Practice variation in physical therapy: Development of a causal model using the disorders adhesive capsulitis of the shoulder and sciatica

Unwarranted variation in hospital utilization is well documented. Proposed explanations for medical practice variation (MPV) include physician practice style, medical resource supply and community socioeconomic status. Causal connections between these factors and MPV have not been tested, nor has the impact of variation on clinical outcomes been thoroughly addressed. Variation in physical therapy services (PTPV) also has been observed but research is limited. The purposes of this study were to: (1) determine to what extent PTPV occurs in the management of two musculoskeletal disorders; (2) examine the relationship between PTPV and practitioner, organizational, and community characteristics; and, (3) explore the impact of PTPV on selected clinical and administrative outcomes.; Data were obtained from Focus on Therapeutic Outcomes, Inc. for PT episodes of care occurring between 1998–2000. Community level data were obtained from the U.S. Census Bureau. The final sample consisted of 1,195 complete adhesive capsulitis patient cases treated by 256 PTs in 145 clinics and 883 complete sciatica patient cases treated by 183 PTs in 115 clinics.; A two-level causal model incorporating the latent constructs PT Practice Style, Rehabilitation Resource Availability, Community Socioeconomic Status, PTPV and Risk-Adjusted Physical Function was developed and tested using data from each diagnostic group. A construct representing administrative outcomes could not be validated and was excluded. Selected patient and clinic characteristics were used as controls.; Variation in PT interventions ranged from 15–217% in the adhesive capsulitis group. PT Practice Style was a significant predictor of PTPV (r = .15) at the practitioner level of analysis; however, Community Socioeconomic Status was the only significant predictor (r = .33) when the data were aggregated to the clinic level. Higher PTPV resulted in lower Risk-Adjusted Physical Function at the PT level of analysis (r = .14).; Variation in PT interventions ranged from 21–190% in the sciatica group. No statistically significant results were obtained for the causal links when the model was retested using the sciatica data.; These findings confirm the existence of PTPV in the management of two musculoskeletal disorders. Predictors of variation appear dependent upon the level of data analysis conducted. Greater variation may result in poorer clinical outcomes for patients with adhesive capsulitis, suggesting that efforts to reduce variation may improve quality of care.