Curious if anyone has worked or is working on this from the payer side.
MA is a weird space, everything pretty much revolves around the MLR. Because of that, payers want to set MLR targets for providers to hit (I’m on the provider side, working with PCPs). But there are some inherent problems with doing that.
Payers often seem to change plan benefits, tweaking ancillary benefits like vision or wellness, that can impact the MLR somewhat significantly. Or payers can bid aggressively to acquire new business, which translates to lower premiums and, for a given MLR target, the ‘benchmark’ comes down in lock step.
And then there is stop-loss, MA doesn’t have it. Currently, we’re being encouraged to purchase stop-loss coverage from a third party, and this fundamentally irks me. Payers literally exist to take on this kind of actuarial risk. Why should they pass this down to providers, and then force providers to obtain coverage for these events? It’s doubly troubling because we often only have a few thousand lives, chopping up the population and insuring at that level is very, very expensive.
I guess my chief complaint (since I’m on my soapbox) is: MLR is NOT a counterfactual. If you want to drive savings, then you need a benchmark that should answer the question: what would these beneficiaries have cost absent the intervention? I don’t believe MLR does that adequately, and not applying stop-loss just adds a lot of noise.
Now, from the payer side, I get that MLR is how the rules are laid out for you. It’s how the game is played, so it makes sense to try to make MLR work for providers. How do we balance these two so that payers and providers are aligned, to the extent possible?
It’s been a long time since I worked on MA, in effect it was called something else at the time, and I cannot remember what it was.
The few employers I worked for had providers receive X% of the Premium from the Government, then left them to their devices. Is that not done anymore? Did providers wise up (or go bottom-up) since then?
Also, smartest thing I ever did was switch employers to avoid working on MA.
I’m in payer VBC, but I don’t work on the MA lines. Our MA side has been pushing for global caps and a large number of our members are in those relationships. This at least takes the premium component out of the equation for the provider, leaving a sole number to haggle.
As for the stop-loss component - I 100% agree with you. Pushing that risk onto providers just isn’t going to end well. I worked some large group previously in my career, so I prefer a pooling approach. I think it’s appropriate in the pooling method for the payer to also add a risk-charge to the dollar amount, but even with that it would be far cheaper than the provider getting reinsurance on their own. It is passed on in practice because payers are offloading their risk and totally fine doing so. I think it will take smart providers pushing back to get that to change - and I’ve approached a few providers with the pooling approach and while many take it, some have rejected it.
As an aside - are you on the SOA VBC subgroup calls? I think this could be turned into a discussion topic there.
anecdote time:
we worked with a provider client who participated with the payer on capped basis. so instead of being attributed $500 PMPM they got $480 PMPM and all individuals with costs above $X would be truncated and excluded from the global performance.
based on history, the real cost of that excess coverage seemed like it should be $40 PMPM, not the implied $20. someone wondered HOW CAN THE PAYER DO THAT?
I asked them how much the payer was keeping on a PMPM basis for the admin/ASO portion. I suspect the contract wasn’t negotiated to the bone if they had that extra amount to “gift” the provider to participate in the contract.
Oh, and when you say ‘pooling,’ do you mean that you can’t let one provider join by themselves because you can’t measure savings on 300 benes? My company is basically an aggregator, so we do the pooling. I’d like to tell you that we won’t sign up for a program with fewer than 5,000 benes, but we aren’t there quite yet. We wouldn’t sign up for a deal with 500 benes unless there was no downside risk, and even then we’d want a plan to get our bene count way up.
Yep. Components as well, such as trend, but it really boils down to the final PMPM number.
Somewhat agreed. I’m not sure the engagement you would get back from others on the call either - people are a little reserved on that call. I did an in-person provider payment boot camp a few years back, and while basic content, it did have more discussion from attendees - format might have been the reason there.
Ah no. I was a little too loose with the term. More like setting a stop-loss limit, say $200,000, and then charging a flat amount for the dollars above that and removing those actual costs from the calculation.
We also do VBC deals with much fewer beneficiaries than I’d personally like - but it can be hard to find providers with sufficient patient volume.
OK, that’s a really interesting concept. I like this idea quite a lot.
I spent seven years in oncology. My first project was to determine sample size for a credible study, I think I told them I wanted 4,000 but I could live with about 3,000. They told me to take a zero off that, our first program had about 260. It was hilarious, our savings were negative for prostate cancer and I was like ‘yeah, that’s Paul, he spent time in the ICU and that tanked us, because we only have 40 prostate patients.’
I work primarily in MA pricing for a payer and have some involvement/oversight of our MA VBC arrangements. My thoughts largely align with yours.
I think MLR can in theory be a counterfactual, but in practice it’s tough to get both sides buying in. Each year, after bid submission, my team puts together projected loss ratios for our MA VBC risk groups based on bid expectations, and each year they’re entirely ignored by our contract negotiators. If we are expecting a profitable year, providers aren’t going to like a corresponding drop in MLR target. If we are expecting an unprofitable year and/or price aggressively in order to grow, providers would certainly like the corresponding increase in MLR target, but we don’t want do that because we’d effectively be sharing “surplus” with providers while actually running at a loss (and also once the target increases it’s hard to bring it back down in future years).
On most of our contracts, we just pass through the reinsurance coverage that we ourselves purchase for our MA business, though the attachment point is too high to be very meaningful for most providers. One of our main providers has opted out of this and taken out their own coverage instead, I think through a captive. We do the sort of internal stop loss pooling that MayanActuary mentions on our VBCs on other lines of business but haven’t gotten any takers on MA.
We’ve strongly recommended to our contracts team a minimum group size of at least 500 members for VBC, to no avail. I think we have some groups taking downside risk who have only 100 members attributed. So dumb.
Oh, I’m shocked that your negotiators would just do whatever they want, lol. I have a biz dev person doing the same, they want to take full cap risk on pools of 500 benes, and I told them they would be doing that over the unambiguous objection of the company’s actuary, and they may do it anyway. I get not wanting to pay out shared savings during a lean year, but from the provider perspective it basically feels like we are paying for your marketing expense. Apologies if that comes off as me being terse, I’m not trying to be. I think we just need to bring both parties together and get some more alignment, I think we have a lot of good physicians that can help reduce cost, just a matter of figuring out how to share in the pie equitably. It’s hard.
We’ve had a payer or two offer to, basically, sell us stop loss. I don’t know if it’s a pass-through as you describe, we’ve found the costs to be very, very high, so it doesn’t feel like they’re being as generous.
I was not aware of this (I’m new to MA for the most part), I very much appreciate this bit of info!
This isn’t quite Wall Street Bets levels of insanity, but it’s not much different than putting $20,000 on black at the casino, lol.
i’ve seen it priced at fair market, favorably priced, and OMG-ridiculously high priced. as Bassooner notes, the actual reinsurance the payer buys is so much higher than what the payer needs (given smaller size and lack of meaningful surplus) that a “pass through” usually isn’t without some projections to price at a much lower attachment
I’m back, with another question for @MayanActuary and @Bassooner , other opinions welcome! In the past, I had generally assumed that MA plan performance would be independent at the contract level. Meaning I assumed that my projections for shared savings would be wrong for Aetna, wrong for Humana, wrong for United, but the errors would be… let’s say normally distributed just for simplicity.
However, in 2023Q4, we saw MA costs increase significantly (you may have noticed this if you track the stock prices of Humana and CVS/Aetna). I probably should have seen this coming but I haven’t had much time to devote to MA. At any rate, pretty much our whole MA book was written down for Q4.
So my thought was to start with some really simple models, using MLR by contract, by quarter. And see if there is a relationship across payers, something like correlation but maybe another metric would be more appropriate. And then separately, measure something like the coefficient of variation at various contract sizes. And from there, build something like a Monte Carlo model that simulates across two things:
Individual contract performance. Basically just variation based on the population size, to account for randomness and high-cost claimants, etc.
A global factor that accounts for what happens when the whole market shifts up/down because of a broad secular trend. Something like Ozempic costs that just hits the system, for example (I understand not all MA plans have Part D risk).
And then generate thousands of scenarios to build up confidence intervals. Does this seem like a reasonable approach?
Seems like a reasonable approach. You could potentially go back and look at some of the MSSP PUFs to see if you can find correlation in those results. But even things like flu seasons being worse can have major impacts across all carriers, so you definitely want that global factor.
And yeah Q4 MA was baaaaad, and I know at least one of the major carriers was caught completely by surprise on this
Thanks for the sanity check. I’m sure there are other approaches and maybe down the line I’ll iterate some more, but I need something at least a step better in 3 weeks.
That’s a good call on the PUFs, I suspect correlation would be pretty comparable between MA and FFS.
In addition to the things like flu season intensity and GLP-1 hitting all carriers’ utilization to varying extents, you also likely have some correlation in unit costs just from the simple fact that lots of carriers’ provider contracts are pegged to the FFS fee schedule, so annual updates and/or policy changes there will hit more or less everyone. Similar deal on the revenue side, with everyone getting hit by changes to the risk adjustment model and benchmarks (though the former is more population/coding-dependent, and the latter is more geographic-dependent).
I had thought that changes to the PFS would be captured in the bids, but now that I think about it, maybe they aren’t published in time.
And I bet on the v28 HCC model, once it’s out providers will find new, innovative ways to upcode… er, ensure that all comorbidities are entered accurately.
This is making me think there is definite correlation (maybe not in the statistical sense of the word, but probably) and maybe more variance than I’d previously assumed. Ugh.
Yeah, changes to the fee schedules are generally captured in the bids when known, but it’s a lot more likely in MA than in other health LOBs that carriers will maintain premiums at the expense of margin / MLR (especially if they’re at $0 premium), so my guess is that the fee schedule changes tend to hit most plans’ loss ratios similarly.
