The Relational Reimbursement Paradox

Or, how Medicaid makes the mental health treatments it pays for worse at helping people...

One of the places alternative psychiatric crisis projects in the US reliably seem to get stuck is in the transition when growing from being funded by grants to being funded by Medicaid reimbursement. I investigated a bit and I have a theory as to why, which I’m dubbing the ‘relational reimbursement paradox’. 

In order to get paid by Medicaid for treating someone who has a mental health problem, you have to tell them what you did. In order for them to price these things, they have created a list of standard ‘services’ they pay for. An alternative to this might be for them to evaluate every bit of work any clinician does, which, as you can imagine, would take a lot of time and effort–possibly as much time and effort as providing the initial service! So they have a standardized list, and the service you bill them for must be one of the things already on the list.

Each item on the list has a description–a specification that defines how a service should be delivered to be considered that particular service. There’s a definition for Cognitive Behavioral Therapy; a definition for case management (like project management but for your social services needs); a definition for ‘staying at a crisis residential service’. There has been a push to make these more specific over time, in part to make Medicaid reimbursement more ‘evidence-based’--that is, if a specific treatment has gone through rigorous scientific research, as it probably has if it got added to the list of ‘evidence-based’ services, then you want to make sure that the thing you are paying for is as similar as possible to the thing that was tested in the research, because that is what was shown to work. 

Any given Medicaid reimbursement fee is probably going to have to pay for a lot more than just the hourly wage of the person doing the service–it probably needs to partially pay for the building the service takes place in, any training the clinician needs, the wages of the person submitting the Medicaid reimbursement, the wages of the administrator who hires all these people and schedules their shifts, etc. There’s generally someone in an organization (who might be the Executive Director if it’s a small nonprofit, but there might be a specific financial analyst in a bigger organization) whose job it is to make sure the budgets line up and that enough revenue is coming in from Medicaid and other sources to pay for all the expenses that are involved in running the program. This is the task of running a business, essentially, even if the organization is a nonprofit. 

In order to sell Medicaid your services you have to prove you’re providing basically one of the services on their list. If your service is a bit unusual, you may find you have to use specific language to persuade them that it is in fact one of the items on their list. The more adaptive, specific, and improvisational your treatment process, the more difficult it is likely to be to describe your practice in language that shows that it is one of items on their list.

Because how successfully an organization can bill Medicaid is a life-or-death financial matter for the organization, the on-the-ground practitioners will likely find themselves under pressure to standardize their treatments to make the process of securing Medicaid reimbursement easier. If the organization has a particularly strong mission that isn’t just ‘maximize the Medicaid reimbursements we get’, the practitioners may end up trying hacky ways to deliver their envisioned service but translate it into Medicaid-ese when they get to documenting. If the organization isn’t quite so strong-willed, practitioners will likely find themselves under pressure to do the service in the way that is most likely to be straightforwardly reimbursed; possibly the service with the best financial return on investment for the organization. How much practitioners are allowed to be driven by mission versus optimizing revenue varies from organization to organization. 

So, if I want to improve the quality of care I give to patients, can’t I just petition Medicaid to add my pet service to their list of services?

Not so fast buddy. 

Firstly, the list of Medicaid-billable services is difficult to add to. This is a multi-year, political process that I will likely detail in a different post. But, just so we’re clear, it’s pretty hard to get something new added at all.

But say you want to do that regardless. You will run into two problems–the social research scaling problem, and the standardization problem, and both of these I would consider sub-problems of the meta-problem I’d call ‘Medicaid can’t understand love/compassion/care/motivation/trust, and it doesn’t trust you’.

What’s the social research scaling problem?

Well firstly, in order to get your service approved, you will need to make sure it is ‘evidence-based’. This just means that enough scientific research has been done on it, research that showed it was effective, for people who trust the field of research psychology (which apparently includes the people at the federal Medicaid bureau but excludes basically everyone else who is familiar with the field of psychological research) to trust it. So you need to get research done on your therapy before you can try to get it approved as a Medicaid-billable service. 

Imagine this example: 

Dr. X, world-famous therapist, invents a new kind of therapy, Xism. Over a decade or two in some small Scandinavian nation she gathers a small group of dedicated true-believer therapists who learn Xism from her. The original crew are dedicated, curious, and fascinated by this work–they spend all their free time thinking about it and learning it and trying it on their friends. They do a small study, and it gets amazing results–everyone recovers from bipolar and schizophrenia! No one needs medications any more! It’s hailed as world-changing, and so they do more studies. There’s a big study funded by the NIMH, which takes years to organize and lots of time by Dr. X and her crew for writing grant applications and organizing experiments. They have to train more therapists and some of those therapists aren’t as smart or as curious as the original crew. The experimental design gives them only a few weeks to train the new therapists so they have to develop a quick standard training process that leaves a lot of nuance out. Plus there’s a lot of time pressure and in order to study it ‘properly’ they have to give everyone the same treatment–even if, as Dr. X and her friends complain, it makes the treatment worse. 

The NIMH study comes out and it’s ok, not amazing. By this point Dr. X is sort of burned out and pretty old, and debates are raging in the Xism community about what does and doesn’t constitute Xism. Some say it’s only true Xism if you do it without wearing pants, or if the therapist and the client meet in a random location. Some people break off and start a school of post-Xism. 

Meanwhile, Xism has begun to be taught in therapy schools. Now it is being learned by burnt-out therapists-in-training with ok-to-dodgy metacognitive skills who don’t actually care about Xism; they just care about getting their degree. They get employed at old-school social services agencies providing once a week Xist therapy for 50 minutes per session to 8 clients a day. Follow-up studies on Xist therapy ‘in the wild’ show it’s no better than a placebo. Industry publication coverage decries ‘the end of Xism’ and features flattering coverage of a new therapy, Yism. And thus the cycle starts again. 

What’s going on here?

There was an attempt to take some improvisational, nebulous, relational process (a therapeutic practice) and standardize it so it could be studied. But the things that were standardized were almost certainly not the key influential factors. Things that don’t get included in such a standardized practice:

• The therapist is highly motivated, likes what they are doing, and is present in the moment, paying deep attention

• The therapist is smart, curious, and cares a lot about their patients

• The therapist responds in a creative, adaptive way to the client, in a way that cannot be specified in advance as if completed by a computer program

• The therapist and client have things in common, are participating freely, and consider each other worthy of respect

• The therapist and client build a relationship over time, building trust and care for each other as specific individuals and not merely as their roles of ‘therapist’ and ‘patient’

Most or all of these factors are present in the initial situation: ‘fancy therapy scene developing a new niche kind of therapy’ and almost none are present in the final situation: ‘therapist stuck in badly-paying job where they feel overworked and their clients are extremely resource challenged, maybe homeless, and are forced to see the therapist for legal reasons’. 

If we could study this set of five characteristics, we might find that they comprise the vast majority of the healing impact of any new psychological therapy. I’m not sure, but I’d bet on it. 

So, back to your mythical mental health organization that wants to bill Medicaid for its new service. You can’t bill Medicaid for ‘our therapists are happy, motivated, well-paid, respond to clients in the moment without a script, and are in a position to build trusting relationships with their clients over time’. There’s a category error–this simply isn’t a ‘service’ and Medicaid only pays for ‘services’. 

So, you have to describe the thing you want to provide as a service, in a way that is as replicable as possible. And this brings us to the standardization problem. 

Almost every change you can make to a ‘practice’ to make it more standardized will take it further away from making therapists happy and motivated to do it, and letting them improvise in the moment. In therapy circles, this process is called ‘manualization’ and many gifted therapists hate it. Their work becomes rote and prescribed. 

Furthermore, to make it attractive to Medicaid to fund over other stuff, you want to make it cheap, which means you want to bill for as little time per patient as you can get away with while still making sure your research shows positive results. Which, predictably, means therapists have less time to build a trusting relationship with each new patient.

So, even if you succeed in getting your pet therapy approved as a service by Medicaid, you will have to do it by a) stripping out the improvisational nature that makes therapists happy, motivated, and better able to respond appropriately to each specific client, and b) shortening it in a way that makes building a trusting relationship over time difficult to impossible. You also ignore the uncomfortable fact that different therapists are just better or worse at helping people improve, for reasons that may have more to do with personality or attitude than any specific trainable skill–you have to bill for excellent therapists and mediocre therapists at basically the same rate because according to Medicaid they’re providing the same ‘service’. (And then your best therapists get frustrated with this and leave to open their own private practice, where they can charge more.) 

(There’s a third, related problem with organizing your service around billing for Medicaid, which is that the open-minded, empathetic, curious souls you hired because of their ability to be with people in extremely difficult situations are probably precisely the people least interested in or able to fill out extensive detail-oriented paperwork on a regular basis. This was certainly a big problem for Soteria Alaska, and seems to be at least somewhat the case for Felton’s early psychosis program.)

I don’t think this paradox is unique to American Medicaid; I suspect it is present to some degree in all healthcare systems where the payer and the patient are different people, the payer has to allocate limited money to fund treatment for a bunch of different individuals and the payer aims to be ‘evidence-based’.

As far as I can tell, there are two initial open problems to make progress on in order to resolve the relational reimbursement paradox:

• Figuring out a way of creating scientifically legitimate knowledge around the quality of therapeutic relationships and therapeutic contexts, in a way that bureaucrats who want everything they fund to be ‘evidence-based’ will trust. It should also be a kind of knowledge-production process that is at least somewhat resistant to being gamed. I doubt this will involve standard randomized controlled trials, although I don’t know enough about research design to suggest alternatives. I would love to talk to anyone who does, though!

• Figuring out a practical method for accounting for and paying for improvisational, nebulous, relational activity in a way that doesn’t constrain the relationship unnecessarily but is legible enough and unhackable enough for Medicaid bureaucrats to trust. There needs to be an alternative to standardization in order to make standardization obsolete. 

The glaring alternative to this is to develop an alternative payer system to Medicaid, but any payer system will face similar constraints in deciding what to pay for and figuring out what to trust, so my hunch is these problems would re-emerge in any new system unless they are addressed directly. 

As will always be the case with these updates, I would love to know if I’m wrong or I’m missing some perspective, particularly if you have experience working in the systems I’m wrong about!