Multivariate Determinants of Early Postoperative Oxygen Consumption in Elderly Patients: Effects of Shivering, Body Temperature, and Gender

With approval from the Committee on Clinical Investigation and written informed consent, 145 patients were enrolled. All three of the following inclusion criteria were required for enrollment: (1) age greater than 60 yr; (2) scheduled lower-extremity vascular, thoracic, or abdominal surgery with an overnight stay in the intensive care unit; and (3) the presence of two or more risk factors for coronary artery disease according to criteria used previously. [10]This group of elderly patients was chosen because they represent those at greatest risk for perioperative complications. Exclusion criteria were a preoperative tympanic temperature of less than 36 degrees Celsius or more than 38 degrees Celsius measured by the infrared technique (Thermoscan, San Diego, CA) or a known thyroid disorder. Patients requiring postoperative mechanical ventilation (n = 34) were excluded from the analysis because they were given a combination of neuromuscular blocking drugs, opioids, and benzodiazepines for sedation, all of which tend to reduce shivering and oxygen consumption. The remaining 111 patients were included. The criteria for removal of the endotracheal tube at the end of the surgical procedure are described below.

Premedication included midazolam (as much as 5 mg) or morphine (as much as 0.1 mg/kg) or both given intramuscularly. A general or an epidural anesthetic regimen was used, in most cases determined by the surgical procedure. General anesthesia, which was chosen for patients undergoing abdominal or thoracic surgical procedures, included intravenous thiopental (4-6 mg/kg); intravenous fentanyl (3-10 micro gram/kg) or morphine (0.1-0.2 mg/kg) or both; intravenous pancuronium or vecuronium; nitrous oxide (as much as 70%); and isoflurane or enflurane (0.2-2.0%). The patients' tracheas were intubated during surgery, and the endotracheal tube was removed at the end of the surgical procedure unless standard extubation criteria (return of consciousness, reversal of neuromuscular blockade, and ability to follow commands and ventilate spontaneously) were not met.

Epidural anesthesia was used for patients undergoing lower-extremity vascular surgery unless contraindicated (anticoagulation, spinal fusion, or patient refusal), in which case general anesthesia was used. Epidural anesthesia was given by delivering lidocaine 2% or bupivacaine 0.5-0.75% through a lumbar catheter to achieve a T6-T10 sensory level. Midazolam (1 mg) or fentanyl (50 micro gram) or both were given intravenously as needed during the intraoperative period to keep the patients comfortably sedated while maintaining responsiveness to verbal commands.

Postoperative analgesia was delivered according to protocol. For patients receiving general anesthesia (n = 71), analgesia was provided during the initial 18-24 h after surgery with intravenous morphine sulfate delivered by patient-controlled analgesia with a background infusion of 1 mg/h, a bolus dose of 1 mg, and a lockout interval of 6 min. Patients receiving epidural anesthesia (n = 23) were given an infusion of bupivacaine 0.0625% with fentanyl (5 micro gram/ml) delivered by patient-controlled analgesia, with a background infusion of 2-4 ml/h, a bolus dose of 2 ml, and a lockout interval of 10 min. For patients undergoing thoracic surgical procedures (n = 17), postoperative analgesia was provided by an intrathecal dose of morphine (0.5 mg) given shortly before induction of anesthesia. This general-and-regional anesthetic technique is referred to hereafter as "combined" anesthesia.

To assess the effects of body temperature on shivering and metabolic demands, patients were randomly assigned to receive either routine thermal care (n = 52) or forced-air warming (n = 59) in the perioperative period. Randomization was stratified by type of surgery.

Routine thermal care was delivered according to the following protocol. This care is referred to as "routine" because it is the normal practice at our institution when extra warming measures are not believed to be necessary. The thermostat in the operating room was set to [nearly equal] 21 degrees Celsius. Intravenous fluids and blood were warmed (BW-5, Fenwal, Deerfield, IL). A heat-moisture exchanger (Thermovent 600, Portex, Keene, NH) was used in the respiratory circuit for patients receiving general anesthesia. After the surgical field had been prepared, the patient was covered above and below the field with one layer of paper drapes. In the postoperative period, one or two warmed cotton blankets were placed over the patient.

Patients in the forced-air warming group were treated as follows. As in the routine care group, the thermostat in the operating room was set to [nearly equal] 21 degrees Celsius; fluids and blood were warmed (BW-5, Fenwal); and a heat-moisture exchanger was used (Thermovent 600, Portex) in the respiratory circuit for patients receiving general anesthesia. Depending on the surgical site, an upper- or lower-body forced-air warming blanket (Warm Touch, Mallinckrodt Medical, St. Louis, MO) was placed over the patient and connected to a forced-air blower (Warm Touch, Mallinckrodt Medical). During the first 2 h of postoperative care, a full-body blanket covered the patient's legs and trunk. During the intraoperative and postoperative periods, both the temperature and air flow were set to "high" or adjusted to "medium" to maintain core temperature at or near 37 degrees Celsius. If core temperature exceeded 37 degrees Celsius, the blower was turned off and the blanket was left in place. After the initial 2 h after surgery, care was similar to that used for patients in the routine thermal care group.

Core temperature was measured with a tympanic membrane thermocouple probe (Mallinckrodt Medical) inserted until the patient reported a slight scratching sound. Core temperature also was measured in the urinary bladder with a catheter that incorporated a thermocouple probe (Mallinckrodt Medical). Urinary bladder temperature was used as the core temperature if the tympanic probe became dislodged during the study; this change was necessary in fewer than 5% of patients. Skin-surface temperatures were measured on the anterior chest, the lateral mid-upper arm, the anterior radial side of the mid-forearm, the index finger (opposite the nail bed), the lateral mid-thigh, and the lateral mid-calf. Mean skin temperature was calculated as 0.3 x (chest temperature + upper arm temperature) + 0.2 x (thigh temperature + calf temperature). [11]Mean body temperature was calculated as (0.66 x core temperature) + (0.34 x mean skin temperature). [12]Temperature data were recorded on admission to the postanesthesia care unit (PACU) and at 30 and 90 min postoperatively.

At 30 and 90 min postoperatively, total-body oxygen consumption was measured by indirect calorimetry (Deltatrac, Sensormedics, Anaheim, CA). Supplemental oxygen was discontinued for at least 5 min before these measurements were taken. A clear plastic canopy was placed over the patient's head, and measurements were obtained every 1 min until a steady state was achieved and maintained. Mean values of five or more steady-state measurements were averaged to obtain total-body oxygen consumption (milliliters per minute). Oxygen consumption data are reported in units normalized for body surface area (in squared meters), calculated as (height [centimeters] x weight [kilograms]) sup -2/3,600. To account for gender differences in body habitus in the comparison between men and women, oxygen consumption data were analyzed as follows: raw data (milliliters per minute), data normalized by weight (kilograms), and data normalized by body surface area (squared meters).

A four-point scale was used to describe shivering at specific intervals postoperatively: 0 = no shivering; 1 = occasional mild tremors of the jaw and neck; 2 = intensive tremors of the chest; 3 = intermittent vigorous generalized tremor; and 4 = continuous violent muscle activity. [13]Scores were assigned by the nursing staff in the PACU every 15 min for the 1st h, every 30 min for the 2nd h, and hourly thereafter until the morning of the 1st postoperative day. These nurses were not blinded with regard to thermal management. The nurses were trained by one of the coinvestigators (S.M.F.) to assess shivering with this scale. The same coinvestigator periodically monitored and retrained the nursing staff to ensure that shivering was evaluated consistently.

Chi-squared and unpaired Student's t tests were used to compare the two groups demographically. Univariate (chi-squared and unpaired Student's t tests) and multivariate (multiple logistic regression) analyses were used to determine the clinical variables that predicted shivering at 30 and 90 min postoperatively. Univariate (chi-squared and unpaired Student's t tests) and multivariate (multiple linear regression) analyses were used to determine the clinical variables that predicted total-body oxygen consumption (milliliters per minute per squared meters) at 30 and 90 min postoperatively. Variables included in the regression models were those that either correlated with outcome on univariate testing (P < 0.20) or that have been correlated with the outcome in previous studies. These variables were sex, core temperature, mean body temperature, shivering, thermal management group, anesthetic technique (regional or general), and age. Regressions were performed by the backward elimination method, retaining variables with P < 0.10. [14]Finally, the relation between mean body temperature and total-body oxygen consumption was assessed by simple linear regression. The direction of this relation was determined by the P value, which was used to determine if the slope was different than zero. All data are given as mean plus/minus SEM, with significance defined as P < 0.05.

Demographic data shown in Table 1compare routine thermal care with forced-air warming groups, shivering with nonshivering patients, and men with women. The routine thermal care and forced-air warming groups were similar with regard to anesthetic technique, surgical procedure, age, sex, height, weight, body surface area, and preoperative core temperature. On admission to the PACU, core temperature was approximately 1.4 degrees Celsius less in the routine thermal care group than in the forced-air warming group. Shivering patients were defined as those with shivering scores greater than 0 at either 30 or 90 min postoperatively, when total-body oxygen consumption was measured. Shivering and nonshivering patients were comparable for all listed demographic parameters except core temperature, which was less in the shivering patients. Men and women were comparable in assignment to forced-air warming, anesthetic technique, surgical procedure, age, and preoperative core temperature. Men, however, were characterized by greater height, weight, and body surface area. Mean age of all patients enrolled was 70 plus/minus 1 yr. The average opioid requirement for analgesia during the initial 24 h after surgery for those receiving intravenous patient-controlled analgesia was 30 plus/minus 3 mg morphine.

Univariate and multivariate predictors of shivering were determined at 30 and 90 min postoperatively (Table 2). Thermal treatment was significant at both times, indicating that patients in the forced-air warming group were less likely to shiver. Male sex was associated with an increased incidence of shivering by univariate testing at 30 min. Decreased core temperature correlated with shivering at 90 min postoperatively by univariate analysis, but age and anesthetic technique were not correlated with shivering at either 30 or 90 min.

Univariate and multivariate predictors of total-body oxygen consumption were determined at 30 and 90 min postoperatively (Table 3). Shivering was correlated with increased oxygen consumption at 30 min, but not at 90 min. Increased mean body temperature was correlated with increased oxygen consumption. Men had a greater oxygen consumption than women. Thermal treatment, anesthetic technique, and age were not correlated with oxygen consumption. Mean total-body oxygen consumption was 139 plus/minus 5 and 132 plus/minus 5 ml *symbol* min sup -1 *symbol* m sup -2 in the routine thermal care and forced-air warming groups, respectively, at 30 min (P = 0.33) and was 126 plus/minus 3 and 125 plus/minus 3 ml *symbol* min sup -1 *symbol* m sup -2 in the routine thermal care and forced-air warming groups, respectively, at 90 min (P = 0.73).

On admission to the PACU, 25% of patients in the routine thermal care group were noted to shiver (score > 0), whereas only one patient (2%) in the forced-air warming group shivered (P = 0.0005) (Figure 1). When a score greater than 2 (whole-body tremor) was used to define shivering, 8% of the routine thermal care group and no patient in the forced-air warming group shivered (P = 0.09). At 6 h postoperatively no patient in either group remained shivering.

A gender comparison for core temperature, incidence of shivering, and total-body oxygen consumption is shown in Table 4. Men had a significantly greater incidence of shivering, 16% versus 4% at 30 min (P = 0.03), despite a mean core temperature that was identical to that measured in the women. At 90 min no significant gender differences in core temperature or the incidence of shivering were noted. Total-body oxygen consumption was analyzed as raw data (milliliters per minute), and as normalized data by weight (kilograms), and by body surface area (squared meters). The men had significantly greater oxygen consumption than the women by all comparisons except for the 90-min measurement normalized by weight.

Total-body oxygen consumption was compared in shivering and nonshivering patients at 30 and 90 min postoperatively. At 30 min oxygen consumption was greater in shivering (180 plus/minus 16 ml *symbol* min sup -1 *symbol* m sup -2) than in nonshivering (130 plus/minus 2 ml *symbol* min sup -1 *symbol* m sup -2) patients, a difference of 38% (P = 0.0001). At 90 min oxygen consumption was similar between shivering (130 plus/minus 9 ml *symbol* min sup -1 *symbol* m sup -2) and nonshivering patients (124 plus/minus 3 ml *symbol* min sup -1 *symbol* m sup -2) (P = 0.50). The highest measured oxygen consumption in any single shivering patient was 292 ml *symbol* min sup -1 *symbol* m sup -2, a value 125% greater than the mean for the nonshivering group.

Simple linear regression was used to assess the relation between total-body oxygen consumption and mean body temperature. The 30- and 90-min data were pooled. When all patients were included the slope of the regression line was positive, indicating a directly proportional relation between temperature and oxygen consumption (P = 0.01). When the 15 patients who shivered at 30 or 90 min postoperatively were excluded from the analysis the slope of the regression line was positive and increased (P = 0.0003).

In this series of elderly patients undergoing surgery, total-body oxygen consumption was 38% greater in shivering than in nonshivering patients, a difference that was statistically significant, but much less than the 300-400% that is often cited. [5-8]Even when data from shivering patients were included, total-body oxygen consumption was directly proportional to body temperature, perhaps because of the low incidence and low intensity of shivering and the reduction in basal metabolic rate with hypothermia. Men had a greater incidence of clinically recognizable early postoperative shivering compared with women, despite similar core temperatures. Male sex was also correlated with increased oxygen consumption, which may reflect their greater lean body mass, increased shivering, or both.

The relatively small difference in total-body oxygen consumption (38%) that we measured in shivering and nonshivering patients is less than has been reported previously in studies of younger shivering patients (100-730%) [1,2]or volunteers who shivered during cold exposure (200-300%). [15]Most likely, the advanced age of our patients (mean 70 yr) contributed to the attenuated shivering response and the relatively small increase in oxygen consumption that was measured. This finding is consistent with previous studies on thermoregulation in elderly persons. Kurz et al. [16]demonstrated a lower vasoconstriction threshold during general anesthesia in elderly patients (age 60-80 yr) undergoing surgery, indicating a less active thermoregulatory response. Frank et al. [17]found that during epidural anesthesia in elderly patients, vasoconstriction did not occur despite core temperatures less than those of younger patients with vasoconstriction. Elderly patients have been shown to rewarm more slowly in the postoperative period compared with younger patients. [18,19]In studies in volunteers, elderly persons have been shown to have attenuated vasoconstriction [20]and shivering [21]responses during cold exposure.

The administration of opioid analgesia in the current study may provide an explanation for the relatively small increase in total-body oxygen consumption with shivering. Although meperidine appears to reduce shivering to a greater extent than other opioids, [2,5]fentanyl and morphine also have significant antishivering effects. [5,22]All patients in the current study received fentanyl or morphine or both intraoperatively or postoperatively as part of the general anesthetic protocol, for sedation during regional anesthesia, or for postoperative analgesia. In previous studies that measured metabolic demands in shivering patients anesthetic regimens were not controlled by protocol. In these studies patients may or may not have received opioid analgesia.

In 1968, Bay et al. [1]measured oxygen consumption postoperatively in 11 shivering and 8 nonshivering patients after general anesthesia. These persons were relatively young (mean age 47 yr), and received little or no opioid analgesia intra- or postoperatively. Mean total-body oxygen consumption was 531 plus/minus 76 ml/min and 230 plus/minus; 14 ml/min in the shivering and nonshivering groups respectively, representing an approximately 130% greater mean oxygen consumption in the shivering patients. One patient in the shivering group had a 486% increase in oxygen consumption above baseline, which may explain the often-quoted "up to 400% increase" in oxygen consumption with shivering. These investigators may have overestimated the metabolic demands of postoperative shivering for several reasons. First, oxygen consumption was measured by collection of expired respiratory gases using the "modified box-bag system" and a face mask with an airtight seal. Interestingly, the authors commented on the difficulty of maintaining the face mask's seal in these sometimes uncooperative shivering patients, a factor that may predispose to artifactual measurements. Second, patients were selected for inclusion in the study because they were noted to be shivering postoperatively. This criterion may have resulted in selection bias, because only patients with the most significant shivering may have been included, leading to an overestimation of the physiologic changes in the average shivering patient. Despite these limitations, the authors hypothesized that the metabolic demands of shivering may be poorly tolerated in patients with cardiopulmonary disease because oxygen delivery might not be adequate to meet oxygen demands. Because the elderly population is at the greatest risk for cardiac and other morbidity, it may be inappropriate to estimate their metabolic demands based on studies performed in younger persons.

In a more recent study, MacIntyre et al. [2]enrolled 14 patients (age unspecified) observed to be shivering when admitted to the PACU after minor surgical procedures. Anesthetic technique and opioid administration were not standardized and not specifically described. Total-body oxygen consumption (measured by indirect calorimetry) increased an average of approximately 400% above a "calculated basal oxygen consumption" in these shivering persons. Of the 14 patients, one had an oxygen consumption of 1,790 ml/min, a 730% increase above the expected basal rate! As in the study by Bay et al., [1]limitations may have included selection bias and the face mask technique for collecting respiratory gases. Although these authors hypothesized that hypoxemia may result from shivering, the partial pressure of oxygen in arterial blood in shivering and nonshivering patients has been shown to be similar. [1]To our knowledge, the only study that has demonstrated hypoxemia with shivering was done in patients who did not receive supplemental oxygen [23]and oxygenation was assessed by pulse oximetry. This method is probably unreliable because of motion artifact in shivering patients. Arterial blood samples were not analyzed for partial pressure of oxygen.

Our results indicate a reduced metabolic rate in patients with mild to moderate hypothermia during the early postoperative period, even when shivering patients were included in the analysis. This finding is the opposite of what would be expected if the majority of hypothermic patients shivered and increased their metabolic rate significantly. Our findings are consistent with previous studies in nonshivering anesthetized patients, in which a 7-8% decrease in metabolic rate was noted for each degree Centigrade reduction in body temperature. [24]In our analysis, we plotted oxygen consumption as a function of mean body temperature rather than core temperature. This is physiologically more appropriate because the core represents only a portion of the body's mass.

The clinical variable most highly correlated with the absence of postoperative shivering was the use of forced-air warming. This finding is consistent with previous studies that have demonstrated decreased shivering with skin-surface warming. [4,25,26]Cheng et al. [27]recently assessed the relative thermoregulatory contribution of core and skin-surface temperatures in volunteers. They determined an approximate 20% skin and 80% core contribution. In addition, these contributions were determined to be "linear" over a range of skin-surface temperatures from 31 degrees Celsius to 37 degrees Celsius. These findings indicate that for every 5 degrees Celsius increase in skin-surface temperature, the thermoregulatory input is equivalent to a 1 degree Celsius increase in core temperature. Thus if a patient's core temperature is within 1 degree Celsius of the shivering threshold, shivering may be "turned off" by warming the skin.

In the current study anesthetic technique, age, and core temperature were not predictors of shivering. Because only elderly patients were included (age > 60 yr), it is not surprising that no correlation was identified between age and shivering. Previous studies in younger patients have reported a 20-60% incidence of shivering, [18,28-30]a higher incidence than we determined in our elderly patients. Although previous studies have shown that patients are more likely to shiver after regional than after general anesthesia, [18,19]we found no significant correlation between anesthetic technique and shivering. However, our findings may be relevant specifically to elderly patients receiving opioid and benzodiazepine sedation.

The results in the current study suggest a more intact shivering response in men that in women, as evidenced by a greater incidence of shivering the greater total-body oxygen consumption in men despite similar core temperatures in both sexes. The increased incidence of clinically detectable shivering might be explained by the greater lean body mass in men than in women. Because shivering is the involuntary contraction of skeletal muscles, and men have a greater lean body mass regardless of age, [31]one would expect shivering to be more easily detected in men. It is more difficult to explain the gender difference that was observed for total-body oxygen consumption, which appears to be independent of body temperature or shivering. When the metabolic data were adjusted for weight, or body surface area, men consistently had a greater total-body oxygen consumption than women.

The relatively low incidence and low intensity of shivering and the small increase in oxygen consumption does not mean that patients should be allowed to become hypothermic during the perioperative period. In fact, we have previously shown mild postoperative hypothermia (< 35 degrees Celsius) to be associated with a two- to threefold increase in the incidence of early postoperative myocardial ischemia in patients at high risk. [32]Subsequently, we have shown in patients after surgery and in volunteers that 1-2 degrees Celsius reduction in core temperature is associated with a 100-500% increase in norepinephrine concentration, peripheral vasoconstriction, and hypertension. [33,34]These findings alone indicate that hypothermia during the perioperative period should be avoided in patients at high risk.

The following limitations of the current study should be considered. We conclude that the relatively small increase in total-body oxygen consumption associated with shivering may be related to the advanced age of our patients or to the administration of opioids for analgesia. However, because we did not include younger persons or patients not receiving opioids, our conclusions are based on comparisons with previously published studies. To identify the specific effects of age and opioids on shivering and oxygen consumption would require further investigation using a different study design. A second limitation may be the somewhat heterogeneous patient population that was studied. The enrolled patients underwent a variety of surgical procedures and anesthetic techniques, although the anesthetics were standardized by protocol. However, the validity of our findings is supported by the large number of patients included in the study and the relatively even demographic distribution of surgical and anesthetic variables in the routine thermal care and forced-air warming groups. Moreover, both univariate and multivariate analyses were used to account for potential confounding variables. Finally, total-body oxygen consumption was measured at two specific times during the early postoperative period, rather than obtaining continuous data. Measurements were made at 30 and 90 min postoperatively, because the 1st 2 h after surgery represent the most active period for rewarming and shivering. [35]Thus, it is unlikely that oxygen consumption would have been significantly greater at times other than the ones we chose to obtain these measurements. Finally, it is possible that we would have observed a greater incidence and intensity of shivering (and consequently a greater increase in oxygen consumption) if the patients were allowed to develop a greater degree of hypothermia. Although the range of core temperatures was 33.1-39.2 degrees Celsius at the end of the surgical procedure, the mean core temperature in those receiving "routine thermal care" was 35.3 degrees Celsius. We believe that this is the degree of hypothermia that clinicians typically see when patients receive what we describe as routine thermal care.

In conclusion, we have shown in a large series of elderly patients that the clinical variables associated with increased total-body oxygen consumption in the early postoperative period were the presence of shivering, increased mean body temperature, and male gender. Although the increased oxygen consumption in shivering patients was statistically significant, oxygen consumption in the average shivering patient was only 38% greater than in nonshivering patients, a difference that may have little or no clinical relevance. This relatively small increase in metabolism with shivering may be explained by an age-related impairment of the thermoregulatory response, the antishivering effects of opioid analgesia, or both. We conclude from these findings that in elderly patients undergoing surgery, the adverse effects of perioperative hypothermia are not likely to result from the previously reported large increase in metabolism associated with shivering, and it is likely that other mechanisms are implicated.

The authors acknowledge Susan Kelly, Bernard Jiang, Trevor Myers, Peter Wu, and Sandra Hebert for their assistance with data collection. They are thankful for support from the nursing staff in the Acute Care Unit and Surgical Intensive Care Unit and from the house staff who cared for patients enrolled in this study. They also thank Constance Bourke for editorial assistance.