E. Schneider1, J. Canner1, F. Gani1, C. Wolfgang1, M. Makary1, M. Weiss1, G. Spolverato1, Y. Kim1, A. Ejaz1, T. Pawlik1 1Johns Hopkins University School Of Medicine,Surgical Oncology,Baltimore, MD, USA
Introduction: Most studies on readmission report only data on the initial readmission to the index hospital within 30-days of surgery. These data may underestimate the actual impact of readmission, as patients who undergo complex procedures may be readmitted beyond 30-days and/or at other hospitals. We therefore sought to define the incidence of early versus late readmission of patients undergoing pancreaticoduodenectomy (PD) requiring re-hospitalization at any hospital.
Methods: Patients discharged after PD (ICD-9-CM procedure code 52.7) between 2010-2012 were identified from the Truven Health Market Scan database, which is a large convenience sample of individuals covered by employer-provided healthcare. Determinants of early (≤ 30 days) or late (31-90 days) readmission were identified and analyzed.
Results: A total of 2,243 eligible patients underwent PD during the study period. Mean (SD) patient age was 54.8 (8.4), 51.6% of the patients were male, and 85.0% had a Charlson Comorbidity Index of 2 or greater. The mean (SD) length-of-stay was 12.7 (11.6) days; 89.2% of patients were discharged home, 5.0% were transferred to another facility, and 1.6% died in hospital. Among 2,209 patients discharged alive, 450 (20.4%) had an early readmission while 165 (9.4%) had a late readmission. Among patients who were readmitted, fewer than 11 patients (<1.8%) had multiple readmissions. Common causes of readmission were similar among patients experiencing an early versus late readmission (post-op infection: 22.0% vs. 5.5%, P<0.001; dehydration: 5.6% vs. 4.9%, P=0.73). Median length-of-stay was longer for early vs. late readmission (5 vs. 3 days, respectively, P=0.002) and no in-hospital mortality occurred among patients readmitted either early or late. While early readmissions were more likely to occur at the index hospital (index hospital: 94.4% vs. non-index hospital: 5.6%), patients who had a late readmission tended to be re-admitted more often to a different hospital than where the PD had been performed (index hospital: 90.3% vs. non-index hospital: 9.7%)(P=0.06).
Conclusion: One-in-three readmissions occurred beyond 30-days, with 10% of late readmissions occurring at non-index hospitals among patients undergoing PD. Assessment of only 30-day same hospital readmissions may underestimate the true incidence of re-hospitalization following PD.