ture), can therefore be measured to quantify the contribution of vasoconstriction to
thermoregulation. Typical skin-temperature gradients are between 2 and þ2  C; gra-
dients 4  C identify significant vasoconstriction. The relationship between several in-
dices of peripheral cutaneous perfusion is indicated in Figure 1.
Arterio-venous shunts located primarily in fingers and toes are the site of thermo-
regulatory vasoconstriction. These shunts are controlled by centrally-mediated alpha-
1 adrenergic receptors; however, constriction is synergistically augmented by local hy-
pothermia via alpha-2 adrenergic receptors.9 Diameter of open shunts is approxi-
mately 100 mm; consequently, they carry 10,000 times as much blood as a given
length of 10-mm capillary.10
Nonshivering thermogenesis is an important thermoregulatory defense in infants11,
but contributes little in children and adults.12 The response is mediated by beta-3 ad-
renergic nerves that terminate on brown fat.13 The macroscopic brown coloration of
this specialized adipose tissue results from its enormous mitochondrial density. Brown
fat is equipped with a unique uncoupling protein that allows direct transformation of
substrate into heat.14 Shivering is an involuntary muscular activity that increases met-
abolic rate two- to three-fold.15
Shivering increases metabolic heat production by z200% in adults, but this increase
is relatively small and surprisingly ineffective when compared with that produced by
exercise (which can increase metabolism by 10-fold). Shivering does not occur in new-
born infants, and probably is not fully effective until several years of age. The rapid
tremor (z50 Hz) and unsynchronized muscular activity of thermogenic shivering sug-
gests no central oscillator. Superimposed on the fast activity, there may be a very slow
(4–8 cycle/minute), synchronous ‘‘waxing-and-waning’’ pattern which is presumably
centrally mediated. Because other rhythmic muscular activities may be clinically similar,
tremor should be considered normal thermoregulatory shivering only when: 1) mean
 Physiology of Thermoregulation 631

body temperature is below the threshold for shivering; 2) tremor is preceded by pe-
ripheral cutaneous vasoconstriction and nonshivering thermogenesis; and 3) tremor
patterns match those produced by centrally mediated shivering.
Thermoregulatory responses are diminished and the risk of hypothermia is in-
creased by age, infirmity, and medication. For example, decreased muscle mass, neu-
romuscular diseases, and muscle relaxants all inhibit shivering which will increase the
minimum tolerable ambient temperature. Similarly, anti-cholenergic drugs inhibit
sweating, decreasing the maximum tolerable temperature.

Practice points

 Core body temperature is tightly regulated around 37  C

 Slight daily variations are due to circadian rhythm
 Processing thermoregulatory information occurs through three pathways:
B Afferent thermal sensing from cold and warm receptors in the
periphery
B Central regulation in the hypothalamus.
B Efferent responses

The skin surface, deep abdominal and thoracic tissues, spinal cord, hypothalamus,
and other portions of the brain each contribute very roughly 20 percent to au-
tonomic thermoregulatory control

 Thermoregulatory responses are characterized by the threshold, gain and max-

imum intensity
B Thresholds are defined by the core temperature triggering each ther-
moregulatory defense (at a given mean skin temperature)
B Gain characterizes the extent to which response intensity increases
with further deviation from the triggering threshold
- A small increase in core temperature triggers sweating and active
vasodilation (warm response threshold)
- A small decrease in core temperature triggers active arterio-
venous shunt vasoconstriction
- Further decrease in core temperature triggers shivering

Temperatures between the initial warm- and cold-response thresholds define the
interthreshold range

THERMOREGULATION DURING ANESTHESIA AND SURGERY

General anesthesia

General anesthesia eliminates behavioral thermoregulatory compensations, leaving

only autonomic defenses to environmental perturbations.
632 A. Kurz

Thermoregulation
All general anesthetics are known to significantly impair thermoregulatory responses.
Anesthetic-induced thermoregulatory inhibition is dose-dependent, and impairs vaso-
constriction and shivering about three times as much as sweating. General anesthetics
linearly increase the warm-response thresholds.16–18 Opioids16 and the intravenous
anesthetic propofol17 linearly decrease the vasoconstriction and shivering thresholds.
In contrast, volatile anesthetics, such as isoflurane18 and desflurane19, decrease cold
responses non-linearly (Figure 2). In contrast to other opioids and the anesthetic drugs
meperidine possesses additional anti-shivering action and inhibits shivering twice as
much as vasoconstriction.20 It has been hypothesized, that the special anti-shivering
effect of meperidine may be primarily related to meperidine‘s activity at k-opioid re-
ceptors. This theory is supported by the facts that moderate-dose naloxone only par-
tially blocks the anti-shivering effect of meperidine and that butorphanol (also a partial
k-opioid receptor agonist) inhibits shivering better than fentanyl . The only drug tested
so far, which does not effect thermoregulatory responses is midazolam.20
Both halothane21 and isoflurane22 impair thermoregulatory vasoconstriction in in-
fants, children, and adults to a comparable extent. On the other hand thermoregulatory

Figure 2. Anesthetic-induced inhibition of thermoregulatory control is usually the major factor determining
perioperative core temperature. Concentration-dependent thermoregulatory inhibition by desflurane (halo-
genated volatile anesthetics), alfentanil (a m—agonist opioid), dexmedetomidine (an alpha-2 agonist), and
propofol (an intravenous anesthetic). The sweating (triangles), vasoconstriction (circles), and shivering
(squares) thresholds are expressed in terms of core temperature at a designated mean skin temperature
of 34  C. Anesthesia linearly, but slightly, increases the sweating threshold. In contrast, anesthesia produces
substantial and comparable linear or non-linear decreases in the vasoconstriction and shivering thresholds.
Typical anesthetic concentrations thus increase the interthreshold range (difference between the sweating
and vasoconstriction thresholds) approximately 20-fold from its normal value near 0.2  C. Patients do
not activate autonomic thermoregulatory defenses unless body temperature exceeds the interthreshold
range; surgical patients are thus poikilothermic over a 3 to 5  C range of core temperatures.
 Physiology of Thermoregulation 633

responses are significantly delayed in the elderly subjecting this patient population to in-
traoperative hypothermia.
Thus, during anesthesia, the interthreshold range (core temperatures not triggering
thermoregulatory defenses) increases approximately 20-fold from its normal value
near 0.2  C. As a result, anesthetized patients are poikilothermic over an approxi-
mately 4  C range of core temperatures. Within this range, patients are poikilothermic
and body temperature changes are passively determined by the difference between
metabolic heat production and heat loss to the environment.
In patients receiving volatile anesthetics both gain and maximum response intensity
of sweating and active vasodilation are well preserved.23 Desflurane, however, reduces
the gain of arterio-venous shunt vasoconstriction three-fold, without altering the max-
imum intensity (Figure 3).24 It seems likely that the thermoregulatory effects of general
anesthetics are primarily central since anesthetics of widely different types produce
similar thermoregulatory inhibition. However, the possibility of peripheral inhibition
has not been eliminated.
Higher vasoconstriction thresholds are observed in anesthetized individuals who
are subject to rapid core temperature perturbations, but the magnitude of the in-
crease has yet to be quantified. Similarly, the extent to which intraoperative thermo-
regulatory responses depend on the direction of temperature change remains unclear.
The ratio of cutaneous to core thermal input to autonomic thermoregulatory

1.5

1.0
Flow (ml/min)

0.5

0
34 35 36 37
Core Temperature (°C)

Figure 3. Gain of vasoconstriction: Finger blood flow, as determined using volume plethysmography, with-
out (open circles) and with (filled squares) desflurane administration. Values were computed relative to the
thresholds (finger flow ¼ 1.0 ml/min) in each subject. Flows of exactly 1.0 ml/min are not shown because
flows in each individual were averaged over 0.1 or 0.05  C increments; each data point thus includes
both higher and lower flows. The horizontal standard deviation bars indicate variability in the thresholds
among the volunteers; although errors bars are shown only at a flow near 1.0 ml/min, the same temperature
variability applies to each data point. The slopes of the flow vs. core temperature relationships (1.0 to
z0.15 ml/min) were determined using linear regression. These slopes defined the gain of vasoconstriction
with and without desflurane anesthesia. Gain was reduced by a factor of three, from 2.4 to
0.8 ml min1  C1 (P < 0.01).
634 A. Kurz

responses ranges from 5–20%*, and it is unknown if the ratio remains similar in anes-
thetized individuals. But once triggered, the intensity of arterio-venous shunt vasocon-
striction during anesthesia is similar to that in unanesthetized individuals.
Both core and skin temperatures contribute to steady-state thermoregulatory con-
trol. Skin and core temperatures contribute linearly to control of vasoconstriction and
shivering in men, and the cutaneous contributions average z20% in both men and
women. The same coefficients can thus be z20% used to compensate for experimen-
tal skin temperature manipulations in men and women (Figure 4).25 However, there is
also a dynamic component that provokes especially aggressive defenses against rapid
thermal perturbations. The dynamic component potentially complicates interpretation
of thermoregulatory studies and slow induction of therapeutic hypothermia. Onset of
vasoconstriction and shivering occurred at similar mean-skin temperatures when the
skin was cooled at between 2 and 6  C/h. Surface cooling at a rate of 6  C/h can thus
be used in thermoregulatory studies and for induction of therapeutic hypothermia
without provoking dynamic thermoregulatory defenses.
In anesthetized, hypothermic adults total body oxygen consumption does not in-
crease significantly, indicating that nonshivering thermogenesis is not functional during
general anesthesia. Since nonshivering thermogenesis is of little importance in normal
humans, it is not surprizing that its influence during anesthesia is also minimal. When
anesthetized infants undergo vasoconstriction, oxygen consumption increases

Vasoconstriction Shivering
38 #1 #2

36

34
Core Temperature (°C)

38 #3 #4

36

34

38 #5 #6

36

34
30 34 38 30 34 38
Skin Temperature (°C)

Figure 4. Skin core contribution: Core and skin temperatures at the vasoconstriction and shivering thresh-
olds were linearly related in men. The correlation coefficients (r2) averaged 0.90  0.06 for vasoconstriction,
and 0.94  0.07 for shivering. The extent to which mean skin temperature contributed to central thermo-
regulatory control (ß) was calculated from the slopes (S) of the skin temperature vs. the core temperature
regressions, using the formula: ß ¼ S/(S1). Cutaneous contribution to vasoconstriction averaged 20  6%,
which did not differ significantly from the contribution to shivering: 19  8%.
 Physiology of Thermoregulation 635

simultaneously, indicating that nonshivering thermogenesis remains intact, and that its
gain is approximately normal.
Temperature variation within surgical incisions has an unknown clinical effect. All
formulas for determining mean body and average skin-surface temperature were de-
veloped in normal subjects; consequently, no equations for mean body temperature
include compensation for surgical incisions of different sizes, in different locations.
Since thermal receptors are widely distributed, it is likely that the afferent input
from tissues exposed to cold by large incisions contributes significantly to total central
(hypothalamic) input and alters thermoregulatory thresholds. Patients having small in-
cisions (at a given anesthetic concentration) require lower mean body temperatures
to trigger thermoregulatory vasoconstriction than those requiring large incisions.
Collectively, these studies indicate that general anesthetics increase the interthres-
hold range from a normal value <0.6  C to approximately 4  C. It remains possible
that body temperature remains accurately sensed during anesthesia, but that temper-
atures within the interthreshold range simply are not integrated to initiate regulatory
responses. However, once body temperature deviates sufficiently from normal to trig-
ger thermoregulatory responses, the gain and maximum intensity of these effector re-
sponses remains nearly normal. Markedly altered thermoregulatory thresholds with
relatively well preserved gain and maximum intensities contrasts starkly with anes-
thetic effects on several other homeostatic systems.

Heat balance

Thermal steady state is defined by heat loss to the environment equaling metabolic
heat production. Thus, over the long term, heat loss must equal heat production to
maintain body temperature. However, body temperature and tissue heat content is
not uniformly distributed: thermoregulation keeps core temperature nearly constant,
whereas peripheral tissues usually are maintained at a lower temperatures by tonic
vasoconstriction.
In practice, vasomotion alters the heat content of peripheral tissues, yet the temper-
ature of vital organs remains unchanged, because the periphery acts as a thermal buffer.
This allows individuals to lose heat in a cold environment or absorb heat in a warm en-
vironment. This strategy minimizes the need for other autonomic responses which may
be costly in terms of metabolic needs, use of resources, or behavioral requirements. Be-
cause of their large mass, the legs probably constitute most of the peripheral thermal
buffer. The capacity of the peripheral compartment is approximately 150 kcal (i.e.,
body heat content can change this amount without altering core temperature).
Usually a 2–4  C core-to-peripheral temperature gradient is maintained by tonic
thermoregulatory vasoconstriction, resulting in the uneven distribution of body
heat.26 Induction of general anesthesia reduces the vasoconstriction threshold to
below body temperature, thus opening arterio-venous shunts. The resulting core-
to-peripheral redistribution of body heat decreases core temperature 1–1.5  C during
the first hour of general anesthesia.26 Net loss of heat to the environment contributes
little to this initial decrease (Figure 5).
Redistribution hypothermia is difficult to treat, but can be prevented by cutaneous
warming before induction of anesthesia.27 Pre-induction warming only slightly in-
creases core temperature (which remains well regulated), but markedly increases pe-
ripheral compartment temperature. Because heat only flows down a temperature
gradient, redistribution is prevented in proportion to the reduction in the core-to-
peripheral temperature gradient.
636 A. Kurz

Loss

80

Heat
60
(kcal/h)

Production 40

0
Temp (°C)

-1 Mean Body
²

Redistribution
-2

Core
-3

-3 -2 -1 0 1 2 3
Time (h)

Figure 5. Changes in body heat content and distribution of heat within the body during induction of general
anesthesia (at elapsed time zero). The change in mean body temperature was subtracted from the change in
core (tympanic membrane) temperature, leaving the core hypothermia specifically resulting from redistribu-
tion. Redistribution hypothermia was thus not a measured value; instead, it is defined by the decrease in core
temperature not explained by the relatively small decrease in systemic heat content. After one hour of an-
esthesia, core temperature had decreased 1.6  0.3  C, with redistribution contributing 81 percent to the
decrease. Even after three hours of anesthesia, redistribution contributed 65 percent to the entire
2.8  0.5  C decrease in core temperature. Data obtained from Matsukawa et al.26

In the subsequent few hours, core temperature usually decreases at a slower rate.
This decrease is nearly linear and results simply from heat loss exceeding metabolic
heat production.28 It has been attributed to undressing patients in a cool environment,
anesthetic-induced vasodilation (which increases skin temperature), evaporation of
surgical skin preparation solution, loss of heat from surgical incisions, and anes-
thetic-induced reduction in metabolic rate. Approximately 90 percent of all heat is
lost via the skin surface, with radiation and convection usually contributing far more
than evaporative or conductive losses.
After 3 to 5 hours of anesthesia, core temperature often stops decreasing. This
core-temperature plateau may be a simple thermal steady-state, with heat loss equal-
ing heat production. This sort of steady-state plateau is especially likely in patients who
are well insulated or effectively warmed. In patients becoming sufficiently hypothermic,
however, the plateau results from re-activation of thermoregulatory vasoconstric-
tion16,18,20 which decreases cutaneous heat loss and constrains metabolic heat to
the core thermal compartment.28 Intraoperative vasoconstriction thus re-establishes
the normal core-to-peripheral temperature gradient by preventing loss of centrally-
generated metabolic heat to peripheral tissues. From a clinical point of view, an active
core-temperature plateau is potentially dangerous because mean body temperature
and body heat content continues to decrease, although core temperature remains
 Physiology of Thermoregulation 637

constant. Because vasoconstriction is effective, intraoperative core temperature rarely

decreases the additional 1  C necessary to trigger shivering.
BMI substantially influences intraoperative core temperature changes, and this effect
depends on the phase of hypothermia.29 During the initial phase, the amount of redistri-
bution hypothermia is inversely proportional to the percentage body fat
(DTC ¼ 0.034$BF  2.2, r2 ¼ 0.63) and the weight-to-surface area (Wt/SA) ratio
(DTC ¼ 0.052$Wt/SA  3.35, r2 ¼ 0.66). During the second phase, the core cools line-
arly, and the cooling rate is inversely proportional to the weight-to-surface area ratio
(Rate ¼ 0.035$(Wt/SA)  2.2, r2 ¼ 0.29). And, during the final phase, thermoregulatory
vasoconstriction is effective in virtually all patients independent of their morphology, and
produces a four-fold reduction in the core cooling rate. (Figures 6–9)
Taken together, these studies suggest that intraoperative hypothermia develops in
three phases. Initially, core temperature decreases rapidly when anesthetic-induced in-
hibition of tonic thermoregulatory vasoconstriction allows core-to-peripheral redistri-
bution of body heat. Although difficult to treat, redistribution hypothermia can be
prevented by warming peripheral tissues before induction of anesthesia. Secondly,
a slower, linear decrease in body temperature results from heat loss exceeding met-
abolic heat production. Because nearly all heat is lost from the skin surface, adding cu-
taneous insulation will decrease the rate of cooling; similarly, sufficient active warming
will increase body temperature during this phase. And in the final phase, core

110 *
*
* Loss
(kcal/h)

90
Heat

***
70
**
Production
50
*
* 1.0
***
² Temp (°C)

Core 0.5
* 0.0
*** -0.5
*
Mean Body -1.0
***
Constraint

20
**
(kcal)

10
0
-10
-2 -1 0 1 2 3
Elapsed Time (h)

Figure 6. Plateauphase: Vasoconstriction decreased cutaneous heat loss (adjusted for evaporative and re-
spiratory loss) z25 kcal/h. However, heat loss always exceeded heat production. Consequently, mean body
temperature, which decreased at a rate of z0.6  C/h before vasoconstriction, subsequently decreased at
a rate of z0.2  C/h. Core temperature also decreased at a rate of z0.6  C before vasoconstriction, but
remained virtually constant during the subsequent three h. Since mean body temperature and body heat con-
tent continued to decrease, constraint of metabolic heat to the core thermal compartment contributed to
the core-temperature plateau. That is, vasoconstriction re-establishing the normal core-to-peripheral tem-
perature gradient by preventing metabolic heat (which is largely generated in the core) from escaping to pe-
ripheral tissues. Constrained heat is presented cumulatively, referenced to vasoconstriction at elapsed time
zero; asterisks (*) indicate values significantly different from those obtained at elapsed time zero.
638 A. Kurz

temperature stops changing after 3–4 h of anesthesia. This plateau can be passive in
patients remaining relatively warm, or may be accompanied by thermoregulatory va-
soconstriction which decreases cutaneous heat loss and sequesters metabolic heat
to the core.

Practice points

 Hypothermia is common during anesthesia and surgery

 Most anesthetics and narcotics effect central and peripheral thermoregulatory
responses
 Anesthetics increase the sweating threshold and markedly decrease the vaso-
constriction and shivering threshold thus increasing the interthreshold range
 An increased interthreshold range makes patients poikilothermic and thus sus-
ceptible to hypothermia
 Intraoperative hypothermia develops in a very characteristic pattern:
B During induction of anesthesia heat redistributes from the body to the
periphery causing an initial drop in core temperature of 1–1.5  C
B During the following 3 hours core temperature linearly decreases due
to heat loss exceeding metabolic heat production
B After 3–5 hours of anesthesia core temperature stops dropping. This
core temperature plateau is due to peripheral vasoconstriction prevent-
ing loss of centrally generated metabolic heat to peripheral tissues.

Research agenda

 Test drugs, which might impair thermoregulatory control without having anes-
thesia-related side effects. This is especially important if hypothermia is needed
for therapeutic reasons.
 Evaluate and quantify the effect of operating room environment perioperative
heat loss.
 Quantify heat loss from the wound during different types of surgery.

Regional anesthesia

Regional anesthesia impairs both central and peripheral thermoregulatory control. As

a result, hypothermia is common in patients given spinal or epidural anesthesia. In pa-
tients becoming sufficiently hypothermic, shivering may again appear — and is often
disturbing to both patients and medical staff.

Thermoregulation
Neural mediation affects all thermoregulatory responses (except during fever, circu-
lating factors normally contribute little to thermoregulatory control). Consequently,
 Physiology of Thermoregulation 639

0.0

² Core Temp (°C/h)

-0.5

-1.0

-1.5

-2.0
10 20 30 40 50
Body Fat ( )

Figure 7. Effect of body mass on redistribution: The amount of redistribution hypothermia [reduction in
core temperature during the first hour of anesthesia (DTC)] was inversely proportional to the percentage
body fat (BF): DTC ¼ 0.034$BF  2.2, R2 ¼ 0.63. The 95% confidence interval for the slope was 0.025 to
0.043  C/%.

nerve blocks prevent regional manifestation of the major thermoregulatory de-

fenses including sweating, vasoconstriction, and shivering. Spinal and epidural anes-
thesia disrupt nerve conduction to more than half the body. This peripheral
inhibition of thermoregulatory defenses is a major cause of hypothermia during re-
gional anesthesia.
Regional anesthesia, however, also impairs central thermoregulatory control.
That spinal or epidural anesthesia would inhibit central control was surprising be-
cause regional anesthesia has no direct central effect. Inhibition is similar with spi-
nal and epidural anesthesia, and does not result simply from recirculation of local
anesthetic to the brain. Instead, it appears that the regulatory system misinterprets
skin temperature in blocked areas as being abnormally elevated.29,30 This apparent
(as opposed to actual) elevation in leg skin temperature fools the regulatory sys-
tem into tolerating lower-than-normal core temperatures before triggering cold
defenses.
Typically, the vasoconstriction and shivering thresholds are reduced approximately
0.5  C whereas the sweating threshold is elevated approximately 0.3  C.30,31 The

0.0
² Core Temp (°C/h)

-0.5

-1.0

-1.5

-2.0
30 35 40 45 50 55
Weight-to-Surface Area Ratio (kg/m2)

Figure 8. Effect of body mass on linear phase: The core cooling rate during the second (linear decrease)
phase was inversely proportional to the weight-to-surface area (Wt/SA) ratio, although the relationship
was weak: Rate ¼ 0.035$(Wt/SA)  2.2; R2 ¼ 0.29. The 95% confidence interval for the slope was 0.017
to 0.053  C/%.
640 A. Kurz

²Core Temp (°C)

Heat Balance
0
* * * Redistribution
* *
-1 * * * * *
* * *
*
Measured
-2
-3 -2 -1 0 1 2 3
Elapsed Time (h)

Figure 9. Redistribution hypothermia during induction of regional anesthesia: To separate the contributions
of decreased overall heat balance and internal redistribution of body heat to the decrease in core temper-
ature, we divided the change in overall heat balance by body weight and the specific heat of humans. The
resulting change in mean body temperature (‘‘heat balance’’) was subtracted from the change in core tem-
perature (‘‘measured’’), leaving the core hypothermia specifically resulting from redistribution (‘‘redistribu-
tion’’). After one h of anesthesia, core temperature had decreased 0.8  0.3  C, with redistribution
contributing 89% to the decrease. During the subsequent two h of anesthesia, core temperature decreased
an additional 0.4  0.3  C, with redistribution contributing 62%. Redistribution thus contributed 80% to the
entire 1.2  0.3  C decrease in core temperature during the three h of anesthesia. The increase in the ‘‘re-
distribution’’ curve before induction of anesthesia indicates that thermoregulatory vasoconstriction was con-
straining metabolic heat to the core thermal compartment. Such constraint is, of course, the only way in
which core temperature could increase while body heat content decreased. Induction of epidural anesthesia
is identified as elapsed time zero. Asterisks (*) identify values differing significantly from time zero.

result is a 3-fold increase in the normal inter-threshold range. Spinal anesthesia re-
duces the shivering threshold in direct relation to the number of dermatomes blocked.
Thus extensive spinal blocks impair central thermoregulatory control more than less
extensive ones. Clinicians can thus anticipate more core hypothermia during extensive
than during restricted blocks.32
Heat flow and distribution during regional anesthesia is comparable to general an-
esthesia. Core hypothermia during the first hour after induction of epidural anesthesia
results largely from redistribution of body heat from the core thermal compartment
to the distal legs. Even after three hours of anesthesia, redistribution remains the ma-
jor cause of core hypothermia. Redistribution contributes proportionately more to
core hypothermia than previously reported during general anesthesia because meta-
bolic rate was maintained during epidural anesthesia. Despite the greater fractional
contribution of redistribution, epidural anesthesia decreases core temperature half
as much as general anesthesia because metabolic rate was maintained and the arms
remained vasoconstricted.31
Furthermore thermoregulatory responses during regional anesthesia are delayed
even further in the elderly patient population. Not only do regional anesthetics delay
the vasoconstriction and shivering threshold, but they also decrease the gain and max-
imum intensity of shivering. Clinically, regional anesthetic-induced thermoregulatory
inhibition is frequently compounded by concomitant administration of sedatives and/
or general anesthesia (Table 1).16
Although regional anesthesia typically causes core hypothermia, patients often feel
warmer after induction of anesthesia.33,34 Increased thermal comfort, like inhibition of
autonomic defenses, presumably results from the thermoregulatory system
 Physiology of Thermoregulation 641

Table 1. Thresholds During Spinal and Epidural Anesthesia in Volunteers.

No Anesthesia Epidural Spinal
Sweating 36.9  0.2 37.1  0.3 37.4  0.3*
Vasoconstriction 36.6  0.2 36.5  0.2 36.7  0.3
Shivering 35.7  0.4 35.4  0.5 35.5  0.5
Sweating-to-Constriction 0.3  0.1 0.6  0.2* 0.6  0.1*
Vasoconstriction-to-Shivering 0.9  0.4 1.2  0.5* 1.2  0.4*
Sweating-to-Shivering 1.2  0.4 1.7  0.5* 1.9  0.4*
The sweating, vasoconstriction, and shivering thresholds under three different conditions: no anesthesia,
epidural anesthesia, and spinal anesthesia. Also shown are the sweating-to-vasoconstriction (interthres-
hold) range, the vasoconstriction-to-shivering range, and the sweating-to-shivering range. All values are
expressed in  C. Asterisks (*) indicate statistically significant differences vs. no anesthesia.

misinterpreting skin temperature as being elevated in the blocked area. Because core-
temperature monitoring remains rare during spinal and epidural anesthesia, and be-
cause patients often fail to recognize that they are cold, undetected hypothermia is
common during regional anesthesia.

Heat balance
Core hypothermia is comparable during regional and general anesthesia.35 As dur-
ing general anesthesia, the initial hypothermia results from a core-to-peripheral re-
distribution of body heat.31 same comments as above In this case, however,
redistribution results primary from peripheral rather than central inhibition of tonic
thermoregulatory vasoconstriction. Although arterio-venous shunt vasodilation is
restricted to the lower body, mass of the legs is sufficient to produce substantial
core hypothermia. Subsequent hypothermia results simply from heat loss exceeding
heat production. Patients given spinal or epidural anesthesia cannot, however, de-
velop a regulated core-temperature plateau because vasoconstriction remains pe-
ripherally impaired.36 Consequently, hypothermia tends to progress throughout
surgery.
Patients becoming sufficiently hypothermic during spinal or epidural anesthesia
shiver. Shivering is disturbing to patients and caregivers, but produces relatively little
heat because it is restricted to the small muscle mass cephalad to the block. Shivering
can be treated by skin-surface warming37, or administration of clonidine (75 mg intra-
venously)38 or meperidine (25 mg intravenously).39 Meperidine is considerably more
effective than equianalgesic doses of other opioids39; its special anti-shivering action
may be mediated by kappa opioid receptors.40
Taken together, these studies indicate that normal thermoregulatory shivering is in-
duced by core hypothermia during regional anesthesia. Hypothermia results when
sympathetic nerve block obliterates tonic thermoregulatory vasoconstriction, allowing
redistribution of heat from the warm core to cooler peripheral tissues. Despite the
core hypothermia and shivering, many patients feel warmer after induction of regional
anesthesia, apparently because perceived skin temperature is elevated. Shivering can
be prevented by maintaining normothermia; as during general anesthesia, redistribu-
tion hypothermia is difficult to treat, but can be prevented by peripheral tissue warm-
ing before induction of anesthesia.
642 A. Kurz

Practice points

 Hypothermia is common during regional anesthesia

 Neuraxial aesthesia causes central, peripheral and behavioral impairment of
thermoregulatory responses
 Neuraxial anesthetics slightly increases the sweating threshold and decreases
the vasoconstriction and shivering threshold thus increasing the interthreshold
range
 Intraoperative hypothermia develops in a very characteristic pattern:
B During induction of neuraxial anesthesia heat redistributes from the
body to the periphery causing an initial drop in core temperature of
1–1.5  C
B During the following 3 hours core temperature linearly decreases due
to heat loss exceeding metabolic heat production
B Sympathectomy prevents the core temperature plateau making patients
under regional anesthesia even more prone to perioperative
hypothermia.

Summary

Temperatures throughout the body are integrated by a thermoregulatory system that

coordinates cold- and warm-defenses. These effective responses usually keep core
temperature within 0.2  C of time-adjusted normal values. General anesthesia pro-
duces marked and dose-dependent inhibition of thermoregulatory control, typically in-
creasing the sweating and vasodilation thresholds approximately 1  C and reducing the
vasoconstriction and shivering thresholds approximately 3  C. As a result, the inter-
threshold range increases roughly 20-fold, leaving patients poikilothermic over an ap-
proximately 4  C range of core temperatures. Regional anesthesia also impairs
thermoregulatory control, producing both peripheral and central inhibition.
The combination of anesthetic-induced thermoregulatory impairment and expo-
sure to cold operating room environments makes most surgical patients hypothermic.
Hypothermia results initially from a core-to-peripheral redistribution of body heat,
and subsequently from heat loss exceeding heat production. Patients becoming suffi-
ciently hypothermic during general anesthesia develop a core-temperature plateau
when arterio-venous shunt tone is reestablished.

REFERENCES

1. Satinoff E. Neural organization and evolution of thermal regulation in mammals – several hierarchically
arranged integrating systems may have evolved to achieve precise thermoregulation. Science 1978; 201:
16–22.
2. Jessen C & Feistkorn G. Some characteristics of core temperature signals in the conscious goat. Amer-
ican Journal of Physiology 1984; 247: R456–R464.
3. Shomaker TS & Bjoraker DG. Measurement offset with liquid crystal temperature indicators (abstract).
Anesthesiology 1990; 73: A425.
4. Marks LI & Gonzalez RR. Skin temperature modfies the pleasantness of thermal stimuli. Nature 1974;
247: 473–475.
 Physiology of Thermoregulation 643

5. Lopez M, Sessler DI, Walter K et al. Rate and gender dependence of the sweating, vasoconstriction, and
shivering thresholds in humans. Anesthesiology 1994; 80: 780–788.
6. Vassilieff N, Rosencher N, Sessler DI et al. The shivering threshold during spinal anesthesia is reduced
in the elderly. Anesthesiology 1995; 83: 1162–1166.
7. Buono MJ & Sjoholm NT. Effect of physical training on peripheral sweat production. Journal of Applied
Physiology 1988; 65: 811–814.
8. Detry J-MR, Brengelmann GL, Rowell LB et al. Skin and muscle components of forearm blood flow in
directly heated resting man. Journal of Applied Physiology 1972; 32: 506–511.
9. Flavahan NA. The role of vascular alpha-2-adrenoceptors as cutaneous thermosensors. News in Physi-
ological Sciences 1991; 6: 251–255.
10. Hales JRS. Skin arteriovenous anastomoses, their control and role in thermoregulation. In Johansen K &
Burggren W (eds.). Cardiovascular Shunts: Phylogenetic, Ontogenetic and Clinical Aspects. Copenhagen:
Munksgaard, 1985, pp. 433–451.
11. Dawkins MJR & Scopes JW. Non-shivering thermogenesis and brown adipose tissue in the human new-
born infant. Nature 1965; 206: 201–202.
12. Brück K, Baum E & Schwennicke HP. Cold-adapative modifications in man induced by repeated short-
term cod-exposures and during a 10-day and -night cold-exposure. Pflügers Archiv 1976; 363: 125–133.
13. Takahashi H, Nakamura S, Shirahase H et al. Heterogenous activity on BRL 35135, a ß3-adrenoceptor
agonist, in thermogenesis and increased blood flow in brown adipose tissue in anaesthetized rats. Clin-
ical and Experimental Pharmacology & Physiology 1994; 21: 539–543.
14. Ricquier R, Casteilla L & Bouillaud F. Molecular studies of the uncoupling protein. FASEB Journal 1991; 5:
2237–2242.
15. Just B, Delva E, Camus Y et al. Oxygen uptake during recovery following naloxone. Anesthesiology 1992;
76: 60–64.
*16. Kurz A, Go JC, Sessler DI et al. Alfentanil slightly increases the sweating threshold and markedly re-
duces the vasoconstriction and shivering thresholds. Anesthesiology 1995; 83: 293–299.
*17. Matsukawa T, Kurz A, Sessler DI et al. Propofol linearly reduces the vasoconstriction and shivering
thresholds. Anesthesiology 1995; 82: 1169–1180.
*18. Xiong J, Kurz A, Sessler DI et al. Isoflurane produces marked and non-linear decreases in the vasocon-
striction and shivering thresholds. Anesthesiology 1996; 85: 240–245.
19. Annadata RS, Sessler DI, Tayefeh F et al. Desflurane slightly increases the sweating threshold, but pro-
duces marked, non-linear decreases in the vasoconstriction and shivering thresholds. Anesthesiology
1995; 83: 1205–1211.
*20. Kurz A, Sessler DI, Annadata R et al. Midazolam minimally impairs thermoregulatory control. Anesthesia
and Analgesia 1995; 81: 393–398.
21. Bissonnette B & Sessler DI. Thermoregulatory thresholds for vasoconstriction in pediatric patients
anesthetized with halothane or halothane and caudal bupivacaine. Anesthesiology 1992; 76: 387–392.
22. Sessler DI, Hynson J, McGuire J et al. Thermoregulatory vasoconstriction during isoflurane anesthesia
minimally decreases heat loss. Anesthesiology 1992; 76: 670–675.
23. Washington D, Sessler DI, Moayeri A et al. Thermoregulatory responses to hyperthermia during iso-
flurane anesthesia in humans. Journal of Applied Physiology 1993; 74: 82–87.
*24. Kurz A, Xiong J, Sessler DI et al. Desflurane reduces the gain of thermoregulatory arterio-venous shunt
vasoconstriction in humans. Anesthesiology 1995; 83: 1212–1219.
*25. Cheng C, Matsukawa T, Sessler DI et al. Increasing mean skin temperature linearly reduces the core-tem-
perature thresholds for vasoconstriction and shivering in humans. Anesthesiology 1995; 82: 1160–1168.
*26. Matsukawa T, Sessler DI, Sessler AM et al. Heat flow and distribution during induction of general an-
esthesia. Anesthesiology 1995; 82: 662–673.
*27. Hynson JM, Sessler DI, Moayeri A et al. The effects of pre-induction warming on temperature and
blood pressure during propofol/nitrous oxide anesthesia. Anesthesiology 1993; 79: 219–228.
28. Kurz A, Sessler DI, Christensen R et al. Heat balance and distribution during the core-temperature pla-
teau in anesthetized humans. Anesthesiology 1995; 83: 491–499.
29. Kurz A, Sessler DI, Narzt E et al. Morphometric influences on intraoperative core temperature
changes. Anesthesia and Analgesia 1995; 80: 562–567.
*30. Kurz A, Sessler DI, Schroeder M et al. Thermoregulatory response thresholds during spinal anesthesia.
Anesthesia and Analgesia 1993; 77: 721–726.
644 A. Kurz

31. Matsukawa T, Sessler DI, Christensen R et al. Heat flow and distribution during epidural anesthesia.
Anesthesiology 1995; 83: 961–967.
*32. Leslie K & Sessler DI. Reduction in the shivering threshold is proportional to spinal block height. An-
esthesiology 1996; 84: 1327–1331.
33. Sessler DI & Ponte J. Shivering during epidural anesthesia. Anesthesiology 1990; 72: 816–821.
34. Glosten B, Sessler DI, Faure EAM et al. Central temperature changes are not perceived during epidural
anesthesia. Anesthesiology 1992; 77: 10–16.
35. Bredahl C, Hindsholm KB & Frandsen PC. Changes in body heat during hip fracture surgery: A com-
parison of spinal analgesia and general anesthesia. Acta Anaesthesiologica Scandinavica 1991; 35: 548–552.
36. Shimosato S & Etsten BE. The role of the venous system in cardiocirculatory dynamics during spinal and
epidural anesthesia in man. Anesthesiology 1969; 30: 619–628.
37. Sharkey A, Lipton JM, Murphy MT et al. Inhibition of postanesthetic shivering with radiant heat. Anes-
thesiology 1987; 66: 249–252.
38. Capogna G & Celleno DIV. Clonidine for post-extradural shivering in parturients: a preliminary study.
British Journal of Anaesthesia 1993; 71: 29–295.
39. Guffin A, Girard D & Kaplan JA. Shivering following cardiac surgery: Hemodynamic changes and rever-
sal. Journal of Cardiothoracic and Vascular Anesthesia 1987; 1: 24–28.
40. Kurz M, Belani K, Sessler DI et al. Naloxone, meperidine, and shivering. Anesthesiology 1993; 79:
1193–1201.