Exploring the Perspective of CMS Regarding Home Health Agency Profits Through CMS Data and the 2024 Proposed Rule.
In this article, I will use data provided by CMS through their proposed and final rules along with the cost report data I discussed in my initial blog article to explore the impact of Medicare PDGM reimbursement for home health agencies and the analytics that drive this reimbursement model. Most importantly, we will discuss assumptions or insights associated with this data and its impact on the home health industry.
PDGM
For background, I will provide a brief explanation of how PDGM works. As part of the Medicare revenue cycle process, home health agencies provide data for each 30 day period of patient services. This is done through the submission of the insurance claim or invoice for these services and OASIS (Outcome and Assessment Information Set) data that measures the functional ability of the patient being treated through the services described in these claims. Soon, Value-Based Payments will be used to increase or decrease these payments by provider based on agency patient treatment quality scores created through patient surveys and quality metrics from claim data, as defined by CMS.
These four data sources, claims, OASIS, cost reports and quality metrics, are all collected and published by CMS quarterly, at least as they apply to Medicare Part A (institutional) activity. This data is either available to anyone for download or available to researchers through agreements with CMS.
CMS uses a prospective payment method to reimburse home health agencies based on this claim, cost report and OASIS data. The formula used to determine payments is called PDGM (Patient Driven Groupings Model). Each year, CMS compiles the cost report data and clinical data obtained in claims to refine the CMW (Case Mix Weight) applied to each of the 462 groups representing each possible combination of the components of this formula. These component values are identified in this diagram from the 2024 proposed rule:
This weight is then multiplied by a base payment rate determined by CMS each year and published in the final rules.
The base payment includes calculations to compensate for economic market basket indexes and an adjustment called the behavioral adjustment that is intended to maintain revenue neutrality compared to previous annual spending by CMS for home health services. Finally, the payment is multiplied by the wage index associated with the location of the services to compensate for the variations in wages for clinicians across the country. This is the method used to calculate payments for 30 day periods where the visits are above the LUPA visit threshold, for those below these thresholds, the claims are paid based on a per visit rate adjusted by the discipline of the clinician and the wage index of the service area.
It is the stated goal of CMS to use PDGM to limit spending on the home health industry as a whole while still allowing most agencies to operate with a reasonable profit, keeping these services available to those Medicare patients that need them. These goals are supported by regulations referenced in the rules that describe these mandates as CMS has interpreted them.
Similar reimbursement formulas are used by Medicare for all other institutional provider types (hospitals, skilled nursing facilities and hospices) with variations on these formulas specific to each provider type and how they provide their services and pay for them. Each different prospective payment methodology is designed to measure cost and deliver payments to these healthcare providers with consistent profit margins for all types of healthcare services.
Updating PDGM Through the Proposed and Final Rules
Each year, CMS refines these formulas and publishes their intended changes for the upcoming year. They communicate these changes and the data used for their assumptions in a document called the proposed rule. For home health, this is normally published around July of each year. The proposed rule for the upcoming year, 2024, was published on 7/10/23. Like other proposed government regulations, there is a comment period. When it is over (8/31/23), CMS reviews these responses and publishes a final rule. The final rule for the current year, 2023, was published on 11/4/22. The final rule includes many of the public comments, CMS responses, and any final changes from CMS in response to the comments or additional data they have compiled since the proposed rule. When the final rule is published, the reimbursement rates and other regulatory requirements are set for the upcoming year.
The proposed rule is published through the Federal Register which documents all regulations created by the federal government. Each year when these rules are published, I read them. It is a daunting task. The home health proposed rule for 2024 is 164 pages filled with many charts, references to other documents, data, acronyms, and technical language. However, there is a lot of thought and work put into these documents by CMS and a lot to be learned if you can spare the time to digest and interpret them. If you are in a line of business that is affected by these regulations, it is time well spent.
How CMS Leverages Data to Support Their Position and Proposed Rule Changes
What I hope to demonstrate in this article is the power that comes from the data included in these regulations and the leverage it gives to the parties that know how to use it to represent their point of view. Up until now, this has been predominantly CMS. However, the data that CMS uses is available to researchers who want to explore these issues themselves. I am one of these people.
When you analyze large sets of data for insights with BI (Business Intelligence), you can often make mistakes. These generally fall into two categories, mistakes in math/logic and mistakes in explaining what the numbers mean. I will refer to the second issue as assumptions.
Assumptions are made when you apply the mathematical results and visualizations of your data to support an existing point of view or develop a new point of view based on conclusions about what you feel the numbers represent. When mistakes are made regarding these assumptions, it might be because you are looking for numbers that support your point of view and ignore those that don’t or because you develop insights based on the data that are not actually supported by or connected to the data.
With proper quality control, math mistakes can be found and eliminated. However, false or misleading assumptions can be more difficult to detect, especially when those looking for them share the same perspective as those that made the mistakes. What I intend to demonstrate is that the assumptions made by CMS using this data are not necessarily supported by their data or that the data they provide could mean something different when viewed from a different perspective.
As I mentioned earlier, one of the goals of CMS is to maintain reasonable profit margins for home health providers. Each year, they provide calculations in the proposed rule that detail the costs of services provided by agencies and compare them to Medicare revenue paid for the same period.
In the proposed rules since PDGM began, CMS has made repeated references to the downward trend in visits made to home health patients. They associate this drop in visits with efforts by home health agencies to improve profit margins under PDGM compared to the previous PPS.
They document this visit decrease in table B2 from the 2024 proposed rule:
This assumption by CMS has driven them to redevelop the concept of the behavioral adjustment as a measurement between what is paid currently under PDGM each fiscal year and what would have been paid under the previous reimbursement model which paid extra for some therapy visits. I have a lot to say on this topic. For now, it is relevant because CMS is proposing a significant drop in Medicare payments to home health agencies for 2024, a net negative 2.2% decrease. This includes a decrease of 5.1% due to this behavioral adjustment.
CMS is responsible for making sure that these annual adjustments do not threaten the availability of healthcare in the home for Medicare beneficiaries. For this reason, each proposed rule includes language intended to assure stakeholders that changes in the rule will not adversely affect these providers. During the comment period, stakeholders present their own point of view on the proposed changes. Afterward, CMS responds to these public comments as a method of defending their position concerning these rule changes.
In the 2024 proposed rule, we can find this language, in part, under Section B “Monitoring the Effects of the Implementation of PDGM”.
In the 2023 proposed rule, CMS calculated a margin for home health agencies during 2021 at 34%, in the 2024 proposed rule, they used the same methodology to calculate an estimated margin for 2022. This time it was 45%. These are the margins reported by CMS to MedPAC who advises congress on these issues.
Table B4 is intended to validate the cost per period value used in this 45% margin calculation for 2022:
When you are looking at data like this, it is important to remember what the provider of the data is trying to accomplish. In this case, CMS is trying to demonstrate a large margin (45%) that home health agencies enjoyed collectively in 2022. What we will discover is that this data is flawed and does not accurately represent the financial position of the home health industry.
Using CMS Data to Analyze CMS Conclusions Regarding Home Health Profits
Since CMS used cost reports for this data, and I have access to this data, I should be able to reproduce these numbers.
In the first column of Table B4, we have 2021 average costs per visit. In the cost report, we can find these values on Worksheet C. This is what this section of Worksheet C looks like on a blank cost report form and where these values would be found:
In Sisense, I built a table that allows me to select any cost report worksheet and view the numeric data from a specific cost report. This is what the data looks like for this section of Worksheet C using a report from a single agency.
I can also look at these values collectively and calculate each of them at an industry level. In Table B-4, CMS used cost report values for (CPV) Cost Per Visit from 2021. When I get averages from the cost reports in 2021 for all agencies, this is what I get compared to the results from CMS:
My results are similar to those from CMS, but not exactly the same.
In the 2024 proposed rule, CMS uses cost report data from 2021 to estimate expenses for 2022 and claim data from 2022 to measure visit activity. To determine the estimated cost per period for 2022, they take the cost per visit from 2021 in the first column of Table B4 and increase it by 2.6% in the second column, they then multiply it by the average visits per 30-day period by discipline in the third column (from 2022 claims) to come up with the projected cost per 30-day period for 2022.
When CMS applies this logic to their data, they come up with an average cost per period of $1402.27. When I use the same logic with my calculated costs per visit, I get $1430.63. Once again, these numbers are close, but do they reflect reality?
When CMS published the proposed rule for 2024 on 7/10/23, they had access to nearly all cost reports for 2021, but less than half of the cost reports for 2022 had been processed. Since then, there have been additional updates to the cost report data and now over 75% of the cost reports for 2022 are available. Although not all cost reports are processed, each one in the data is complete and there are enough of them to provide statistically significant results that will be very similar to the final results concerning averages when this data for 2022 is complete.
This gives us the opportunity to compare the estimate for 2022 calculated by CMS against the actual 2022 costs. Repeating the same process in Sisense, I created the costs per visit values for 2022 as we did for 2021.
What we get is an actual average per period cost for 2022 of $1523.24. This is 8.63% higher than the CMS estimate of $1402.27 per 30-day period, a significant difference.
In Table B4, CMS took the costs from 2021 and multiplied these costs by 2.6% to get costs for 2022. The 2.6% comes from the increase in home health payments by CMS to providers from 2021 to 2022. Although this might accurately represent the increase in revenue between these two years, it is not an accurate method of estimating a future change in cost.
When I see data like this that many people have looked at, but not questioned, I bear down even harder when exploring other data from the same source. Sometimes I try to imagine the meeting, at CMS in this case, where someone said “we need to talk about the healthy margins home health agencies have under our program, how can we estimate this for 2022 when all we have is 2021 cost report data?” At that point, someone said, “how about we just take 2021 costs and increase them by the same percentage as our payment increase?”. I find it puzzling that everyone involved in the review of this data at CMS agreed that this would provide an accurate estimate of increases in expenses. As it turns out, the actual increase using this method of measurement was 8.63%, over three times the CMS estimate.
This is an example of an assumption error that I described earlier. In this case, it appears that CMS came up with this estimate because it followed some semblance of logic in the mind of the people presenting the data and it fit well into the CMS narrative regarding home health profit margins. In my mind, it was not the best solution in the first place and it was way off when compared to the actual data it was intended to estimate.
This leads us to an important principle regarding this type of “predictive analytics''. Whenever you develop tools that use current data to predict future data, you should evaluate their performance when the future data becomes current data. Most businesses perform this task when they prepare financial statements comparing budgeted to actual revenue and expenses. They review their budget figures for accuracy and the lessons learned improve future budgeting. In this case, CMS created estimates of future home health margins in last year’s proposed rule for 2023 as well as the current rule for 2024. These results are used to update CMW for PDGM payments that will be implemented in 2024.
In this same section of the proposed rule for 2024, we find further issues with these CMS estimates of home health margins. In the two paragraphs above Table B4, they explain how they calculated the margins, taking the cost report estimate per period and subtracting it from the base payment for that year. In the paragraph referencing the 2023 proposed rule, they explain the calculation and the result. They compare the base payment of $1901.12 which is approximately 34 percent more that the estimated average cost of $1420.35. In the 2024 proposed rule, they come up with the results discussed from Table B4, comparing an estimated expense of $1402.27 per period and a base payment rate of $2031.64 coming to a 45% margin.
When people look at financial data like this that includes revenue, expenses and a margin expressed as a percentage, most people would assume that this margin is a profit margin. Let’s look again at the math using the calculation from the 2024 proposed rule. Margin (45%) = (Revenue ($2031.64) - Expenses ($1402.27) ) / Expenses ($1402.27). The math is correct, it works out to 44.88%. The only problem is that this is not how profit margins are calculated. To calculate profit margin you divide the difference between revenue and expenses by the revenue, not the expenses. If you use these same numbers properly, you come up with 25.29% gross profit for 2021 and 30.98% for 2022. In this case, gross profit uses only the cost of clinician visits as the basis for expenses.
Here is a comparison between the CMS margins provided with the proposed rule and the actual gross profit margins using their estimated costs and the actual costs for 2021 and 2022:
CMS states in these same referenced sections that they presented this data, including their “margin” calculations, to MedPAC in the spring of 2023, as they do every year. This is the organization responsible for informing congress on Medicare issues. In the 2023 final rule, MedPAC stated in its comments and in their March 2022 report to congress that the behavioral adjustments would not result in a loss of access to home health care for Medicare beneficiaries. Their report included a recommendation to CMS to reduce the 2023 base payment by 5%. These comments are based on these margins for free standing agencies averaging more than 20% from 2001 - 2020.
In this same MedPAC report, they state that home health agency margins under Medicare increased under PDGM, from 15.4% in 2019 to 20.2% in 2020. They projected margins for home health agencies to be roughly 17% in 2022. These appear to be projections for profit margins including all eligible expenses.
In response to the CMS request for comments in the proposed rule regarding how to apply payments toward the temporary adjustments (total HHA liability for overpayments based on the behavioral adjustment), MedPAC recommended an additional reduction in Medicare base payments for several years, starting in 2023 at the rate of $502.5 million per year to reduce the outstanding temporary adjustment liability for HHAs. CMS did not implement this recommendation in 2023 and they do not propose to do so in 2024.
Why is this important? As it applies to the calculation of home health profit margins, I have found no significant mistakes in the CMS math, just a difference in the interpretation of these numbers and what they mean. In the home health proposed rules, CMS focuses most of their research, calculations and assumptions on two issues.
The calculation of the behavioral adjustment and its impact on annual permanent and temporary adjustments.
The calculation of margins to demonstrate the impact of these adjustments and other PDGM factors on the financial health of home health agencies.
Each year, CMS challenges the industry to reproduce the calculations related to the behavioral adjustment. As far as I know, no one has attempted this or at least published the results or provided them as a comment to CMS. My guess is that if someone did, they would probably end up confirming most of the numbers provided by CMS, but we might find assumption errors that might alter the results or how they are perceived, just as I have illustrated with the second issue, profit margins.
In my next article, we will continue where we are leaving off, with an assumed average home health agency gross profit of 25% for 2022 under Medicare PDGM. What is the actual profit margin? How will these margins be affected by the proposed cuts for 2024? With Sisense and the cost report data, we will answer these questions and more.
Comments