A taxonomy for classification of stroke rehabilitation services

Objective: To develop a taxonomy for use in measuring stroke rehabilitation services. Design: A cross-sectional study using facility-level survey data and extant data files. Setting: Veterans Administration medical centers (VAMCs). Variables: (1) A list of rehabilitation characteristics, including personnel, physical facilities, coordination of care, and hospital characteristics; and (2) a classification or typology of VAMCs according to the type of postacute stroke care on-site. Main Outcome Measures: Data sources included extant Veterans Administration (VA) computerized databases, VA central office administrative files, and 2 mailed surveys to VA rehabilitation medicine services and stroke acute care services. The rehabilitation taxonomy was derived using 2 methods that assess face and construct validity, respectively: (1) an expert panel rating, using a modified Delphi process, of the clinical importance of each of the rehabilitation characteristics; and (2) a comparison of rehabilitation characteristics across the different types of VAMCs. Variables were included in the final taxonomy if the expert panel reached consensus that the variable was clinically important, or if there were statistically significant differences in these characteristics across the different types of medical centers. Results: Of 67 possible rehabilitation characteristics, a multidisciplinary expert panel reached consensus about the likely clinical importance of 21 rehabilitation characteristics, 11 of which showed statistically significant differences across different types of VAMCs. An additional 9 variables that lacked expert panel consensus differed significantly among the differ ent medical centers. These 30 variables represent a preliminary taxonomy of key rehabilitation characteristics. Among the 20 variables that varied significantly across the different types of medical centers, 18 showed a pattern with the greatest amount of resources and organizational sophistication being found in VAMCs with rehabilitation units, followed by medical centers with geriatric units, and the least amount of resources and organizational sophistication was seen in medical centers whose postacute care services were limited to nursing home or intermediate care. Conclusion: Thirty rehabilitation characteristics had face validity and/or construct validity, and can be considered to represent a preliminary taxonomy for measuring stroke rehabilitation services. This study also shows that there are significant differences among hospitals in resources and organization of care deemed to be important for stroke patients.