Gains reported in SSI prevention and detection

Interventions, culling patient data prove effective

A comprehensive infection control program to prevent surgical site infections (SSIs) following coronary artery bypass grafts (CABGs) can save both lives and money, researchers report.

In particular, deep chest SSIs — which entail a higher risk of patient mortality — were reduced significantly after an infection control program was implemented by Samuel McConkey, MB , fellow in infections disease at Washington University School of Medicine in St. Louis, and colleagues at Barnes Jewish Health Center. The study was presented recently in Denver at the Infectious Disease Society of America (IDSA) conference.1

"A lot of people find it difficult to put in practice a program like this," McConkey tells Hospital Infection Control. "We were very fortunate to have very good relationships and a good liaison between the surgical teams, the OR staff, and the infection control practitioners. If there is an antagonistic relationship, nothing will work. We worked with the surgeons and the OR staff and identified the measures together. Clearly, having low wound infection rates is in everybody’s interest."

The researchers followed a prospective cohort of patients undergoing a total of 2,231 CABG procedures before, during, and after implementation of the control program (April 1991-December 1994). Infection control efforts included a number of practices supported in the medical literature, including prospective surveillance, reporting surgeon-specific SSI rates, hair clipping rather than shaving, limitation of operating room traffic, and minimization of flash sterilization.

"We also changed things like bathing," he adds. "[Patients] had to bathe in chlorhexidine the night before and the morning of surgery."

Among the most important components was implementation of a surveillance system that used both microbiology lab data and ICP observations to identify SSIs, he explains. Confidential surgeon-specific rates were complied and reported quarterly for four sites: deep chest, superficial chest, deep leg, and superficial leg.

"[We] communicated how many site infections they had at each of those four sites and then showed the average rates for the other surgeons," McConkey says. "So each individual surgeon saw what the general trend was, and his or her own specific trend."

While such rate reporting has been shown to reduce infections, the simultaneous implementation of multiple interventions makes it impossible to single out the efficacy of individual measures, he notes.

"We are not able to really say which of the 10 interventions we did truly made a difference," he says. "We can see that our rates came down over the period that we introduced these things, and having adjusted it for multiple patient factors, it still came down quite significantly."

Indeed, rates for deep chest SSIs declined from 2.6% in 1991 to 1.6% in 1994. Over the same period, leg SSIs decreased from 6.8% to 2.7%. Median length of stay went from nine to seven days. The differences could not be explained by factors related to the patients, surgeons, or type of surgery.

"It could have been potentially [with health care delivery changes] that we just ended up with healthier, younger patients who get less infections," he notes. "We found that when we adjusted for patient factors, there is still quite a significant downward trend."

In addition, the authors conducted a case control study in 1995 looking at the cost factors. Despite statistical limitations, the findings provide a general indicator that efforts to prevent such infections can be cost-effective, as developing an SSI was associated with substantial increase in expenditures. For example, the cost of hospitalization for a deep chest SSI was $87,463 compared to $35,218 for non-infected patients.

Still, the cost data should be considered only a crude overall comparison rather than a true measure of the attributable costs of an SSI when all other patient variables are accounted for, he cautions. For example, many patients who developed SSIs also had diabetes or renal failure, or were elderly.

"We know that all of those factors on their own — without getting a wound infection — will increase your costs," McConkey says. "I can’t say that all of the difference in costs between those who got wound infection and those who didn’t is due to the wound infection."

Nevertheless, there are patient outcome incentives beyond the immediate financial benefit, he reminds.

"The deep chest infections have about a 20% to 30% mortality," he says. "If we can decrease the rate of deep chest infections by 30% to 40%, that is hopefully going to translate into significant mortality differences as well. Clearly, if you can show a decrease in rates of infection, that is very likely to translate into cost [savings]. But actually translating the cost-benefit is not an easy business."

The program did not include a formal post-discharge surveillance component (i.e., patient contact at 30 days), but two systems were in place that made it unlikely that many SSIs would be missed, he notes. First, the cardiothoracic surgeons routinely had their patients return to the hospital for post-op visits, so the physicians would be among the first to detect a potential wound infection, he notes. Secondly, the hospital computer system routinely raises red flags based on microbiology results like the presence of Staphylococcus aureus on a wound swab, he adds. The median time to infection was around six to seven days, though some occurred as late as 30 to 40 days.

A database for post-discharge SSIs

The difficulty of detecting post-discharge SSIs has been well-established, but another study presented at the IDSA conference reported that reviewing commonly available patient data could capture many infections. Kenneth Sands, MD, MPH, hospital epidemiologist at Beth Israel Jewish Medical Center in Boston, and colleagues evaluated the use of claims-based outpatient data on antibiotic dispensing and hospital readmissions as a methodology for identifying SSIs.2

The study population was a cohort of 4,086 procedures among which 96 post-discharge SSIs had been identified previously by extensive review of outpatient records. Through a process called "recursive partitioning," the researchers identified which indicators most efficiently select out cases complicated by a post-discharge SSI. They found that 77% of post-discharge SSIs can be detected by red-flagging medical records for two markers: prescription of anti-staphylococcal antibiotics within 30 days after discharge and hospital readmission associated with a diagnosis of skin/soft tissue infection. The process has a predictive value of 24%, which means about one-quarter of the suspected infections will be confirmed as a SSI, he says.

"Seventy-five percent of them still wouldn’t be a wound infection," he explains "That seems like a lot, but it’s still better than having to look at every outpatient chart. What it basically allows you to do is look at [data] that represent only 7% of all outpatient records, and [find that] one in four of the people will have a wound infection."

Such data are commonly available in managed care organizations, as has been recently emphasized by the Centers for Disease Control and Prevention. Indeed, Sands’ data represent some of the Harvard Pilgrim Health Plan findings that have raised hopes at the CDC that health care delivery changes may open up new opportunities in surveillance. (See HIC, December 1998, p. 179.)

"It’s not that fancy," Sands tells HIC. "We’re basically talking about billing data and pharmacy dispensing data, which is something that most HMOs track now. . . . It’s very expensive to do surveillance and collect data, so if you can use data that’s being collected by somebody else for another reason — what could be better? We acknowledge it is not perfect. You will never get an exact infection rate just by looking at billing data, but you might get something that is close enough that you can use it to track infections, focus your program, or find a problem area that you want to more closely examine."

References

1. McConkey SJ, L’Ecuyer PB, Murphy DM, et al. Results of a comprehensive program for reducing surgical site infections in coronary artery bypass surgery. Abstract 620 Sa. Presented at the annual meeting of the Infectious Diseases Society of America (ISDA). Denver; Nov. 12-15, 1998.

2. Sands K, Vineyard G, Platt R. Detection of surgical site infections by outpatient pharmacy dispensing records and hospital discharge diagnoses. Abstract 611 Sa. Presented at the annual meeting of the Infectious Diseases Society of America (ISDA). Denver; Nov. 12-15, 1998.