Zero-Toll Medicine

How crypto and AI can fix our broken health insurance system

The U.S. health insurance system is stupid, immoral, and infuriating. It is time to get rid of it altogether and replace it with an intelligent, modern, and efficient infrastructure befitting the American people and the 21st century. Incrementalism is not the answer, and solutions that amount to adding further fragmentation and complexity (including Medicare Advantage, ACOs, delegated managed care, and integrated managed care) ultimately obscure the need to burn it down and start afresh. I propose in this essay a novel model, “Zero-Toll Medicine,” which leverages advances in AI and blockchain technology to deliver health insurance free of toll-taking intermediaries like private insurance companies and PBMs. This scheme aims to put patients in control of their own healthcare spending, eliminate vast amounts of administrative cost and complexity, and drive far more efficient market-based competitive pricing. It is rooted in libertarian and socialist idealism alike, exploiting the virtues of decentralized decision making and markets while realizing the civilized goal of healthcare as a fundamental right of American citizenship.

Why Do We Have Health Insurance?

Insurance exists primarily to pool risk. We do not know at birth who will require substantial medical care over his or her lifetime, but we can be certain that many of us will. And as the philosopher John Rawls argued, we should design our society such that we would be ok being born into any particular lot in life, since none of us chooses his or her circumstances at birth. Pooling our resources to guarantee basic medical care to everyone no matter where we’re born or how our lives unfold should be uncontroversial given our enormous societal wealth.

Limits on this commitment must arise from the fact that medical care can be very costly, and so we are unlikely to agree as a nation to cover unlimited care for every individual without regard to how much longevity or happiness it might buy. Therefore, rationing or “managing” the amount of care is inevitable. Furthermore, we cannot be held hostage to medical care at any price, and so should desire that medicine exist within the market economy, subject to competitive forces that optimize supply via price signals.

These three goals: 1) risk pooling 2) rationing care and 3) competitive pricing, are in theory the functions of our existing insurance system. But it fails at all three.

1. It fails to pool risk efficiently because, unlike in other insurance markets, we (rightly) prohibit insurers from discriminating on the basis of pre-existing conditions, or pricing policies on the basis of individual risk (apart from generic characteristics of a patient like geographic location and age). In principle, if carriers had the ability to discriminate, they could generate profits from careful underwriting and risk prediction. But in our system, they effectively must cover any patient who subscribes. Furthermore, if we seek to maximally spread out costs, we should have the largest risk pool possible, one that includes everyone. Our current system, by contrast, splits the population into thousands of smaller risk pools, among them: employees of individual companies (self-insured employer plans), poor people in each state (Medicaid plans), retirees (Medicare), and so forth.
2. It fails to ration care ethically because this rationing is largely done at the insurance plan level. Pre-authorizations, claim denials, and benefit design are managed by plans, which amounts to their “playing doctor.” This is not a job that insurers are qualified to do, and they do not do it well. Moreover, evidence-based medicine demands that a “standard of care” dictate what care is appropriate and medically necessary, and this standard should be universal vs. varying arbitrarily across different insurers. The care that is appropriate for a patient has nothing to do with who insures them.
3. It fails to drive optimal competitive market pricing and supply because it aggregates both supply and demand into oligopolistic blocs and removes decision making power from the actual providers and users of care. On the demand side, each insurer effectively becomes a representative of tens of thousands of patients (and their employers, who hire plans and bear costs). On the supply side, large health systems and private-equity-financed provider roll ups are increasingly the counterparties in price negotiations. In other words, large insurance companies negotiate against large providers to set prices. Unfortunately, this means relatively little true competition exists on either side. In many markets, there are only one or two dominant health systems, and they therefore have tremendous leverage when negotiating with insurers. If a plan wants to operate in that geographic market, it has little choice but to ultimately contract with these health systems and take the prices on offer. In addition to the limited choice that this model affords market participants, it also leads to opaque “omnibus” pricing schedules highly abstracted from the actual costs of underlying services. Specifically, an insurer will contract with a health system for the full range of its services - surgeries, inpatient stays, hospital-administered drugs, etc. The insurer will be optimizing to minimize its overall expected costs, and the provider will be optimizing to maximize its overall expected profits. This means that different payers negotiating with the same provider may end up with very different prices for the same services. In other words, the same product (say, a joint replacement) will have a different price depending on the patient’s insurance company. This excessive complexity begets excessive administrative operations. Every time a patient comes in for a service, providers have to figure out who is insuring the patient and what the negotiated price will be for that insurer. Then the provider will need to bill that payer for the service and manage collections. In several states, including California, if you go to a hospital, you will discover that the anaesthesiologists, physicians, and hospital itself are all different commercial entities, each with its own contracts with your insurance company, tripling the administrative billing apparatus required to get paid for your care. If this seems like it doesn’t make sense, that is because it doesn’t.

Subsidizing Subsidies

What I have described so far are failures of our employer-based private insurance system, which covers just over half of Americans. The other side of our system is public insurance, which covers 36% of the population in total. Among those publicly insured are 93% of those 65 and older (Medicare), 36% of children (Medicaid), and 16% of working-age adults (Medicaid). The public system is itself a patchwork of government and private entities. For example, Medicaid is run by the states, but outsources substantially to private insurance companies in what are known as MCO arrangements (managed care organizations). Similarly, Medicare is run by the federal government, but a majority of beneficiaries now choose Medicare Advantage plans, which are also outsourced plans run by private insurers. Even Traditional Medicare (the original plan that is managed by CMS and not outsourced) outsources its claims processing and assorted functions to 12 “MAC”s (Medicare Administrative Contractors), privately-owned and somewhat mysterious vendors that also engage in care rationing across Medicare patients (for advanced diagnostics, as one example).

While there are imaginable efficiencies derived from these outsourcing arrangements, you can be sure that there is also a lot of graft and toll taking. If the government lacks discipline in spending taxpayer money, private companies one further layer removed are often even less scrupulous. This is particularly true if there is little competition for these contracts. To take one well-publicized example, Medicare Advantage plans have pressured physicians they control to overdiagnose patients with chronic diseases in order to receive larger payments without delivering correspondingly valuable incremental care. The entire premise of Medicare Advantage is in fact questionable insofar as the government pays plans a premium beyond typical Medicare costs to manage each patient. The last government MedPac estimate is that we are spending $83 billion more than we would in Traditional Medicare to cover Medicare Advantage members.

In spite of this waste, Medicare and Medicaid still pay artificially low prices to care providers. And the majority of hospitals, including all non-profit hospitals, must accept these prices in order to qualify for their tax exemptions. The result is that many providers serve public-insurance patients at a financial loss, and therefore need to make up for it with private insurance patients. In other words, private-insured patients provide an enormous subsidy for public-insured patients. In fact, private payer prices are oftentimes negotiated explicitly as multiples of Medicare rates. This also drives a perverse incentive for hospitals to manage their “payer mix” in favor of patients with private insurance. While many hospitals take seriously their obligation to the common good, it is simply a fact that their financial health is imperiled if they take care of too many poor people.

The subsidies run in the other direction, as well. Since employer-based insurance is a tax deductible expense, all taxpayers are subsidizing private insurance. And as economist Uwe Reinhardt pointed out, since employer-based insurance is a form of compensation, the deductibility of these premiums amounts to a regressive tax policy because the tax avoided by high wage earners (on “insurance compensation”) would have been at a higher rate than the tax avoided by low wage earners.

In short, employers subsidize public-insured patients, taxpayers subsidize employer-insured patients, and we redistribute wealth from lower to higher income workers in the process. These convoluted cash flows makes it exceedingly difficult to calculate precisely who is getting screwed most. But the short answer is: all of us.

Radical Simplification

A smart and modern healthcare system would achieve the following goals:

1. Enable every American citizen to afford a satisfactory level of medical care
2. Ensure that medical care is evidence-based and transparent
3. Create robust competition among healthcare providers and pharmaceutical companies to win the business of patients, driving efficiency, patient experience, and innovation

Decentralized ledgers and smart contracts, technologies originated in the cryptocurrency industry, enable a new model of American healthcare capable of meeting these goals with a radical new open insurance and payments system, “Zero Toll Medicine” (ZTM). Here are the building blocks of this system:

1. A new, open blockchain protocol, ZTM, on which every US citizen and healthcare provider would have a wallet. This wallet would function as a bank account specifically for government-funded healthcare spending and receipts.
2. A minimal federal government health insurance agency, which could replace CMS. This “agency” would primarily comprise a computer codebase governing its interactions with the ZTM protocol and participants, but would inevitably require some number of staffers to manage activities in which human judgment is unavoidable.
3. A ZTM stablecoin pegged to the US Dollar.
4. A library of composable smart contracts enabling programmable payments in the stablecoin.
5. A computer-readable standard of care.

The aim of this system would be to directly fund patient wallets when they require basic medical care, and to enable them to shop and directly pay providers and product manufacturers for this care with no insurance companies, PBMs, or other intermediaries. The system would also allow patients to further fund their wallets with their own money (or supplementary insurance payouts) in order to spend it on services and products beyond basic care.

Basic Care

The term “basic care” is of course ambiguous and loaded. There is no alternative but for the country to determine what level of care it deems worth covering for every person, and this is a political and ethical decision, not a technical matter. “Basic care,” then, amounts to whatever care can be paid for with the country’s total public health insurance budget divided by the number of its citizens, with the caveat that spending would end up unevenly allocated according to individual patient needs (as with payouts for all insurance).

In 2023, our country as a whole spent an estimated $4.9 trillion on healthcare. And at that year’s level of taxation, the federal government took in about $4.5 trillion of total revenues, of which it spent roughly $2 trillion on Medicare and Medicaid. In other words, the government currently covers about 40% of our total healthcare budget and employers/employees cover about 60%. That latter 60% is misleading since, again, taxpayers subsidize healthcare costs through the tax deductibility of employer-sponsored premiums and subsidies for exchange plans (Obamacare), to the tune of $300bn per year. In other words, we currently socialize about 50% of total healthcare spending while employees bear the remaining 50%. I say that employees bear that share because the premiums their employers pay on their behalf would otherwise be paid to them as wages.

If we were to eliminate employer sponsored insurance, then, 1) wages would go up, 2) government revenue would go up since these incremental wages would be taxed, and 3) employees would be left without insurance, but with more income available to pay for their healthcare needs out-of-pocket. Assuming healthcare spending remained constant, and all incremental tax revenue went to public healthcare spending, the 50/50 split between socialized and privatized healthcare cost-bearing would remain the same.

With the government covering 50% of costs, but expanding coverage to all Americans (not just those currently covered by Medicare and Medicaid), we could only afford to cover a level of care substantially lower than that supported by Medicare today. But as Amy Finkelstein and Liran Einav point out in their excellent book, Medicare today is “Cadillac” health insurance, and represents a dubious choice society is presently making to give the elderly sterling coverage instead of providing everyone with basic coverage. The latter should be our goal, and the level of this basic coverage can be increased should we wish to socialize more than 50% of health spending. I personally would advocate something more like 70-80% of cost coverage for everyone (members of congress enjoy 72%), but the redistributive implications of such a choice would depend on the tax strategy funding it, and are difficult to project. Ultimately, society a a whole pays for 100% of heathcare costs, so the share borne by the public sector is simply a matter of how much we choose to insure each other vs roll the dice on our individual future costs.

As Finkelstein and Einav also argue, basic care should begin with foundational primary care for every citizen. Today, we spend a mere 5% of our healthcare budget on primary care, and this is one area in which higher spending is likely warranted and could bring down overall costs. The paradigm of preventive medicine requires investing in the health of patients before they are sick or injured, and like education, pays off over long periods of time and in diffuse, difficult-to-calculate ways. In today’s private-insurance model, the frequent movement of patients between insurers (typically when they change jobs) disincentivizes investment in primary and preventive care since its payoffs are assumed to be too far in the future to generate an ROI. Moving to a public insurance model eliminates this irrational short-termism.

Zero-Toll Medicine would engage each citizen as a stakeholder in the healthcare costs of the entire society, and could provide direct incentives for individuals to proactively invest in their health. For example, each year, every citizen could receive a payment to cover annual primary care costs. When the individual spent this on a wellness exam, labs, etc., he or she could receive an automated tax credit of an amount sized to drive maximum compliance. Various other “gamification” mechanisms could be considered to drive this sort of virtuous behavior at scale.

Bundled Stablecoin Payments

Beyond primary care, patients would receive so-called “bundled payments” sized to cover the projected reasonable costs of basic care for most conditions or needs. In a non-acute context, patients would receive a diagnosis from their primary care physician, and this diagnosis, entered into the ZTM protocol, would instantly trigger a bundled payment to their wallet corresponding to the “episode of care” associated with it. For those unfamiliar, an episode of care is a concept utilized in today’s so-called “value-based” healthcare models, and describes a set of medical services and products required to treat a condition over a pre-determined period of time. For example, if a patient needs a joint replacement, an episode of care covering this could encompass pre-surgery appointments, the surgery itself, the medical devices used in the surgery, post-surgical doctors appointments, and physical therapy services for a couple of months. A “prospective bundled payment” would be a payment made prior to all of this care that is sufficient to pay for it.

Bundled payments are today utilized in a variety of insurance models with the goal of aligning incentives between those who pay for healthcare and those who deliver it. The majority of insurance arrangements today remain “fee for service,” wherein providers bill insurance for each individual service and product they deliver. Many have concluded that this incentivizes over-delivery of care by rewarding providers when they rack up as many charges as they can, and bundled payments arose as a means of setting a fixed budget for an episode of care. These models typically allow providers to profit if they are able to treat patients for less than the budget, generally by letting providers pocket the difference between their costs of care and the bundle amount. This can, unsurprisingly, create the opposite incentive, to penny-pinch on care and risk under-treatment. Compensatory payment structures are meant to overcome this by, for example, tying final “bonus” payments to measurable patient outcomes, so that docs maximize their profits only when they have rendered verifiably high quality care.

In today’s value-based care models, this kind of incentive structure tends to flow downward from the ultimate source of insurance funding. Medicare Advantage plans, as the most prominent example, receive per-patient fixed annual payments from the government to manage the care of individuals who enroll in them. This amounts to their taking on the “risk” of those patients’ annual costs, since if a patient costs more to care for, the insurer will need to cover it, but will do so at a loss. They may then enter into agreements with physician groups or health systems to delegate this risk.

The aspiration of any of these models is that, ultimately, someone - a physician, a group of physicians, or an insurance company - will have a financial motive to be cost-conscious while taking good care of patients. Zero-Toll Medicine embodies the same logic, but puts the power of value-based medicine in the hands of patients themselves!

Power To The Patients

Concurrent with receiving a bundled payment, the patient and their physician would also receive an automated report summarizing the diagnosis and the standard of care for treating it. This report on the standard of care would clearly lay out for the patient the evidence supporting the recommended course of treatment and the projected itemized costs for obtaining this treatment, which would add up to the total bundle amount. Based on the location of the patient, the report would also generate a list of providers in the area, along with the prices they charge for the recommended services and the outcomes their patients have achieved in the past. Based on this outcomes data, the patient could also receive a prediction of the range of outcomes they might expect, and the probabilities of these outcomes. Importantly, this report would also include information on complication rates and malpractice incidents associated with providers, empowering patients with visibility into the risk profile of the providers they consider.

The same framework would apply to pharmaceutical products. There would be no insurance plan drug formulary, no PBM, no rebates, and no convoluted discount networks. The patient would be free to choose among the drugs approved by the FDA and indicated by the standard of care for their condition, and could purchase these directly from a pharmacy or pharmaceutical manufacturer.

This would put patients in a position of power to decide where they want to spend their healthcare dollars, and would dramatically invert today’s hierarchy, forcing physicians and pharma companies to appeal directly to patients to win their business.

Furthermore, this model would lead to “one price for one product,” since each provider or product manufacturer would be competing in an open market for the business of millions of individual patients. Real price discovery would occur as patients determined where they could get the best value for their money, and attempts at predatory pricing would simply result in low sales.

Open Information

The model I’ve described can only work if information that is currently secret and siloed becomes public and open. This includes the prices that providers charge, which they have in the past brazenly refused to publish even when the federal government mandated it. More controversial than prices are data on the number of procedures doctors have done, complication rates, malpractice, and patient outcomes.

But Zero-Toll Medicine can drive this information sharing simply by requiring it of any network participant. For a hospital, physician, device company, diagnostic lab, or pharma company to receive payments through the ZTM system, an ongoing real-time “credentialing” process could be mandatory. Specifically, to be approved on the network, a medical provider would need to verify its qualifications to deliver specific types of care, and then would need to feed into the network real-time data regarding patient outcomes to remain approved. This credentialing could likely be almost entirely automated.

Infinitely Programmable Healthcare

The most innovative aspect of Zero-Toll Medicine is programmability, which is uniquely possible with a crypto protocol. By programmability, I mean that smart contracts can be composed to endow payments with limitless custom logic. Some examples:

• If a patient receives a bundled payment for a particular episode, those stablecoins could be made spendable only for a set duration of time, and on providers specifically credentialed for the associated services. This would prevent, for example, fraud in which a patient uses a payment intended for a surgery and spends it on cosmetic Botox injections over several years.
• A provider like a hospital could take on an entire episode of care from the patient and commit to certain outcomes. The patient’s payment for this could be held in escrow until the patient or a 3rd party oracle on the network validated that the outcome was achieved.
• Similarly, a pharmaceutical company could offer a drug on a pay-for-performance basis, giving patients the choice between a lower up-front price and a higher performance-based price which they only pay if particular outcomes are achieved.

It is worth mentioning that the volume of transactions and speed required to support ZTM may be infeasible with today’s incumbent “L1” blockchains. There are ongoing efforts, including Improbable’s Somnia, to create “gigachains” with sub-second finality and low enough gas costs to unlock this kind of highly scaled use-case.

Private Funding

These ideas merely scratch the surface of automated logic that could be deployed over the ZTM network. Practically, there would be two types of funds flowing through it: government insurance-funded stablecoins and patient-funded stablecoins. The latter would arise to the extent that patients desired to spend more on their healthcare than the government’s basic care budget would allow. Say a patient required an inpatient surgery and the government bundle for this was only sufficient to cover a shared hospital room, the patient could put additional funds into her wallet to pay for a private room. This would result in a wallet balance comprising some amount of “public stablecoins” and some amount of “private stablecoins.”

Many benefits of this network, and most obviously comprehensive visibility into healthcare spending, result from having all healthcare spending take place on the network. And so a major question would be how to incentivize all private spending to be done on network vs off network. One possible solution would be to make private spending tax advantaged. In the surgery example I used, a patient could fund private stablecoins into their wallet, and upon spending them on the surgery, would receive a tax credit or deduction. Again, programmable logic could determine what healthcare spending would benefit from this treatment: perhaps a private hospital room purchase would not but paying for a higher priced surgeon with better expected outcomes would.

There is precedent for this sort of tax-advantaged healthcare spending in today’s HSAs. But rather than deduct contributions to an account (saving), ZTM would only provide a tax benefit upon spending on qualified products and services.

From the standpoint of providers and manufacturers, payments received through the ZTM network would seamlessly integrate public and private funded stablecoins. And recipients could cash-out into USD at any time.

Finally, supplementary insurance could easily coexist with this model. Patients desiring additional coverage beyond the basic care government coverage could purchase plans independently. But these plans would pay out benefits directly to patients or into their ZTM wallets. The goal again would be to ultimately push all spending into the network, irrespective of its funding source.

Summary

Zero-Toll Medicine proposes eliminating America’s existing public and private insurance systems and replacing them with a programmable digital infrastructure for public insurance to cover all citizens, yet compatible with private spending, as well. This infrastructure would enable Americans to freely shop for their medical care with no intermediaries constraining the choices they could make.

By putting patients in control of all healthcare spending, ZTM would lead to maximally competitive market price setting and replace the oligopolistic and opaque pricing schemas that hamstring medicine’s affordability today. By allowing patients to decide what they value, prices would converge with patient demand, not the demands of today’s corporate intermediaries like insurers and PBMs who imperfectly represent patient interests due to their own financial incentives.

ZTM’s programmability would create a dynamic infrastructure for democratically-determined priorities and regulations constraining publicly-funded healthcare spending, and allow instantaneous enactment of these rules. It would create total transparency around usage of healthcare products and services, the quality and cost-efficiency of this care, and its safety.

While such an ambitious revolution in our system is difficult to imagine, business as usual will bankrupt our country and continue eroding the public’s confidence in government and medicine. The sacred cow of healthcare reform efforts has unfortunately been our for-profit insurance industry, which extracts rather than adds value. It is high time to dissolve it for good and reclaim America’s leadership among advanced nations in the innovative provision of public health

Postscript: Q&A w Peter Kolchinsky

My friend Peter Kolchinsky , founder of RA Capital, raised helpful questions in response to this essay. I share them below with my responses.

• PK: What if the algorithm doesn't allocate enough credits to allow someone to afford the nuanced approach that escaped codification in the algorithm?

DAW: The approach I’ve proposed would leverage existing ontologies for estimating the services and costs appropriate for managing a diagnosis. For example,

• PK: What if there are complications?

DAW: To receive payments on the ZTM protocol, physicians would need to report outcomes into the network, which would make them fully transparent. Patients, regulators, and others could directly view any provider’s complete history of services provided and corresponding outcomes. This would make it straightforward to generate comparative performance analytics on providers and products in real-time. Critically, complications and patient harm would become matters of public record. There could also be Amazon-like patient ratings for providers or individual services rendered.

• PK: What if physicians judgement results in upcoding and issuance of additional credits that enrich the doctor?

DAW: As I mentioned in the essay, this behavior is currently a scourge on the Medicare Advantage market. While dishonesty will always be impossible to eliminate wholesale, ZTM would put physician diagnosis patterns into the open. Whereas today, this sort of fraud has been visible only to individual insurance plans or CMS, with ZTM it would all be publicly visible. De-identified patient characteristics (demographics, etc) could be viewable for a provider’s patient population and the provider’s diagnostic and prescribing behavior could be compared to that of other providers. Bots on the network could even be programmed by independent citizens to constantly look for fraud on the network, and win bounties when they found it. I have separately proposed the creation of a new category of labor that I call the “medical accountant,” which would in my “

• PK: What use is the credit to a patient if they don't spend it? Can they pocket the money?

DAW: The goal would be for patients to spend as little of their bundled payments as they can to achieve the targeted outcomes of the care the bundle covers. To incentivize spending as little as possible, any bundled payments that are not spent should benefit the patient somehow. I have proposed as one possibility that this could result in tax credits, but other incentives could be considered. As with capitated payments in today’s value-based models, this runs the risk of patients not taking care of themselves in order to get financial benefits. These incentives therefore would need to be contingent on patients actually purchasing the care they had been recommended, and achieving solid, measurable outcomes. This is a tough needle to thread, but all things considered, I believe that putting patients in charge of these decisions will work better than putting intermediaries in charge of them. Patients have the most at stake when it comes to their own health, so while they may behave irrationality, so too do physicians, hospitals, and insurance companies, to whom we entrust these sorts of choices today.

• PK: What if patients don't spend credits on prevention and get sicker and have to be issued more credits? Will we regret letting the patient make a poor choice that sticks us all with the cost? Maybe better to have just said "You should get this preventative treatment and we'll pay for it.... if you don't get it, then you lose out, you don't get to keep any credits, and we'll all be worse for it... in fact, can we pay you to go through with the preventative treatment?"

DAW: As with the previous question, tuning incentives to drive the behavior that is best for patients and society at large will be a constant challenge and opportunity. ZTM provides a framework for programmatically optimizing and evolving these incentives over time.