The Effect of Postoperative Skin-Surface Warming on Oxygen Consumption and the Shivering Threshold

Summary

Cutaneous warming is reportedly an effective treatment for shivering during epidural and after general anaesthesia. We quantified the efficacy of cutaneous warming as a treatment for shivering. Unwarmed surgical patients (final intraoperative core temperatures ≈35°C) were randomly assigned to be covered with a blanket (n=9) or full-body forced-air cover (n=9). Shivering was evaluated clinically and by oxygen consumption. Forced-air heating increased mean-skin temperature (35.7±0.4 °C vs. 33.2±0.8°C, P< 0.0001) and lowered core temperature at the shivering threshold (35.7±0.2 °C vs. 36.4±0.2°C, P< 0.0001). Active warming improved thermal comfort and significantly reduced oxygen consumption from 9.7±4.4 to 5.6±1.9 mL·min⁻¹·kg⁻¹(P=0.038). However, duration of shivering was similar in the two groups (37±11 min [warming] and 36±10 min [control]). Core temperature thus contributed about four times as much as skin temperature to control of shivering. Cutaneous warming improved thermal comfort and reduced metabolic stress in postoperative patients, but did not quickly obliterate shivering.

Perioperative hypothermia causes severe complications and prolongs hospitalization. It is also associated with impaired drug metabolism, delayed postanaesthesia unit discharge, negative nitrogen balance, thermal discomfort, and shivering-like tremor [1]. Some postoperative shivering-like tremor has an abnormal clonic pattern [2] and some is non-thermoregulatory [3]; however, most is normal shivering in response to hypothermia, which can be prevented by maintaining intraoperative normothermia [4] or administering pharmacologic treatments [5–9].

Thermal input is integrated at numerous levels within the nervous system, but the hypothalamus is the dominant controller [10]. Hypothalamic responses are based on thermal input from the core component (brain, spinal cord, deep tissues) and skin surface [11, 12]. Mean skin temperatures contribute ≈20% to the control of shivering with [13] or without [14] general anaesthesia. An alternative to pharmacologic treatment of shivering in hypothermic patients is, thus, skin-surface warming [15]. This concept was confirmed in patients who shivered after surgeries mostly conducted with neuraxial anaesthesia [16, 17]. However, in these studies, skin temperature [16] or core temperature [17] is lacking; therefore, it is impossible to determine the extent to which skin warming compensates for core hypothermia.

Forced-air warming systems increase skin temperature 2 to 3°C [18]. However, this amount of skin warming should compensate for a reduction of only 0.4 to 0.6°C in core body temperature [14]. Reports that cutaneous warming is an effective treatment for perioperative shivering thus contrast with studies in volunteers showing that skin temperature contributes only 20% to control of shivering. We therefore tested the hypothesis that cutaneous warming improves thermal comfort and reduces the intensity and duration of postoperative shivering. A secondary goal was to quantify the relative contributions of mean-skin and core temperature in postoperative patients.

Methods

With approval from the Ethics Committee of Hôpital Ambroise Paré and informed consent, we studied 18 ASA 1 and 2 male patients. They were aged 18-40 years and were recovering from knee or shoulder arthroscopy. All were hospitalized in an ambulatory unit. None was obese; febrile; taking beta-receptor blockers or alpha-2 receptor agonists; or had a history of thyroid or neuromuscular disease, dysautonomia, paracetamol allergy, or Raynaud's syndrome. Approximately two-thirds of the patients smoked cigarettes.

Protocol

Anaesthesia was induced, without premedication, with 2–3 mg/kg propofol and 2 μg/kg remifentanil; the patients’ tracheas were intubated without administration of a muscle relaxant. Ventilation was mechanically controlled (end-tidal PCO₂ of 30–35 mmHg). The fresh gas flow was maintained at 2 L/min via a semi-closed circle system without airway heating or humidification. Anaesthesia was maintained with isoflurane (0.8-1.2%) in 65% nitrous oxide. Isoflurane was altered to maintain mean arterial blood pressure within 20% of baseline. No muscle relaxant or additional opioid was given. After surgery, isoflurane was discontinued.

To avoid the confounding effects of postoperative pain on shivering-like tremor, an interscalene block was performed under nerve-stimulator guidance with 30 ml of 0.5% bupivacaine in patients scheduled for shoulder arthroscopy. Similarly, 20 ml of bupivacaine was administered intra-articularly at the end of the surgery in patients having knee arthroscopy. Passive insulation was restricted to a single layer of surgical draping. Patients were not actively warmed during anaesthesia unless tympanic membrane temperature decreased to less than 35°C. Intravenous fluids were not warmed.

Patients were transported to the postanaesthetic care unit where they were randomly assigned to active rewarming (forced-air) or passive insulation. Randomization was based on computer-generated codes that were maintained in sequentially numbered opaque envelopes. The forced-air cover (Bair Hugger; Augustine Medical, Eden Prairie, MN) was positioned over the body immediately after the arrival in the unit. The warmer was set to “high” (≈43°C). Patients assigned to passive insulation were covered with a single cotton blanket. Patients were extubated when spontaneous respiration was sufficient. Postoperative pain was treated, when necessary, by intravenous proparacetamol (2 g).

Measurements

Ambient temperature was measured by a thermocouple positioned well away from heat-producing equipment. Core temperature was measured at the tympanic membrane. Mean skin temperature (MST) was calculated from ten sites, using the formula based on surface area [19]:

$$\begin{array}{l}\text{MST}(°\text{C})=0.06\cdot {\text{T}}_{\text{Forehead}}+0.09\cdot {\text{T}}_{\text{Arm}}+0.06\cdot {\text{T}}_{\text{Forearm}}+0.045\cdot {\text{T}}_{\text{Hand}}+\\ 0.19\cdot {\text{T}}_{\text{Back}}+0.095\cdot {\text{T}}_{\text{Chest}}+0.095\cdot {\text{T}}_{\text{Abdomen}}+0.19\cdot {\text{T}}_{\text{Thigh}}+0.115\cdot {\text{T}}_{\text{Calf}}+0.06\cdot {\text{T}}_{\text{Foot}}\end{array}$$

The probes were positioned on the non-operative side of the patient. Temperatures, measured using Ellab thermometers and probes (Ellab, Inc., Copenhagen, Denmark), were recorded at 10-minute intervals during surgery. In the postanaesthetic care unit, all temperatures were electronically recorded at 10-second intervals until the end of the study.

Fingertip blood flow was evaluated using forearm minus fingertip, skin-surface temperature gradients [20]. Gradients exceeding 0°C indicated vasoconstriction because this gradient corresponds to onset of the core-temperature plateau [21].

Shivering was evaluated by clinical assessment and oxygen consumption (DeltaTrac metabolic monitor; Datex, Inc., Finland). Measurements began within 5 minutes of extubation. The system was used in canopy-mode with measurements averaged over 1-minute intervals and recorded every minute. This device has been extensively validated [22, 23].

Shivering was evaluated clinically using a 3-point scale: 0 = no shivering, 1 = moderate shivering, 2 = intense shivering. Absence of visible shivering and a 20% decrease in oxygen consumption, lasting at least 5 minutes, identified shivering cessation. An investigator blinded to skin and patient core temperature determined the time of shivering cessation.

Pain and cold sensation were assessed by the patients at 10-minute intervals, starting shortly after extubation, using a 5-point verbal scale: 0 = none to 4 = intense.

Heart rate and pulse oximeter saturation were monitored continuously. Arterial pressure was determined at 5-minute intervals during surgery and at 10-minute intervals postoperatively. End-tidal carbon dioxide and isoflurane concentrations were quantified using a Cato gas analyzer (Dräger, Germany). Concentrations were sampled from a catheter inserted several cm through a set of nasal prongs; aspirated gas was returned to the Deltatrac canopy.

Data Analysis

The sample size was based on the expected differences in the core temperature at shivering cessation between unwarmed and warmed patients. Assuming a standard deviation of 0.2 °C for the unwarmed patients, 9 patients per group were required to detect a mean difference of 0.4°C or larger between groups with at least 80% power.

The equations for calculation of the proportionality constant (β) have been described previously [14]. Briefly, the contribution of mean skin temperature to central control of shivering can be expressed by the equation:

$$Threshol{d}_{MBT}=\beta T_{skin}+(1-\beta )T_{CORE}$$

where threshold_MBT is the shivering threshold in terms of mean body temperature, T_skin is mean skin temperature, and T_core is core temperature, all in degrees Celsius. On the other hand, the relation between core temperature and mean skin temperature is linear at the time of shivering cessation and a regression equation could be established (T_core = ST_skin + K , where S is the slope and K is the intercept). By rearranging the equations, the proportionality constant is calculated: $\beta =\frac{\text{S}}{\text{S}-1}$)

The derivation of these equations has been described, and the linear assumption is justified [14][24]. The correlation coefficient (r) for the skin-versus-core regression indicated the extent to which the shivering threshold was a linear function of skin and core temperatures.

Demographic and morphometric characteristics of the patients, as well as their anaesthesia management, hemodynamic responses, maximal oxygen consumption recorded during shivering, and thermoregulatory responses were compared using two-tailed, unpaired t tests. Cold sensation and pain scores were compared using Fisher exact tests. Results are presented as means ± SDs unless otherwise indicated; P < 0.05 was considered statistically significant.

Results

Morphometric and demographic characteristics of the patients in the groups were similar (Table 1). Duration of surgery, anaesthetic management, and intraoperative hemodynamic responses were also similar in the groups. Core temperatures were near 35°C at the end of surgery; mean skin temperatures were near 33°C (Table 2). Only one of nine unwarmed patients and three of nine warmed patients required proparacetamol treatment for postoperative pain.

Table 1.

Morphometric and Demographic Characteristics.

	Unwarmed	Warmed
Age (yr)	29 ± 5	32 ± 7
Height (cm)	176 ± 6	179 ± 6
Weight (kg)	68 ± 9	74 ± 9
Arthroscopy (Knee/Shoulder)	7/2	6/3

Data are presented as means ± SDs. There were no statistically significant differences between the groups.

Table 2.

Environmental, Anaesthetic Management, Haemodynamic, Respiratory, and Thermoregulatory Data.

	Unwarmed	Warmed
Pre-operative Core Temperature (°C)	36.6 ± 0.2	36.6 ± 0.2
Pre-operative Mean Skin Temperature (°C)	33.1 ± 0.8	33.0 ± 0.9
Duration of Surgery (min)	83 ± 33	91 ± 41
Propofol (mg/kg)	5.0 ± 2.1	3.4 ± 0.8
Remifentanil (μg/kg)	2.7 ± 1.1	2.4 ± 0.7
Intraoperative End-tidal PCO2 (mmHg)	34 ± 1	34 ± 3
Intraoperative End-tidal Isoflurane (%)	1.2 ± 0.3	1.1 ± 0.3
Intraoperative Mean Arterial Pressure (mmHg)	68 ± 8	73 ± 9
Intraoperative Heart Rate (beats per min)	71 ± 17	71 ± 13
Administered Intraoperative Fluid (L/h)	0.6 ± 0.4	0.5 ± 0.3
Final Intraoperative Core Temperature (°C)	35.1 ± 0.4	35.2 ± 0.4
Final Intraoperative Mean Skin Temperature (°C)	32.7 ± 0.7	32.9 ± 0.5
Postoperative Ambient Temperature (°C)	20.8 ± 0.9	20.8 ± 0.8

Data are presented as means ± SDs. There were no statistically significant differences between the groups.

All patients shivered, and all were vasoconstricted (skin-temperature gradient >0°C) during shivering. Core temperature, hemodynamic responses, and expired gas concentrations were comparable when shivering began. Mean skin temperature, by design, was significantly greater in patients assigned to active warming. During shivering, maximal VO2 was significantly greater in the unwarmed than warmed patients: 9.7 ± 4.4 versus 5.6 ± 1.9 mL·min⁻¹·kg⁻¹ (P=0.021, Table 3). Hemodynamic, respiratory, and metabolic data were comparable at shivering cessation; however, skin temperature was significantly greater and core temperature was significantly less in the warmed patients (Table 3). Unwarmed patients felt significantly colder during shivering and at shivering cessation than those who were actively warmed (Table 4). Pain scores in the two groups did not differ significantly at any time. Time to shivering cessation was similar in the groups.

Table 3.

Haemodynamic, Respiratory, Metabolic, and Thermoregulatory Data during Shivering.

	Unwarmed	Warmed
Core Temperature at Onset of Shivering (°C)	35.2 ± 0.4	35.2 ± 0.3
Mean Skin Temperature at Onset of Shivering (°C)	32.7 ± 0.9	34.2 ± 0.7	P=0.013
End-tidal PCO2 at Onset of Shivering (mmHg)	41 ± 8	42 ± 3
End-tidal Isoflurane at Onset of Shivering (%)	0.23 ± 0.12	0.24 ± 0.34
Mean Arterial Pressure at Onset of Shivering (mmHg)	97 ± 6	96 ± 7
Heart Rate at Onset of Shivering (beats per min)	85 ± 18	82 ± 13
Maximal VO2 during Shivering (ml·kg−1·min−1)	9.7 ± 4.4	5.6 ± 1.9	P=0.038
Time to Shivering Cessation (min)	37 ± 11	36 ± 10
Core Temperature at Shivering Cessation (°C)	36.4 ± 0.2	35.7 ± 0.2	P<0.0001
Skin Temperature at Shivering Cessation (°C)	33.2 ± 0.8	35.7 ± 0.4	P<0.0001
Skin-Temperature Gradient at Shivering Cessation (°C)	3.3 ± 1.1	2.9 ± 1.6
Mean Arterial Pressure at Shivering Cessation (mmHg)	93 ± 14	86 ± 33
Heart Rate at Shivering Cessation (beats per min)	80 ± 14	77 ± 11
End-tidal PCO2 at Shivering Cessation (mmHg)	37 ± 4	36 ± 4
End-tidal Isoflurane at Shivering Cessation (%)	0.02 ± 0.01	0.02 ± 0.02
VO2 at Shivering Cessation (ml·kg−1·min−1)	4.3 ± 2.2	2.6 ± 1.2

Data are presented as means ± SDs.

Table 4.

Thermal Sensation and Pain.

		Unwarmed	Warmed
Cold Sensation During Shivering (# of patients)*	None	1	1
	Mild	0	3
	Moderate	1	2
	Considerable	0	3
	Intense	7	0
Cold Sensation at Shivering Cessation (# of patients) †	None	1	6
	Mild	3	3
	Moderate	3	0
	Considerable	1	0
	Intense	1	0
Pain During Shivering (# of patients)	None	7	5
	Mild	1	1
	Moderate	1	2
	Considerable	0	1
	Intense	0	0
Pain at Shivering Cessation (# of patients)	None	4	5
	Mild	3	2
	Moderate	1	2
	Considerable	1	0
	Intense	0	0

Cold sensation and pain were assessed using a 5-point verbal scale. The scores were statistically different for cold sensation during shivering (*P=0.001) and cold sensation at shivering cessation (^†P=0.048) in the unwarmed and warmed patients.

Warming the skin with forced-air heating increased mean-skin temperature ≈2°C which nearly halved oxygen consumption and improved thermal comfort. However, active warming reduced the shivering threshold only ≈0.4°C. Mean-skin and core temperatures showed a linear relation at shivering cessation: T_core = −0.20 T_Skin + 42.8, r = 0.71. The cutaneous contribution to postoperative shivering, ß, thus equalled 17% (95% confidence interval: 8–23%; Fig. 1).

Fig. 1.

Mean-skin and core temperatures at shivering cessation. There was a linear relation between mean-skin (Tskin) and core (Tcore) temperatures in °C at the shivering threshold: Tcore = -0.2.Tskin + 42.8, r = 0.71. This corresponds to a proportionality constant (ß) of 17%. The straight line indicates the regression slope; the dashed lines show the 95% confidence intervals.

Discussion

Although active warming improved thermal comfort and nearly halved oxygen consumption, shivering persisted for about 36 minutes with or without forced-air warming. In contrast, pharmacologic treatments usually stop shivering rapidly [25]. Cutaneous heating may nonetheless be a useful adjunct even in patients who are treated pharmacologically.

Although shivering lasted about the same amount of time in each group, it was more intense in the unwarmed patients. Consequently, the unwarmed patients stopped at a higher core temperature because they generated more heat and more rapidly rewarmed the core. Shivering is an effective way of increasing core temperature [26], but is probably sub-optimal in postoperative patients because they find shivering uncomfortable and shivering per se may increase pain by stretching surgical incisions. Pain scores were similar in each of our groups, and few patients reported more than mild pain. The results may have differed considerably in surgical patients who have greater pain because pain per se [27] aggravates nonthermoregulatory shivering-like tremor [3].

The relative contribution of skin temperature to the control of shivering in our patients was 17%, which does not appreciably differ from 19 ± 8% without anaesthesia [14] or 18 ± 10% during isoflurane anaesthesia [13]. Available data thus suggest that the relationship between skin and core contributions to control of shivering remains constant under a variety of circumstances. That the proportionality constant is so well preserved suggests that the relative importance of skin and core inputs to thermoregulatory control is a fundamental characteristic of the system.

A potential difficulty with using cutaneous warming to treat postoperative shivering is that forced-air [18] and carbon-fibre [28] heating increase skin temperature roughly 2-3°C. Since skin temperature contributes about 20% to control of shivering, a 2°C cutaneous temperature increase will stop shivering only when core temperature is within ≈0.4°C of the shivering threshold.

The shivering threshold in unanaesthetised adults is 35.5 ± 0.5°C [29]. However, the threshold is markedly decreased by anaesthetics [29], and opioids [24]. In this study, residual isoflurane concentrations were barely detectable and did not differ in the two groups. Skin-surface warming is therefore likely to prove effective when core temperature exceeds 35°C and patients have considerable residual anaesthesia. In contrast, the treatment will likely fail when patients are considerably colder and unprotected by residual anaesthesia. Clinicians thus need to evaluate patients who shiver postoperatively and make individualized decisions about the likely efficacy of cutaneous warming. It is probable that cutaneous warming was effective in previous reports [16, 17] because the patients were nearly normothermic.

A limitation of our protocol is that the upper chest and face may be more sensitive than other areas of the body [30]. Facial temperature was included in our current study and the mean skin temperature calculations discussed above [13], but the face was usually excluded from active warming. A given increase in mean skin temperature will therefore be proportionately more effective when facial warming is included. This may be an important distinction between our current results and those reported previously [15-17]. Although we evaluated numerous potentential confounding factors in Tables 1 and 2, we elected not to compensate for multiple comparisons because we were largely concerned about type 2 statistical errors.

In summary, active cutaneous warming improved thermal comfort and markedly reduced the intensity of shivering. However, the duration of shivering was similar with or without active cutaneous warming. Forced-air heating increased mean-skin temperature ≈2°C, an amount that reduced the shivering threshold by only ≈0.4°C.