What is the HEI?
The HEI is a bonus ranging from 0 to 0.4 added directly to each plan’s raw star rating based on two things: (1) the percentage of members that either are disabled or have Low Incoming Subsidy or Dual Eligibility. and (2) the number of measures in which the plan scored well for this subset of members.
How Important Is It?
CMS will drop the legacy Reward Factor and replace it with the HEI Reward. It's of a similar magnitude (0-0.4 on your final Star Rating) but the swap is estimated to cut Medicare Stars bonuses by $5.13B.
Don't lose the forest from the trees. This is a seismic shift in Stars. Plans rarely reach 5 Stars without the Reward Factor; we showed in a 2023 Blog Post that of the fifty-seven 5-Star plans in 2023, only one would have been 5-Stars without the Reward Factor.
When?
Right Now! The HEI reward will be implemented with the 2027 Star Ratings – beginning with dates of service in 2024.
Why the HEI Reward?
Page 506 of the 2024 Final Rule, from CMS:
- CMS developed a health equity index (HEI) factor to reward contracts for obtaining high measure-level scores for the subset of enrollees with specified SRFs.
- Their intent in implementing an HEI reward is to improve health equity by plans to perform well among enrollees with specified SRFs.
- The CAI is designed to improve the accuracy of performance measurement, while not masking true differences in performance between contracts
- In contrast, HEI reward is specifically designed to create an incentive to reduce disparities in care.
- HEI reward, therefore, would not replace the CAI but rather assist plan sponsors in better identifying and then addressing disparities in care provided to members with a particular SRF, with the ultimate goal of reaching equity in the level and quality of care provided to enrollees with SRFs.
- There would be no changes to the current CAI with the implementation of the HEI reward.
In other words the CAI adjusts for an unfairness in certain Stars measures stemming from high risk members, whereas the HEI Reward incentivizes plans specifically to enroll and improve care for these members.
Some Musing
Plans have always had 2 ways to get higher Star ratings:
A) Work with case management, benefit design, network management, operational excellence, etc. to get members to close gaps in care, live healthier and have a higher opinion of their plan and providers. - OR -
B) Bias who joins your plan. Use geography, benefits design or marketing such that you have more members who are predisposed to follow doctors' advice and more naturally inclined to give high marks on CAHPS surveys.
B is much easier. 10 or 15 years ago, when Stars were new and I worked for a C-SNP, CMS was philosophically opposed to risk adjustment in Stars even as it was patently obvious that plans with certain membership demographics were going to struggle. They gave 2 reasons. First, SNPs and other plans with high risk members were already compensated with higher HCC scores elsewhere in the payment model. Second, adjusting Star ratings lowers the demand for high quality care -- but precisely for the members who have the highest need. As Stars have played out, this has become untenable. It's become quite clear that plans in the midwest do better on Stars just due to culture. Plans with high concentrations of group retirees -- combining extra funding from the pension with a population that naturally skews away from poverty. That's a perfect setup for Stars, then the Reward Factor adds to it. "Since you have a high Star Rating, here's a bonus to your Star Rating."
CMS is taking away this added bonus while at the same time attempting to counteract the incentives for cherry picking healthy members. It offers a 3rd choice:
C) Enroll members with high needs. Do everything on #A to perform well on this membership.
Will it work? It's a complicated calculation; perhaps too complicated but it's a powerful nudge from CMS. The losers from this are obvious, but it's less obvious who will win. Only the plans that have high risk members also can operate at in the top tier of measure performance. This hasn't always been the case with Medicaid focused companies.
What’s the Impact in Dollars?
Page 15 of the 2024 Proposed Rule CMS says:
The HEI reward provision, which would replace the current reward factor, is expected to result in net savings of between $680 million in 2028 and $1.05 billion in 2033, resulting in a ten-year savings estimate of $5.13 billion.
By “savings” they mean less money to plans and members due to lower Star Ratings. This is largely because 4.5 Star plans get a 70% reimbursement from CMS and 4.0 Star plans get 65%.
Also note that this is “net savings” – losing Reward Factor will cost plans Stars money and gaining the HEI will boost different plans.
HEI Related Data
The Proposed Rule introduces the concept of “Social Risk Factors” (SRF) to Stars. That sounded fancy and new, but there are only two SRFs – Dual/LIS eligible members (indicating poverty) and Disability Status. These are already familiar in Stars as the inputs to the Categorical Adjustment Index (CAI).
CMS had been laying the groundwork for the HEI for several years now. We can see this in how HEDIS rates are reported. In addition to the overall rates, plans must report the following subgroup “Stratifications” for some measures for Medicare populations:
- Non-LIS/DE, Nondisability
- LIS/DE
- Disability
- LIS/DE and Disability
- Other
- Unknown
- Total Medicare
See where this is leading? Combine the rates from #2, 3 and 4 and you have a HEDIS Rate for just the “SRF” subpopulation. For any non-HEDIS measures with member-level data, CMS has similar ability to report subgroup scores for CAHPS® and for HOS®.
The other input that we need for the HEI is simply the percentage of members that have an SRF, in other words what percentage of the members in the plan have LIS/DE or are disabled.
HEI Reward Calculation
First we calculate a “HEI Score” as a decimal number between -1 and 1. Then we multiply that score by 0, 0.2 or 0.4 based on the percentage of “members with SRF”.
Part 1 – HEI Score
- Calculate the measure rates for the SRF members
- Include any case mix adjustment in the measure (w/some exceptions)
- Exclude contract-scores with low denominator or low statistical reliability
- Sort the contracts by each measure and assign 1/0 or -1:
- 1: Top Third Score on this Measure
- 0 : Middle Third
- -1 : Bottom third
- Take a weighted average. This gives the “HEI Score” between -1 and 1.
- Skip any contracts where HEI Score is negative
Part 2 – HEI Reward
- Calculate the percentage of members with SRF in all contracts
- Calculate the median % SRF and ½ that median
- Using the median and ½ median
- Calculate HEI Reward Factor which is 0, 0.2 or 0.4 times the HEI Index:
- If %SRF < ½ Median: No HEI Reward Factor
- If %SRF < Median: 0.2 * HEI Score
- If % SRF >= Median: 0.4 * HEI Score
- Add the HEI Reward Factor to the plan’s weighted average as you normally would with the legacy Reward Factor.
For any longtime Stars person, all of this should feel vaguely familiar. It’s really an amalgam of three existing parts of Stars. It uses the similar data to the CAI, calculates a weighted average of 1/0/-1 for each score just like the Improvement Measures but then comes up with a score that gets applied just like the legacy Reward Factor
A Quirk
HEI and legacy Reward Factor feel similar; each one has two tiers of bonus at 0.4 and 0.2. But there’s a huge difference in how those tiers are set.
The legacy RF bonus was only available to plans with a “high average” and a “relatively high average” – basically a 4.2 and a 3.8 raw score. If you’re already at 4.2, what does adding 0.4 do? It shoots you up to 4.6 and puts you very, very close to the 4.750 you need for 5 Stars. The net effect is that it it’s lowering the bar to 5 Stars. The “relatively high” group only gets 0.2 and doesn’t have as decisive an impact, but it is a similar effect. The takeaway message is that legacy Reward Factor’s just lowers the bar to 5 and 4.5 Stars.
HEI Reward Factor also has 0.4 and 0.2 tiers, but it’s not dependent at all on the raw average. You can get this reward at any overall star rating. In practice, this means that the HEI RF is now going to lift some plans up into 4 Stars (maybe some into 3.5 or even 3 stars as well) but won’t have such a large impact at the high end.
Timeline
HEI reward includes two years of data, the earliest measurement year data it can use is from 2024 and 2025, and the earliest it can be implemented is the 2027 Star Ratings.
The contract percentages of enrollees with SRFs included in the HEI will be based on enrollment in the most recent of the two years of data used to calculate the HEI.
For example, if the HEI includes data from measurement years 2024 and 2025, CMS would use enrollment from 2025. The percentage of enrollees with SRFs would include any enrollees who are LIS/DE or have a disability. This is treated as one group of enrollees with SRFs.
Discussion
This is a very large change in Stars – the big news when HEI was announced in the Proposed Rule was simply that the Reward Factor was going away to be replaced by something that’s going to have a different impact on different plans.
Another quirk of the HEI is that it is not a “step function”. Instead of always getting 0, 0.2 or 0.4, the HEI can be any number in between. This is a huge departure for Star ratings which tend to have “yes/no” quality to them.
In the Proposed Rule, CMS discussed a number of other possible SRFs that they considered, including a geographic “Area Deprivation Index” that measures “income, employment, housing, education, social environment, and readmissions” but they weren’t able to find an easy to use indicator with predictive value that they could apply. What this suggests to me is that the proposed rule is just the start; CMS is trying to come up with a structure that allows them to build over time. This is going to be a challenge for them, however, because they will either have to collect member level data or require plans to apply the SRF logic before submitting rates.
Disability status in the HEI is a flag indicating disability at the time of enrollment into Medicare. CMS is asking for other ways of identifying disabled members.
Another Quirk from 5th Grade Math
The HEI Reward is a single number summarized across all measures, but there's a quirk in the math that makes each measure's HEI bonus almost freestanding. You can give it a value and calculate ROI per-measure independent of performance on other measures. Think back to 5th grade Math.
Remember the "distributive" property: A*(B+C) = A*B + A*C (yes, I looked it up to write this)
The same holds if “A” is involved in division or is really complex.
So consider the HEI calculation:
“Sum the 1/0/-1 indicators for each measure multiplied by each measure weight. Then divide the sum by the total measure weight. Then multiply this value by 0, 0.2 or .4”
Write this out mathematically like this:
This can be rearranged a little:
Everything on the right side has nothing to do with your performance on any measure. It's set by the demographics of your plan and the measures you're eligible for. Let's just treat it as a constant and call it "PH" for "Plan's HEI Eligibility and Weight Multiplier".
Here's where the distributive property comes into play. You can use the multiplier on each measure instead of adding them up first:
Why is this important?
It means that you can isolate individual measure impact in the HEI Reward. As long as your overall HEI Reward is on track for break-even (i.e. not in the bottom third for most measures) and you qualify for the HEI RF at all (i.e. your % LIS/Dual/Disabled > about 20%), then it’s easy to calculate what would happen to your Star rating by improving C01 from the bottom third to the middle third or the middle third to the top third.
This means that the HEI gives you another way to set targets with a clear ROI for each measure. This didn’t exist with the legacy RF or the improvement measure in which the value in Stars of each measure is dependent on what happens with every other measure. (You still have to get from Stars to dollars but that’s a topic for another tutorial.)
Measures Included
CMS outlined some guidelines on how they will select measures for the HEI each year but noted that they will not be able to finalize it each year until they’ve collected and analyzed the data:
In general, measures from HEDIS, HOS, and CAHPS would be included unless they meet one of the exclusion criteria, as previously described. Additionally, medication adherence, MTM Program Completion for CMR, and Statin Use in Persons with Diabetes measures would be included as long as they meet the requirements for inclusion for more than 25 percent of contracts.
This is as-of the 2023 Proprosed Rule. CMS has not updated this information.
An Example from CMS
Key References
- 2024 Proposed Rule (Starting Page 175)
- 2024 Final Rule (Starting Page 504)
- CMS Health equity webinar (Starting Page 23)
- 2025 Proposed Rule (Recent proposed rule had no major changes to already finalized provisions)
About Baltimore Health Analytics
BHA is the leader in Medicare Star Ratings strategy and analysis. We help health plans and partners understand how to operate in this very dynamic and very important area. If you find this article interesting, you may like our Stars Planner (free or premium version), our tutorials, and other content on our website. If you want help digging deeper into your own contracts, please contact us for more information.
You can reach BHA at info@BaltimoreHealthAnalytics.com or go to www.BaltimoreHealthAnalytics.com and register an account to use the tools.
