Uncomfortable Arithmetic — Whom to Cover versus What to Cover
Posted Dec 17 2009 11:33am
Katherine Baicker, Ph.D., and Amitabh Chandra, Ph.D.
Much of the current debate about expanding health insurancecoverage avoids addressing an uncomfortable trade-off: witha limited budget, making benefits more generous means beingable to cover fewer people. Moreover, designing insurance benefitsthat are limited to coverage of higher-value care but are extendedto more people will generate greater improvements in healththan providing unlimited care for fewer people. Policymakersand patient advocates are reluctant to acknowledge that in aworld of scarce resources it will not be enough to eliminatewaste: we will have to make active choices in our public insuranceprograms between increasing the number of people covered andincreasing the generosity of that coverage.
Table 1 illustrates the most basic of these choices: the moregenerous the insurance policy, the fewer the people who canbe covered with a given budget. It shows the amount that itwould cost to cover a certain number of people with policiesof a certain level of generosity (as indicated by the per-personpremium). We chose these values on the basis of the distributionof premiums for individual coverage in the employer-sponsoredhealth insurance market today, using data for employers withmore than 50 workers from the 2008 Medical Expenditure PanelSurvey (MEPS) conducted by the Agency for Healthcare Researchand Quality.1 The median premium was $4,200, the premium atthe 25th percentile was $3,500, and the premium at the 75thpercentile was $5,100. This dispersion reflects many factorsbesides the generosity of policies, including geographic variationand enrollee characteristics (although basing the analysis onpremiums paid by larger employers mitigates the effects of thesecharacteristics). One could also think of the less generouspolicies as reflective of the typical premiums of a decade ago(for example, the 25th-percentile premium in 2008 was similarto the average premium in 2000, which was $3,500 after adjustmentfor inflation).
This analysis demonstrates an obvious trade-off: a fixed budgetof $180 billion per year could cover 30 million people witha policy whose annual premium was $6,000, or it could covermore than 50 million people with a $3,500 policy. With a fixedbudget, the “cost” of moving from providing a plan at the 25thpercentile of generosity to one at the 90th percentile is leaving20 million people uninsured. Of course, some might say thatthe health care budget should not be fixed: we should spendas much as it takes to cover everyone. However, this argumentneglects the financial reality that with uncapped benefits,health care programs could grow so quickly that there wouldbe no public funds left for anything else, from food to housingto education.
This discussion implicitly assumes that more expensive healthplans offer benefits that improve health more, but there isdisagreement on this point. Some would argue that more expensiveplans provide virtually no advantage over less expensive ones,and that eliminating duplicated tests, unnecessary procedures,and therapies with unproven benefit would be sufficient to stemthe growth of health care spending and provide care for theuninsured. But even after such waste is eliminated, we willstill be confronted with the same stark choice between usingpublic resources to provide all people with some care and usingthose resources to provide some people with all care (or usingsome of those resources for things like education).
Our very real need to decide how generous publicly subsidizedinsurance policies should be highlights a related trade-off:when dollars are focused on covering higher-value care, theyproduce greater aggregate health gains than when they are spreada cross care of mixed value. Evidence suggests that the benefitsthat are gained from various kinds of health care spending varywidely.
Table 2 presents examples of health care interventions and theirassociated benefit. Here, too, we have simplified a complexlandscape: the effectiveness of all these interventions clearlyranges beyond the categories in which we have placed them, butthe examples illustrate the point that dollars can be stretchedto produce much greater improvements in health if they are focusedon certain uses rather than others. The table shows the reductionin the number of quality-adjusted life-years (QALYs) gainedas we cover services that are less cost-effective: an annualbudget of $180 billion can offer a gain of at least 1.8 millionQALYs if those resources are devoted exclusively to servicesthat cost less than $100,000 per QALY. But if the thresholdis raised to include less effective services so that the averagecost per QALY gained is $300,000, the same budget will resultin only 600,000 additional QALYs. (Of course, cost-effectivenessshould be only one of many criteria used to design public healthinsurance programs: the purpose of health insurance is to reducefinancial uncertainty and increase access in the case of largeand uncertain medical expenses.)
Although the benefits of successfully targeting limited resourcescould be dramatic, the mechanisms by which spending might betargeted toward the highest-value uses are complex. One stateoffers a potential template: in Oregon, a commission that includesboth patients and providers ranks treatments according to theireffectiveness, with the goal of having the public insuranceprogram cover only services whose value is above a certain threshold.2 In practice, however, there has been very little limitingof services on the basis of these rankings — a fact thathighlights the tremendous political difficulties of making suchtrade-offs explicit.
Mandating what is covered and what is not isn’t the only approachfor increasing the reach of limited public dollars. Competitionamong private plans for enrollees (who could receive government-subsidizedvouchers based on their income and health risks) is anotherstrategy for moving people into plans that offer higher-valuecare. Lessons from the behavioral economics literature, however,imply that unregulated competition alone is unlikely to resultin patients’ choosing the highest-value plans, suggesting thatthere is a powerful role for more nuanced plan design. All thesestrategies, however, will involve implicit or explicit trade-offsbetween the generosity of subsidies and the number of peoplewho are eligible for them, as well as the resources that willbe available for other public programs.
Unfortunately, the mere recognition of the existence of trade-offsdoes not tell us how best to make them. There are no easy solutionsin which all people receive all care that might potentiallybenefit their health. There is only 100% of Gross Domestic Productto go around, whereas we could theoretically spend a virtuallyunlimited amount of money on health care. As medical technologyadvances, there will continue to be new treatments that willoffer incremental improvements in health at increasingly highcosts, and we will have to decide how to allocate scarce resourcesamong treatments and among people. To date, there has been littledebate in Congress about the generosity of public benefit packages,except for whether such benefits should cover abortion. Buteventually, we will have to engage in the difficult discussionsrequired to choose whom and what our public insurance programsshould cover. Some might call this rationing, but the realityis that millions of Americans now have no access to lifesavingmedical technologies at the same time that public resourcesare being devoted to covering less-effective therapies for less-seriousconditions. We find that sort of rationing hard to justify.
Financial and other disclosures provided by the authors areavailable with the full text of this article at NEJM.org.
From the Harvard School of Public Health, Boston (K.B.); and the John F. Kennedy School of Government, Harvard University, Cambridge, MA (A.C.).
This article (10.1056/NEJMp0911074) was published on December 16, 2009, at NEJM.org.