HOT SPRINGS, VA – Using risk adjustment alone to compare surgical site infection rates among hospitals may not be valid, judging by a retrospective analysis of data on colorectal procedures.
An examination of data submitted on 336,190 cases by 237 hospitals participating in the National Surgical Quality Improvement Project (NSQIP) in 2009 revealed that a patient’s underlying disease state is an important contributor to the risk of surgical site infection, said Dr. Robert Cima.
Dr. Cima and his colleagues at the Mayo Clinic Rochester, Minn., identified 24,673 colorectal procedures, using CPT codes. Of those procedures, 6,324 patients had colon cancer, 2,061 had benign neoplasm, 2,859 had rectal cancer, 4,821 had diverticular disease, 764 had ulcerative colitis, 862 had regional enteritis (Crohn’s disease), and 6,982 had other conditions, which included perforation, volvulus, intestinal obstruction, rectal prolapse, fistula, vascular insufficiency and unspecified neoplasms.
"Comparisons between hospitals could be misleading if they don’t include case mix," Dr. Cima said.
They specifically analyzed colorectal procedures because they are associated with the largest rate of surgical site infections (SSI; 5%-30%) and those infections led to a longer length of stay, higher costs, and higher mortality.
To get a baseline rate of SSIs for comparison purposes, the authors queried the data set to determine which procedure had the lowest rate. Benign neoplasms had the lowest overall rate, and because they were not caused by underlying disorders or malignancies, that rate was chosen as the reference, Dr. Cima said at the annual meeting of the Southern Surgical Association.
The authors conducted a regression analysis using the same factors used in the NSQIP risk adjustment: age, body mass index, American Society of Anesthesiology classification, wound classification, and relative value units (RVU) used, which are a surrogate for CPT codes. Odds ratios for SSIs were then calculated for each procedure.
Overall, 13.5% of patients developed an SSI. Rectal cancer patients had the highest overall odds ratio at 1.9, followed by Crohn’s disease, at 1.7. Rectal cancer patients also had the highest risk of superficial incisional SSI at 1.6; diverticular disease patients had a 1.6-fold higher risk than did those with benign neoplasms.
The patients at highest risk for deep incisional infections were those with ulcerative colitis (OR, 2.4), followed by rectal cancer (2.1). Both of those conditions also put patients at higher risk for organ/space infections (OR, 2.2 and 2.1).
The difference in infection rates by underlying disease led the researchers to question whether case mix might have an effect on the infection rates at individual institutions. Looking at the Mayo Clinic’s profile, compared with all NSQIP facilities’ case mix, they found that there was a higher percentage of patients with the higher-risk conditions: rectal cancer, regional enteritis, and ulcerative colitis.
The NSQIP data include a disease diagnosis, but it does not factor that into risk-adjusted figures, which will soon be publicly reported, noted Dr. Cima. Comparisons between hospitals could be misleading if they don’t include case mix, he said. And, there are implications for quality improvement, said Dr. Cima.
"If we had just looked at the overall surgical site infection rate and tried to design studies or processes to just reduce it without consideration of case mix, we would not be able to really determine whether or not we can drive it down," he said.
"This is a provocative and important study," said Dr. Danny Jacobs, chairman of the department of surgery at Duke University, Durham, N.C., in commenting on the paper. And, he said, it raises questions about whether the risk-adjustment system being used by NSQIP is adequate.
Dr. Susan Galandiuk, a professor of surgery at the University of Louisville (Ky.), questioned whether infection rates weren’t also influenced by the duration of surgery and whether patients were given appropriate antibiotic prophylaxis.
Dr. Cima said that while many studies have shown that infections rise as the length of surgery increases, it was not clear whether that was a factor in his study.
He agreed with Dr. Jacobs that the paper raised questions. The data do "go to the very heart of using data sets to drive quality improvement," he said.