Risk Stratification for Surgical Site Infections in Colon Cancer | Colorectal Cancer | JAMA Surgery | JAMA Network
[Skip to Navigation]
Sign In
Table 1.  Baseline Characteristics (N = 1481)
Baseline Characteristics (N = 1481)
Table 2.  Association of Baseline Characteristics With Surgical Site Infection Rates and Overlap
Association of Baseline Characteristics With Surgical Site Infection Rates and Overlap
Table 3.  Association of the Number of Risk Factors With the Risk of Surgical Site Infections
Association of the Number of Risk Factors With the Risk of Surgical Site Infections
1.
Ju  MH, Ko  CY, Hall  BL, Bosk  CL, Bilimoria  KY, Wick  EC.  A comparison of 2 surgical site infection monitoring systems.  JAMA Surg. 2015;150(1):51-57.PubMedGoogle ScholarCrossref
2.
Shih  T, Nicholas  LH, Thumma  JR, Birkmeyer  JD, Dimick  JB.  Does pay-for-performance improve surgical outcomes? an evaluation of phase 2 of the Premier Hospital Quality Incentive Demonstration.  Ann Surg. 2014;259(4):677-681.PubMedGoogle ScholarCrossref
3.
Anthony  T, Murray  BW, Sum-Ping  JT,  et al.  Evaluating an evidence-based bundle for preventing surgical site infection: a randomized trial.  Arch Surg. 2011;146(3):263-269.PubMedGoogle ScholarCrossref
4.
Lutfiyya  W, Parsons  D, Breen  J.  A colorectal “care bundle” to reduce surgical site infections in colorectal surgeries: a single-center experience.  Perm J. 2012;16(3):10-16.PubMedGoogle Scholar
5.
Zimlichman  E, Henderson  D, Tamir  O,  et al.  Health care-associated infections: a meta-analysis of costs and financial impact on the US health care system.  JAMA Intern Med. 2013;173(22):2039-2046.PubMedGoogle ScholarCrossref
6.
Kirkland  KB, Briggs  JP, Trivette  SL, Wilkinson  WE, Sexton  DJ.  The impact of surgical-site infections in the 1990s: attributable mortality, excess length of hospitalization, and extra costs.  Infect Control Hosp Epidemiol. 1999;20(11):725-730.PubMedGoogle ScholarCrossref
7.
Engemann  JJ, Carmeli  Y, Cosgrove  SE,  et al.  Adverse clinical and economic outcomes attributable to methicillin resistance among patients with Staphylococcus aureus surgical site infection.  Clin Infect Dis. 2003;36(5):592-598.PubMedGoogle ScholarCrossref
8.
Anderson  DJ, Podgorny  K, Berríos-Torres  SI,  et al.  Strategies to prevent surgical site infections in acute care hospitals: 2014 update.  Infect Control Hosp Epidemiol. 2014;35(suppl 2):S66-S88.PubMedGoogle ScholarCrossref
9.
Tang  R, Chen  HH, Wang  YL,  et al.  Risk factors for surgical site infection after elective resection of the colon and rectum: a single-center prospective study of 2,809 consecutive patients.  Ann Surg. 2001;234(2):181-189.PubMedGoogle ScholarCrossref
Original Investigation
July 2017

Risk Stratification for Surgical Site Infections in Colon Cancer

Author Affiliations
  • 1Division of General and Gastrointestinal Surgery, Massachusetts General Hospital, Harvard Medical School, Boston
JAMA Surg. 2017;152(7):686-690. doi:10.1001/jamasurg.2017.0505
Key Points

Question  How can the risk of surgical site infections be stratified using readily available clinical characteristics?

Findings  In this cohort study, surgical site infection risk was scored and stratified based on the number of high-risk characteristics such as smoking history, alcohol abuse history, type 2 diabetes status, obesity, operation duration more than 140 minutes, or nonlaparoscopic approach. Surgical site infection incidence rose from 2.3% among patients with 0 risk factors to 13.6% among patients with 4 or more risk factors.

Meaning  This scoring method may provide a simple way to use readily available characteristics to stratify patients by surgical site infection risk.

Abstract

Importance  Surgical site infections (SSIs) feature prominently in surgical quality improvement and pay-for-performance measures. Multiple approaches are used to prevent or reduce SSIs, prompted by the heavy toll they take on patients and health care budgets. Surgery for colon cancer is not an exception.

Objective  To identify a risk stratification score based on baseline and operative characteristics.

Design, Setting, and Participants  This retrospective cohort study included all patients treated surgically for colon cancer at Massachusetts General Hospital from 2004 through 2014 (n = 1481).

Main Outcomes and Measures  The incidence of SSI stratified over baseline and perioperative factors was compared and compounded in a risk score.

Results  Among the 1481 participants, 90 (6.1%) had SSI. Median (IQR) age was 66.9 (55.9-78.1) years. Surgical site infection rates were significantly higher among people who smoked (7.4% vs 4.8%; P = .04), people who abused alcohol (10.6% vs 5.7%; P = .04), people with type 2 diabetics (8.8% vs 5.5%; P = .046), and obese patients (11.7% vs 4.0%; P < .001). Surgical site infection rates were also higher among patients with an operation duration longer than 140 minutes (7.5% vs 5.0%; P = .05) and in nonlaparoscopic approaches (clinically significant only, 6.7% vs 4.1%; P = .07). These risk factors were also associated with an increase in SSI rates as a compounded score (P < .001). Patients with 1 or fewer risk factors (n = 427) had an SSI rate of 2.3%, equivalent to a relative risk of 0.4 (95% CI, 0.16-0.57; P < .001); patients with 2 risk factors (n = 445) had a 5.2% SSI rate (relative risk, 0.78; 95% CI, 0.49-1.22; P = .27); patients with 3 factors (n = 384) had a 7.8% SSI rate (relative risk, 1.38; 95% CI, 0.91-2.11; P = .13); and patients with 4 or more risk factors (n = 198) had a 13.6% SSI rate (relative risk, 2.71; 95% CI, 1.77-4.12; P < .001).

Conclusions and Relevance  This SSI risk assessment factor provides a simple tool using readily available characteristics to stratify patients by SSI risk and identify patients at risk during their postoperative admission. Thereby, it can be used to potentially focus frequent monitoring and more aggressive preventive efforts on high-risk patients.

Introduction

Surgical site infections (SSIs) feature prominently in surgical quality improvement1 and pay-for-performance measures.2 Multiple approaches are used to prevent or reduce SSIs, often grouped in standardized protocols or evidence-based care “bundles.”3,4 The focus on reducing SSI rates is prompted by the heavy toll it takes on patients and health care budgets. Patients who develop an SSI see their admission times considerably lengthened5 and have an increase in mortality,6,7 while the total cost effect of SSIs in the United States is estimated to range between $3.5 to $10 billion.8 With an overall estimated incidence of 160 000 to 300 000 cases annually,8 the extrapolated total costs per SSI case range between $11 500 to more than $60 000.

Surgery for colon cancer is not an exception, as increasingly complex and diverse protocols and preventive bundles are implemented to standardize care and as preventing wound infections has become one of the primary goals. Standardized methods cover a range of preventive measures, including perioperative prophylactic antibiotics, specific evidence-based choices in preoperative surgical site preparations and postoperative wound dressing, and accelerated discharge pathways to isolate patients from the strongest risk bearing factor for SSI: being in a hospital. As these protocols are becoming increasingly comprehensive and expensive, a maximum obtainable effect will be reached, meaning that the unavoidable a priori risks of SSI become increasingly important, if only to recognize those at the highest risk to concentrate the efforts on them. This article identifies a risk stratification score based on preoperative and operative variables to identify the a priori risk of an SSI for individual patients.

Methods
Patients

A cohort that encompassed all surgically treated patients with colon cancer at Massachusetts General Hospital from 2004 through 2014 (n = 1481) was retrospectively included in a prospectively maintained survival and outcomes database after receiving institutional review board approval from Massachusetts General Hospital and Partners HealthCare. Informed consent was deemed not applicable because of the retrospective nature of the database and the nonidentifiable data used. Data on patients were collected from the research patient data repository and records from our health care network. Because of the significant differences in treatment approaches, especially in terms of adjuvant (radio) chemotherapy and nonabdominal (perineal) approaches influencing the risks of (surgical site) infections, we exclusively focused on colon cancer and did not include patients with rectal tumors. Colonic tumors were defined as any tumor proximal to the rectosigmoid junction. Surgical site infections were defined as a clinically reported infection of any site incised during the primary surgical procedure that was culture-confirmed or visually ascertained, either during admission or within 30 days of discharge. As this is a retrospective series, a reported SSI diagnosis was made through either wound culture or clinical pattern recognition by a physician, which was usually a combination of abnormal redness, warmness, and discharge that was often combined with fever, unexplained leukocytosis, and/or delayed healing and consolidation. Through the time span of this cohort, no significant changes were made in the preventive measures to combat SSI.

Of this cohort, a range of baseline characteristics were reviewed to assess whether they were significantly associated with the incidence of SSI. The factors reviewed were age (> or <65 years), if a participant was currently smoking, if a participant had ever smoked, alcohol abuse or a history thereof, type 2 diabetes, obesity (defined as a body mass index [calculated as weight in kilograms divided by height in meters squared] more than 30), an operation duration of more than 140 minutes, and a nonlaparoscopic approach (including lap-assisted and hand-assisted procedures). The thresholds for age and operation duration were selected, as they were expected to be close to the population median. The overlap between all pairs of factors was assessed using correlation coefficients to avoid including factors that measure the same risk effect. Subsequently, to validate a stacked score based on the number of risk factors present, the rates of SSI and the relative risks of its occurrence compared with the remaining population were also assessed based on the number of high-risk baseline characteristics present.

Statistical Analysis

The statistical analysis was performed using IBM SPSS Statistics, version 22.0 (IBM Corp). A 2-tailed P value of .05 or fewer was considered the threshold for statistical significance. Descriptive statistics illustrated the frequency of specific baseline characteristics; this was expressed in outcome percentages for dichotomous variables, and means or medians with standard deviations and interquartile rates, respectively, for continuous variables. A cross tabulation followed of the identified risk factors to show overlap, expressed both as percentage rates and Φ correlation coefficients (r). Subsequently, associations between the number of high-risk characteristics (0-1, 2-3, and ≥4) and SSI rates were measured using an χ2 test.

Results
Baseline Characteristics

Among the 1481 participants, 90 (6.1%) had SSI. Baseline characteristics are shown in Table 1. Surgical site infection incidence was not statistically associated with median age, emergency admissions, or comorbidity, expressed as either a Charlson Comorbidity Score or an American Society of Anesthesiologists score. Patients with SSIs were significantly more often men (66.7% vs 48.2%; P = .001), white (96.7% vs 88.75%; P = .02), and obese (median body mass index, 30.3 vs 26.3; P < .001).

SSI Risk Factors: Effect and Relative Overlap

Several lifestyle choices were associated with a higher incidence of SSI. The statistically significant factors included whether a patient had ever smoked (7.4% vs 4.8% P = .04), alcohol abuse or a history thereof (10.6% vs 5.7% P = .04), type 2 diabetes (8.8% vs 5.5%; P = .046), obesity (11.7% vs 4.0%; P < .001), and an operation duration more than 140 minutes (7.5% vs 5.0%; P = .05). A clinically meaningful trend was also found between the difference of SSI incidence in laparoscopic vs nonlaparoscopic approaches, including lap-assisted and hand-assisted procedures (laparoscopic, 4.1%; nonlaparoscopic, 6.7%; both P = .07).

In terms of mutual correlations, all potential risk factors were paired and none had a correlation coefficient of more than r = 0.16, with the strongest overlap being between open approaches and a history of smoking with nearly 40% (n = 580) of patients having a history of smoking and undergoing an open procedure. This overlap likely results from the high prevalence of these factors. As three-quarters of patients underwent a nonlaparoscopic procedure and half of patients had a history of smoking, an overlap between these factors is unavoidable. Table 2 lists all SSI risk factor incidences and the relative overlap between them.

SSI Risk Score

The number of risk factors was significantly associated with an increase in the rate of SSIs (P < .001). While the overall population had an SSI rate of 6.2%, patients with 2 risk factors or fewer had a lower risk than the overall population, while patients with 3 factors or more had a significantly higher risk. Patients with 0 or 1 risk factors (n = 427) had an SSI rate of 2.3%, equivalent to a relative risk of 0.4 (95% CI, 0.16-0.57; P <.001) compared with the rest of the population. Patients with 2 risk factors (n = 445) had an SSI rate of 5.2% (relative risk, 0.78; 95% CI, 0.49-1.22; P = .27). Patient with 3 risk factors (n = 384) had an SSI rate of 7.8%, corresponding to a relative risk of 1.38 (95% CI, 0.91-2.11; P = .13). Last, the 198 patients with 4 risk factors or more had a 13.6% SSI rate and a relative risk of 2.71 (95% CI, 1.77-4.12; P <.001). Details on the SSI risk stratification are shown in Table 3.

Discussion

This article analyzes a broad range of readily available baseline and surgical factors to determine how their compounding effect would be associated with risks of SSI. Considering the great financial and patient morbidity consequences of SSI, it makes sense to maximize these efforts. The risk score identifies patients for whom frequent monitoring and more aggressive preventive efforts would be the most useful, thereby maximizing the returns on attempts to reduce SSI incidence.

An analysis of baseline factors demonstrated that lifestyle factors such as a history of smoking and alcohol abuse drastically increased the risk of SSI, as did comorbid diabetes and obesity. Interestingly, an operation duration of 140 minutes or more was related to a 50% increase in SSI rates. When compounding these factors in a risk score, it was revealed that patients with 1 risk factor or fewer were 2.5 times less likely than the remainder of the population to have an SSI. Conversely, patients with 4 risk factors or more had 2.7 times the risk of developing an SSI.

This method of risk stratification advantageously uses factors that any baseline assessment of a surgical patient will discuss and provides an easy method to stratify risk and selectively focus preventive efforts and postoperative methods on patients with a significantly higher risk. Using this score, efforts can be concentrated on the tier that has a 1 in 6 chance to develop an SSI, rather than those with a 1 in 40 chance, without any need for additional diagnostics or interventions. Using this approach, higher-risk patients could be selected to benefit from additional preventive measures, including additional antibiotic prophylaxis, placement in isolation or low-patient-count rooms, and more frequent monitoring of temperatures and/or inspections of the surgical site to ascertain a timely response when the first symptoms of SSI appear. Last, with the introduction of “care bundles,”3,4 it might also be possible to tailor their elements to specific risk strata, potentially avoiding undertreatment and overtreatment.

Limitations and Further Research

Because our center is on the low end of the SSI rate,9 this also means that differences in SSI will be lower in absolute terms when compared with other centers, which may also mean that the gaps are smaller. Validation in centers with higher SSI rates may also be needed to confirm these proportions.

Rectal operations will most likely have even higher rates of SSI9 and would also benefit from a similar risk score, possibly dichotomized among patients with or without a perineal approach. Our database did not contain enough rectal cancer cases to validate any of the risk factors, which meant that a similar analysis was not possible. Future research using a similar approach for cancers of the rectum will be useful.

Conclusions

This risk assessment factor provides a simple tool to stratify patients based on their baseline characteristics and operative factors to predict the relative risk of developing a postoperative SSI. It offers a method to focus preventive efforts on high-risk patients using readily available characteristics that do not require any additional effort to verify. Using this score, efforts can be used in a more targeted manner to optimally use available resources to prevent or detect SSI in a timely way.

Back to top
Article Information

Corresponding Author: David. L. Berger, MD, Massachusetts General Hospital, Wang 460 15 Parkman St, Boston, MA 02114 (dberger@mgh.harvard.edu).

Accepted for Publication: January 25, 2017.

Published Online: April 12, 2017. doi:10.1001/jamasurg.2017.0505

Author Contributions: Dr Berger had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Amri, Dinaux, Bordeianou, Berger.

Acquisition, analysis, or interpretation of data: Amri, Dinaux, Kunitake, Berger.

Drafting of the manuscript: Amri.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Amri.

Obtained funding: Amri.

Administrative, technical, or material support: Amri, Dinaux, Kunitake.

Supervision: Amri, Bordeianou, Berger.

Conflict of Interest Disclosures: None reported.

References
1.
Ju  MH, Ko  CY, Hall  BL, Bosk  CL, Bilimoria  KY, Wick  EC.  A comparison of 2 surgical site infection monitoring systems.  JAMA Surg. 2015;150(1):51-57.PubMedGoogle ScholarCrossref
2.
Shih  T, Nicholas  LH, Thumma  JR, Birkmeyer  JD, Dimick  JB.  Does pay-for-performance improve surgical outcomes? an evaluation of phase 2 of the Premier Hospital Quality Incentive Demonstration.  Ann Surg. 2014;259(4):677-681.PubMedGoogle ScholarCrossref
3.
Anthony  T, Murray  BW, Sum-Ping  JT,  et al.  Evaluating an evidence-based bundle for preventing surgical site infection: a randomized trial.  Arch Surg. 2011;146(3):263-269.PubMedGoogle ScholarCrossref
4.
Lutfiyya  W, Parsons  D, Breen  J.  A colorectal “care bundle” to reduce surgical site infections in colorectal surgeries: a single-center experience.  Perm J. 2012;16(3):10-16.PubMedGoogle Scholar
5.
Zimlichman  E, Henderson  D, Tamir  O,  et al.  Health care-associated infections: a meta-analysis of costs and financial impact on the US health care system.  JAMA Intern Med. 2013;173(22):2039-2046.PubMedGoogle ScholarCrossref
6.
Kirkland  KB, Briggs  JP, Trivette  SL, Wilkinson  WE, Sexton  DJ.  The impact of surgical-site infections in the 1990s: attributable mortality, excess length of hospitalization, and extra costs.  Infect Control Hosp Epidemiol. 1999;20(11):725-730.PubMedGoogle ScholarCrossref
7.
Engemann  JJ, Carmeli  Y, Cosgrove  SE,  et al.  Adverse clinical and economic outcomes attributable to methicillin resistance among patients with Staphylococcus aureus surgical site infection.  Clin Infect Dis. 2003;36(5):592-598.PubMedGoogle ScholarCrossref
8.
Anderson  DJ, Podgorny  K, Berríos-Torres  SI,  et al.  Strategies to prevent surgical site infections in acute care hospitals: 2014 update.  Infect Control Hosp Epidemiol. 2014;35(suppl 2):S66-S88.PubMedGoogle ScholarCrossref
9.
Tang  R, Chen  HH, Wang  YL,  et al.  Risk factors for surgical site infection after elective resection of the colon and rectum: a single-center prospective study of 2,809 consecutive patients.  Ann Surg. 2001;234(2):181-189.PubMedGoogle ScholarCrossref
×