Variable post-CABG infection rates may hinge on hospital, surgeon-level factors

Hospital-acquired infection rates after CABG differed by 18.2 percent in an analysis of 33 hospitals in Michigan that was published online July 1 in Circulation: Cardiovascular Quality and Outcomes. Patient mix did not explain the variation.

Terry Shih, MD, of the University of Michigan in Ann Arbor, and colleagues turned to the Michigan Society of Thoracic and Cardiovascular Surgeons Quality Collaborative to design the study. The collaborative uses the Society of Thoracic Surgeons data collection protocols, and the 33 participating hospitals in Michigan submit data to both groups. Their goal was to compare observed and expected post-CABG rates of hospital acquired infections.

The defined infections as the presence of pneumonia, sepsis or septicemia, harvest or cannulation site infection, deep sternal wound infection and thoracotomy or parasternal site infection. They used a model to determine predicted rates of infection with predicted probabilities based on patient characteristics. Hospitals fell into one of three categories by infection rate: low, medium or high.

Shih and colleagues identified 20,896 cases of isolated CABG between 2009 and 2012 for the study. Patients who had infections were more likely to be older, with a higher body mass index and more comorbidities.

They found the overall rate of hospital-acquired infection post-CABG group was 5.1 percent. Rates by individual type of infection ranged from 3.1 percent for pneumonia to 0.02 percent for isolated thoracotomy. Pneumonia dominated as infection type for both low- and high-rate hospitals, but it accounted for only 1.5 percent of infections at low-rate facilities as opposed to 8.4 percent at high-rate hospitals.

The predicted rate of infections varied by 2.8 percent due to patient case mix. But the observed rate varied by 18.2 percent.

“In our analysis, 4 centers seem to be high outliers for observed HAI [hospital-acquired infection] rates,” they observed. “However, the predicted rate of HAI does not differ markedly from other centers, suggesting that this difference in observed HAI rates is not because of differences in the case mix. Institutional or surgeon-level factors may play a role in explaining this variation.”

Given that pneumonia is the predominant infection type, they recommended preoperative preventive interventions such as respiratory muscle training and smoking cessation as well as postoperative approaches. Checklists and computer prompts may be helpful, they argued, but a culture that emphasizes safety and invests in infection control efforts and the use of multidisciplinary care teams adds value, too.