Understanding Variation in Postacute Care: Developing Rehabilitation Service Areas Through Geographic Mapping

Abstract

Objective: The aims of the study were to demonstrate a method for developing rehabilitation service areas and to compare service areas based on postacute care rehabilitation admissions to service areas based on acute care hospital admissions.

Design: We conducted a secondary analysis of 2013-2014 Medicare records for older patients in Texas (N = 469,172). Our analysis included admission records for inpatient rehabilitation facilities, skilled nursing facilities, long-term care hospitals, and home health agencies. We used Ward's algorithm to cluster patient ZIP Code Tabulation Areas based on which facilities patients were admitted to for rehabilitation. For comparison, we set the number of rehabilitation clusters to 22 to allow for comparison to the 22 hospital referral regions in Texas. Two methods were used to evaluate rehabilitation service areas: intraclass correlation coefficient and variance in the number of rehabilitation beds across areas.

Results: Rehabilitation service areas had a higher intraclass correlation coefficient (0.081 vs. 0.076) and variance in beds (27.8 vs. 21.4). Our findings suggest that service areas based on rehabilitation admissions capture has more variation than those based on acute hospital admissions.

Conclusions: This study suggests that the use of rehabilitation service areas would lead to more accurate assessments of rehabilitation geographic variations and their use in understanding rehabilitation outcomes.

FIGURE 1.

Cohort flow diagram with exclusion criteria.

FIGURE 2.

Rehabilitation service areas based on Ward’s clustering algorithm (22 numbered areas) with corresponding HRRs (dark lines). White area denotes unpopulated areas.