eFigure 1. Multivariable Model Predicting SSI, Including Temperature Nadir as the Temperature Variable
eFigure 2. Multivariable Model Predicting SSI, Including Mean Intraoperative Temperature as the Temperature Variable
eFigure 3. Multivariable Model Predicting SSI, Including Percentage of Time at the Nadir as the Temperature Variable
eFigure 4. Multivariable Model Predicting SSI, Including Percentage of Time <36°C as the Temperature Variable
Baucom RB, Phillips SE, Ehrenfeld JM, Muldoon RL, Poulose BK, Herline AJ, Wise PE, Geiger TM. Association of Perioperative Hypothermia During Colectomy With Surgical Site Infection. JAMA Surg. 2015;150(6):570-575. doi:10.1001/jamasurg.2015.77
Maintaining perioperative normothermia has been shown to decrease the rate of surgical site infection (SSI) after segmental colectomy and is part of the World Health Organization’s Guidelines for Safe Surgery. However, strong evidence supporting this association is lacking, and an exact definition of normothermia has not been described.
To determine whether intraoperative hypothermia in patients who undergo segmental colectomy is associated with postoperative SSI.
Design, Setting, and Participants
In a retrospective cohort study at a single tertiary-referral hospital, 296 adult patients who underwent elective segmental colectomy from January 1, 2005, through December 31, 2009, were included. Exclusion criteria included postoperative stoma, emergent or urgent operation, and diagnosis of inflammatory bowel disease.
Perioperative temperature was measured continuously, and 4 possible definitions of hypothermia were explored, including temperature nadir, mean intraoperative temperature, percentage of time at the temperature nadir, and percentage of time with a temperature of less than 36.0°C.
Main Outcomes and Measures
The primary outcome measure was 30-day SSI. Secondary outcome measures included clinical leak, return to the operating room, and nasogastric tube placement (a surrogate for ileus).
The mean (SD) findings were as follows: intraoperative temperature, 35.9°C (0.6°C); temperature nadir, 34.3°C (2.8°C); percentage of time at the nadir, 4.7% (10.8%); and percentage of time with a temperature of less than 36.0°C, 49.9% (42.0%). The rate of SSI was 12.2% (n = 36). There was no statistically significant difference in temperature measurements between the patients who developed an SSI and those who did not. Logistic regression models evaluated each exposure measure and its effect on SSI, adjusting for body mass index, smoking status, and sex. The adjusted analyses revealed no association between intraoperative hypothermia and 30-day SSI (odds ratio, 1.17; 95% CI, 0.76-1.81; P = .48). Increased body mass index (odds ratio, 1.39; 95% CI, 1.10-1.76; P = .007) was significantly associated with SSI in all 4 logistic regression models.
Conclusions and Relevance
Patients who underwent segmental colectomy and sustained a period of intraoperative hypothermia were no more likely to develop an SSI than those who were normothermic.
In 2009, the World Health Organization published the Guidelines for Safe Surgery and described “maintenance of normothermia during surgery” as 1 of 10 objectives aimed to reduce the incidence of postoperative surgical site infection (SSI).1 These guidelines did not explicitly define normothermia. The Agency for Healthcare Research and Quality (AHRQ) has established a similar quality metric for hospitals using a cutoff of 36.0°C, measured either 30 minutes before the end of the operative procedure or 15 minutes after its conclusion.2 These guidelines were based largely on 3 studies,3- 5 one of which involved a population of patients who underwent colorectal operations. Since that time, hospitals have invested time, effort, and money implementing warming measures for patients during the perioperative period.
Recent investigations6,7 have brought into question the effectiveness of these efforts on reducing SSIs, including evidence from the colorectal and general surgery populations. In 2010, authors from Vanderbilt University Medical Center published a 1-year retrospective analysis6 of patients who underwent elective segmental colectomy, examining the relationship of perioperative temperature on postoperative outcomes. The current study aimed to determine whether a more appropriate definition of hypothermia exists and to examine its effect on 30-day SSI in patients who undergo elective colectomy.
A retrospective cohort study (performed with the approval of the Vanderbilt University Human Research Protection Program and Institutional Review Board, with a waiver of patient informed consent) was conducted that included adult patients who underwent elective segmental colectomy by 1 of 4 board-certified colorectal surgeons (R.L.M., A.J.H., P.E.W., and T.M.G.) at a single institution from January 1, 2005, through December 31, 2009. Patients were excluded who underwent urgent or emergent procedures, had a diagnosis of inflammatory bowel disease, or had a postoperative stoma created. We hoped by these criteria to exclude patients who had ongoing inflammation or active intra-abdominal infections that could raise their core body temperature and increase the risk of SSI. The population included patients with a diagnosis of colorectal cancer, endoscopically unresectable colon polyp, or diverticular disease. To identify the cohort for analysis, all segmental colectomies were identified by Current Procedural Terminology codes (44140, 44141, 44143-44147, 44160, 44204-44208), and International Classification of Diseases, Ninth Revision codes (153.0-153.9, 154.0-154.2, 154.8, 211.3, 211.4, 562.10-562.13) were used to further identify our target population. Finally, operative notes were reviewed to identify patients who had unplanned stomas placed during the procedure. All of the surgeons have a standardized practice of preoperative bowel preparation, a perioperative antibiotic regimen, and anastomosis creation (ie, the wound is protected, the gown and gloves are changed after the anastomosis is performed, and separate instruments are used for the creation of the anastomosis). In this patient population, the surgeons at our institution routinely close the skin incisions primarily.
Operative details, including procedural codes, date of service, length of stay, and indication for surgery, were obtained from the Vanderbilt Procedural Outcomes Database, an institutional database that includes clinical and administrative information for patients who undergo a variety of procedures. For the purpose of this analysis, cases in which a laparoscopic procedure was converted to an open procedure were considered open procedures. Perioperative temperature data were obtained from the Perioperative Data Warehouse, an institutional data repository that includes robust perioperative clinical details about patients, including continuous temperature data, smoking history, American Society of Anesthesiologists classification, age, sex, comorbidities, procedure duration, antibiotic administration, and intraoperative blood transfusion.
The primary outcome was 30-day SSI, as defined by the National Surgical Quality Improvement Program.8 Postoperative SSI was determined from the institutional National Surgical Quality Improvement Program data and included superficial incisional, deep incisional, and organ-space infections. During the study, National Surgical Quality Improvement Program data captured 100% of colectomy cases at our institution. Secondary outcomes, including postoperative nasogastric tube placement, return to the operating room, and clinical leak, were obtained from review of the electronic medical record. Clinical leak was defined in 1 of 2 ways: (1) radiologic evidence of fluid surrounding the anastomosis that required antibiotics, percutaneous drainage, or a return to the operating room or (2) clinical evidence of abdominal sepsis that required a return to the operating room, with or without preoperative imaging, with intraoperative confirmation of anastomotic leak.
Intraoperative core body temperature measurements were recorded continuously by an automated system, and measurements were obtained from either an esophageal or a bladder temperature probe. To eliminate erroneous temperature readings resulting from a dislodged probe, temperature measurements greater than 3 SDs from the mean were excluded from analysis. The standard of care for all patients included the use of forced–warm air blankets (Bair Paws and Bair Hugger; 3M) and intravenous fluid warmers (Baxter Healthcare).
Given the simplistic nature of the AHRQ quality metric, we hypothesized that periods of intraoperative hypothermia, which might affect the rate of SSI, were not adequately captured by 1 temperature measurement at the end of every procedure. Therefore, in addition to evaluating the quality metric, we aimed to explore the effect of other potential definitions of hypothermia on the rate of SSI. A review of the published medical literature was conducted, and meetings were held with experts in the field at Vanderbilt University Medical Center to determine potential variables to include in a more robust analysis. Discussions focused on current practices of maintaining hypothermia at our institution, clinically relevant time points for monitoring core body temperature, and mechanisms by which hypothermia might impair wound healing and therefore be clinically significant. As a result of these discussions, 4 variables were chosen to explore the effect of hypothermia on patient outcomes in addition to the current AHRQ quality metric: temperature nadir, percentage of time spent at the temperature nadir, percentage of time spent at a temperature of less than 36.0°C, and mean intraoperative temperature.
Descriptive statistics were calculated for the overall patient population and included demographics, comorbidities, operative characteristics, temperature data, and postoperative outcomes. Cases in which a laparoscopic procedure was converted to an open procedure were considered open procedures. Univariate analyses evaluating differences in each of the temperature variables, final intraoperative temperature, first postanesthesia care unit temperature, and the AHRQ quality metric were conducted. Four multivariable models were then constructed, adjusting for body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared), smoking status, and sex, each including a unique temperature variable. The dependent variable in each model was development of 30-day SSI (dichotomous). All statistical analysis was performed using R, version 2.15.2 (The R Foundation).9
A total of 868 patients underwent elective segmental colectomy between January 1, 2005, and December 31, 2009. After excluding patients with a diagnosis of inflammatory bowel disease and those who underwent emergent procedures, defined using International Classification of Diseases, Ninth Revision, codes for colon cancer, rectal cancer, colon or rectal polyp, diverticulitis, or diverticulosis, 350 patients remained. Review of operative notes revealed 54 patients who had unplanned stomas placed, leaving 296 patients who were included in the analysis. The mean age of the population was 61.8 years, and 51.7% were male. Other demographic information, comorbidities, and operative characteristics can be found in Table 1. Of note, there were 15 procedures that began laparoscopically and were converted to open procedures, for a 6.4% conversion rate. Overall, 275 of the 296 patients (92.9%) attained the AHRQ quality metric and had a final intraoperative temperature of 36.0°C or greater.
The rate of 30-day SSI was 12.2% (n = 36). The rate in the group who attained the AHRQ quality metric was no different from the rate of those who were considered hypothermic by the quality metric (12.4% vs 9.5%; P >.99). Table 2 compares the temperature measurements of patients who developed a 30-day SSI with those of patients who did not. There was no statistically significant difference in mean intraoperative temperature, temperature nadir, percentage of time spent at the temperature nadir, or percentage of time spent hypothermic between the 2 groups. The mean final intraoperative temperature was 36.3°C in the group who developed SSIs compared with 36.1°C in the group without SSIs (P = .48). The mean postanesthesia care unit temperature was also similar between those who developed SSIs and those who did not (36.6°C vs 36.4°C; P = .15). In addition, there was no difference in mean intraoperative temperature between laparoscopic and open operations (36.0°C vs 36.0°C; P = .93). When several known risk factors for SSI were compared between those who developed SSI and those who did not, the only statistically significant difference was in BMI (31.7 for those who developed an SSI vs 28.5 in those who did not develop an SSI; P = .04) (Table 3). There were 33 patients (11.1%) who required placement of a nasogastric tube postoperatively, 14 (4.7%) who developed a clinical leak, and 19 (6.4%) who required a return to the operating room postoperatively. Owing to the relatively small number of these events, subgroup analysis was not conducted for these secondary outcomes.
To evaluate the effect of each potential definition of hypothermia on SSI, 4 multivariable regression models were constructed. Each model included a unique temperature variable. The other covariates remained the same for each model and included BMI, smoking status, and sex. Diabetes mellitus was highly correlated with BMI; therefore, only 1 of these covariates was included in the final models. A summary of the 4 models can be found in Table 4.
The first model, depicted in eFigure 1 in the Supplement, includes temperature nadir as the temperature variable. In this model, BMI is the only significant risk factor for SSI (P = .009). The odds ratio depicted represents the increased odds of SSI for a 5-point increase in BMI. Female sex appears to be associated with decreased odds of SSI, although this association was not statistically significant (P = .07). In addition, smoking appears to increase the odds of SSI, although this outcome was not statistically significant (P = .22). The model shown in eFigure 2 in the Supplement includes mean intraoperative temperature as the temperature variable. Similar to the first model, BMI is associated with increased odds of SSI (P = .01). Smoking status, sex, and temperature were not statistically significant risk factors for SSI.
eFigure 3 in the Supplement depicts the model that includes the percentage of time a patient spent at the temperature nadir as the temperature variable. A longer time spent at the temperature nadir is not significantly associated with an increased risk of SSI, and BMI is again the only significant risk factor for SSI (P = .009). The fourth model is demonstrated in eFigure 4 in the Supplement and includes the percentage of operative time the patient spent with a temperature of less than 36.0°C as the temperature variable. An increased percentage of operative time spent hypothermic was not associated with increased odds of SSI. In all the models, BMI and sex were the most important covariates for predicting SSI, and the temperature variable was the least influential in all models.
This study of 296 patients during a 5-year period failed to demonstrate an association between hypothermia and 30-day SSI in patients who underwent segmental colectomy. Regardless of how perioperative hypothermia was defined, temperature did not predict SSI in multivariable analysis and was the least influential variable in each of the models assessed.
These findings further support those of Geiger et al,6 which demonstrated no correlation between hypothermia and SSI in similar patients during a 1-year period. They are also similar to findings of a recent study that evaluated more than 1000 patients who underwent colorectal surgery using continuously obtained perioperative temperature measurements and found no correlation between hypothermia and SSI.10 Furthermore, our data indicate that multiple definitions of hypothermia also fail to yield associations with SSI. In addition to data published regarding the colorectal surgery population, recent data from patients who underwent ventral hernia repair have also demonstrated no correlation between perioperative hypothermia and 30-day SSI.7
Our study does stand in contrast to the 1996 study by Kurz et al,4 which included 200 patients who underwent colorectal resection for cancer or inflammatory bowel disease. Based on this randomized study, active warming has been used perioperatively in an effort to reduce SSIs. However, several potential limitations to that study exist. First, although the normothermic and hypothermic groups had similar patient characteristics and comorbidities, the number of patients per group with a postoperative stoma is unknown. Also, the patients included in the study had a much lower mean BMI (24.5 kg/m2) than the typical population at our tertiary-referral center, where elevated BMI clearly affected the incidence of SSI. Third, the patients in the hypothermia group required a greater volume of blood transfusions intraoperatively, a risk factor for SSI that may introduce bias.10 Finally, the total number of SSIs was relatively small, and an increase of only 2 more SSIs in the normothermic group would have made the study findings statistically insignificant. Several other studies have also demonstrated a correlation between hypothermia and SSI. However, these have not specifically evaluated the colorectal population and might not be generalizable.3,5,11
Our study is limited by its retrospective nature and relatively small number of events. However, the number of patients and events in our study is larger than the study on which many of the national and international guidelines have been based. Other factors appear to contribute much more than perioperative temperature to the development of a postoperative SSI. We used strict inclusion and exclusion criteria that limited the number of patients included in analysis. However, we believed this step was necessary to reduce bias given the higher-risk population seen at our tertiary care center. Because we were unable to adjust for every possible confounding variable, we chose those with the strongest known associations with SSI. Blood transfusion was not included in the models because it was not significantly different between groups and applied to only 5 patients. A second limitation is the fact that patients now routinely undergo active warming perioperatively. Although patients still spent an average of 50% of the operative time with core body temperatures of less than 36.0°C, the average temperature at the end of the procedure was greater than or equal to 36.0°C to comply with the AHRQ quality metric.2 This likely limits the temperature variation between patients and could diminish an effect that hypothermia might have on SSI. Finally, although we attempted to exclude patients with any ongoing inflammatory response through our exclusion criteria, there might have been a small number of patients with ongoing inflammation and therefore increased core body temperature. We expected that most of these patients would have received a diverting stoma or undergone emergent procedures, but if any remained in the cohort, their risk of SSI was likely above average.
One strength of our study is that the temperature data available through our institution is robust, and most patients have measurements taken every minute intraoperatively. This practice allows for detailed analysis that includes temperature measurements throughout the operation rather than just the patient’s core temperature at the conclusion of the procedure. Our findings call into question the actual effect of efforts to maintain perioperative normothermia on reducing postoperative SSIs. Although the number of events in our study population was relatively small (36 [12.2%]), this population consisted of surgical patients with higher-than-average risk for developing SSIs. If during 5 years no association was found between perioperative hypothermia and SSI, the clinical significance of any effect the temperature might have seems low. Other factors, such as BMI, appear to be of much greater importance to reduce the risk of SSI.
Although the findings in this study suggest that perioperative normothermia does not decrease the rate of SSI, we acknowledge that maintaining perioperative normothermia is likely important for other physiologic reasons. Anesthesiology publications suggest that hypothermia can impair drug metabolism, affect cardiac morbidity, and lead to coagulopathy.12 The most dramatic drop in patient temperature occurs on induction of anesthesia, and prewarming has been shown to minimize this initial redistribution hypothermia.13 If this is the time that the patient is most at risk, then thermoregulatory efforts should focus on maintaining normothermia during that particular portion of the procedure and quality measures should reflect this focus. In this era of cost containment, efforts that generate costs but do not lead to improved outcomes need to be identified and reduced.
In summary, using robust perioperative temperature measurements and 4 definitions of perioperative hypothermia, this study of patients who underwent elective segmental colectomy without creation of a stoma failed to demonstrate an association between perioperative temperature and 30-day SSI. In the 4 models analyzed, temperature was the least influential variable when predicting SSI. Quality measures should accurately reflect the optimal timing of active patient warming to reduce waste in the health care system. Attention needs to be given to other potential risk factors for SSI to decrease the incidence of this morbidity in the colorectal surgery population.
Accepted for Publication: November 4, 2014.
Corresponding Author: Rebeccah B. Baucom, MD, Division of General Surgery, Vanderbilt University Medical Center, D-5203 Medical Center N, 1161 Medical Center Dr, Nashville, TN 37232 (email@example.com).
Published Online: April 22, 2015. doi:10.1001/jamasurg.2015.77.
Author Contributions: Drs Baucom and Geiger had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Baucom, Phillips, Ehrenfeld, Poulose, Wise, Geiger.
Acquisition, analysis, or interpretation of data: Baucom, Ehrenfeld, Muldoon, Herline, Wise, Geiger.
Drafting of the manuscript: Baucom, Phillips, Geiger.
Critical revision of manuscript for important intellectual content: All authors.
Statistical analysis: Baucom, Phillips, Poulose.
Administrative, technical, or material support: Ehrenfeld.
Study supervision: Ehrenfeld, Poulose, Herline, Wise, Geiger.
Conflict of Interest Disclosures: Dr Poulose reports receiving grant support for unrelated research from Bard Davol and having a consultancy relationship with Ariste Medical. Dr Herline reports receiving royalties and holding stock and/or stock options with Pathfinder Therapeutics. The nature of the relationship is outside the realm of the article. Dr Wise reports receiving unrelated research support from Cancer Prevention Pharmaceuticals. No other disclosures were reported.