This past week the TX HHSC released an exclusive report to SHC in which they gave their interpretation of a requisitioned Texas A&M study that was suppose to have objectively compared Medicaid reimbursement rates for pediatric acute therapy services between those of Texas and the other states of the nation. The actual study was not released (to be requested by QDR through the Open Records Act) and so, we do not have the exact wording and composition of the study or of supporting citations, assumptions, or goals. Instead, an insider TX HHSC group (Strategic Decision Support) interpreted the A&M study results by way of a summary report. A copy of that report is linked here Review of Texas Medicaid Acute Care Therapy Programs – Final.
Tables were given comparing Texas Medicaid therapy services reimbursement rates to those of the California (Medi-Cal), Arizona (AHCCCS), Minnesota (MinnesotaCare and PMAP), and Florida (AHCA, KidCare, and SMMC) Medicaid programs. Additional tables compared Texas rates to Truven-blended 11-state (median) rates whose composition did not mention what states were involved.
The study and its interpretations were, on the surface, so flawed, that it may be difficult to determine where to start its critique. Many issues are equally deficient and important. This is our preliminary analysis of the study and its interpretations. We will follow this with a more detailed and robust statistical analysis in the near future:
(1) The 11-state Truven-blended block showed insufficient, selectively biased, and subjective sampling. The states involved in that blend were not revealed. This may be because this particular 11-state combination was optimal for the purposes of the agenda of the 83rd Texas legislature’s directive to reduce Medicaid rates to those of Medicare. There may also be missing or censored data within that blended panel. Additionally, the study utilized the mean estimator for the central tendency of each CPT rate, while the HHSC report on the study used the median estimator presumably because the distribution of such rates was skewed (asymmetrical), non-normal, and sparsely sampled. This is a double-edged sword and can be used to directly influence the agenda of the study before the results are calculated and/or interpreted. Monetary descriptive statistics (as in this case) rarely use median estimators as the mean may be more directly meaningful. The dispersion of rates among the 11-state Truven survey needs to be examined in order to utilize the right type of estimators. The small sample size may be a problem (power and robustness) for any type of parametric estimation or testing.
Median estimators are a special and extreme version of what are called alpha-trimmed mean estimators where a percentage, alpha, of outliers are omitted from the calculation. If the 11-state blend contained a few very high reimbursements for a CPT code, they would be eliminated from the calculation. This would give a false sense of the other state’s reimbursement policies. The typicality of the central tendency for a reimbursement rate is subjectively presented by using a median estimator. The mean estimator would contain a numerical global central tendency without weighing one state over another.
There are families of statistical estimators for the central tendencies (and dispersion) that are more robust than both the median and mean in the case of asymmetrical distributions, but the skewness, kurtosis, modalities, and other general characteristics of the distribution should be revealed before selecting such a family of estimators. For example Winsorized or alpha-trimmed (alpha<.5) means and Hodges-Lehmann-Sen location estimator families are more robust estimates for central tendencies in asymmetric distributions. Combinations of such families and bootstrapping (taking repeated means of subsets and weighing their iterations) can also be used to optimize this robustness with respect to the kurtosis, skewness, modalities, and general shape of particular distributions. More legitimate weighing of the states’ rates can be done before smoothing out to estimate a central tendency based on severity of diagnoses, economic conditions, and states’ budget solvency, among other notable monetary and human-equitable factors.
(2) There was data missing from the comparative tables between Texas and the four aforementioned states’ therapy services CPT reimbursements (grayed out cells). Was this because of insufficient data or because those states do not reimburse for such codes? If the later is the case, then after briefly reviewing those states’ fee schedules, that would be erroneous. Would those data be favorable to the Texas rates, (i.e., produce ratios closer to or under 1.00)?
(3) The study did not consider important confounding factors such as demographic and geographic healthcare epidemics, subepidemics, and syndemics (Texas border conditions), diagnoses and within those diagnoses, severity levels of chronic diagnoses such as autism, developmental delay, cerebral palsy, etc. All this data could have been extracted from the claims and/or clinical documentation that would have had to be included in claims for at least, the MCO claims. It is a well known clinical phenomena that the severity level of chronic diagnoses presented by pediatric clients greatly influence the intensity and duration levels of therapy treatments, hence the resultant cost of care. Texas presents with a disproportionate number of severely disabled and poor pediatric clients. The Texas border elevates these numbers due to long occurring economic and healthcare conditions. This number should be harmonized with those of other states’ data. It was not even approached in the study.
(4) The study does not mention the fact that within the past 3 years (after late 2011), the number of cases of therapy services, the number of therapy services provided, spent money on CORF/ORFs, and cost per client have all decreased or leveled out in Texas Medicaid due to rate cuts and MCO utilization discounted rates. Their own data suggest that unit therapy costs have gone down universally since the last legislative session in 2013. By utilizing the interval of time between 2009 and 2014, the emphasis was put on the increase of spending that was directly caused by a nature increase in the Medicaid population. They also emphasized that the number of claims per client has increased in managed care from 2009 to 2014. This is directly due to the increase in managed Medicaid population shifting from traditional Medicaid. The corresponding decrease in traditional Medicaid utilization should have been reported alongside that number. This was plain and simple, deceptive statistics.
Clients and providers have simply given up on the massive bureaucracy of the increasing burden of the continually evolving authorization process for therapy services and sub-Medicaid MCO effective rates that approach Medicare rates for some CPT codes. Unnecessary therapy has not been the issue – it has always been decreased access to therapy services issues due to logistical and economic problems. These problems have only been exacerbated. No statistics were collected on the number of Medicaid-eligibles with legitimate problems that have not received therapy services or wait too long for such.
(5) By directly mentioning the mandate from the 83rd Legislative session with Rider 51 specifically singling out therapy services (consideration of the service delivery model and to develop a “more appropriate” fee schedule) with an overall 2-year reduction goal of $4B in state general revenue funds and $5.61B in Federal funds with the results of this study – the implied burden is placed on therapy services rather than the equitable and reasonable smearing of savings across all Texas healthcare providers and suppliers. This is another example of selective bias.
(5) The study suggests that in the other states involved, the various delivery models of therapy were reimbursed equally. Not all states in the nation follow this policy. There are at least 39 states and US territories that reimburse for PT/OT Medicaid services and 40 states and US territories for ST Medicaid services according to Kaiser. Authorization policies differ as well as benefit duration limits.
(6) No longitudinal substudies were involved that would show that consistent therapy regimens for Medicaid pediatrics would shorten any followup or long term usage of those services, hence saving state’s taxpayers a significant amount that would outpace any increase in patients with the current therapy fee schedule and hence lessen the overall financial burden on the Medicaid program and making moot the directive of reducing Medicaid rates to Medicare rates while simultaneously maintaining adequate access to effective therapy services. Moreover, pediatric therapy services demand more attention, intensity of treatment, and oversight that other age groups within the confines of the authorization limitations. The therapist provides more expertise and work density per unit for pediatric chronic cases than in other scenarios. In the whole, their efficiency is higher in those cases than in other types of service groups, (i.e. the Texas taxpayer receives more worth per unit paid for). The Truven 11-state blend combined adult and pediatric services (Medicare, Medicaid, commerical, etc) and hence equated adult and pediatric therapy services, a misinformed assumption with overtures of dictating therapy policy without a therapy license.
(7) While we all want effective and efficient therapy protocols and treatment plans, there is no mention of what those might be in the study. There is good reason for this. No one knows the answer to this. It is a continuing research process. Efficacy studies in therapy are rare and have not been adequately designed. Texas legislators do not plan to spend money on such studies.
(8) Without complete raw data on all states, none of either the interpretation or the Texas A&M Study itself can be taken seriously. Each state that provides Medicaid therapy services for pediatrics has a multitude of ways of presenting such services including through education budgets and combined budgets with other state programs. These methodologies were not exposed in the study and would lay bare where the money comes from, following the claim and reimbursement paths. This is the only way to compare one state’s Medicaid therapy services budget with another.
(9) The study and interpretation do not mention that by 9/2016, all Texas Medicaid will be managed by MCOs and hence the lowest denominator of reimbursement will be presented. So, by stating that the lowest individual therapist rates were used as a conservative figure, it is being disingenuous. All Medicaid therapy rates will be at the lowest rate by 9/2016 or possibly sooner.
(10) The study listed the various delivery models for therapy in Texas and accumulated their respective rates. However, the study was not cognizant (nor the agency interpretation) that assistant therapists, even those properly supervised (per board registration) by a licensed therapist, cannot practice therapy in a home setting per DADS rules and would be subject to possibly heavy monetary fines, recoupment, and criminal prosecution (practicing home health care without a license). This restriction may not apply to other states. Hence they did not properly qualify the two versions of home health therapy delivery models as compared to other states’ services.
(11) The most dangerous thing in the study may be that there was an indication by this group that they have already supplied options for legislators and HHSC rate analysts to adjust rates and change authorization policies to meet the objectives of SB 1 Rider 51. This “options” report was not supplied nor was there any indications that it would be made available to the public. What is the methodology used for producing these options? Additionally, MCOs are already utilizing rigorous authorization processes that strictly supersede Texas Medicaid authorization processes. MCOs are already, on average, reimbursing the therapy industry as a whole (invariant to delivery model) at or near the lowest rate. Again, by 9/2016, all Medicaid will be MCO-managed. Each MCO has a standardized authorization form. However, each MCO is free to exercise their own authorization process that they have decided is in line with their respective at-risk policies and capitation agreement with the State. Texas cannot standardize the overall authorization process because the MCOs are private organizations that have agreed to manage case loads for Medicaid based on their own authorization process portfolio, capitation agreement, and ensuing internal actuarial risk assessment.
(12) The study cites various numbers for claims paid, per claim payment, Medicaid population, and claims per Medicaid client. They also cite the projected savings that SB 1 Rider 51 authors stated, that is $36.8M/$51.7M (state/federal) for 2014-2015. The 9/1/2013 HHSC rate reductions and authorization policy changes were estimated to save $18.1M/$25.4M. The 83rd Legislation then assumed that another $18.7M/$26.3M would be saved with further anticipated rate adjustments (reductions) and authorization policy changes made during 2015 (mid-year as such). However, the actual real-time savings were not stated, (i.e., how much has HHSC saved since 9/1/2013 to present?). These numbers are easily computable, especially the savings achieved from 9/1/2013 to 12/21/2014) – done by comparing the total Medicaid therapy services cost outlay using the pre-9/1/2013 rates versus the post-9/1/2013 rates. Hence, we have no idea if the savings have overshot their objective or have been underestimated or are somewhere in between.
That this group would be willing to further recommend reduce rates, tighten the authorization process, and possibly harmonize rates across delivery models without these numbers, is at the least, irresponsible, and at worst, possibly fraudulent. If the group’s recommended options include a further reduction of rates that would lower all Medicaid therapy service CPT reimbursements to near or at their Medicare equivalents, putting them below the other state’s rates (on average), an overly zealous overshot would happen. The wording of SB 1 Rider 51, taken literally by HHSC, would lead to the demise of Texas Medicaid therapy providers and an obvious access to pediatric and other Medicaid-eligibles therapy care problem. So the real question is: “is there a gap in the SB 1 Rider 51 budget objectives and the already saved budget since 9/1/2013 for therapy services?” This study and its interpretation did not address this. That is the most important question at this juncture and not the question of blended Medicaid therapy service rate comparisons (especially when erroneously compared to adult Medicare and commercial cases and mixed states rat estimates).
At some point, we must also question how any rate analysis is done for therapy services. It does not follow the RBRF (resource-based reimbursement fees methodology) under the RBRF (Resource-Based Reimbursement Fees) TMRM (Texas Medicaid Reimbursement Methodology) as directed in the Texas Administrative Code (TAC Section 355.8085). The RBRF formulary is simple, but requires subjectively selected parameters (similar to the CMS Medicare formulary) such as the RVUs (RVUw, RVUu, RVUm) and a conversion factor (CF) that is suppose to be economically reviewed every year and therein lies the wiggle room. The RVUs are economically driven based mostly on cost-of-goods, while the CF is driven by State and Federal government leading economic indicators, (i.e., PCE, etc.). So, while prices have increased, reimbursements have decreased. Hence, the only way a reduction can be had is by tweaking these parameters based on State budget reductions. The answer lies in the following determination by HHSC – based on the Texas static fee schedules for therapy services for the various delivery models, therapy service CPTs are NOT RBF (resource-based fees) and hence are not calculated based on resource economic information of costs, worth, etc. They are calculated based on access-based historical motives (ABRF). But this means that they should be based on access to therapy care statistics. There are no access to therapy care statistics that have been presented either in this study or interpretation or in any other recent determinations for rate adjustments. This is a convenient absence.
We have identified multiple types of biases in this report. The following are examples of these:
(a) selective bias – certain states, CPT codes, and programs highlighted or picked
(b) design bias – study group programs are not matched properly
(c) bias in selection of subjects – again only certain states, codes, and programs were selected
(d) length bias – selection of certain lengths of time in comparing costs that optimized pre-existing agenda
(e) Bias due to concomitant and confounding factors – oversight of obvious confounding factors not accounted for
(f) Bias in assessment – some data are more reliable (if at all) than others
(g) Observer bias – study researchers may have been influenced by political and economic motivations because they were paid by like-minded and associated groups
(h) Outlier bias – what was the criteria to label a rate as an outlier in order to define the median rates?
(i) Bias in analysis – analysis slanted to support a particular hypothesis or agenda
(j) Bias due to lack of power – sample size is inadequate to statistical significance of rate differences
(k) Interpretation bias – the tendency to interpret results in favor of a particular hypothesis or agenda, ignoring opposite evidence or not trying to prevent loss of robustness of study.
(l) Reporting bias – Study done by humans – tendency to suppress contradictory evidence by not mentioning contrary statistics.
(m) Bias in presentation – Scale for graphs or tables chosen to depict small changes to appear large and within 5-year length of cost reporting, later years not mentioned as cost-saving years.
(n) Publication bias – tendency to report only agenda agreeing results – positive towards affirming preconceived agenda.
There are countless other design, sampling, and interpretation flaws in the study and in the interpretations that will be exposed in followup posts. In the meantime, it is imperative that the therapy industry unite in the cause to expose this report and study for what they are – an agenda-directed diatribe to satisfy certain political motives to minimize Medicaid in Texas. This is also coupled with the movements by some to start accepting some form of Federal ACA Medicaid Expansion monies to offset shortages in the Texas Medicaid coffers to come. Medicaid expansion in Texas would mean that the Federal matching dollars shortfall would instantaneously shrink or be eliminated with respect to Medicaid therapy services costs. Additionally, through the Texas Open Records Act, the original study should immediately be made available to the public so that sufficient time can be had to scrutinize it and present resulting evidence to legislators and to the HHSC rate analysis team and commissioner before legislation is haphazardly introduced and passed in the current session and/or the HHSC rate analysis team releases reimbursement rate reductions (rate adjustment and authorization policy changes) to be implemented usually by September 2015.
We would encourage everyone with a passing interest in preserving the therapy industries to contact your federal, state and local legislators and their respective staffs and educate them on how therapy is practiced, billed for, and controlled in Texas and to point to them the egregiousness of this study and the threat it poses to the Medicaid children and other Medicaid-eligibles of Texas and to the various therapy industries and the economies that they help support and grow within the state. The collapse of these industries will directly lead to a domino-effect on the state that will cause (1) massive unemployment, (2) the large exodus of professional licensed therapists from the state, (3) a pandemic-like lack of access to therapy care, (4) thousands of business closings, and (5) an over all ensuing and large decrease in state and municipal tax revenues. Especially important is to visit your state officials locally or in Austin as the current legislative session is in full swing and bills may be amended to accommodate a naive acceptance of this misguided and harmful report.