Using Rehabilitation Services Areas (RSAs) to Characterize Post-Acute-Care Utilization in Stroke

PURPOSE: Geographic variations exist in care use and health outcomes (Reistetter et al., 2015). Three tools currently available to examine geographic variations are Hospital Referral Regions (HRRs), Hospital Service Areas (HSAs), and Primary Care Service Areas (PCSAs). These geographic boundaries are developed for healthcare sectors other than post-acute rehabilitation. HRRs and HSAs are designed for acute-care hospital services while PCSAs are for primary care physician services. No geographic boundary or tool has been designed for describing post-acute rehabilitation services, where the Institutes of Medicine cited 73% of variation in Medicare spending occurs (Institute of Medicine, 2013). Our research team developed Rehabilitation Services Areas (RSAs) to address this gap. We presented information on stroke— a large subset of patients receiving post-acute rehabilitation care—to demonstrate the process and need for RSAs. Findings will increase occupational therapists’ knowledge about small-area regional differences in rehabilitation services provision for persons living with stroke.

DESIGN: Retrospective secondary analyses of 2013–2014 Medicare claims data were conducted to estimate and compare post-acute care utilization across three geographic boundary tools: HRRs, HSAs and RSAs. The HRRs and HSAs are currently the most commonly-used geographic tools. We linked 100% Centers for Medicare and Medicaid Services (CMS) data from the Medicare Provider and Analysis Review, Master Beneficiary Summary, and Provider of Service with U.S. Census Bureau ZIP Code Tabulation Areas (ZCTA) file. We identified patients with stroke using ICD-9 codes 430-438 (McCormick, Bhole, Lacaille, & Avina-Zubieta, 2015).

METHOD: To develop the RSA, we employed a Ward’s clustering algorithm (Shwartz, Payne, Restuccia & Ash, 2001) to identify geographic areas served by inpatient rehabilitation facility (IRF), skilled nursing facility (SNF), long-term care hospital (LTCH), and home health agency (HHA). We then mapped these market areas using ArcGIS 10.3 to visualize the patterns of care. Additionally, we explored three types of regions (urban, suburban and rural) to examine our primary outcome—post-acute care utilization for individuals with stroke. Finally, we used Intraclass Correlation Coefficients (ICCs) to compare the variations explained by each of the three geographic tools, where tools with higher ICCs represented better coverage of variance in the geographic areas.

RESULTS: RSA showed the highest ICCs in suburban and rural regions compared to HRR and HSA (suburban: 23.5%, 22.2% and 14.5 % for RSA, HSA and HRR, respectively; rural: 19.0%, 18.1% and 9.6% for RSA, HSA and HRR). However, for the urban region, HSA had the highest ICC compared to RSA and HRR (11.3%, 8.9% and 5.7% for HSA, RSA and HRR).

CONCLUSION: This study’s findings suggested that the RSA best explained the variance of post-acute care utilization in the suburban and rural regions for Medicare beneficiaries with stroke. HSA performed better than HSA in urban area, suggesting a need for future study to examine the function of RSA in higher populated areas.

IMPACT STATEMENT: There has been limited study of geographic variations in rehabilitation provision. Most studies primarily applied hospital area boundaries to explore post-acute care. Understanding the use of RSAs will add value to the post-acute rehabilitation services that occupational therapists provide. Our study is the first to characterize the use of rehabilitation services areas to characterize post-acute care utilization and address the value of post-acute rehabilitation services. Our finding will potentially guide future post-acute policies based on these boundaries.

References

Reistetter, T. A., Kuo, Y. F., Karmarkar, A. M., Eschbach, K., Teppala, S., Freeman, J. L., & Ottenbacher, K. J. (2015). Geographic and facility variation in inpatient stroke rehabilitation: multilevel analysis of functional status. Arch Phys Med Rehabil, 96(7), 1248-54. doi: 10.1016/j.apmr.2015.02.020.

Institute of Medicine (2013). Variation in health care spending: Target decision making, not geography. The National Academies Press, Washington, D.C.

McCormick, N., Bhole, V., Lacaille, D., & Avina-Zubieta, J. A. (2015). Validity of diagnostic codes for acute stroke in administrative databases: A systematic review. PLoS One, 20, 10(8): e0135834. doi: 10.1371/journal.pone.0135834.

Shwartz M, Payne, S. M. C., Restuccia, J., D., & Ash, A. S. (2001). Does it matter how small geographic areas are constructed? Ward’s algorithm versus the plurality rule. Health Serv Outcomes Res Methodol, 2 (1), 5-18.