PS 4-40
APPROACH TO EVIDENCE-BASED OPERATIONAL MODELING OF A HIGH-VOLUME SURGICAL HOSPITAL
Surgical site infections (SSIs) after orthopedic surgery have large personal and societal costs despite low incidence rates. Implementing strategies to alter modifiable SSI risk factors can be disruptive and costly in a hospital environment; using Evidence-Based Operational Modeling (EBOM) to estimate their impact may improve patient care and hospital efficiency. We used data from a high-volume surgical hospital to establish the evidence base for such simulations.
Method:
We evaluated electronic medical record data on 7,420 patients undergoing spine surgery and 34,965 patients undergoing elective hip and knee total joint replacement (TJR) surgeries from January 1, 2008 to June 30, 2013. Patients who had SSIs were identified by the hospital infection control department. Differences between cases and controls were compared using t-tests, Fisher’s exact tests or chi-square tests, as appropriate.
Result:
The average age of spine cases was 58.6 years, (SD 16.1), 63% were women, and mean BMI was 29.3, (SD 7.0); the average age of TJR cases was 64.4 years, (SD 10.0), 54% were women, and mean BMI was 32.6, (SD 8.4). Patients with SSIs after all surgery types were more likely to be overweight (BMI>30); had uncomplicated diabetes, hypertension, and ASA score above 2 (p-value <0.05 for all). Patients with SSI also had longer procedure durations (290 vs. 224 minutes for spine, 92 vs. 82min for TJR) and longer length of stay (median 7 vs. 4d for spine and 4 vs. 3d for TJR, p<0.001 for all). Spine surgery SSIs were more frequent among female patients (p=0.039), among those with paralysis or rheumatological diseases (p<0.01), and for non-cervical-level surgery (p<0.001), and had longer stays in the OR before and after surgery (63 vs. 54min, p<0.001). TJR SSIs were more common among patients who self-identified as minority (p=0.003), had chronic pulmonary disease, coagulopathy, or depression (p<0.01) or had acquired immune deficiency syndrome (p=0.045).
Conclusion:
Despite the low rate of SSIs in this patient population, we characterized a variety of patient and facility-level factors that distinguish cases complicated by post-operative infection. Ongoing process modeling using operative management data will provide an overall patient flow framework into which these data may be used to power EBOMs of surgical patient flow and outcome.