This Is What Health Insurance Should Look Like

Laboratories (Photo credit: SLU Madrid Campus)

When you're shopping for insurance coverage, what do (should) you think about? You might look at the price of coverage - in the case of the ACA, this was the biggest factor explaining why the uninsured opted against buying coverage. So the answer was greater subsidies, and greater regulation of insurance prices.

If you're just switching plans, you'll probably check out the physician network to see if your doctors are covered. If you're a bit more savvy, you'll pay attention to the deductibles and co-pays, and maybe check out the formulary if you take prescription drugs. Benefits might come into play too - a survey of Californians buying coverage on the individual market found a strong preference for more-generous benefits. Suffice to say, your decision won't be simple and involves a large number of considerations.

In an attempt to open up access to affordable coverage, the ACA imposes significant restrictions on most of these fronts. Premiums can only vary within a narrow age-band; deductibles and co-pays are limited statutorily; and an expansive minimum benefit requirement limits variation of benefits. (Some states, like New York, are also attempting to regulate networks.) So when you're shopping for coverage on the new individual market, it's unlikely that you're really getting the coverage that you want. At best, you're getting a second- or third-best policy. But even before the ACA, it would probably be very difficult to get the policy you really 'want.'

While plans certainly made clear their drug formularies, physician networks, and deductibles, there were still massive information asymmetries. You might have bought a plan that excluded coverage of some expensive oncology drugs - only to be diagnosed with lung cancer in the middle of your policy year. Alternatively, if you were on a low-deductible employer-sponsored plan, and you present with lower-back pain, an x-ray would normally suffice but you and your doctor may opt for an MRI instead - which costs more and provides more information, but offers little additional clinical benefit. So long as the MRI does no harm, you'd be indifferent between the treatments, but all else equal, you shouldn't 'want' the MRI any more than the x-ray.

Ironically, the ACA lumps both cost-effective care and care of marginal utility into its basic benefits package - called essential benefits, this includes 10 categories of care that must be covered by all plans. Conservative approaches, on the other hand, tend to favor high-deductible plans - expansive benefits coupled with high-deductibles would discourage the use of low-value treatments, but without significant improvements in transparency, it's difficult for consumers (outside of specific kinds of technology, like generic drugs) to make those kinds of nuanced cost-benefit decisions.

But in a recent paper (highlighted at The Upshot by Austin Frakt and Amitabh Chandra), UCLA Law School Professor Russell Korobkin offers another way to think about health insurance:

This article proposes a new paradigm for rationalizing health care expenditures called 'relative value health insurance,' a product that would enable consumers to purchase health insurance that covers cost-effective treatments but excludes cost-ineffective treatments.

The idea is fairly straightforward. Instead of ranking (as the ACA does) plans by the actuarial value (the share of spending that they cover on average), plans would be sold based on the value of treatments that are covered. Some plans would cover only the highest value treatments, while others would cover all treatments - the advantage is that consumers could purchase coverage without knowing ex-ante that they might need treatment for a particular condition, but still safe in the knowledge that some level of treatment would be covered.

Korobkin, Frakt and Chandra all do a good job explaining the potential for adverse selection under such an arrangement, as well as possible solutions, so I won't rehash those concerns.

Another important point, however, deserves some more attention.

Distinguishing high-value treatments from low-value ones requires significant investments into 'comparative effectiveness research' (CER). CER pits different treatments against one another to help determine the cost-benefit of using particular treatments over alternatives. The ACA allocates $500 annually to CER (and ARRA before it allocated $1.1 billion), and establishes the Patient-Centered Outcomes Research Institute (PCORI) to conduct the research. Here, as Frakt and Chandra argue, the research becomes similar to a public good. No individual insurer has an incentive to invest in this research because once the research is completed, it becomes available to all insurers. This is why public funding is so necessary.

Yet, this seems a bit incomplete. For starters, data sharing among competing firms isn't entirely a novel concept. Pharmaceutical companies routinely share data on websites like ClinicalTrials.gov, as well as directly with academic researchers. More recently, the National Institutes of Health (NIH) convinced 10 pharmaceutical companies to pool data on Alzheimer's, Type 2 diabetes, rheumatoid arthritis, and lupus to explicate basic disease biology - helping them all develop more effective treatments. While there's certainly a collective action problem when it comes to CER, it doesn't appear insurmountable - particularly if collaboration would allow insurers to put pressure on providers, hospitals, and drugmakers.

Moreover, we shouldn't take for granted that CER is in fact a public good. In a 2011 paper, Scott Harrington of The Wharton School at UPenn notes that there are many examples, in and out of health care, of firms that perform functions similar to CER and charge for access to the information. In the same paper, Harrington points out that one of the major impediments to private CER is simply the lack of demand.

The tax-advantaged system of employer-sponsored coverage makes employers much less price-sensitive to health care costs, and therefore less interested in cost-effective treatments, while Medicare and Medicaid explicitly do not use comparative effectiveness when deciding what treatments will be covered. Paring back the tax code's inefficiencies, and perhaps offering a CER tax credit, could be a less distortive option to incentivize CER. Also, though somewhat circular, Korobkin's proposal for relative value health insurance (RVHI) could, in theory, create exactly the demand needed to stimulate private CER.

Even if the obstacles for privately-funded CER are indeed significant, public funding likely need not be massive. Private CER databases already exist. Massachusetts's General Institute for Clinical and Economic Review conducts annual appraisals of various procedures. Tufts University's Cost Effectiveness Analysis Registry does the same, and also allows some limited searching of their registry by the public.

The prime example of a publicly funded CER database is England's National Institute for Health and Care Excellence (NICE) which reviews most new drugs and technologies, and publishes guidelines on efficacy and cost-effectiveness. Certainly, U.S. officials could negotiate use of NICE's database as a baseline for more serious public investment into CER. This doesn't mean that we'd be stuck with the effectiveness thresholds that the U.K. uses - around $50,000 per quality-adjusted-life-year (QALY). Different plans would use different thresholds, which would in turn determine the drugs and treatments covered under the plan.

Perhaps the most exciting thing about Korobkin's RVHI is the potential to mesh with future developments in electronic health records and mountains of genotypic and phenotypic data. Different treatments can often have different cost-effectiveness levels for different people - this can mean two people with the same exact policy can have different treatments covered based on the cost-effectiveness of these treatments given their biology.

For instance, someone with with HER2-positive breast cancer would likely see significant benefit from Herceptin, but not much help from hormone therapy. For such a policyholder, Herceptin would be much higher-value than hormone therapy and so would be covered by skimpier plans.