J. K. Canner1, S. Pourzal1, H. AlSulaim1, K. E. Steele1 1Johns Hopkins University School Of Medicine,Department Of Surgery,Baltimore, MD, USA
Introduction: Hospitals accredited by the Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP) are required to report readmission rates. However, such rates may not account for readmissions to other hospitals if follow-up data are incomplete. We calculated a more accurate readmission rate using a nationally representative database that tracks readmissions at all hospitals within the same state. We then investigated factors associated with readmission to a different hospital.
Methods: The new Nationwide Readmissions Database (NRD) from the Healthcare Cost and Utilization Project (HCUP) contains data on inpatient hospital stays from 21 states that collect data linkage information sufficient for identifying readmissions. We identified patients in the NRD admitted for elective bariatric surgery in 2013 and collected patient demographics, including age, gender, insurance status, and residence, as well as clinical information such as length of stay, Charlson comorbidity index, and APR-DRG severity score. We calculated the proportion of patients readmitted within 30 days after initial discharge and also recorded the APR-DRG severity and elective status for the readmission and whether the readmission was to the same hospital.
Results: A total of 61,220 NRD patients underwent elective bariatric surgery in 2013. Of these, 3,860 (6.3%) were readmitted within 30 days. Of those readmitted, 693 (18.0%) were to a different hospital. Patients readmitted to a different hospital were more likely to be covered by Medicare (OR=1.46; p=0.036) or Medicaid (OR=1.62; p=0.011) than be privately insured, less likely to live in a medium-sized metro area than in a large metro area (OR=0.61; p=0.013), and less likely to have their surgery at a teaching hospital than at a non-teaching hospital (OR=0.71; p=0.021). Readmission to a different hospital was strongly associated with higher APR-DRG severity (OR=1.69; p=0.001) and non-elective status at readmission (OR=2.28; p=0.002). Patient age, sex, income level, out-of-state residence, comorbidities, type of surgery and length of stay were not associated with location of readmission. These relationships persisted with multivariable analysis, with the exception of Medicare coverage.
Conclusion: Failure to account for readmissions to different hospitals may underestimate readmission rates by at least 18%. Patients with more severe complications are the most likely to be readmitted to different hospitals. A better understanding and accounting for all readmissions may improve the care and safety of the bariatric surgical patient. Further research using data sets with more detailed geographic information may reveal the role of distance as a factor in readmission location.
