Quality-adjusted output cost economies of US adult versus paediatric physical therapy production

Abstract

With rapid population ageing and continuing proliferation of diseases limiting physical mobility and flexibility, timely research is needed on the operational economics of US physical therapy clinics at the disaggregated level (e.g., adult versus paediatric). This is because of potential variations in the cost economies, reimbursement structures and quality-adjusted outcome mandates of the reformed health care system. As such, the production technology structure and resource use flexibility implications in adult compared with paediatric physical therapy would tend to differ. Specifying a Generalized Translog (GTL) cost-minimization model, the core innovation in this article is investigating the cost structure differences between ‘for-profit’ adult and paediatric physical therapy operations. Using a uniquely rich data set of 4552 bi-weekly, site-specific operations across 27 US states with measurements on output, labour inputs (clinical, technicians, support staff) and capital, we separately model the economic contents of adult and paediatric physical therapy clinics. Results suggest that paediatric clinics do, indeed, have a statistically different operational cost structure compared with adult centres. The estimated factor demand elasticities of pair-wise factor substitutions (own and cross-price, Morishima and Shadow) and scale economies also differ, among other technological characteristics. These results have operational policy implications.

Keywords:

• translog cost model
• physical therapy industry
• technological characteristics
• adult versus paediatric costs

JEL Classification:

• I12
• I11
• D24

Acknowledgements

This is a revised version of an earlier paper presented at the 86^th annual conference of the Western Economics Association International (WEAI) in San Diego, CA, USA (29 June 2011–3 July 2011). The authors thank, without implicating, the conference participants, Dr. Victor Omotunde and Dr. Cyril F. Chang for their insightful comments on earlier drafts. The standard caveat applies, however.

Notes

¹ Scientific and technical progress is likely to permit expanded treatment of the more disabling conditions currently untreatable. The associated efficiency cost effects of new and innovative technological changes would depend on the economics of cost-effectiveness and regulations (Chandra and Skinner, 2012), and their relationships with the high-grade labour types necessary in the physical therapy production process.

² Since 2010, the percentage of US residents without health insurance has consistently declined, and the rate of escalation in annual health insurance premiums for individuals declined more for the 2011–12 period than in 2010–11 when the ACA began (Kaiser Family Foundation, 2012). These factors are increasing health care access and utilization, including for physical therapies.

³ Demographic and life-style trends favouring the physical therapy industry operations include general population growth with advancing age (baby boomers), rising interests in health, the growth of sports medicine and increased injuries from improper fitness regimen methods (US Physical Therapy, 2012).

⁴ For example, lower back pain, spine, shoulder and rotator cut-off injuries, knee disorders, a multiplicity of sprains and strains, heart transplants, and hip resurfacing or replacements.

⁵ Physical therapists (PTs) are part of a health care team that may encompass doctors, nurses, physical therapy aides (PTAs), social workers, occupational therapists, vocational counsellors and social workers.

⁶ Let f(x) have n + 1 continuous derivatives on [a, b] for some n > 0, and let x, x_oϵ[a, b].

Then

⁷ The additional variables collected included the quantity and type of CPT code units performed, and also the patient focus of each clinic. However, this data was not available for all time periods of the previous dataset. Therefore, this analysis uses a sub-sample of the previous dataset.

⁸ The less preferred results of the computational algorithms based on the arithmetic and harmonic mean concepts as alternative expansion points are not presented here but available from the authors upon request.

⁹ Costs are nondecreasing in all input factor wages, homogenous of degree one in input factor prices, and the negative semi-definite Hessian matrix in the adult and paediatric models signals concavity in factor prices. Cross-price effects are also symmetric, and the own-price effects are nonpositive at the expansion point.

¹⁰ After adjusting output for quality, the output measure decreased from 176.27 gross patient visits for adult clinics and from 314.47 for paediatric clinics.

¹¹ The estimates for both adult and paediatric clinics are found to be statistically different from those obtained in an earlier model fitted to a similar dataset but which did not adjust the output measure for quality.

¹² Respectively, 0.9684, 0.9673, 0.8358, 0.8125, and 0.8125 for adult-centred operations, and 0.9242, 0.9244, 0.9170, 0.9064, and 0.9065 for the paediatric clinics.

¹³ Past studies of health care production in economics (e.g., on physicians, physician assistants and nurse practitioners, pharmacists and pharmacy technicians; dentists, dental assistants and dental hygienists; mental health physicians and nonphysicians, etc) reported estimates of labour–labour substitution possibilities. See, for details, Folland et al. (2013, p. 108)