In our last article, we explored CMS’s efforts in the 2024 proposed rule to describe the profitability of home health agencies in an effort to establish the financial health of the industry under their current reimbursement methodology, PDGM, and after their proposed cuts for 2024.
Their logic uses visits from 2022 claims and costs per visit from 2021 as a method of calculating expenses per 30-day period for the financial year under review (2022) for the 2024 proposed rule. The base rate paid for 2022 is used as a measure of revenue. Because their calculated visits per period have declined since PDGM began, their estimates paint a rosy picture of increasing home health agency margins under PDGM.
They use these results to support their theories under the behavioral assumption that agencies would work to improve these margins under PDGM initially by managing PDGM factors to increase their revenue and then visits to reduce their expenses. In this article, I intend to explore this expense issue with cost report data. To do this, we need to take a look at the actual expenses from the cost reports and compare them to these estimates by CMS.
Using Cost Report Data to Check CMS Assumptions
We have cost report data for 2020 and 2021 under PDGM, CMS considers these years to be complete. When the 2024 proposed rule was developed early this spring, less than half of the 2022 cost report data had been processed. CMS deals with this by using the cost report data from the prior year since it is more comprehensive. In the current proposed rule, they use 2021 cost report data to estimate 2022 costs (see Table B4). They then combine this with visits from 2022 claim data which they consider to be complete at the time that the proposed rule was developed. The estimated 2022 cost data becomes the basis of cost data used to develop the PDGM weights for 2024. Each year, when they refine this data, they advance one year forward in the data used for the proposed rule and repeat this process.
In reality, even the claim data from 2022 was not 100% by the time they started on the proposed rule for 2024. However, it was close enough. Based on my experience with this claim data, CMS had all of the first three quarters of 2022 when they began working on the proposed rule, but the final quarter was not released until after the proposed rule was nearly finished. To deal with this, during the comment period, they update their claim data with the final quarter’s claim data and if it changes the results, their calculations are updated in the final rule.
So far, we have not seen this happen. This is because even 75% of the claim data is enough to create a clear picture of the final result when examining the data through averages and percentages and not totals for the period. Revenue in the final quarter would not change the outcomes since the payments are set annually and do not normally change through the year. On the expense side, expense per period would not change unless home health service patterns were significantly different over the final three months of the year. This is why CMS relies on these per period estimates for their calculations instead of cost report totals.
Cost reports are filed by agencies after the end of their fiscal year. For most agencies, this is also the calendar year. Like the claim data, there is a delay between the end of the year and the availability of this data. For cost reports, the delay is even longer. It takes nearly a year after the end of a calendar year for nearly all of the cost reports for a year to be processed. This database is updated quarterly so the largest batch of cost reports processed each year is in the release in July of processed reports through the end of June. By then, reports should have been processed for all agencies for the previous calendar year that had a fiscal year end of 12/31, but some are still missing because they are in review by the MAC, being refiled, or are simply late.
My version of the cost report data includes the data for 2022 unavailable to CMS when the proposed rule was published. This data includes the update from July (all cost reports processed by 6/30/23 released 7/17/23). This is still the most current version of the data until it is updated again sometime later this month (10/23). This data includes only free standing agencies. CMS explains that they represent better data quality than agencies that are part of hospitals or other healthcare organizations because they represent the entire picture of business revenue and expenses unaltered or subsidized by other provider type relationships. I would agree.
Each provider is required to submit a single cost report per year, but some providers have split reports due to accounting issues like changes in fiscal year or acquisitions. We can count these reports by year and the number of unique agencies submitting them. To clean this data, I have omitted some invalid cost reports such as those that include no line items or no costs. CMS describes similar steps in the proposed rules to clean this data prior to use. Using Sisense, here are the results:
You can see that 2020 is finalized and even though 2021 is considered complete by CMS, there is still some minimal updating going on. For 2022, if we can expect a totals similar to 2021, about 80% of the final reports are included in this version of the data as of the July update. This makes this data as “complete” as the 2022 claims data used by CMS initially in the 2024 proposed rule (75% of 2022 claims).
If you use this data to calculate totals by year, the results will be distorted by this missing 20% of the cost reports for 2022. In this format, the data can also be distorted by other factors such as the growth or decline of the overall services provided by the home health industry year to year.
Using Census From the Cost Reports as a Unit of Service
To deal with this, we need to duplicate CMS’s logic of looking at this data using units other than annual totals. In the proposed rule, CMS uses the claim data to determine the number of 30 day service periods per year. They then use this unit of “30-day period” to calculate averages for visits and costs. This allows you to create reasonably accurate ratios. As long as the data you have access to is statistically significant, it does not have to be final.
In this chart from the 2024 proposed rule, CMS calculates these 30-day periods for 2018 - 2022 in Table B1. They then use this unit to measure visits by discipline per 30-day period in Table B2. This makes this data easier for stakeholders to comprehend and CMS can accurately measure this claim data for 2022, even though this data is incomplete compared to prior years.
In the cost report data, we do not have 30-day periods. However, there is another unit of service measure available that can be applied to data that is included in the cost reports, patient census.
Using this metric, we can compare the level of services by year using both units, 30-day periods and census. With this approach, we can see an increase in overall home health services from 2020 to 2021 and a decrease from 2021 to 2022. For the CMS 30-day period units, the 2022 decline is 8.01%. Using census as a unit, the decline is 8.48%. Although the numbers don’t exactly match, they both appear to provide the same general ratios, as you would expect.
CMS uses Chart B1 and the 30-day periods as the units for subsequent measures of revenue, expenses and margins like those in Chart B4, discussed in my last article. In the remainder of this article and the next one, I will use census to measure the same metrics, and some new ones.
Before I continue, I want to revisit the subject of data assumptions that I discussed in the previous article. In that article, I talked about how these assumptions influenced the conclusions shared by CMS in the 2024 proposed rule in support of their position regarding home health profit margins. I want to be clear that although I question these assumptions, I found no significant errors in the calculations. I also want to make it clear that any assumptions I make should be treated with the same level of skepticism by you. For CMS, their purpose was to support the position that the proposed payment cuts in the rule did not threaten access to home health services for Medicare beneficiaries. I am going to attempt to show how the same data demonstrates a completely different picture of home health profitability under Medicare Part A and overall.
It is entirely possible, and even likely, that I will make some assumptions that are flawed in an effort to support this position. My point is that regardless of motive, these mistakes can be made and the numbers I provide you can be inaccurate or not actually support my conclusions. This is particularly true because I am performing this analysis on my own. For this reason, I depend on the readers of these articles to analyze the data from both sources and draw your own conclusions. If you find what you believe to be mistakes in either my calculations or the assumptions, I am depending on you to let me know. If you want to try to reproduce these numbers, I can show you how.
For example, here is some data and an associated assumption. This data on 30-day periods and census shows an increase in services from 2020 to 2021 and a decline in services from 2021 to 2022. Why did this happen?
I spent some time working on this question over the last year. My theory is that this was related to Covid. In 2020, about April, there was a nationwide cancellation of elective surgeries as the acute care institutions became maxed out dealing with the pandemic. For home health agencies, the most common primary diagnosis for a patient is recovery from knee replacement surgery. This has been true for many years. When these elective surgeries were eliminated, it had a negative impact on the home health business for the remainder of 2020 and part of 2021.
During this period, there was pent up demand for these surgeries that were postponed and the home health market expanded as hospitals caught up on elective surgeries in 2021. In 2022, as the acute care backlog was processed, home health services began to return to the pre-pandemic “normal”. This explanation makes sense to me and it might make sense to you, but that does not mean that it is the entire or even dominant explanation of this data.
Using census as a unit of measure, let’s apply this unit to visits to see if we get similar results as CMS did with their 30-day period in Table B2 for the PDGM years of 2020, 2021 and 2022.
We can find total visits in the cost report data for each year. When PDGM began in 2020, CMS also redesigned the home health cost reports starting the same year. I am assuming these events are connected. In the new version of the cost reports, they collect this data by three financial classes or sources of reimbursement, Medicare, Medicaid and Other insurance. I am not aware of the purpose of this redesign, but it turns out that this new data gives us additional insights that are still unavailable for other provider type cost reports.
Here are the CMS visits per 30-day period provided in the 2024 proposed rule and the total visits and Medicare visit averages per census under the first 3 years of PDGM from the cost reports. For the CMS data in their proposed rule, only Medicare Part A data is used. This is identified as “Medicare” in the cost reports, but it does not include Medicare Advantage (MA) which is in the “Other Insurance'' and agency totals.
In this spreadsheet, both methods show that home health utilization increased from 2020 to 2021 and then decreased from 2021 to 2022. Both show a decrease in visits per unit from 2020 to 2021, but the cost reports show that the visits per census stabilized and even went up slightly in 2022 while CMS says they declined using data from cost reports combined with claims.
Let’s take a look at another assumption from CMS regarding expenses per visit. There are two factors involved in this statistic, the decrease in visits per unit as described by CMS and the cost reports and the increase in cost per visit from year to year.
In the proposed rules, CMS estimates the visit cost per unit by discipline as we described in the previous article and then multiplies this cost by the visits per discipline for each 30-day period. In the cost report data, we can apply a more direct approach. We can take total costs for all disciplines and divide it by the census to come up with a cost per census. In this case, to keep it simple, we will look initially only at the total wages paid per year for all clinicians visiting patients. We will use the same data source used by CMS in their calculations, the wages from Worksheet C of the cost report. I will refer to these as “Direct costs”. These costs do not include employee benefits, transportation costs, and other costs associated with home health patient visits, but should prove reliable to demonstrate changes in percentages and ratios from year to year associated with these clinician costs.
Here is what the direct cost per census looks like and the changes year to year per census.
If we go back to the estimated costs per 30-day period from the 2024 proposed rule, CMS estimates these as $1420.35 in 2021 and $1402.27 for 2022. This is a decrease in 2022 compared to 2021 of 1.27%. When we look at direct cost per census (wages only), this increased from 2021 to 2022 by 5%.
Why would Medicare show an estimated decline in costs per unit while the cost reports show an increase in wages? Here is my theory (assumption). When CMS estimates their costs in the proposed rule, they calculate a cost per visit by discipline and then use average visits by discipline from the claims and multiply them by these costs to come up with a cost per period.
Period Visit Cost = Cost Per Visit x Visits Per Period
When Medicare makes their estimates using this formula, they make one significant false assumption. Any time there is a change in visits, in this case, when visits are going down, there is a direct relationship between the reduction in visits and a reduction in visit costs for the agency. This is not how it works in the real world.
If a visit by a clinician to a patient is canceled or otherwise does not happen, the clinician still gets paid the same. These individuals are paid based on a salary or hourly rate for the time they are available to work, not the work they complete. Reductions in visits do eventually result in a decrease in expenses, but there is not a direct relationship. In order to realize savings at the agency level, these clinician costs have to go down. This only happens when you adjust how many clinicians FTEs you have providing visits. Agencies do this when visits have decreased to a level for a certain discipline that can no longer justify the same number of employees in that discipline.
Unless you want to put yourself in a position where your patients are underserved for that discipline, you must wait to reduce this cost until the overall reduction in visits is greater than the visit productivity per period of a single clinician. Then, you can let that clinician go and realize these savings in the future.
For this reason, when your market is expanding, your costs per unit decrease even if wages remain the same because your sales are increasing while your wages are not. When your capacity to serve your customers is maxed out or you hire in anticipation of growth, you hire another person to expand your capacity. Immediately, your costs per unit increase, but they decline again as your business continues to grow.
When your market is shrinking, the opposite occurs, your wages are stable, but your units sold are decreasing (census or 30-day periods). With each decrease in units sold, your cost per unit increases until the unit decrease is large enough that you reduce your staff to compensate. When you do, your cost per unit immediately decreases, but then continues to increase as your market continues to decline with the same level of staffing.
If you were to chart this data at the agency level, you would not see a straight line of variable cost reduction based on visits, like a downward ramp, but stair steps of expenses going down at the general rate of your service period (sales) decline, but at a higher average per unit cost than what is project by CMS when this is summarized for a year.
CMS calculates similar levels of industry growth from 2020 to 2021 and contraction from 2021 to 2022 that I do using their 30-day periods compared to census. However, their estimates assume an economy of scale at the industry level that does not match what is happening at individual agencies. The individual agency cost reports include this factor. This is why, in the previous article, when I calculated cost per visit using actual 2022 cost report data compared to the CMS estimate for 2022, I got a larger amount per visit.
If my assumptions are correct, these CMS estimates of costs per period will always be lower than actual costs per period in a declining market.
Overall, this is an example of how business intelligence can be applied to data to support an argument, only this is from the other side of the negotiating table, from the perspective of HHAs instead of CMS. Each of these percentage points associated with home health margins represent hundreds of millions of dollars. The difference in estimated cost increases per period by CMS for 2022 (2.6%) and the actual increase in period costs in the cost report data (8.63%) is 6%. This is larger than the cut in the 2024 proposed rule for the behavioral adjustment (5.1%).
These costs are used for the proposed CMW for 2024 and the per visit LUPA payments, if they are underestimated, they reduce overall revenue for HHAs not by altering the base payment, but by lowering all the 462 CMW and LUPA rates using the faulty visit expense data.
Using this actual cost report data per census, we can get a clearer picture of revenue, expenses and profit margins for HHAs for the first three years of PDGM. So far, we have been using this data to check CMS expense estimates, but we can also use this data to measure actual overall performance at the industry level by averaging cost report KPI (Key Performance Indicators) by census. With the new format of HHA cost reports where the data is collected by financial class, we can measure performance for each financial category and get an overall picture of profit margins for the industry for all types of reimbursement, not just Medicare Part A.
In my next article using the cost report data, we will continue this approach to further explore financial metrics for Medicare Part A and the dark side of the home health business, “Other” insurance as described in the cost reports, more familiar to HHAs as Medicare Advantage.
Comments