The New York Times set out on Sunday, November 15, 2009 to evaluate on its editorial page proposals to address what they described as the worrisome and “relentless rise in health care costs and health insurance premiums.” Their review of ten cost control approaches was largely a mix of wishful thinking, naïve optimism, and a misguided but well-intentioned embrace of faddish ideas. In an April issue of the Annals of Internal Medicine, my colleagues Joe White of Case Western University and Jonathan Oberlander of the University of North Carolina and I argued against the faddish embrace of managed competition, preventive medicine, electronic medical records, and research on comparative effectiveness on the grounds that there was no empirical support for their roles as cost savers. Here I want to concentrate on one recommendation the Times (and their economic columnist David Leonhardt) support: the proposal to tax expensive, so-called Cadillac health insurance plans.
The Times endorses the Senate Finance proposal for an “excise tax on health insurance plans that cost more than $8000 for an individual or $21,000 for a family,” on two grounds. They first claim that such a tax would prompt insurers to design plans less than the threshold amounts. This, even if done to avoid the threshold, in itself has no necessary effect on total costs. Secondly, the Times makes the conventional micro- economic argument that having the insured pay more out of pocket would make them more cost-conscious consumers of medical care. In their words, “enrollees would have to pay more money for many services out of their own pocket, and that would encourage them to think twice about whether an expensive or redundant test was worth it.”
It is the claim about patient cost-sharing that is crucial and the astonishing feature of this line of argument is how odd it is and how weak is the evidence used to support it. The assumption from which this proceeds is that we should be critical—even contemptuous—of the wish for first-dollar, extensive health insurance. Why else use the slogan, Cadillac coverage, when the proper description is broad coverage of health risks with no deductibles or co-insurance to speak of? One would have hoped that those journalists supporting reform—and cost containment—would have noted that our Canadian neighbors have precisely that coverage for hospital and physician services and spend roughly 40 percent less per capita while using more physician visits and bed days than do American patients. (Note too that the US and Canada had identical spending in 1971, when Canadian national health insurance began in full, another indicator that enacting broad coverage cannot explain our cost differences.) Nor, incidentally, do any other of our trading partners rely on significant patient cost-sharing as a major cost control instrument. Where, then, does this confident support of a faddish idea come from? (fn.Their research includes the older book, Patient Cost Sharing: Snare or Delusion, 1981) and 7 review articles from the Center last decade.
The short answer is economic theology, not social science evidence. The standard teaching in American economic courses not only embraces market allocation of goods and services generally, but regularly extends the scope of goods and services that should be allocated by the willingness and ability to pay. Since the fall of Communism, few challenge the basic claim that for most ordinary goods and services, market allocation works better than its competitors. But there is no reason to believe this applies to most medical care. Not only do we go to physicians to find out what to make of our health circumstances, but we rely on them overwhelmingly for guidance about what to do. The asymmetry of the bases of judgment is overpowering, even when the Internet offers all manner of ambiguous information about our health and its care. The only tolerable basis for using patient cost-sharing as an allocational tool would be if citizens spending their own money can choose wisely, separating the useful from the useless or harmful medical intervention. Does the evidence support this assumption?
The short answer is no, but there is much to dispute about the major empirical basis in American reform writing for the support of patient cost-sharing. Though neither the Times editorial board nor Leonhardt invoke the Rand Health Insurance work of the 1970s, that social experiment has provided the seeming authority to the argument that requiring patients to pay for some of their health care does not harm their health and is therefore worth requiring. In fact, the experiment—on which I worked briefly in the summer of 1973—does not support that claim. For those who experimentally paid a co-insurance rate of 25 percent, there was a reduced rate of using medical care compared to those with first-dollar coverage. Having people pay to get medical care does reduce use; that much is established by much research. But the crucial other finding was that services cost sharing dissuaded was not useless and the care sought useful. Rather, cost-sharing blocked both useful and wasteful care in roughly equal proportions. Moreover, the fact that in a three-year experiment, large differences in health outcomes did not show up provides little grounds for thinking nothing is lost when care is rationed by the ability and willingness to pay.
The extensive research on the distributive consequences of patient cost-sharing provides further grounds questioning the conclusion that bearing costs when ill or anxious is a worthy policy. For Health Services Research in the Work over many years at the University of British Columbia—by Professors Robert Evans, Morris Barer, and Greg Stoddard—shows that lower income citizens and families disproportionately ration their care more than their higher income counterparts. Since the care postponed has not been shown to be wise, the distributive consequences alone support the attack on the Cadillac coverage argument.
So much then for the empirical findings within the health economics field in the United States. Elsewhere, there is largely contempt for the argument on which the Cadillac excise tax is based. In the United Kingdom, care is free at the point of service and, for the first 50 years of the National Health Service, it was the model instance of aggressive cost control through overall budgeting. One does not have to celebrate what the NHS has done to notice that it has relied on direct budget limits to control outlays. Equally, Canadian provinces have paid lower prices for the services they fund and used the same budget limits to enforce cost control goals. The central lesson, as we argued earlier, of international experience is that effective, defensible cost control requires strong government leadership. This experience reflects the obvious, but often forgotten axiom that a dollar of expenditure for medical care is necessarily a dollar of income for the medical care industry. That means cost control must be costly to someone and in turn will always be controversial. This is true whether the foregone expenditures are administrative, wasteful, useful or useless.
It is this hard truth that has led so many to try to find appealing seemingly costless means of controlling inflation in American medicine. But if the United States is to control its costs, it will have to learn from those that have done so in the last half century. That will mean price restraints, spending targets, and insurance regulation. Painless savings are an illusion and an excise tax on Cadillac coverage is one example of precisely such an illusion.