Risk-adjustment for determining surgical site infection in colon surgery: are we using appropriate variables?

Sydne Muratore, University of Minnesota Department of Surgery; Catherine Statz, University of Minnesota Department of Surgery; JJ Glover, University of Minnesota Department of Surgery; Mary Kwaan, University of Minnesota Department of Surgery; Greg Beilman, University of Minnesota Department of Surgery

Surgical site infection (SSI) rates after colon surgery are being used as a hospital quality measure in the Medicare Hospital Value-Based Purchasing (VBP) program. Reliable reporting and risk adjustment are increasingly debated as VBP assigns greater weight to clinical outcomes.

We expect to identify a discrepancy between two discrete risk-adjustment models generated through the Center for Disease Control’s (CDC’s) National Health and Safety Network (NHSN).

We used prospective, clinical data on complex SSIs reported to NHSN from 2012-2014 for a metropolitan health system (6 hospitals, 1,596 procedures). We compared expected quarterly SSIs generated by risk-adjustment models utilized by Centers for Medicare and Medicaid Services (CMS) versus NHSN. The CMS model includes age and ASA classification. The NHSN model includes age, ASA classification, operative length, endoscope/laparoscope, medical school affiliation, hospital bed size, and wound classification. Comparisons were performed using nonparametric statistics.

Patients who developed SSIs had the following NHSN risk factors: 92% open procedures, 49% no medical school affiliation, 85% age < 75, 46% ASA > 2, 21% contaminated/dirty wound classification, and 87% OR length ≥100 minutes. Comparison of expected complex SSI rates in colon surgery for all hospitals (1,596 procedures) showed lower rates predicted by CMS vs NHSN (4.54 vs 4.84, p = 0.003). This difference was again evident when comparing rates within an academic medical center (640 procedures; 1.86 vs 2.25, p = 0.003) and a community hospital (469 procedures; 1.30 vs 1.42, p = 0.005).

The current risk adjustment model using only age and ASA classification by CMS predicts consistently lower rates of expected SSI in colon surgery cases throughout a metropolitan health system when compared to expected rates using expanded criteria supported by NHSN. This may lead to lower reimbursement rates due to use of limited risk factors. Further efforts at elucidating appropriate risk adjustment measures without unnecessarily burdening hospitals with expensive data collection is vital to ongoing efforts at quality improvement in US hospitals.