One of the most common side effects of PH is postoperative shivering. Initially believed to result from dissociation of spinal reflexes from the cerebral cortex, it is now believed that most postoperative shivering is thermoregulatory in origin. Recent studies suggest that, on average, postoperative shivering increases oxygen consumption by approximately 40% [12]. Although there are few data to suggest that this increase in oxygen consumption is associated with perioperative morbidity, postoperative shivering can be uncomfortable for patients and can be easily treated with small doses of narcotics (particularly meperidine) [13]. In addition, skin-surface warmers can also be used to raise the skin temperature and optimize thermal comfort.

Perioperative cardiac events are associated with in-hospital mortality of up to 25% and increase length of stay by an average of 11 days [14]. PH is associated with increased serum catecholamine levels, vasoconstriction, and increased blood pressure, which, in turn, can lead to increased cardiac demand, ischemia, and morbidity. These cold-induced cardiovascular effects are more prominent after anesthesia once the adrenergic response is no longer suppressed, particularly in high-risk patients [15]. The relationship between PH and perioperative cardiac morbidity was analyzed in a trial by Frank and colleagues [16] in which 300 high-risk patients undergoing thoracic, abdominal, or peripheral vascular surgery were randomized to receive warmed intravenous (IV) fluids alone or warmed IV fluids and intraoperative/postoperative active warming via warm forced-air devices. Mean temperatures on arrival to the intensive care unit were significantly higher in the normothermia group (36.7°C) compared with the control group (35.4°C; P <.001). Although there was no difference in intraoperative cardiac events, the normothermia group had lower rates of electrophysiologic (7% vs 16%) and morbid cardiac events (1% vs 6%) during the subsequent 24-hour follow-up period. Although a subsequent trial comparing forced-air devices versus circulating-water blankets resulted in higher core temperatures in the forced-air device group (36.4°C vs 35.6°C), rates of postoperative cardiac events at 24 hours were similar [17]. Cardiac events were a secondary outcome, and, as such, the trial was not specifically powered to detect a difference.

It is well documented that PH can also adversely affect coagulation, resulting in surgical bleeding. Although PH can impair activity of various temperature-dependent factors in the coagulation cascade, it is often missed in the clinical setting because coagulation laboratory studies (ie, PT, PTT) are usually performed at a temperature of 37°C [18]. Hypothermia also impairs platelet function, likely because of reduced levels of thromboxane A2 [19]. Mild hypothermia was shown to increase perioperative bleeding and transfusion requirements in a recent randomized trial involving patients undergoing hip arthroplasty [20]. Perioperative blood loss was significantly higher in hypothermic patients (final intraoperative temperature 35.0°C) compared with normothermic patients: 2.2 L versus 1.5 L (P <.001). Seven out of 30 patients with hypothermia required blood transfusions, compared with only 1 normothermic patient. The investigators concluded that, in patients undergoing hip arthroplasty, typical PH increases perioperative blood loss by approximately 500 mL.

SSIs are the third leading cause of nosocomial infections, accounting for 14% to 16% of all hospital-acquired infections, and are the leading cause of nosocomial infection among surgical patients [21]. SSIs are a major cause of postoperative morbidity and are associated with a twofold to 12-fold increased risk of postoperative mortality [22,23]. Furthermore, SSIs increase postoperative length of stay by an average of 4 days and result in attributable direct costs of up to $8000 per case [24], and, as such, represent a significant burden to the health care system.

Little is known about the direct effects of PH on the immune system. Incubation of leukocytes at low temperatures suppresses migration and the mitogenic response, whereas increases in temperature lead to enhanced interleukin (IL)-1 activity [25,26]. Hypothermic laboratory animals are more susceptible to bacterial infections compared with normothermic animals [27]. In patients having surgery, PH suppresses mitogen-induced activation of lymphocytes, reduces the production of IL-1 and IL-2, and impairs neutrophil oxidative killing in the intraoperative period [28]. PH has been hypothesized to predispose patients to SSIs by triggering thermoregulatory vasoconstriction, which may decrease partial pressure of oxygen in tissues, impair oxidative killing by neutrophils, and interfere with collage deposition resulting in impaired wound healing [29–35].

Several randomized controlled trials have analyzed the effect of active warming to prevent PH on SSI rates. In a seminal 1996 clinical trial, Kurz and colleagues [36] randomized 200 patients undergoing colorectal surgery to either active warming using forced-air warmers and intravenous fluid warmers to maintain patients’ core temperatures near 36.5°C (the normothermia group) or routine intraoperative thermal care (the hypothermia group). Six percent of patients in the normothermia group developed SSIs, compared with 19% of patients in the hypothermia group (P = .009). In addition, mean final intraoperative core temperature in the normothermia group was 36.6°C, compared with 34.7°C in the hypothermia group (P = .002). Although the data are compelling, serious concerns have been raised regarding some of the methods used in this trial [37,38]. Patients in the control arm received forced air at ambient temperature, which may have effectively cooled patients to lower than expected temperatures, increased the resulting difference in mean temperatures between study arms, and exaggerated the apparent effect of the active warming. In addition, patients in both groups received almost 4 days of postoperative systemic antibiotics, 35% of patients in the hypothermia group received intraoperative transfusions (a known risk factor for SSIs) compared with only 22% of patients in the normothermia group, and length of hospital stay was atypically long (approximately 2 weeks) in both groups [39–41]. Furthermore, SSI (the primary outcome of interest) was defined by positive aerobic or anaerobic cultures from pus that was aspirated or expressed from the surgical incision during the first 15 days after surgery (rather than by one of the more standard definitions of SSI used by the World Health Organization or the Centers for Disease Control and Prevention) [42].

More recently, Melling and colleagues [43] analyzed the effect of preoperative local warming (via noncontact radiant heat) or systemic warming (via warmed forced-air devices) versus routine perioperative thermal care on SSI rates in 421 patients undergoing clean surgery (ie, breast surgery, elective herniorrhaphy, or varicose vein surgery). Wound infections were defined as erythema or purulence within 6 weeks of surgery that lasted more than 5 days and required antibiotic therapy. SSI rates were significantly lower in patients who received preoperative local warming (4%) or systemic warming (6%) compared with nonwarmed patients (14%, P <.001). The study did not include patients having gastrointestinal surgery. More importantly, it did not report the resulting mean temperatures of the 3 study groups, and, as such, there was no direct evidence that the observed reduction in wound infection rates was caused by prevention of PH [44].

Although the trial data cited earlier suggest that active warming during surgery may reduce SSIs, there is limited evidence that perioperative normothermia, in and of itself, is associated with lower rates of SSI. Barone and colleagues [37] retrospectively reviewed the records of 150 consecutive patients who underwent colorectal surgery in a 30-month period. Although 1 or more warming modalities were used in all patients, approximately 33% of patients were hypothermic (defined as T <34.3°C) during the intraoperative or immediate postoperative period. Despite the extreme definition of hypothermia used in this study (even lower then the mean final intraoperative temperature [T <34.7°C] of the hypothermia group in the trial by Kurz and colleagues [36]), the rates of wound infections (defined as suppuration requiring removal of sutures) was identical at 6% in both the normothermia and hypothermic patient populations.

More recently, Lehtinen and colleagues used a nested, case-control study design to analyze the association between perioperative normothermia and incisional SSIs after gastrointestinal surgery [45]. Cases consisted of all patients having consecutive gastrointestinal surgery enrolled in a single institution’s American College of Surgeons National Surgical Quality Improvement Program database during a 3-year period who developed SSIs. When cases and control were compared with respect to recorded perioperative core temperatures, median lowest intraoperative temperatures and final intraoperative temperatures were, paradoxically, slightly higher in cases compared with controls. The percentage of patients with intraoperative normothermia was significantly higher among patients who underwent emergency surgery (Fig. 1), and the percentage of patients with lowest intraoperative temperature greater than 36°C also increased with increasing wound class (ie, patients with more contaminated wounds were more likely to have been normothermic during surgery; Fig. 2