The Obstacles To Value-Based Care
Get Out-Of-Pocket in your email
Talking about value
The idea of paying for quality in healthcare instead of volume is older than your mom (gottem). But considering we’ve been talking about it for so long, why do we continue to see most spend in fee-for-service vs. value-based care? We’ve had more success creating names and acronyms for these reimbursement programs than actually implementing them.
First I’ll talk a bit about the key components of value-based care programs, operational challenges/considerations for companies that want to enter them, and then some musings about why it’s been hard to shift the industry as a whole towards it.
Value-based care mechanics
For simplicity’s sake, I think there are really four dimensions to a value-based care contract.
Type of risk - Value-based care contracts can generally be in “upside only” or “upside-downside” risk. This determines whether you only get bonuses if things go well, or do you also get penalized if you mess up. Usually being in an upside-downside contract will have higher rewards for the upside vs. upside-only. Either contract can also have payouts held until quality metrics are hit as well.
Voluntary or not - This is more specific to government programs, but one question is whether a value-based care program should be opt-in or required. If not required, then healthcare orgs will only opt-in if the financial incentives are high enough relative to their normal fee-for-service business which can be tricky and probably won’t lower costs. If required, it’ll be way harder to actually get buy-in to get it approved cause the providers will fight you.
Patient population - Which patients exactly fall into the bucket of value-based care? This requires some careful design because you need to know when you’re in charge of quarterbacking care for a patient since they might be seeing many different providers. Also, if a program is targeting a certain disease type and severity, then you need to know what data to look out for to identify those patients and you want to make sure there are the right incentives in place so that providers/payers aren’t gaming the system. If you provide no incentives, providers/payers will take the easiest cases that are low-risk and avoid sick patients. If you provide too many incentives, they’ll make patients look as sick as possible to get more money.
Measured Metrics + Benchmarking - You need to set a baseline to understand if things are improving. First you need to establish agreed upon metrics that are considered “high quality” and how they’re measured. This is tricky because in many cases the metric is a proxy that you can measure sooner for an outcome you’re actually trying to measure later (e.g. management of blood sugar levels today which will most likely lead to less diabetes related complications later). Then you need a few years of historical data to establish a benchmark for that metric that you want a provider to be better than. The benchmark might be set on a per provider basis where you’re looking at just their historical data, or it might be a benchmark generated from an aggregate number of similar healthcare organizations (normalized for patient population, geography, provider size, etc.). Benchmarks can also be prospective so the provider knows a target price they need to hit or retrospective which is more accurate but tougher on providers who aren’t sure what they’re going to get paid.
Time covered - For what period of time is the provider on the hook for the patient’s health? If you go for knee surgery, it probably doesn’t make a lot of sense for you to get penalized for a normal wear and tear issue 20 years later. Providers might be on the hook for a patient during an “episode” of care with defined start and end times, or maybe for their total cost of care every year but it resets every year.
Payment cadence - When do you get paid? For a lot of providers/payers, they might need upfront capital to make improvements needed to deliver on these value-based care promises (tech upgrades, staffing upgrades, etc.). But you also want to make sure that most of the payment is coming from when they actually hit the quality metrics. And is it the same payment amount each time?
There are way more considerations when thinking about value-based contracts, but these are usually the bigger sticking points. Some common models that you might see that combine these:
Quality bonuses (sometimes called Pay for Performance) - This is the most straightforward, and it usually consists of bonuses going to providers based on whether they took certain actions or not. New York has had a P4P (pay-for-performance because it was created in the 90s when replacing “for” with “4” was socially acceptable) quality bonus system for a while now. The program will give a % bonus above the premiums paid to managed medicaid health plans if they do certain actions. These are usually consensus measures of “good” care. Many plans will use HEDIS measures which include things like cancer screenings, vaccination rates, etc., or they’ll create their own. Medicare Advantage plans get Star Ratings which result in bonuses for the plan for higher ratings that are based on a composite of quality metrics including customer service, plan responsiveness, screenings/test completed, member complaints/churn, etc.
Bundled payments - For an episode of care that’s relatively well-defined and predictable (e.g. a surgery), there’s a good understanding of what types of things should be included in the “bundle” that treats the episode and what quality looks like. Bundled payment programs will typically reimburse as normal and at the end of the year compare what you spend vs. the target price (benchmarking what a “target” price is is the hard part). For providers below that amount and that hit certain quality metrics, they’ll get a bonus. I wrote a bit about the lower-extremity joint replacement bundle here.
Shared Savings/Risk - Similar payment style as bundled payments, providers will charge a fee-for-service and get paid bonuses (or penalties if it’s a shared-risk contract) if they meet certain cost and quality benchmarks that are adjusted for the types of patients they care for. This is usually not a defined episode of care where the services are predetermined like a bundle, but is for a general population with different disease issues that require more long term care management.
Capitation - In a capitation arrangement a provider will get paid a flat rate that’s usually adjusted in some way to give providers more cash if they’re seeing sicker patients. It’s the provider’s goal to take care of the patient underneath that flat rate. Many primary care providers targeting Medicare Advantage patients are in some type of agreement like this. All the others are de-capitated 😂😂 sorry please don’t close the tab.
Challenges for companies
I genuinely think a lot of companies would ideally like to move to some form of value-based contracting, but there is no shortage of operational challenges in the actual implementation.
Problems with benchmarks
It’s hard to get into a value-based care contract if you have no historical data to show for how you perform. You can’t really do any type of value-based care reimbursement except for ones that have a benchmark and reimbursement methodology from aggregate performance data. This is one of the reasons why many new companies will start in regular fee-for-service contracts to build up some performance data before slowly adding more “risk” into their contracts.
Benchmarks can also be problematic if something changes in the regular treatment of a disease. For example, when a new drug hits the market that’s very expensive a provider would be discouraged to use it since the cost of care would go up a lot compared to the benchmarks. In models like the Oncology Care Model this was a huge problem, so CMS added a “novel therapies adjustment” to account for that, but many people in the model claimed it wasn’t enough.
Patient and intervention attribution
It’s really tough to figure out if a patient is actually “yours” or not. Sometimes you’ll need the patient to consent that they’re your patient, sometimes you’ll use claims to figure it out, and sometimes a healthcare organization will auto-assign a patient and it’s their job to get the patient to see them. Just try and catch me!!!
Beyond just whether or not they’re your patient, you’ll need to prove that what you specifically are doing improves their health. This is easier for areas where there’s a well-understood relationship between an intervention and an outcome (e.g. physical therapy adherence and pain reduction/range of motion). But for areas where it’s harder to measure or the benefits are way down the road (e.g. mental health, primary care, etc.) then it’s way harder and why you see these services usually under some “total cost of care” model that includes all of your care.
Patients can end up leaving your care for a whole host of reasons like if they move or if they change jobs/insurance coverage status. Unfortunately this can happen before any meaningful change can be measured and why companies try to optimize for 1 year increments or specific “episodes” of care to measure value.
This is one of the reasons why there’s been an effort to push more risk to providers through programs like Direct Contracting. Patients are more likely to stay with providers vs. insurers. Also patients are more likely to have better outcomes if they have a long time relationship with their provider but are more likely to have Stockholm Syndrome if they’re in a long time relationship with their insurance plan. Incentivizing provider relationships should probably be the goal.
More likely than not, you’re not the first to approach a large healthcare company about setting up some sort of value-based care contract. The reality is that payers, providers, pharma, etc. have a million contracts with different clauses that they even find hard to keep track of. Probably in a file cabinet somewhere dark. Companies proposing their own value-based care contract can bump into these other contracts that might make it more difficult for them.
One example are payment issues like “most favored nation” clauses that guarantee a buyer the best contracted rate across all customers or any reference-based pricing payment model. The low upfront payouts of value-based care models would end up messing up other contracts.
Another example would be companies that provide service offerings where the patient might overlap. For example a company might offer a care navigation solution for patients or the PCP might be at-risk for the patient’s health. If a patient develops a musculoskeletal disorder, they might get handed off to a different service provider that specializes in end-to-end musculoskeletal care and has their own care coordinators. Figuring out the handoff + and who gets paid might get tricky in that scenario.
Data sharing from third-parties
Once you know a patient is yours, you’re going to want a lot of the historical data about the patient. However, getting things like historical claims data about a patient population or data from the patient’s other providers is a huge pain. It requires payers to be responsive, consistently send data back to providers around things like ongoing medical costs for a given patient/patient population, and tightly integrate with the providers that they’re pushing the patient risk onto. It also requires a patient’s other providers being both willing and able to send the data to get a patient’s full history, which is necessary to figure out the actuarial risk of the population you’re taking on. As you can imagine neither of these happen smoothly.
You can get some of this information through things like Blue Button and some of the new interoperability players like Particle Health, Health Gorilla, etc. but there’s still a heavy reliance on claims data to do this which is tougher in commercial contracts and hard to get real-time visibility into. Providers can’t manage patients correctly or understand their future spend if they don’t have this information.
Data capture and reporting
This is so much more of a pain in the ass than most companies realize. This comes down to two things. The first is whether you’re getting all the data necessary from the patient and capturing it in their record. This can be difficult if you’re a physician that’s time-strapped and have more pressing things to get through during a visit than checking vaccinations and doing less relevant screenings for example. The second is how the data you captured is actually getting reported to whoever is paying you. This requires resources to structure the data into submittable formats, which usually requires expertise in cleaning said data. If I search “risk adjustment outsource” there are literally four ad slots of companies offering services for this.
Explaining things to patients
A tricky operational challenge is explaining to patients who are in value-based care programs why you’re making certain decisions. This usually doesn’t happen explicitly - almost no one is telling patients, “hey you’re actually in a capitated model and I get paid a flat fee per head so be cool.” But the reality is that you’re going to be steering the patient's care in some way, and it might be very different from what they’re used to (e.g. if they were previously going to the emergency room as their regular place of care). Messaging this to patients and actually getting them to go somewhere different can be tricky.
Industry level questions about value-based care
Beyond just the operational issues of getting into a value-based care arrangement, there are larger and more ideological questions about value-based care that the industry is still grappling with.
Adjusting for the risk of your patient population
A core component of value-based care is that you need to set up some sort of incentive system to take on riskier patients, otherwise companies will just cherry pick for the less sick patients, which defeats the spirit of these reimbursement systems. This incentive system is called risk-adjustment. However, if you have too many bonuses for taking sick patients, then you get companies that try to game this system by making them look sicker and you’ll likely end up overpaying for those patients.
This is a huge debate in Medicare Advantage specifically where the bonuses for taking on sick patients are high; so firms invest a lot of money into making sure their patients look exactly as sick as they are (or sometimes even sicker than that). The government agencies like MedPAC, health insurance lobby, and the architects of these systems go back and forth about whether these risk adjustment bonuses are paying way too much or not.
The reality is that there’s no “correct” way to incentivize taking sicker people, though maybe the magnitude of incentives could be adjusted. I think it’s also a question of where the extra dollars you get for taking sick people go - if it goes mostly to supplemental benefits or less premiums is that bad? What about more $ to marketing/ad spend? What about just giving me some money, please? Some guardrails on the spend might be helpful.
Already sick vs. “rising-risk”
A core issue of most value-based care models is that they’re designed to treat patients that are already sick instead of at-risk for becoming sick. This is partially because it’s easier to pay for a reduction in something vs. determining whether your intervention stopped a disease from happening at all. It’s easier to show that you prevented someone from ending up in the ED from a heart disease- related complication vs. developing heart disease in the first place.
It’s also partially because for historically undertreated areas, the comparison baseline is going to start really low (e.g. mental and behavioral health issues). This will mean that patients who look “healthy” but are actually undertreated turn out to be much more expensive than you’ll get reimbursed for via traditional risk-adjustment.
But beyond anything it’s because no one is actually incentivized to take long-term healthcare outcomes into account, a running theme of this newsletter. Instead every insurance carrier assumes you’re going to churn and eventually the result is Medicare absorbing the costs when you get old.
The inputs vs. the outcomes
I think there’s a reasonable debate about whether we should be measuring the outcomes of patients' health or whether we should be measuring whether providers did things we associate with “quality” care. Sometimes even if you do all the right things, patients will still get sick. Should companies get penalized for that?
Most contracts will have incentives that include both inputs and outcomes, but have different weights between the two and the associated tradeoffs. If you put more weight on the inputs of what a provider does there’s less incentive to innovate on the care model to see if there’s a better way to do things for cheaper and still achieve good outcomes. But if you focus mostly on patient outcomes, you might end up penalizing companies for things out of their control.
The amount of things you need to report to take part in value-based care programs sucks. In a lot of cases you need data from other parts of the patient record that you don’t have access to or you need to invest very seriously in data cleaning and keeping up to date with rule changes in reporting requirements. In fact if you look at surveys, the biggest barriers to taking part in value-based care programs tend to be infrastructure + capturing data. This study estimated a cost of $12,811 per physician to participate in MIPS, a Medicare focused value-based care program. Both of these issues disproportionately end up hurting small practices the most and end up increasing administrative requirements just to take part.
Who’s allowed to participate?
Who exactly should be the target for value-based care arrangements? Should it be bloated health systems that we know could run more efficiently? Should it be small independent practices? Or should it be new companies that could build natively to the value-based care arrangement?
I think it’s hard to change your workflows from fee-for-service into value-based care later. They’re just completely different from the tools you use to the clinical workflows you build. There’s a reason why Kaiser Permanente, which was built as a payer-provider from the beginning to take on patient risk, looks very different from large fee-for-service health systems which try to tack on their own insurance plans. In my opinion, providers have to build a product natively to a value-based care arrangement to be successful in one.
The reality though is that it’s difficult for smaller and newer players to enter the value-based care arrangements. They might not have historical data for benchmarks, they don’t have the resources to clean the data or wait to get paid, they don’t have enough patient lives to be considered, and they can’t coordinate care as easily once it leaves their domain.
This has meant that most value-based care arrangements have targeted large health systems. However, this presents a separate problem: there are fewer and fewer standalone health systems to test out value-based care arrangements. Different value-based care arrangements start crashing into each other at the same hospital and create confounding variables, or these fewer health systems dedicate less resources to making the programs successful since they do pretty well in regular fee-for-service land.
What is “value”?
There’s actually a larger question here which is *rips gravity bong*, like, what is value? If you were to ask an insurance company vs. a physician what “high value care” is they would answer completely differently. Sure, there are some areas of consensus (e.g. commodity labs and ancillary services should probably be done at the lowest cost place) but for the most part there isn’t an agreed upon definition of value. When I ask you what good quality care means what comes to mind?
There’s a separate post to be written about the success/failures of different attempted value-based care programs, but for now I wanted to outline the components and potential issues at a higher level. If you’ve been a part of a value-based reimbursement model and wanna chat about your experiences in it, hit me up! Would love to learn more from you.
Nikhil aka. “Putting the based in value-based care”
Thanks to Sandy Varatharajah and Laura Carlson for reading drafts of this piece
If you’re enjoying the newsletter, do me a solid and shoot this over to a friend or healthcare slack channel and tell them to sign up. The line between unemployment and founder of a startup is traction and whether your parents believe you have a job.