Multivariable Predictors of Postoperative Surgical Site Infection after General and Vascular Surgery: Results from the Patient Safety in Surgery Study
“We envision a system wherein an individual patient’s risk factors will be entered into their record as part of their history, physical examination, and preoperative evaluation. The SSI risk index analysis would then be electronically calculated and the individual patient’s risk for SSI (and potentially other complications) would be provided to the surgeon, along with a history and physical document.”
—Neumayer et al.1
Funded by the Agency for Healthcare Research and Quality, Grant 5U18HS011913 (“Reporting System to Improve Patient Safety in Surgery”).
A total of 184,120 patients, from fiscal years 2002 through 2004, undergoing vascular and general surgical procedures at 14 private-sector hospitals and 128 Department of Veterans Affairs (VA) medical centers.
Who Was Excluded:
Patients with preoperative wound infections or sepsis, nongeneral surgery or nonvascular procedures, and patients with missing data were excluded. A total of 20,750 patients were excluded: 19,558 with a preoperative superficial, deep wound infection or preoperative sepsis; 938 due to procedures not performed by vascular or general surgeons; 254 due to missing data.
Each facility had a trained clinical nurse reviewer who prospectively collected data about patient characteristics, intraoperative variables, and postoperative adverse outcomes. Data was collected 30 days after the index operation at VA hospitals and 40 days in private-sector hospitals, with data collection performed once every 8 days.
Surgical site infection (SSI) was the primary endpoint studied. Preoperative demographics, medical risk factors, preoperative laboratory values, type of procedure, and relative value units (RVU) of the operation as a proxy for surgical complexity were examined.
Patients were grouped into 7 procedure categories based on the body part or system involved:
1. Integumentary and musculoskeletal system
2. Thoracoabdominal aneurysm, embolectomy/thrombectomy, venous reconstruction, and endovascular repair
3. Aneurysm, blood vessel repair, thromboendarterectomy, angioscopy, angioplasty and atherectomy, bypass and composite grafts, other artery and vein
4. Respiratory system, hemic and lymphatic systems, mediastinum and diaphragm
5. Mouth, palate, pharynx, and esophagus
6. Stomach, intestines, appendix and the mesentery, rectum and anus, liver, biliary tract, pancreas, abdomen, peritoneum, and omentum (nonhernia)
7. Hernioplasty, herniorrhaphy, herniotomy, and endocrine
Stepwise regression analysis was performed, with all independent variables initially included in the statistical analysis as potential risk factors. A scoring system was created. The odds ratio of developing a SSI for each variable was rounded to the closest whole number and became the scoring point assigned to that potential risk factor.
A predicted SSI rate was produced from this logistic regression model, then compared with the actual observed SSI rate using a chi-square goodness of fit test. This scoring system was also tested against the National Nosocomial Infections Surveillance system risk score, a system originally developed by the Centers for Disease Control and Prevention and private-sector collaborators, a risk score calculated using the American Society of Anesthesiologis (ASA) class, operating room time, and wound classification.
SSI occurred in 7,035 (4.3%) of the patients in the study. Twenty-eight preoperative risk factors and 10 preoperative laboratory values were associated with an SSI on bivariate analysis.
Patients who developed an SSI were more likely to be older and male, have a higher ASA class, have >2 alcohol drinks per day, be smokers, be diabetic, have low hematocrit, have high white blood cell count, be inpatient, have emergency surgery, undergo general anesthesia, have cancer, have weight loss >10%, have low albumin, have other comorbidities, have a wound class other than clean, have a complex procedure, and have surgery duration >1 hour.
Fourteen variables were independently associated with SSI on regression analysis: age over 40, smoker, >2 alcohol drinks/day, use of steroids, diabetes, dyspnea, recent radiation therapy, serum albumin <3.5 mg/DL, total bilirubin >1, ASA class >2, emergency surgery, type of surgery (gastrointestinal procedures and procedures on the skin and musculoskeletal system had the highest risk), and work RVU >10.These variables were utilized to calculate a SSI risk score, divided into three categories: low risk (score of 1–5), medium risk (score of 6–8), and high risk (score >8).
SSI risk score accurately predicted actual SSI rate in each category, and when compared with the National Nosocomial Infections Surveillance score, was found to more accurately predict SSI.
Thirty days after the index operation at VA hospitals and 40 days in private sector hospitals.
Criticisms and Limitations:
• The model is limited by the absence of the use and timing of prophylactic antibiotics.
• The study did not incorporate organ space infections.
• Data beyond 30 days for patients with prostheses inserted at the surgical site were not collected.
This was one of the first major studies to delineate the key risk factors for developing a SSI. Key variables independently associated with SSI are age over 40, smoker, > 2 alcohol drinks/day, use of steroids, diabetes, dyspnea, recent radiation therapy, serum albumin <3.5 mg/DL, total bilirubin > 1, ASA class > 2, emergency surgery, type of surgery (gastrointestinal procedures and procedures on the skin and musculoskeletal system highest risk), and surgery complexity. This tool can be used by clinicians to preoperatively predict SSIs. The SSI risk index model can be utilized to assist surgeons and patients in preoperatively optimizing modifiable factors, such as nutrition status, timing of surgery, control of diabetes, and smoking and alcohol cessation.2
Clinical Case: Surgical Site Infection Risk
A 55-year-old man presents to the emergency room department with right upper quadrant pain, nausea, and vomiting for the past 4 hours. He has no significant past medical history, is a nonsmoker, and occasionally drinks a glass of red wine once per week. His vital signs are within normal range. On physical exam he has a positive Murphy’s sign. He has no abdominal scars. A transabdominal ultrasound identified numerous stones within the gall bladder and pericholecystic fluid. Lab results revealed an elevated white blood cell count of 14. Bilirubin and albumin are normal. Based on the risk index model developed from the Patient Safety in Surgery Study, what is the patient’s predicted risk of developing a SSI?
• A. 1%
• B. 3%
• C. 5%
• D. 7%
B. 3%. Based on the history, physical exam, imaging, and labs, this individual has cholecystitis. Antibiotic therapy and laparoscopic cholecystectomy are warranted. Based on Table 6 from the study, the following points will be assessed in the patient:
• Type of operation:
• Stomach, intestines, and others: 2 points
• Work RVU >17 points: 4 points
• Age ≥40 years old: 1 point
The score can be used to assess the patient’s risk using Table 7 from the study:
• Score of 7: Medium risk
• Predicted SSI risk in this patient: 3.07%
1. Neumayer L, Hosokawa P, Itani K, El-Tamer M, Henderson W, Khuri S. Multivariable predictors of postoperative surgical site infection after general and vascular surgery: results from the Patient Safety in Surgery Study. J Am Coll Surg. 2007 Jun;204(6):1178–87.Find this resource:
2. Centers for Medicare and Medicaid Services. http://www.cms.hhs.gov/physicianfeesched/pfsrvf/list.asp. Accessed September 10, 2002, for 2002 data; June 3, 2003, for 2003 data; and November 8, 2004, for 2004 data.