Study Suggests Previous Hospital Readmission Rates for Three Common Conditions are Poor Predictors for Future Readmission
As a result of high hospital care costs and their prevalence, hospital readmissions have come under scrutiny as a potential source of cost savings. This study sought to assess whether historic hospital readmission rates predict risk-adjusted patient readmission — and to measure the costs of readmission. Using data from VA acute care hospitals from 2005-2009, investigators focused on three common conditions for hospitalization among Veterans: acute myocardial infarction (AMI), community acquired pneumonia (CAP), and congestive heart failure (CHF). Readmissions were defined as subsequent non-transfer, all-cause admissions to a VA hospital within 30 days of discharge from an index admission. Patients were excluded if they died during the admission, were transferred to another hospital, or had a hospital length of stay less than 6 hours. Costs were taken from VA Decision Support System and included all costs of treatment involved in the patient inpatient care. Investigators also analyzed characteristics of episodes of care, patient demographics, and comorbid conditions for Veterans with AMI (n=15,603 admitted to 35 VA hospitals), CAP (n=63,679 admitted to 101 VA hospitals), or CHF (n=64,391 admitted to 90 VA hospitals).
- Previous hospital readmission rates are poor predictors of readmission for future individual patients, so policies using these meaures to guide subsequent reimbursement might prove problematic.
- Patients who are readmitted do have substantially higher episode costs, even after conventional risk adjustment. Being readmitted increased total episode cost by 53% for Veterans with AMI, 83% for Veterans with CAP, and 80% for Veterans with CHF. These costs are net of the fixed cost of occupying the hospital bed for additional days.
- To the extent that some patients may seek care outside the VA healthcare system after their index hospitalization, this study may have under-counted the cost of readmission. Also, this study did not include the costs of outpatient care surrounding the index hospitalization.
- Investigators did not include detailed clinical data in their models, raising concerns about potential unobserved patient-level heterogeneity that might affect readmission. However, this concern is limited in light of previous studies that show additional clinical data have low predictive value.
- Study findings may provide bounds for how reimbursement by non-VA payers would need to be modified if moving to bundled payments for episodes of hospitalization.
- As private sector hospitals are exposed to increasing payment policy pressure related to readmission penalties and payment bundling, these organizations will have an increased incentive to steer dual VA-Medicare eligible patients, particularly the clinically complex, toward VA providers.
This study was funded by HSR&D (IIR 08-067). Drs. Hockenberry, Vaughan-Sarrazin, and Kaboli and Mr. Glasgow are part of HSR&D Center for Comprehensive Access and Delivery Research and Evaluation located in Iowa City, IA. Dr. Burgess is part of HSR&D Center for Organizational Leadership and Management Research in Boston, MA.
Hockenberry J, Burgess J, Glasgow J, Vaughan-Sarrazin M, and Kaboli P. Cost of Readmission: Can the VHA Experience Inform National Payment Policy? Medical Care 2013 Jan;51(1):13-19.