This post describes my theory regarding the pivot by CMS from the original BA adjustments to the new one.
In my last post, we looked at the original CMS home health Behavioral Adjustment (BA), what it predicted and why. With four years of PDGM under our belt, we can measure the accuracy of these predictions. As described in my last post, these initial HHA revenue increase projections made by CMS never happened.
Based on actual data provided by CMS in the subsequent proposed and final rules, of the 8.389% in projected revenue increases in 2020, only the 0.25% revenue increase from higher comorbidity actually occurred.
To review, here are the three behavior assumptions that made up the original CMS home health behavioral adjustment in the 2020 final rule:
To gain a better understanding of what happened next with the BA, it helps to put on your CMS hat and view the impact of this original BA from their perspective.
CMS and their vendor, Abt Associates, had worked on the PDGM payment model for several years. Originally, it was to be introduced in 2019, but it was delayed by CMS to allow more time for HHAs and their software vendors to prepare. CMS used this time as well to refine this payment model and their assumptions.
CMS is a large organization, like most large business entities, they are subdivided into many departments with unique objectives and workflows. These departments have their own goals that may sometimes be in conflict with other departments within CMS or with the greater goals of the organization.
The CMS departments and vendors that were and are involved in the development of PDGM and the home health BA have a primary goal of managing home health spending by CMS to maintain budget neutrality. Like most of the rest of us, their performance is measured by their ability to achieve their goals.
Years of effort was put into PDGM to make sure that reimbursement mirrored cost for each of the PDGM groups. In my opinion, this went very well since the subsequent data shows that PDGM payments tend to produce relatively similar profit margins per claim, on average, even with the significant relative differences in payment for each PDGM group.
To achieve the budget neutrality goal for 2020, when CMS changed to PDGM, all it took was establishing a base payment per 30-day period that would match the previous total CMS home health payments when applied to the new PDGM claims. The only remaining part of the formula was the potential behavioral adjustment. From the perspective of CMS, if these behaviors could be predicted and quantified in the formula, CMS would maintain budget neutrality even as these behaviors were implemented by HHAs.
Each year that PDGM was under development, CMS refined the payment model from 2017 - 2019 using actual claim data paid under the previous model, CMS published the results through the Limited Data Set (LDS) annually as a file of simulated PDGM claims each of these three years. This data included the actual payments made through 60-day claims and the claims split as 30-day claims and paid under the future PDGM model. These files included results before and after the BA for each claim. You can see in these individual claims the extent that they predicted that each of these BA adjustments would occur.
As I discussed in my last post, there was no apparent problem with the math or logic in calculating the BA and budget neutrality including this behavior with this test data. The only problem was the assumptions themselves, how they would be implemented by providers, how often they would happen and the degree that they would influence revenue and budget neutrality.
CMS develops proposed rules for all provider types separately. They even have proposed and final rules for other related organizations like Medicare Advantage plans. These rules each have their own development workflows and cycles that are dependent on annual adjustments based on data collected by CMS. For home health, the proposed rule is developed using data from claims, OASIS, HHQRP and cost reports.
The proposed rule for home health is normally published in July. This document is very extensive with substantial technical detail on every aspect of CMS regulations regarding home health operations and reimbursement. To prepare the document, CMS must begin their full effort early in the spring. This presents a timing difficulty regarding the data they use to support the development of the PDGM weights and the base payment.
CMS claim data is collected quarterly by another CMS department and is then available within CMS to other departments needing this data. It is published in the LDS at about the same time and is available externally to industry stakeholders. The normal time frame for claim and OASIS data availability is four to six months after the end of the quarter. The reason for the delay is that the CMS revenue cycle process includes claim rejections, claims filed late, resubmitted claims, and other issues. This “turbulence” in the revenue cycle process prevents many of these claims from being processed in a timely manner. Before you can rely on this claim data to represent activity for a given period, you have to have enough clean and finalized claims to reasonably guarantee statistically significant results when using this data. CMS does not publish these quarterly claim data sets, internally or externally, until over 92% of these claims submitted for the quarter are finalized. Later, a year after the end of a calendar year, they publish 100% finalized claim data for the prior year.
For cost reports, the problem is more severe. These are annual documents provided by HHAs after the end of their fiscal years. Like the claim data,, the availability of the cost report data is delayed by submission and processing issues. It is also updated quarterly by CMS, but a large portion of HHAs use the calendar year as their fiscal year. This means that it takes until the July quarterly update, including most of these year end reports, to get a statistically significant number of cost reports for the prior year.
In April, when they begin articulating the upcoming proposed rule internally, CMS must use cost reports from two years earlier, but they can use claim data from the prior year since the first three quarters will be available to them and the final quarter will be visible between the proposed and final rules.
Jumping back in time to the 2020 proposed rule developed in 2019, the final weights implemented the first year of PDGM were developed with 2018 claim data and cost reports from 2017. Because of this data timing issue, when they would develop the proposed rule for 2021 the following year, they would have seen the first three quarters of claim data for 2020 and none of the cost reports. This limited data would be their only indicator of the actual HHA behaviors in response to PDGM.
Each of the three initial BA assumptions can be measured easily and accurately with claim data. This means that as CMS began preparing the proposed rule for 2021, they had already seen two or three quarters of 2020 claims. Even without the entire year of claim data, they could already see that their predicted behaviors were not happening in the data they could see.
CMS had weeks, maybe months, to review this data and develop policy for the 2021 proposed rule to adjust the BA. Instead of questioning their logic or assumptions, CMS doubled down. They continued the original BA adjustment for 2021 with no changes from 2020. They insisted that these behaviors were still going to happen, they were just delayed. They probably believed this to be true.
Moving forward to the next year, April of 2021, when the 2022 proposed rule was being developed. The situation had changed as CMS now had access to at least 18 months of claim data under PDGM. At this point, even the most die hard BA people within the CMS proposed rule development team would have realized that their assumptions were not going to happen.
It is not hard to imagine how this might have been viewed within CMS. They had fought hard against the home health industry for over two years insisting that these behaviors would occur. The data told a different story. This was not simply an error in judgment or predictive analytics, it represented a financial liability for CMS representing billions in unpaid revenue to HHAs. It is not difficult to believe that some people in this CMS department were in serious trouble with jobs and vendor contracts on the line.
The data time lag now worked to the benefit of CMS as it prepared for the 2022 proposed rule. They had months to digest the claim data and develop a strategy to somehow extract themselves from the box they had put themselves in with the original BA. Their solution had to deal with 4.3% in revenue they had already inappropriately withheld from HHAs over the first two years of PDGM and the debt it created. They could no longer claim that these behaviors were “pending” after two years of PDGM data with no indication of these upcoding and LUPA revenue enhancements by HHAs. This would have to be addressed for 2022 in the proposed rule, punting again was not an option.
Unrelated to the revenue increasing strategies they predicted from HHAs, PDGM contributed to a reduction in visits and the associated costs for HHAs as CMS implemented their goal of the dissociation of therapy visits from the new payment model. This was not a new or unexpected result of the introduction of PDGM. The same thing has happened with the introduction of each new prospective payment model by CMS for all provider types.
PDGM is not the first prospective payment model deployed by CMS. Hospitals went to DRGs in the 1980s. The goal was to try to control healthcare spending. When hospitals went from an itemized and transactional cost based system to a fixed payment per discharge, behavior changed dramatically for hospitals attempting to maintain their margins. Length of stay in hospitals went down significantly as well as total census. Many procedures that previously required an overnight stay became outpatient procedures. These new reduced costs flowed through the DRG payment model as weights in future DRGs, just as they do for PDGM, through the cost report data.
It is difficult to measure these savings due to the many different factors influencing healthcare costs in any setting. However, it is generally accepted and documented in studies that Medicare spending on acute care was lower under DRGs than it would have been without it.
Each of these new payment models is introduced with its own specific and unique legislation that defines the payment model goals and the regulatory authority of CMS to make sure that these goals are met. For some of this legislation, this includes a methodology for correcting problems with the payment models due to incorrect assumptions, data or new environmental factors.
These new payment models include regulatory language to allow for prospective changes to the formula based on future assumptions using existing data. Each provider type has the same data timing issues CMS deals with in home health. Without this language, CMS would not be able to include vital assumptions of future metrics like estimating future costs to calculate appropriate base payment amounts.
Some of these payment model regulations include language that would allow for retroactive changes to reimbursement if CMS determined that errors in the payment formulas created inaccurate payments at an industry level.
This language was included with the Inpatient Prospective Payment System (IPPS) implemented by CMS in 1983, the legislation that created DRGs. DRGs have been around for forty years now and have reshaped the operational efficiency of acute care. These actions by providers, influenced by the payment model, not only reduced costs, but they increased margins for the hospitals that could adapt quickly to DRGs. During this forty year history, CMS never took the position that any changes in length of stay and other cost reductions represented behaviors that threatened acute care budget neutrality when compared to prior years or previous payment models.
The retroactive changes allowed for in the IPPS legislation were not intended for small descrepencies in behavior. During the history of this payment model, the actual implementation of these retroactive processes were very rare. In 2021, CMS made a retroactive change in reimbursement for hospitals due to the impact of COVID-19. CMS has also made retroactive changes when errors were found in wage index data and some DRG weights. CMS never made any retroactive payment changes for hospitals intended to share in the savings created from any strategy they leveraged to reduce costs.
In the Skilled Nursing Facility (SNF) provider environment, they recently experienced their own migration to a new payment model, the Patient-Driven Payment Model or PDPM. Like PDGM for home health, this new model disconnected reimbursement from the volume of therapy services provided. Like PDGM, PDPM implementation instigated a reduction in the therapy services provided by SNFs as they no longer influenced reimbursement.
SNFs provide many of the same services provided by HHAs like PT and OT therapy. In the SNF revenue cycle process, these services are measured in minutes. This study by CMS states that they confirmed that when PDPM was implemented, therapy minutes per patient decreased from 91 minutes per resident per day to 62 minutes the first year (30%). Group therapy increased from 1% to about 30%. A second study I found confirmed this behavior and calculated a 19% reduction in therapy services in the first five months of PDPM. This study cited many more studies also verifying this trend.
These SNF behavior changes are directly attributable to the change in the payment model and the efforts of SNFs to maintain their margins by providing more cost efficient care. Unlike home health, CMS did not take any action to recover these savings by comparing payments under PDPM to the previous RUG model and labeling these cost saving measures as behavior changes affecting budget neutrality.
So the question is, why is home health different? Why are their savings from visit reductions considered a liability owed to CMS when no other providers have been subjected to this adjustment when the same behaviors occurred after the implementation of their prospective payment models?
There is a lot of focus from industry leaders and advocates on how these behavioral adjustments are calculated by CMS and the accuracy of the data. If these reductions threaten the industry financially and their fairness given MA reimbursement. I believe that these are the wrong questions, the question should be why are reduced costs considered a reduction in budget neutrality only for home health providers?
I believe that the answer lies in the development of the 2022 proposed rule and the failure of the original BA. If these original CMS predictions had turned out to be accurate, or even in the neighborhood of accurate, there may have never been the permanent and temporary BA adjustments we see today.
Instead, as CMS began to develop the 2022 proposed rule, they began discussing how to deal with the looming debt caused by their inaccurate predictions. How would they address the issue? How could they explain these failed predictions and their seemingly inept understanding of how HHAs actually worked?
If you make a mess and you can’t figure out how to clean it up, one way to deal with it is to create a much larger mess on top of it. Suddenly, the original mess is irrelevant. Any effort to clean up the second mess will take care of both of them. If you would be blamed for the first mess, but not for the second one, your problem might go away. An example would be if someone burned down a house to cover up the existing evidence of another crime. I am not suggesting that the current BA is criminal, but I do believe it was created by CMS, at least in part, as a method to cover up their initial BA failure
In 2021, when CMS reviewed the prior year in home health, a normal part of the proposed rule workflow, the only changes in behavior that had any significant financial impact to HHAs was the reduction in visits. Most of these visits were therapy visits in the first year of PDGM. At some point during the process, someone at CMS suggested the answer. What if we could classify this behavior as an overlooked and unanticipated response to PDGM that needed to be addressed in the payment formula retroactively?
The money associated with these visit cost savings was far greater than the amount assessed against HHAs under the original BA. Because this visit reduction effort began immediately when PDGM was implemented, they could go back and recalculate what would have been paid under their previous payment model with the same visits and define that as “budget neutral”. Without the therapy incentive payments, relative revenue dropped dramatically in this previous payment model compared to the actual PDGM payment model which had no such incentives.
They could leverage their retroactive recalculation superpower, intended to fix substantial errors in the formula, to calculate the difference between payments using the same claim data with both payment models. The PDGM payments for 2020 and 2021 included the BA, but even with this reduction, the PDGM total was far greater than the previous model payments using the same claim data. The reduction in visits, motivated by PDGM, would substantially lower this total paid to HHAs if this previous payment formula was still in effect.
This would not only erase their previous problem by replacing it with a new BA in excess of the old one, but it would substantially reduce CMS home health spending since the beginning of PDGM and in the future. Finally, it would create a new BA and make the original BA irrelevant, erasing their debt associated with the original BA and replacing it with a debt owed by HHAs to CMS, flipping the script. The original BA would cease to exist and any uncomfortable conversations about its validity and the associated debt to HHAs would evaporate.
It would have been very interesting to sit in on these meetings when this idea was presented and listen to the related discussions. I would imagine that not everyone believed it could work. CMS had never previously used this regulatory power to retroactively classify any type of effort by providers to reduce costs in response to a new payment model as a payment model error that needed to be corrected. This was probably viewed as a “hail mary” effort within CMS, but their previous mistakes had left them with few options to save their reputations and possibly their jobs.
As the 2022 proposed rule was developed, the decision was made to implement a new definition of the BA and budget neutrality retroactive to the beginning of PDGM, 1/1/20. CMS calculated claim data using both payment models and described their methodology and the results. The total difference was described as the “temporary adjustment”. They took the position that the original payment methodology, applied to current claims, was the new definition of budget neutrality. The difference in reimbursement in these models was to be the new BA.
Unlike the original BA, this BA did not require any predictions of behavior, it could be calculated accurately using claim data and even adjusted after data updates. There would be no debate about predictive behavior and the accuracy of these predictions. Finally, like the original BA, they calculated a permanent adjustment that would be applied to claims in 2022 to collect on part of this debt. Once again, they cut this in half between the proposed and final rules to appease HHAs, but the unapplied portion of the new permanent adjustment remained in the temporary adjustment totals, to be collected later.
This behavior of reducing costs by providers was present in all previous prospective payment models and was never previously discouraged by CMS. In fact, this behavior was the intent of these models and was previously considered beneficial to both CMS and the American taxpayer as they drove down the costs of providing quality healthcare under CMS health plans.
As CMS mentioned in their analysis of the payment model for SNFs introduced just one year before PDGM, as long as quality of care is maintained, retroactive adjustments in response to reduced care and costs are not needed, but future payment increases might be delayed or phased in.
Similar to the SNFs, HHAs have maintained their quality measures as PDGM was implemented. There is no evidence that these therapy reduction behaviors in either setting have reduced the quality of care and CMS has not taken that position. However, HHAs have been penalized for these cost reductions and SNFs have not. Neither have any other providers.
In my opinion, this is the key to dealing with this BA problem and ending it. The solution is not in arguing with CMS about the accuracy of these new BA calculations, but in contesting the decision made by CMS to uniquely punish HHAs for improving operational efficiency. The circumstantial evidence that this might have happened to erase the impact of CMS’s own previous predictive errors in the original BA should contribute to this argument.
In my next post, we will look at the actual cost of the BA and what it would look like if it was fully paid by HHAs or eliminated.
Comentários