D. Metcalfe1, O. A. Olufajo1, C. Zogg1, A. J. Rios Diaz1, A. H. Haider1, M. B. Harris2, M. J. Weaver2, A. Salim1 1Harvard Medical School,Center For Surgery And Public Health,Boston, MA, USA 2Brigham & Women’s Hospital,Department Of Orthopedic Surgery,Boston, MA, USA
Introduction:
30-day hospital readmission is used as a quality metric in some pay-for-performance frameworks, such as the CMS Readmissions Reduction Program (CRRP). There are plans to extend the CRRP to selected surgical populations. However, the odds of unplanned readmission have been shown, in some surgical settings, to be associated with lack of insurance and Black race. These characteristics are also associated with greater odds of injury, which raises the possibility that trauma centers will be unfairly penalized by extension of the CRRP to include injured patients.
This study characterized the reasons for, and factors associated with, unplanned 30-day readmission of orthopedic trauma patients. We also sought to understand whether it is sufficient to limit measurement of readmissions to the hospital at which patients were initially treated by exploring the proportion that were readmitted to other hospitals.
Methods:
Hospital admissions for fracture and/or dislocation (ICD-9-CM 800-839) were extracted from the California State Inpatient Database (SID) 2007-2011, which is an all-payer dataset that captures 98% of hospital admissions. Isolated rib, skull, and facial fractures were excluded. Unplanned readmissions to any hospital in California were tracked using a unique statewide identifier. Multivariable logistic and generalized linear regression models were used to identify independent associations with readmission. The covariates within these models were age, sex, race, payer status, admission source, weekend admission, Injury Severity Score (ISS), Charlson Comorbidity Index (CCI), hospital bed size, trauma center designation, and teaching hospital status.
Results:
There were 416,568 orthopedic trauma admissions to 391 different hospitals. The population was predominantly older (mean age 63.9, SD 23.6), white (71.5%), male (59.5%), and funded by public insurance (63.9%). Severely injured patients (≥15) accounted for only 3.3% cases. 27,008 (6.5%) were readmitted within 30 days, 27.6% of which to a different hospital. Factors significantly associated with 30-day readmission were older age (>65 adjusted odds ratio 1.32, 95% CI 1.24-1.41), Black race (aOR 1.18, 1.10-1.26), public insurance (aOR 1.40, 1.27-1.54), greater comorbidity burden (CCI >2 aOR 1.09, 1.77-1.89), and spine fracture (aOR 1.38, 1.31-1.45). Major reasons for readmission included cardiopulmonary disease (25.9%), procedural complications (12.8%), and musculoskeletal problems (8.5%).
Conclusion:
Many orthopedic trauma readmissions are for cardiopulmonary disease and potentially unrelated to the quality of their index hospitalization. Penalties for unplanned readmissions risk unfairly penalizing hospitals that serve disadvantaged communities and treat a high proportion of trauma patients. Future work should aim to determine the proportion of readmissions that are truly avoidable given optimal trauma and medical care.