Here is one Michael Cannon would prefer to forget. The problem with the ideologues is that they learn to reference their papers after they’ve written theM. So often, the articles do not say anything about what is being referenced. This is why I do not often use material from think tanks unless it is clear to me they do not have a political agenda and treat facts respectfully, with reason and an absence of rhetoric. From The Physician Executive in September in 2007. I will never delve this deeply into ideological clap-trap ever again!
Dear readers, I need your help.
As you may know, I am a proponent of a non-dogmatic approach to policy debate and would like to see some truly conservative approaches to health care reform. I despise the tools of rhetoric and the use of formal logical fallacies that characterize the current crop of so-called conservatives.
Yesterday (via InsureBlog), USA today published an op-ed by Michael Cannon of the Cato Institute, an organization which I usually find provocative and challenging, but not thoroughly manipulative nor responsible for shoddy scholarship. I reviewed the articles which Mr. Cannon offers as references and have trouble connecting the articles to the point being made. There are also some logical inconsistencies.
Here is a systematic breakdown of what I found:
1. US Census Bureau. Nothing wrong here, the Bureau’s number may very well bear re-examination since all surveys have strengths and weaknesses. There is no such thing as methodological perfection.
2. Agency for Health Research and Quality: “other recent surveys put the number between 19 million and 36 million” for the uninsured. The link takes us to a MEPS survey (Medicare Expenditure Panel Survey is a running survey of medical expenditures using a representative sample of the entire US population) which does not support Mr. Cannon’s statement. The study delves more deeply into the census bureau’s figures by looking at the duration of being uninsured. The census bureau counts people as uninsured if they have been uninsured for any time n the past 12 months. Since the public health concern is identifying a vulnerable population, this is an entirely valid definition. The MEPS survey states “In 2003, 25.4 percent of the population was uninsured at some point during the year, 18.8 percent was uninsured throughout the first half of the year, and 13.6 percent was uninsured for the entire year.” Even math errors on Mr. Cannon’s part does not explain why he is comparing the proportion of American uninsured for the first half, second half and at any time of the year.
3. The next link is used to support the phrase “As many as 20% of the “uninsured” are eligible for government health programs, so in effect they are insured.” This is the most egregious. It comes from data that many who are eligible for Medicaid do not sign up since 20% of those eligible are not signed up for SCHIP. The statement holds true only if all the uninsured are eligible for some kind of government program, which is inconsistent with a seprate implication, presented with no evidence, that so many of the uninsured are illegal. Moreover, it escapes me how someone who is eligible for a program is still covered if they chose not to sign up. How does this address the vulnerability associated with catastrophic health expenditures? Moreover, the study referred to is a sober and numbing methodological comparison of the MEPS and Census surveys, not one of the many studies which have demonstrated repeatedly that under-utilization appears to be the hallmark of programs like SCHIP.
4. Mr. Cannon uses a study by Bundorf and Pauly to support the statement that as many as 75% of the uninsured can afford insurance. The paper is a fascinating and illuminating look at the effect of different definitions of affordability on the population estimate. While 76% is the high end, 31% is the low end of the estimate. Their findings support a statement much different than Mr. Cannon’s, here I quote from Bundorf and Pauly’s conclusion: “Our results demonstrate that lack of “affordability” is an important barrier, but not the only or the major barrier to obtaining coverage for all, or even most, of the uninsured. […]Omitted variables related to health status are potentially of particular importance. If our measures of health status do not capture characteristics of individuals that result in unusually high premiums (potentially due to risk rating of premiums or denials of coverage in the individual market, for example), we may over estimate the affordability of health insurance for high risks. […]Deciding for whom health insurance is affordable is ultimately a normative decision on the part of policymakers and society. We believe that our definitions, however, offer researchers and policymakers a positive empirical framework with which to begin to evaluate this question by basing the definition of affordability on the behavior of other consumers with similar characteristics, rather than an arbitrarily chosen income threshold.” This is very wise, unlike Mr. Cannon’s inexplicable peripatetic diversion.
5. To support the statement that “many economists can find no evidence that it [expanding coverage] is a cost-effective way to improve health” Mr. Cannon uses a non-peer reviewed piece of secondary literature that is actually an interesting review of the literature with respect to causality between insurance and health. The reviewers observe that if the causal chain fails, it may be either health insurance or health care that may not improve the health of the population. That is an established fact, which is not at issue because we are talking about extending health coverage to vulnerable sub-populations. The poor represent the majority of the uninsured unless you believe the prior misinformation. Perhaps the argument against covering the uninsured is being used as an argument against either government run or universal health insurance.
6. A rapid sequence of references asserting that expanding health coverage will not
a. Improve quality: New England Journal article shows that income is more important than race. The study does not address access to which coverage is most relevant.
b. Reduce disparities: Paper argues that reducing poverty is more important to health than improving health care access.
c. Affect life expectancy: A New York Times article about education being related to longevity.
d. Reduce cheating: A Health Affairs analysis of how health care costs for the uninsured are currently distributed. No mention of how not having a program deters cheating on the aforementioned non-existent program, i.e. Mr. Cannon’s argument is nearly circular.
7. The Kaiser Family Foundation says that the average family of four spends $11,000 a year. Individuals are pegged at $4,000. What the average cost per employee is, I just don’t know. Using one number without the other is not an honest presentation of the problem and I may be a little dense here… what was the point? Health care is expensive? We know that.
8. Several correct citations regarding the number of people covered by employer-sponsored insurance, rise in health insurance premiums, a White House press release, Rudy Guliani’s campaign website and a CBO letter.
Please review what you can (not everyone has full text access to Bundorf & Pauly) and let me know if I have mis-read any of Mr. Cannon’s references. Please note the title of the editorial refers to making Americans care about health care. This is a very promising position. I hunger to hear the argument, but am I just dense, or did he completely skirt around the cost, except a passing mention of average family insurance expenditures? I think there are extremely cogent arguments to be made.