How the data revolution makes universal health insurance inevitable

The limit of U.S. healthcare policy as the data revolution progresses is single-payer, government-provided, universal health insurance.

Meaning? Health care delivered by assorted private, public, and not-for-profit medical facilities — just like it is today. Health insurance financed by the federal government and given to everyone. a.k.a. “Medicare for all”. a.k.a. The health insurance system that U.S. Congresspeople enjoy today.

Why? Because predictive analytics increasingly empowers private payers to identify high-risk customers…even before they become high-cost patients. And, because a complex hybrid public-private health insurance system leaves private payers with opportunity to selectively excise those future-high-cost patients from their customer rolls.

Data predicts medical problems

Medical data has existed as long as societies have had professional medical providers. Predictive analytics has been around since the 17th-century invention of inferential statistics and the 20th-century invention of regression analysis. Predictive analytics was later made less time-consuming by software (Excel, SPSS, SAS) that automates the statistical math — and does so in an interface that frees the user from needing to understand the underlying arithmetic.

What’s different now is that we have more voluminous medical data from more sources: data from wearable devices, digitized text chart notes, electronic medical histories, patient-reported information in apps, real-time vital sign monitoring data, lab results, genetics test results, and published medical research insights. We also have cloud-based IT systems to compile those disparate data types into a centrally-accessible and analyzable format. Lastly, we also have more advanced analytics software, which uses data mining, qualitative text analytics, and machine learning techniques, in addition to standard statistical modeling.

Big data- and technology-enabled algorithms deployed today “predict” (i.e., assess the risk of) sepsis, heart failure, stroke, timing to safely remove artificial ventilation, diabetes complications, need for ICU versus standard hospital admission, bleeding during surgery, need for elective infant delivery, and many other medical outcomes. Such “prescriptive analytics” enable higher-quality medical decision-making and early intervention to lower the chance of adverse outcomes.

In addition to data proliferation and analytics evolution, there are transformative medical innovations in the pipeline:

· Things that improve health outcomes, but make care more expensive: advanced prosthetics, 3D printed organs, robotic home care, VR-assisted surgery. Private insurers have an incentive to charge more for people likely to need such treatments. How actionable that incentive is depends on health insurance regulatory policy.

· Things that make care cheaper due to early detection, but identify unhealthy people in so doing: nanopills that detect early signs of cancer and heart disease from inside the body, wearables like contact lenses with streaming data to detect glucose levels or other leading indicators of disease, full-body multi-functional radiologic diagnostic scans. How will private insurers value the avoided costs from early detection compared to the non-zero cost of long-term treatment?

· Things that cure diseases that would otherwise become expensive, but at a high upfront cost: gene therapy, gene editing, “exercise in a pill”. How will private insurers balance known upfront costs against unknown future actual or avoided expenses?

Among all the growing sources of data, payers only have access to patient demographics and individual claims data (procedures, medications, and lab tests ordered), plus published research insights (which link demographics and treatments with outcomes). In contrast, providers have all of that, plus all data any provider generates about a patient (e.g., lab test results) and patient-reported data. The reach of payer predictive analytics is, thankfully, bounded by America’s strict medical privacy laws. However, payers are able to extract ever more predictive value from the increasing volume of data they do have. The power of inference increases as data accumulates.

Health insurance companies are rational

A fundamental premise of the Affordable Care Act (ACA) is non-discrimination based on medical condition. Yet, in late 2016, the outgoing Obama administration allowed a federal agency to issue a regulatory rule that effectively allows private insurers to excise its sickest patients.

Even a simple explanation of the background for this surprising action is necessarily lengthy, but it’s worth pausing to understand. The ACA champion’s anti-ACA rule affects dialysis patients, as follows:

America’s 700,000 patients with End Stage Renal Disease (ESRD) often can’t work, since they sit for hours-long life-saving dialysis sessions several times per week…forever (unless they get to the top of the 100,000-person-long national kidney transplant list). Still, about 60% don’t financially qualify for Medicaid (depending on individual state policy regarding ACA-related Medicaid expansion). All ESRD patients are eligible for Medicare (it’s the only diagnosis-linked eligibility group), but those under 65 aren’t eligible for the Medicare gap coverage (depending on state policy) that makes Medicare affordable.
Meanwhile, dialysis providers earn more from patients enrolled in private insurance plans than from those on Medicare/Medicaid. Therefore, “charitable premium assistance” from dialysis providers to ESRD patients is widespread, though somewhat masked to private insurers by flowing payments through not-for-profit intermediaries. For example, a provider pays ~$6000 in annual private insurance premiums on behalf of a patient, in exchange for ~$100,000 in incremental annual payouts from the private insurer: charity that yields a 1500% financial return on investment (which can be used to cross-subsidize the provider’s losses on Medicare patients).
Historically, 90% of ESRD patients use Medicare and 10% use private insurance. This is a one-time, very weighty decision that frightened patients must make upon initial diagnosis because one of the ACA’s consumer protections is a prohibition on selling private insurance to Medicare recipients. Given that the ACA has successfully made private insurance more affordable with capped annual out-of-pocket expenses, the privately-insured ESRD segment will grow organically (if the ACA indeed remains law). In addition, dialysis providers have a strong financial motivation — and thus a potential conflict of interest with Hippocratic obligations — to steer patients into private insurance.
One of the ACA’s notoriously-numerous loopholes is that it’s legal for private insurers to void coverage for patients receiving charitable premium assistance (unless they have HIV/AIDS) and to reject Medicare-eligible patients. Private insurers have only sporadically followed this practice to date, but are getting more aggressive as their own financial pressures mount. A bipartisan House bill to address the issue failed in 2015.
The Center for Medicare and Medicaid Services (CMS) has authority over healthcare provider facilities that treat Medicare/Medicaid patients (i.e., almost all of them). A December 2016 proposed CMS rule forces healthcare providers to inform private insurers if they’re making charitable premium payments — thereby facilitating private insurers’ expansion of the ACA-sanctioned practice of excising costly ESRD patients from their risk pools.
The ACA has made private insurance preferable to Medicare for many of the ~50% of ESRD patients who are both under 65 and also Medicaid-ineligible. However, Medicare is preferable for at least the other half. Why? ESRD patients on Medicare don’t risk: (a) coverage interruption due to their insurer voiding coverage upon discovering charitable premium assistance, (b) coverage interruption due to provider cessation of charitable assistance post-transplant (when patients still need a lifetime of costly immunosuppressant drugs), © de-prioritization on the national kidney transplant waiting list due to anticipated coverage interruption, or (d) high residual payments for the 65% of their healthcare spending that is unrelated to ESRD (if they’re under 65 and thus can’t get Medicare gap coverage).
Rather than making it easier for private insurers to excise high-cost, chronically ill patients, a more humane and financially efficient solution is:
1. Legislatively close the ACA loophole that legalizes denying private insurance to patients receiving charitable premium assistance and to Medicaid-eligible patients.
2. Legislatively make Medicare gap coverage eligibility congruent with Medicare eligibility
3. Eliminate the existing CMS rule that discourages private insurers from accepting charitable premium assistance.
4. Cancel the December 2016 proposed CMS rule that mandates charitable premium assistance disclosure. [Overturned by a federal court in January 2017]
5. Mitigate provider conflict of interest by mandating patient-specific analysis of the private vs Medicare/Medicaid insurance decision, and by reinforcing provider ethics in favor of patients’ interests over financial gain. [Completed by non-for-profit intermediaries, effective January 2017]
6. Improve the Medicare enrollment process for privately-insured, post-transplant ESRD patients to prevent kidney transplant disqualification by eliminating risk of coverage interruption. (This could take the form of opt-out automatic enrollment, better provider-insurer coordination, and/or a transplant center-led program.)
7. Increase Medicare payouts for dialysis to cover actual provider costs, to reduce financial incentive to steer patients into private insurance.

Imagine that, instead of all this tediously-complex, economically-inefficient horse-trading of patients between Medicare and private insurance, we enact “Medicare for all”.

In the above ESRD example, Obama didn’t write the offending CMS rule; but, the executive branch oversees the CMS, and thus at least tacitly approved the proposed rule. So, the guy whose proudest presidential legacy is the ACA capitulated to weakening it. That’s evidence of how powerful the systemic incentives are to cherry-pick patients based on expected medical cost. Moreover, those incentives aren’t static — they will strengthen further as the cost of care delivery inexorably rises. (The cost of healthcare delivery is a serious, independent issue which influences the health insurance crisis and thus must be addressed in parallel.)

It’s folly to expect private companies to self-limit. The defining objective of a corporation is profit-maximization. To behave otherwise, companies must explicitly bake a social mission into their decision-making framework (e.g., by using a triple bottom line, lower discount rate, more comprehensive risk assessments). Water runs downhill. Such is the argument in favor of market regulation to preserve the public good. Someone got a Nobel Prize in Economics for explaining this (Jean Tirole, 2014): regulation is necessary for a well-functioning market, and effective regulation is necessarily complex and nuanced, adaptive over time, and customized to each industry.

The combination of private provision of health insurance with predictive data led to a classic market failure: Quality fell (insurance policies covering less stuff). Quantity fell (fewer people had health insurance). Someone got a Nobel for addressing that, too (Daniel Kahneman, 2002): regulation of insurance product content and regulatorily-automated insurance enrollment are necessary to address bounded rationality and bounded willpower of consumers. Hence, the ACA. And, hence, the improvements still needed to perfect the ACA.

Predictive data power + Private profit motive = Deepening healthcare crisis

Data is getting more powerful. Incentives are strong. Data will be used with ever-increasing effectiveness to follow those ever-strengthening incentives, i.e., to reduce coverage of the sickest people.

We’ve seen this play out over decades, with the American health insurance system becoming more and more unfair: pre-existing condition exclusions, reproductive and maternity care de-standardization, horror stories of untreated people dying of treatable conditions in the world’s richest country. The patient horror stories are a leading market indicator, akin to oil prices spiking in the run-up to reserves one day reaching depletion, new realtor license applications surging before an asset price bubble bursts, and the proverbial writing on the wall before Babylon fell.

This downward trend has fueled 20 years of attempted health insurance system reform, culminating in the ACA.

Band-aid solutions

1. 1993: First Lady Hillary Clinton leads a task force to identify a solution to the healthcare crisis. Proposes mandatory universal private health insurance coverage, with government-subsidized premiums for low-income individuals. Outcome: Conservatives deny there’s a healthcare crisis and complain this solution is too complex. Liberals say single-payer universal health insurance (i.e., “Medicare for all”) would be better. Nothing is implemented.

2. 2006: Massachusetts Republican Governor Mitt Romney enacts mandatory universal private health insurance coverage, with government-subsidized premiums for low-income individuals. Outcome: Uninsured Massachusetts population drops from 10% to 2%. Continues in force today, with no meaningful opposition.

3. 2010: President Barack Obama copies RomneyCare, enacting mandatory universal private health insurance coverage, with government-subsidized premiums for low-income individuals. Outcome: Conservatives abhor the mandate, but have no counter-proposal. Liberals believe single-payer universal health insurance would be better, but that it’s unrealistic politically. Implemented in 2014 after judicial challenges. Uninsured U.S. population drops from 18% to 10%. Widespread popular support.

Note that, each time, we landed on the same solution…regardless of political party.

The ACA is a valiant effort that has meaningfully improved millions of lives. Obama was right to consider it a great legacy. But, it’s a losing battle against the forces of rational monetary incentive among private companies. The ACA is a makeshift dam. Water runs downhill.

The ACA’s imperfection is that it needs to be more comprehensive and absolute. In economic terms, it’s an “incomplete contract”. If government is to leave provision of health insurance in private hands, it must specify product quality (i.e., the list of minimum coverage requirements for exchange-qualified health insurance plans) and quantity (i.e., insurance available to everyone regardless of pre-existing conditions, age, occupation status, etc.). Economists also note that price caps on policies (i.e., max annual total cost to policyholder of premiums + deductible + copays) are inefficient because of informational asymmetry between government regulators and private insurers (Joseph Stiglitz, 2001 Nobel).

Partisan wrangling to pass the landmark legislation left loopholes for private insurers to excise high-risk people. They can provide limited coverage of medications used by the chronically ill to make plans unappealing to them (not prohibited by the ACA), charge higher premiums for the elderly (allowed by the ACA), classify university hospitals as out-of-network because they offer more advanced and costly treatments (not prohibited by the ACA), sell lower-quality and discriminatory insurance outside the ACA-regulated exchange (allowed by the ACA), and void coverage of policyholders who receive of charitable premium assistance (allowed by the ACA).

First, consumers don’t have perfect information to make a choice (think hundred-page insurance policy documents full of technical jargon). Second, the consequences of choosing badly are very high (buying a policy that doesn’t wind up covering what you thought it did, and thus not being able to afford medical services you need). It’s precisely in such a situation when market intervention is economically and morally justified. Accordingly, the ACA cleverly packaged governmental financial intervention (subsidies to help consumers pay premiums) with regulatory intervention (rules about exchange-qualified insurance policy content and inclusivity + a consumer mandate to purchase insurance).

But, when financial and regulatory interventions the private market fail, it’s time for public provision of the good: single-payer government-provided universal health insurance.

The inevitable solution

Give everyone health insurance. Not just old people (Medicare), really poor people (Medicaid), military and veterans (VA), and US Congresspeople.

“Everyone” means everyone living legally in the US: citizens, permanent residents (i.e., green card holders), and temporary residents (i.e., visa holders). Tourists and undocumented residents continue to finance their own insurance or pay out of pocket. In other words, if you are obliged to pay taxes in the US, you are currently obliged to get insurance under the ACA. Under a future single-payer universal health insurance system, that same group of people would be automatically enrolled.

As part of such a system redesign, it would be imperative to true up government insurance payouts with actual private healthcare provider costs. And, with or without universal health insurance, it is imperative to address skyrocketing healthcare delivery costs. Single-payer insurance would be simpler, but it wouldn’t by any means be simple — and it wouldn’t be successful in a policy vacuum.

The (losing) compassion argument

The argument from compassion and fairness using “patient stories” doesn’t work well on strict ideological conservatives…for the same reason that images of forlorn polar bears on tiny ice floes don’t work. Voluminous academic research in political psychology has shown that conservative brains respond to messages of personal responsibility and purity, not fairness and tolerance. Their primary objective is moral order, not equality.

The psychology of political belief formation explains that conservatives tend to see the world as reductively black-and-white, resist grayscale compromise, heavily discount the future, and reserve loyalty and empathy for a parochially-defined in-group. In practice, they value deterring a small number of free riders more than helping a larger number of qualified beneficiaries of government programs. As I’ve written elsewhere, this can be constructively framed as a conservative policy preference for avoiding Type I errors of over-inclusion (a.k.a. statistical “gullibility”) over avoiding Type II errors of over-exclusion (a.k.a. statistical “blindness”).

Though it may sound harsh to the unfamiliar and to the idealistic, it is uncontroversial to note that compassion is relatively low on conservatives’ list of values. They say it, their voting reflects it, and research confirms it.

The (winning) data argument

The argument that data makes universal health insurance inevitable uses language resonant with conservatives. The inescapable reality of big data and advanced predictive analytics leads organically to a logical conclusion of where it’s all going. No special pleading is necessary.

We’re not talking about moral reframing to improve the argument. Moral re-framing is how pro-intervention environmental pragmatists (e.g., the Pentagon, scientific community, property insurers, 144 Paris Agreement ratifying nations) will eventually win the climate change action debate among American voters. They’ll argue for restoring past environmental purity, protecting America from climate-driven military conflicts, preventing storm damage to private property, and conserving the sanctity of God’s creation — rather than argue for “compassionately” investing in future generations’ welfare, preserving evolutionary biodiversity, or protecting faraway wilderness.

Moral reframing of single-payer universal health insurance might go something like this: “It’s the purest way to guarantee loyal taxpayers the liberty to freely choose where to receive medical treatment. Our current system of many private insurance companies burdens families with complex restrictions on who’s in which network and who can go to which hospital. Single-payer insurance will restore the foundational American ideal of all hard-working individuals having dominion over their personal health outcomes.” That’s true. But, it’s a weak argument.

The data argument is an appeal to logic that doesn’t rely on out-group empathy: prediction gets better, risk pool wheeling and dealing gets worse, and we end up in the 1997 dystopian film “Gattaca”. Political ideology can’t insulate anyone from predictive analytics. Soon enough, data and policy could empower private insurers to cut fat-but-healthy people off for being on a path to expensive diabetes treatment. Only immense wealth can offset being denied insurance and having to pay 10s or 100s of thousands of dollars out of pocket for medical treatment.

Experience over empiricism underlies conservatism’s skepticism toward change. But, at some point, empirical reality becomes so well-documented that it’s undeniable. That’s when we get single-payer universal health insurance.

How soon is “inevitable”?

Hillary Clinton’s 1993 healthcare proposal was met with claims that there was no healthcare crisis. Two decades later when Barack Obama enacted the ACA, no credible voices dared claim that there wasn’t a healthcare crisis. Reality is chipping away at ideology.

Demographics will inexorably shift the country to the left. So goes the platitude of frustrated social progressives. But, it’s taking longer than predicted. The political right is successfully kicking the can down the road with gerrymandering. Similarly, neglected neighborhoods eventually rebound, but that can realistically take more than a generation — longer than beleaguered residents or gentrifying pioneers expect to tough it out. In 2017, there are places only recently recovered from the 1968 riots.

Likewise, trends suggest that data and its consequences will force the issue of universal health insurance…eventually. We may tinker with band-aid solutions for quite a while longer. The pain must become extreme to change the minds of entrenched skeptics. But data — like demographics and economic cycles and water — will ultimately force its own agenda. It’s just a matter of time.