Abstract
Background A Medicare beneficiary's annual outpatient therapy expenditures that exceed congressionally established caps are subject to extra documentation and review requirements. In 2011, these caps were $1,870 for physical therapy and speech-language pathology combined and $1,870 for occupational therapy separately.
Objective This article considers the distributional effects of replacing current cap policy with equal caps by therapy discipline (physical therapy, occupational therapy, and speech-language pathology) or a single combined cap, and risk adjusting the physical therapy cap using beneficiary characteristics and functional status.
Methods Alternative therapy cap policies are simulated with 100% Medicare claims for 2011 therapy users (N=4.9 million). A risk-adjusted cap for annual physical therapy expenditures is calculated from a quantile regression estimated on a sample of physical therapy users with diagnoses and clinician assessments of functional ability merged to their claims (n=4,210).
Results Equal discipline-specific caps of $1,710 each for physical therapy, occupational therapy, and speech-language pathology result in the same aggregate Medicare expenditures above the caps as 2011 cap policy. A single combined-disciplines cap of $2,485 also results in the same aggregate expenditures above the cap. Risk adjustment varies the physical therapy cap by as much as 5 to 1 across beneficiaries and equalizes the probability of exceeding the physical therapy cap across diagnosis and functional status groups.
Limitations One limitation of the study was the assumption of no behavioral response on the part of beneficiaries or providers to a change in cap policy. Additionally, analysis of risk adjusting the therapy caps was limited by sample size.
Conclusions Equal discipline-specific caps for physical therapy, occupational therapy, and speech-language pathology are more equitable to high users of both physical therapy and speech-language pathology than current cap policy. Separating the physical therapy and speech-language pathology caps is a change that policy makers could consider. Risk adjustment of the therapy caps is a first step in incorporating beneficiary need for services into Medicare outpatient therapy payment policy.
Medicare covers 3 types of outpatient (Part B) rehabilitation therapy: physical therapy, occupational therapy, and speech-language pathology. These services are primarily furnished by qualified therapists and therapist assistants under the supervision of qualified therapists. Medicare also recognizes that physicians and certain nonphysician practitioners furnish outpatient physical therapist, occupational therapist, and speech-language pathologist services. Therapist services are provided in multiple settings, including private practices, nursing facilities, hospital outpatient departments (HOPDs), and outpatient rehabilitation facilities (ORFs). In 2011, Medicare paid $5.7 billion for outpatient therapy, which was utilized by 4.9 million beneficiaries (15% of beneficiaries enrolled in the traditional Medicare fee-for-service [FFS] program).1 Table 1 provides additional detail on the utilization of the Medicare outpatient therapy population.
Characteristics of Annual per Capita Outpatient Therapy Utilization Among Medicare Fee-for-Service Beneficiaries, by Users of Each Therapy Discipline, 2011a
Since 1999, Medicare law has placed nearly all outpatient therapy under an annual limitation amount, or cap, which initially was set at $1,500 of allowed charges for physical therapist and speech-language pathologist services combined and $1,500 for occupational therapist services. Therapy caps originated in a 1972 law with a $100 per-beneficiary annual limit placed on the payment to independently practicing physical therapists, whose services were first made available to Medicare outpatients on January 1, 1973.2 The caps were a response to policy makers' concerns that this new source of services would exacerbate the growth of Medicare expenditures on physical therapy.3 A broadening of the cap policy in 1999 was one component of many in the Balanced Budget Act of 1997 intended to restrain spending growth of postacute care.4
However, 11 months after the caps went into effect on January 1, 1999, a new Medicare law imposed the first in a series of moratoria on the cap provision.5 The moratoria resulted from concerns that caps might prevent patients needing large amounts of therapy from accessing medically necessary services, especially if they could not make use of hospital outpatient departments, which were not subject to the caps at that time.6 In 2006, the moratoria were replaced with an “exceptions process” by which providers could formally attest on their claims that services above the cap were medically necessary. Such claims can be paid even though the patient exceeded the cap.7
Many private insurers impose utilization restrictions because of concern that not all services provided under an unconstrained benefit are medically necessary. There is demand for exceptions to a “hard” cap policy where Medicare does not pay for therapy services after a beneficiary has reached the cap, leading to Medicare's exceptions policy, which allows access to needed services above the cap. Exception rates have risen substantially since the policy's initiation. In 2006, 12.6% of therapy users exceeded the physical therapy/speech-language pathology cap ($1,740), and 15.3% exceeded the occupational therapy cap ($1,740).8 By 2010, the rates were 20.5% and 22.6%, for physical therapy/speech-language pathology and occupational therapy, respectively, when the 2 cap levels were $1,860.9
Alongside measures to restrain spending, defining a new basis for therapy payment has been another part of policy making in law affecting Medicare outpatient therapy. Congress mandated that the US Department of Health and Human Services perform utilization studies, conduct research to develop health and functional assessments, and report on potential new approaches to coverage and payment, including consideration of how information on characteristics such as diagnosis and functional status could be used in a revised payment policy.5,10,11 In 2008, as part of a long-term strategy to develop new payment approaches, The Centers for Medicare & Medicaid Services (CMS) began a multiyear project that, for the first time, collected detailed, uniform functional and clinical information on patients receiving Medicare outpatient rehabilitation therapy. The CMS intended that the effort would define measures of therapy need, whose application could eventually lead to alternatives to the therapy caps.12
In this article, we report on several results of that project that address 2 separate, but not distinct, issues. First, we use existing Medicare claims data to simulate alternative therapy cap policies that may be preferable to current policy. To compare cap policies that place the same constraints on total Medicare therapy expenditures, we choose caps in the alternative policies that result in the same aggregate Medicare program expenditures above the caps as current cap policy. That is, we simulate restructured cap policies that, while controlling Medicare expenditures as well as current policy, may be preferable in how they constrain expenditures. Second, we risk adjust the physical therapy cap using individual beneficiary diagnoses and functional status from a sample of Medicare Part B beneficiaries. This approach varies the cap based on patient characteristics rather than being uniform for everyone. For example, a patient with a stroke would receive a higher cap than a patient with an ankle sprain. Although a risk-adjusted cap will not necessarily achieve an ideal therapy payment system, it could represent an improvement in the current payment system.13
Method
The research presented in this article focuses on 2 study objectives: (1) a comparison of current therapy caps with alternative budget-neutral cap payment policies and (2) risk-adjusted therapy caps for patients receiving physical therapy. In the following sections, we discuss the methods for each objective in turn to support the study results and recommendations.
Study Objective 1—Alternative Therapy Cap Simulation Analysis
Study design.
We used Medicare claims for calendar year 2011 to simulate Medicare's current cap policy and possible changes in this policy. We consulted a CMS-sponsored technical panel of rehabilitation, payment, and measurement experts about the structure of the caps. They indicated that the current approach of combining 2 disciplines is inferior to either 3 separate caps or a single combined cap. Most often, panel members favored separate caps, citing, for example, patients needing care from multiple disciplines who could be disadvantaged by combined discipline caps. Those panel members favoring a combined global cap cited the flexibility this option might create in meeting patient needs. We simulated 2 alternatives: (1) 3 separate, equal caps, one for each therapy discipline, and (2) a single combined cap encompassing all 3 disciplines.
Data source.
The data file for this objective contains data from the Medicare Outpatient File for Institutional Claims (FFS claims data submitted by institutional outpatient providers such as hospital outpatient departments), the Medicare Carrier File for Noninstitutional Claims (FFS claims data submitted by noninstitutional providers such as physicians and therapists), and the Medicare Denominator and Beneficiary Summary file (Medicare enrollment information such as birth date/age and sex) for any beneficiary receiving therapy services during calendar year 2011.14 These files identify all professional and institutional outpatient services received and identify the beneficiaries and their characteristics.
Cohort creation.
In the Medicare Carrier File, therapy claims were retained if the specialty code on the claim line was allowed to render therapy services and Healthcare Common Procedure Coding System (HCPCS) codes were classified as therapy, as specified by CMS.*,† In the Medicare Outpatient File, therapy claims were retained if the “Type of Bill” indicated a therapy service provider.‡ The HCPCS codes were the same as the codes used for the Medicare Carrier File, but they also included additional restrictions for skilled nursing facility (SNF) “Always Therapy” codes, which require a facility type of SNF and type of service of inpatient SNF part B (22X) or outpatient SNF (23X).§ A total of 4,898,689 unique Medicare beneficiaries are included in this analysis—88.9% used physical therapist services, 21.9% used occupational therapist services, and 11.4% used speech-language pathologist services. We use this 100% claims file for the descriptive results in Table 1.
Study variables.
The study variables used to address the first objective were the sum of the line-allowed charges (total allowed charge) and line payments (the portion of the allowed charge paid by Medicare) that took place in calendar year 2011. The dates of service and beneficiary identification information also were retained.
Data analysis.
The 2 alternative cap policies were constrained to have the same aggregate Medicare allowed charges, across all sample beneficiaries, above the caps as current policy, assuming no behavioral response to changes in caps on the part of providers or beneficiaries. The restructured caps would constrain Medicare expenditures as well as current policy but might be preferable to current policy in how they constrain expenditures. Alternative caps were identified by calculating beneficiaries' 2011 expenditures on physical therapy, occupational therapy, and speech-language pathology above an initial cap and iteratively adjusting the cap until equality in Medicare expenditures above the caps versus current policy was achieved (across all 3 therapy disciplines in total). The Appendix provides a detailed description of how the alternative caps were calculated. We considered varying the caps by discipline but chose to constrain them to be the same because of the similar expenditure distributions among users of the 3 therapy disciplines.
Study Objective 2—Risk-Adjusted Cap Analysis
Study design.
The goal of this analysis was to determine how beneficiary need for therapy factors could be used to adjust the annual physical therapy expenditures cap. Accomplishing this objective required collecting data on beneficiary need for therapy, linking these data to therapy expenditures, and estimating a statistical model relating the two.
Data collection.
The CMS established a research project in 2008 titled Developing Outpatient Therapy Payment Alternatives (DOTPA) for the purposes of specifying, collecting, and analyzing information on beneficiary need and resource use of outpatient therapy.13 The DOTPA project included both analyses of existing CMS administrative data and real-time data collection from a sample of patients receiving therapy services.
Recruitment of the DOTPA therapy providers begin in late 2010 and ended in early 2012. A total of 162 sites were enrolled into the study. They included 68 private practices, 31 HOPDs, 27 ORFs, 21 nursing facilities, and 15 from other categories. The providers that supplied data volunteered for this study and do not represent a random sample of all therapy providers. Demographic characteristics and expenditure levels are similar among beneficiaries treated by the volunteer provider sample and the national Medicare therapy populations (see Results section). Sample recruitment emphasized achieving broad representation of providers geographically and enough representatives of the provider types associated with high-need patients to ensure reliable data pertaining to such patients.
Beneficiaries were recruited at the 162 sites beginning in March 2011. Patient assessments using the modified Continuity Assessment Record and Evaluation (CARE) assessment were obtained over a 6-month period at the sites and ended in June 2012.13 The original CARE tool, which was developed for use in institutional posthospitalization rehabilitation settings, was modified to focus on often healthier and less impaired community-dwelling patients receiving outpatient therapy. Providers completed an assessment at therapy initiation and at completion of a course of outpatient therapy for a systematic sample of new patients. Only the initial assessment was used in our analysis. A total of 5,822 assessments were obtained from the beneficiary sample for the larger project—5,007 physical therapy, 588 occupational therapy, and 227 speech-language pathology. Additional information on the DOTPA study design is provided in the final report to CMS.13
Cohort creation.
We included community-dwelling patients receiving physical therapy who had initial assessments at any time during the 12-month analysis period March 2011 to February 2012 (the first 12 months of volunteer data collection). This approach reduced the physical therapy sample size from 5,007 to 4,210 patients. To be consistent with the calendar year and the all-provider basis on which the Medicare outpatient therapy caps are applied, we included patients' expenditures from all providers of outpatient physical therapy, preassessment or postassessment, during the 12-month analysis period. We excluded any beneficiaries who were long-term residents of nursing facilities (as indicated by the presence of the 3-month Medicare nursing facility assessment) due to the distinct needs and different assessment instrument used for this population. Because the Medicare outpatient therapy caps are applied uniformly to all beneficiaries, we did not make any provisions in our analysis for beneficiaries who were hospitalized or received postacute care during our analysis period. These decisions were made to simulate a simple, calendar year cap risk-adjustment policy for the purpose of building on Medicare's current policy.
The analytic file included March 2011 to February 2012 expenditures from Medicare claims merged with assessment data for the 4,210 community-dwelling beneficiaries who received Medicare-reimbursed physical therapy from the 162 nationally dispersed providers that volunteered for the study. To be eligible for this analysis, the beneficiaries must have been continually enrolled in Medicare Part B in the time periods described above. They also must have been continually enrolled in the traditional FFS program with Medicare as the primary payer for all months in the study period.
Study variables.
The dependent variable to be predicted by our statistical model was 12-month analysis period Medicare allowed charges for physical therapy, including both Medicare payments and required beneficiary cost sharing. The independent variables in the model included the following demographic information: end-stage renal disease (ESRD) status; age groups 0–64, 65–74 (reference), 75–84, and 85+ years; male, female (reference); and currently aged 65+ years but originally entitled to Medicare by disability (when under age 65 years). Using the primary diagnosis reported (reason for therapy) from the assessment, we grouped beneficiaries into 1 of 12 mutually exclusive diagnosis groups, as shown in Table 2, the reference group being osteoarthritis. The primary diagnosis groups were created by clinician-researchers with backgrounds in physical therapy and rehabilitation nursing based on consideration of clinical similarity, sample sizes, and expenditures across diagnoses. The intent was to define primary diagnosis groups such that within-group homogeneity would be optimized in terms of both clinical characteristics, and resource needs and expenditures.
Quantile Regression Predicting Annual Physical Therapy Expendituresa
Finally, we included a mobility score, based on 13 different clinician-observed items in the assessment measuring different aspects of mobility, such as walking distance, transfers, and stair climbing. In preliminary analyses, the mobility scale was the best single summary measure of patient functional status relevant to predicting physical therapy expenditures.13 The mobility score was created using Rasch analysis, a psychometric, item-response, theoretic scaling technique for multi-item scales.15–17 The final output is a mobility estimate ranging from 0 to 100 describing the patient's level of mobility; higher scores indicate more functional ability. We also analyzed patient self-report of mobility available in the DOTPA data as an alternative to clinician-assessed mobility.13 However, in stepwise regression variable selection analysis, the clinician-observed mobility scale entered the model well before the self-report mobility scale. In addition, patient nonresponse and the validity of responses from respondents who are cognitively impaired or proxy respondents is a concern in payment applications of self-report data. Most Medicare payment systems, such as those for SNFs and home health care, rely on clinician assessment of beneficiary function rather than beneficiary self-report.
Data analysis.
We estimated a statistical model to predict 12-month physical therapy expenditures and used it to calculate a risk-adjusted cap for annual physical therapy expenditures. We estimated the model on the sample of physical therapy users with clinician assessments merged to their claims. Previous work has explored the use of age, sex, geography, diagnoses, pain, postsurgical status, affected area of the body, functional status, and practice type to explain outpatient therapy expenditures.18–27 CMS Part B therapy payment does not account for all such factors. We included in our model individual patient case-mix factors suitable for use in payment, namely demographics, diagnoses, and clinician-assessed patient functional status, which are derived from items on the assessment, as described in the previous section.
We used quantile regression with SAS PROC QUANTREG (SAS Institute Inc, Cary, North Carolina) to predict the quantile of expenditures corresponding to the therapy cap.28 Ordinary least squares (OLS) regression is a less preferred method because OLS predicts mean expenditures. Factors explaining therapy expenditures may have a different incremental impact on the therapy cap than they do on mean expenditures.
The quantile of 2011 annual expenditures corresponding to our simulated 2011 physical therapy–specific cap of $1,710 was determined to be the 70th percentile. Estimates from the quantile regression represent the incremental change in the simulated therapy cap (ie, in the specified percentile of expenditures) due to a 1-unit change in the explanatory variables.
Using all parameter estimates from the quantile regression, we predicted each beneficiary's risk-adjusted annual expenditure cap. We calculated the risk-adjusted cap for each of the 4,210 sample physical therapy beneficiaries and then compared their actual 12-month physical therapy spending with the predicted risk-adjusted cap and with the simulated physical therapy–specific budget-neutral cap from our simulation analysis. We adjusted the estimated standard errors to account for clustering (correlated practice patterns) at the provider level.
Results
Sample Population
Cohort 1.
There were 3,655,812 users of physical therapist services in 2011; the average user was 74 years old, and 65% of users were female. The average age of the 858,189 beneficiaries using occupational therapist services in 2011 was 77 years, and 68% were female. The 423,922 beneficiaries using speech-language pathologist services in 2011 were 78 years old, on average, and 62% were female. Fourteen percent of physical therapy users were currently entitled by disability (disabled) status, and 21% had dual-eligibility (Medicaid) status; the analogous percentages were 13% and 34% for occupational therapy users and 12% and 40% for speech-language pathology users. Although beneficiaries who resided in long-term care facilities for at least 1 month during the year made up only 8% of the physical therapy population, they comprised 28% of the occupational therapy population and 38% of the speech-language pathology population. The average allowed charges for physical therapy in the 12-month period were $1,258 (Tab. 1, which also presents information for occupational therapy and speech-language pathology users). Annual allowed charge variation was driven mostly by variation in therapy days, a 72-fold variation, rather than by variation in allowed charges per therapy day, a 6.5-fold variation. Physical therapy users, on average, received 14 distinct days of therapy during the course of a 12-month period. The duration in calendar days, from first to last therapy day during the year, averaged 73 days, with a median of 38 days.
Cohort 2.
The majority of this cohort (56%) consisted of patients with physical therapy private practice assessments, 24% of the sample had HOPD assessments, 16% had ORF assessments, and 3% had assisted living facility assessments. This is very similar to the makeup of the national Medicare settings for community-based outpatient therapy.1 We compared our physical therapy analysis sample with the national Medicare physical therapy population and found that our analysis sample had a slightly higher mean age (74 versus 73 years), an identical percentage of female users (65%), and a slightly lower percentage of beneficiaries entitled to Medicare by disability (12% versus 14%). Mean allowed charges were higher in the physical therapy analysis sample ($1,488 versus $1,258), and there were fewer dual-eligible beneficiaries than in the general Medicare population (9% versus 21%). The allowed charges were likely higher in our physical therapy analysis sample due to explicit instructions to the reporting providers to exclude any patients receiving 1-day therapy utilization who would not be returning for a subsequent visit. Previous work showed that greater than 10% of Medicare therapy users have therapy utilization lasting only 1 day.13 There were fewer dual-eligible beneficiaries in our physical therapy analysis sample because it included only community-dwelling beneficiaries.
Study Objective 1—Alternative Therapy Cap Simulation Analysis
Results of the cap policy simulation are shown in Table 3. Equal discipline-specific caps that result in the same aggregate Medicare expenditures above the caps as current cap policy are $1,710 each for physical therapy, occupational therapy, and speech-language pathology. For beneficiaries needing only one therapy discipline, this alternative cap policy reduces the 2011 allowed charges cap from $1,870 to $1,710. However, for beneficiaries needing large amounts of both physical therapy and speech-language pathology, the maximum increases from $1,870 combined under 2011 cap policy (physical therapy/speech-language pathology cap in 2011) to $3,420 under the alternative policy (2 × $1,710). For beneficiaries needing large amounts of all 3 disciplines, the maximum increases from $3,740 under current policy ($1,870 [physical therapy/speech-language pathology cap] + $1,870 [occupational therapy cap]) to $5,130 under discipline-specific caps (3 × $1,710).
Alternative Outpatient Therapy Annual Expenditure Caps, Simulated for 2011a
The simulated single combined-disciplines cap resulting in the same aggregate Medicare expenditures above the cap as current cap policy is $2,485. For those beneficiaries needing only one discipline of therapy, the cap rises from $1,870 under current policy to $2,485 under the alternative policy. For beneficiaries needing large amounts of both physical therapy and speech-language pathology, the maximum also rises from $1,870 to $2,485. However, for beneficiaries needing large amounts of all 3 disciplines, the maximum falls from $3,740 ($1,870 [physical therapy/speech-language pathology cap] + $1,870 [OT cap]) to $2,485 (combined cap). Also, maximum spending under a combined cap is only about half of maximum spending under discipline-specific caps: $2,485 compared with $5,130 (3 × $1,710).
Table 3 shows that under the current (2011) cap policy, the annual allowed charges of 21% of outpatient therapy users exceed at least one cap, by an average amount of $2,211. Under equal discipline-specific caps, 23% of outpatient therapy users exceed at least one cap, but by a smaller average amount of $1,991. Under a single combined-disciplines cap, only 16% of outpatient therapy users exceed the cap, but by a larger average of $2,926. A single combined cap flags a smaller proportion of beneficiaries with potentially excessive utilization, but these beneficiaries' spending is farther above the cap, on average.
Study Objective 2—Risk-Adjusted Cap Analysis
Table 2 presents results of the quantile regression explaining annual physical therapy allowed charges on the sample of Medicare physical therapy users with a clinician assessment. The quantile regression model explains a small percentage of the overall variation in annual physical therapy expenditures (5% adjusted R2). Nevertheless, some of the model's explanatory factors can substantially modify the annual physical therapy cap.
The intercept term of $1,620 is the predicted physical therapy annual cap for a 65- to 74-year-old woman, non-ESRD, non-Medicaid, not originally entitled to Medicare by disability, with a primary reason for therapy of osteoarthritis and the highest clinician-observed mobility category (71–100). End-stage renal disease status is statistically significant and adds $1,210 to the predicted cap. Younger age/currently entitled by disability (age=0–64 years) and male sex predict significantly lower annual physical therapy expenditures. Primary reasons for therapy of stroke ($806) and joint replacement ($437) were significant in the model and predicted higher caps. Sprain, strain, bursitis, tendinitis (−$309); genitourinary conditions (−$512); and vertigo (−$973) are all also significant and predict lower caps. The 2 lowest functioning mobility groups predict $678 (0–50) and $555 (51–70) additional cap dollars.
These results indicate that certain diagnoses (primary reason for therapy), demographic factors, and functional status significantly affect annual physical therapy expenditures. For example, the predicted cap for a beneficiary with the omitted categories, including the highest mobility level, and vertigo is $647 ($1,620 [intercept] − $973 [incremental effect of vertigo]). By comparison, the predicted cap is $3,104 for an otherwise similar beneficiary with stroke and the lowest mobility level ($1,620 [intercept] + $806 [incremental effect of stroke] + $678 [incremental effect of lowest mobility category]). In other words, risk adjustment can vary the cap by a factor of nearly 5 to 1 for 2 beneficiaries who would have equal caps under current policy.
We now consider how a risk-adjusted physical therapy cap would affect beneficiaries grouped by diagnosis and mobility status. Table 4 presents the percentage of beneficiaries falling above both the risk-adjusted and non–risk-adjusted caps by diagnosis and mobility groups and the mean annual expenditures associated with each group. Risk-adjusted caps were created using the predicted values from the quantile regression; actual expenditures per beneficiary were then compared with the simulated cap. The non–risk-adjusted cap is our simulated 2011 physical therapy–specific budget-neutral cap of $1,710 (Tab. 3). The overall percentage of beneficiaries exceeding the cap is similar for the risk-adjusted (29.7%) and non–risk adjusted (29.8%) caps. Without risk adjustment, the percentage of beneficiaries exceeding the cap ranges from 5.7% (diagnosis of vertigo) to 47.9% (diagnosis of stroke). Beneficiaries with more severe diagnoses and poorer mobility are more likely to exceed the cap. For example, 44.7% of beneficiaries with the poorest mobility (mobility score of 50 or less) exceed the non–risk-adjusted cap versus only 25.5% of the beneficiaries with the highest mobility (mobility score greater than 70). With risk adjustment, the percentage of beneficiaries exceeding the cap varies only in a tight range of 26.9% to 29.9%. For example, 29.6% of patients with a stroke and 28.6% of patients with vertigo exceed the risk-adjusted cap. Across all 3 mobility levels, from 29.3% to 29.7% of beneficiaries exceed the risk-adjusted cap. In sum, risk adjustment equalizes the probability of exceeding the cap across diagnosis and functional status groups.
Percentage of Beneficiaries Exceeding Risk-Adjusted and Non–Risk-Adjusted Annual Physical Therapy Expenditure Cap, by Diagnosis and Mobility Groupa
Discussion
In this article, we have analyzed 2 refinements to the current Medicare policy of annual expenditure caps for outpatient therapy. Although neither refinement results in an ideal payment system for outpatient therapy, they may be worth considering given the widely recognized shortcomings of the current cap policy, which has existed in various forms for 15 years.
Study Objective 1—Alternative Therapy Cap Simulation Analysis
The 3 disciplines of therapy covered under the Medicare Part B benefit address different needs. Our data show that 82% of outpatient therapy users use only one discipline in a given year, but the multiple disciplines can work in concert to rehabilitate patients with complex multidisciplinary conditions. An important question is whether expenditure limitations should be set separately by discipline or whether a single limit on utilization should apply to all outpatient therapy provided to the patient. A single combined cap helps beneficiaries with a high need for therapy from a particular discipline while limiting those with high demand for multiple disciplines of therapy. A combined cap disadvantages the highest-need beneficiaries (eg, patients who are severely impaired due to a stroke and who require physical therapy, occupational therapy, and speech-language pathology). A combined cap makes some beneficiaries better off and some worse off versus separate, discipline-specific caps; a combined cap is not a “Pareto improvement” making all beneficiaries better off.
The degree of substitutability among the 3 therapy disciplines affects the cost control impact of expenditure caps. If the disciplines are substitutable, separate caps by discipline can be evaded by substituting one discipline for another. For example, in a rural area without a rehabilitation center, the physical therapist treating a patient with a stroke may furnish the necessary self-care training without the ability to bill for these services under the occupational therapy cap. The possibility of substitution argues for a single, combined cap for transparent and effective cost control. The degree of substitutability among the therapy disciplines, however, is not well understood, and further research is needed in this area.
Equal discipline-specific caps for physical therapy, occupational therapy, and speech-language pathology are more equitable to high users of both physical therapy and speech-language pathology. Separating the physical therapy and speech-language pathology caps is a change that policy makers could consider, but they should consider that achieving the same level of cost constraint with discipline-specific caps as with current cap policy would require lowering the cap level. Such a change in policy would benefit some beneficiaries and disadvantage others, as described above. Also, a reconsideration of therapy cap policy would ideally be informed by a better understanding of the patterns of use of single versus multiple therapy disciplines and the substitution among disciplines. Expenditure limits imposed on one discipline could have unintended spillover effects on other disciplines or may not achieve the intended constraint on expenditures. Changes in policy could be especially disruptive if a hard cap were imposed; the current soft cap policy provides a safety valve for Medicare beneficiaries to access needed therapy services.
Study Objective 2—Risk-Adjusted Cap Analysis
Given our findings that diagnosis and functional status are significantly related to variation in annual per capita outpatient therapy expenditures, risk adjustment is an important component to consider in any alternative outpatient therapy reimbursement system.13 In an FFS system of payment, risk adjustment could function as a method of predicting therapy utilization and subsequently targeting review of or increasing provider documentation requirements for beneficiaries with higher-than-predicted utilization (“soft caps”). Risk adjustment would calibrate annual expenditure caps to the specific needs of each patient. If risk adjustment is sufficiently precise, “hard caps” (with no or limited exceptions) may be more feasible, or policies such as higher beneficiary cost sharing or reduced provider payments above the caps could be implemented. Our research shows that risk adjustment dramatically lowers the percentage of beneficiaries with severe diagnoses such as stroke and the lowest functional mobility exceeding the physical therapy cap. Nevertheless, there is substantial variation in therapy expenditures unexplained by the quantile risk adjustment regression. The significant unexplained variation limits the feasibility of imposing binding limits (hard caps) on beneficiary therapy expenditures because of the potential risk that hard caps pose to beneficiary access to needed services.
Moreover, risk adjusting the therapy caps requires collecting data on patient clinical characteristics such as reason for therapy (diagnosis) and functional status (eg, mobility), data that are reported by therapy providers and could be manipulated by them to increase reimbursement (“gaming”). It is not clear if the accuracy gained in setting the caps justifies the cost and burden of collecting this information through an assessment instrument such as the one we used, and auditing and verifying the information to minimize gaming. A lower-cost possibility is to collect clinical information through a claims-based approach. Under the Middle Class Tax Relief and Jobs Creation Act of 2012, a claims-based data collection system for Medicare outpatient therapy services is already in effect.29,30 Therapy patient functional status is reported at various points during an episode of treatment using 42 nonpayable functional “G codes” and 7 severity/complexity modifiers on physical therapy, occupational therapy, and speech-language pathology claims. The G codes cover functioning such as mobility; self-care; carrying, moving, and handling objects; swallowing; voice; memory; spoken language expression; and motor speech. Unlike the research instrument used in our study, the G-code system uses nonstandardized measures, which require assessment of validity and reliability. Additionally, the Medicare Payment Advisory Commission has recommended that therapy claims be required to have a specific medical diagnosis that is the reason for therapy rather than the general V code diagnoses that are often recorded as per current International Classification of Diseases, 9th edition (ICD-9) coding guidelines.1
Limitations
A limitation of our alternative therapy caps analysis is our assumption of no behavioral response on the part of beneficiaries or providers to a change in cap policy. To the extent that beneficiaries or providers can evade therapy caps by substituting services from one therapy discipline for services from another, our simulations may not have identified alternative caps that constrain expenditures to the same extent as current cap policy. For example, with a high degree of substitutability among therapy disciplines, the simulated single combined cap would constrain expenditures more tightly than current policy, and the simulated 3 discipline-specific caps would constrain expenditures less tightly. Taking account of behavioral responses but still requiring equivalent cost constraint could identify different alternative caps than we simulated.
Our analysis of risk adjusting the therapy caps is limited by our sample. Although larger and richer than most previously available data, our sample is nonrandom and not large enough to estimate all risk-adjustment model parameters with ideal precision. It is possible our results could be modified by a larger and more diverse sample. Also, our model explains only about 5% of annual physical therapy expenditure variation. Our simulations show that this percentage is enough to achieve meaningful risk adjustment of the therapy caps for beneficiary need. Clearly, however, other factors are affecting therapy expenditures, such as the setting in which services are provided (eg, private practice versus hospital outpatient department). Further analysis of additional factors affecting therapy expenditures and whether they should be incorporated into therapy payment models would be valuable.
In conclusion, data collection approaches to support risk adjustment of therapy payments need to be further developed, tested, and validated. Risk adjustment models can then be built on the foundation of the newly available data. Future changes to Medicare outpatient therapy payment—whether FFS with annual cap, episode payment, or some other approach—may include risk adjustment of payment in some form. The analysis in this article helps in understanding what a model to risk adjust a separate annual physical therapy cap might look like and what its implications for different types of patients might be. As such, it is a contribution to developing outpatient therapy payment approaches that incorporate beneficiary need for services.
Appendix.
Method of Identifying Alternative Annual Expenditure Limits (Caps)a
a PT=physical therapy, OT=occupational therapy, SLP=speech-language pathology.
Footnotes
Dr Amico, Mr Pope, Dr Meadow, and Dr West provided concept/idea/research design. Dr Amico, Mr Pope, Dr. Pardasaney, Mr Silver, Dr Meadow, and Dr West provided writing. Mr Silver provided data collection. Dr Amico, Mr Pope, Dr Pardasaney, and Mr Silver provided data analysis. Dr Amico, Mr Silver, Dr Dever, and Dr Meadow provided project management. Dr Meadow and Dr West provided institutional liaisons. Dr Pardasaney, Dr Dever, Dr Meadow, and Dr West provided consultation (including review of manuscript before submission). The authors thank Anne Deutsch, Barbara Gage, Melvin Ingber, Alan Jette, Walter Adamache, Janet Valluzzi, and Alex Laberge for their contributions to this work. Any interpretations, opinions, or errors are the responsibility of the authors and not those we acknowledge.
The Centers for Medicare & Medicaid Services (CMS) funded this research. Any interpretations, opinions, or errors are the responsibility of the authors and not those of the CMS.
↵* The provider specialty codes include: 01–31, 33, 34, 36–41, 44, 46, 48, 50, 65–67, 70, 72, 76–79, 81–86, 89–94, and 97–99.
↵† Therapy HCPCS codes were defined as those classified “Always Therapy” codes (a specific therapy service regardless of the provider), including: “92506,” “92507,” “92508,” “92526,” “92597,” “92605,” “92606,” “92607,” “92608,” “92609,” “92618,” “96125,” “97001,” “97002,” “97003,” “97004,” “97010,” “97012,” “97016,” “97018,” “97022,” “97024,” “97026,” “97028,” “97032,” “97033,” “97034,” “97035,” “97036,” “97039,” “97110,” “97112,” “97113,” “97116,” “97124,” “97139,” “97140,” “97150,” “97530,” “97533,” “97535,” “97537,” “97542,” “97750,” “97755,” “97760,” “97761,” “97762,” “97799,” “G0281,” “G0283,” and “G0329” or “Sometimes Therapy” codes (“64550,” “90901,” “90911,” “92520,” “92610,” “92611,” “92612,” “92614,” “92616,” “95831,” “95832,” “95833,” “95834,” “95851,” “95852,” “95992,” “96105,” “96110,” “96111,” “97532,” “97597,” “97598,” “97602,” “97605,” “97606,” “0019T,” and “0183T”) with a required therapy modifier (GP=physical therapy, GO=occupational therapy, or GN=speech-language pathology).
↵‡ Providers included HOPDs, SNFs or other nursing facilities (NFs), comprehensive outpatient rehabilitation facilities (CORFs), outpatient rehabilitation facilities (ORFs), and home health agencies (HHAs). These providers are covered by the following codes composed of facility type in the first digit and type of service in the second digit: 12, 13, 22, 23, 34, 74, and 75.
↵§ SNF “Always Therapy” codes include: “64550,” “92612,” “95834,” “97597,” “90901,” “92614,” “95851,” “97598,” “92520,” “92616,” “95852,” “97602,” “92610,” “95831,” “96105,” “97605,” “92611,” “95832,” “96110,” “97606,” “95833,” “96111,” “97532,” “0019T,” and “0183T.”
- Received September 24, 2014.
- Accepted June 8, 2015.
- © 2015 American Physical Therapy Association