Junkfood Science: Dying without insurance — a myth or fact?

December 16, 2008

Dying without insurance — a myth or fact?

A worrying claim that has circulated for years is that 18,000 Americans die every year because they don’t have health insurance. A thought-provoking analysis revealed that this is little more than an urban legend.

The source of this widely-repeated factoid turns out to be a 2002 report sponsored by Robert Wood Johnson Foundation, the leading organization using it to call for a complete restructuring of our healthcare system and the adoption of universal healthcare by 2010 to solve this problem.

What makes this analysis valuable to read is that if we don’t accurately define and understand the problem we’re trying to solve, then the solution isn’t likely to work. While it seems intuitive, will nationalized health insurance really help to improve health, and especially help those with the poorest health and the least access to care, and reduce health disparities in our country? Or could it actually make things worse for the very people we’re trying to help, as well as cause greater hardship for even more people? By failing to correctly identify the problem, what other, better solutions might we be overlooking? This is not about not caring about the uninsured, but to remind us that sensationalizing an issue with scares and myths won’t lead to better reasoning.

A year and a half ago, the widely claimed numbers of uninsured people were examined and found to have been exaggerations, nonetheless, the proposed solution was to mandate insurance coverage for everyone. The pilot study for that proposal, however, has sadly failed to improve health care and healthcare access for people; is leading to unforeseen intrusions into private lifestyles and doctor-patient relationships, with costly preventive health mandates that aren’t evidence-based but designed to make money for stakeholders; and has proven to be a financial disaster, expanding bureaucracy and requiring increasingly more taxpayer federal funds to keep it solvent.

This latest analysis was written by Greg Scandlen and published at the State Policy Network, an independent nonprofit think tank to explore the soundest information and the most effective solutions for policy issues, while preserving a free society. He examined the RWJF report, Care without Coverage, behind the claim that there are 18,000 deaths a year from lack of insurance. This report had been issued by the private organization, the Institute of Medicine, as part of its Coverage Matters series.

Mr. Scandlen explained that this report hadn’t actually conducted original research, but was a meta-analysis of existing studies. “There is little consistency between these studies in quality or methods, and all are ‘observational rather than experimental,’ as the IOM report acknowledges,” he wrote. They were epidemiological studies reporting correlations that cannot identify the cause for observed relationships between insurance and health. “In other words, there may be some other condition that leads people to be both uninsured and in poor health,” he said.

Very few of these underlying studies adjusted for socio-economic differences in the populations, which is a major problem since income and education are closely associated with health outcomes regardless of the kind of insurance coverage involved. We know, for instance, that people on Medicare all have the same insurance coverage, but 54% of those under the poverty level report themselves as being in poor or fair health, as compared to only 25% of those above 200% of poverty. The “universal” systems in Canada, the United Kingdom, New Zealand, and Australia report even bigger income-based disparities with lower-income adults over three times as likely as higher-income adults reporting themselves to be in “fair or poor health.”

While IOM did not correct for income differences in the populations (insured vs. uninsured), it had available a proxy for income… IOM might have studied people on Medicaid as being both low-income and fully-insured. But this result did not suit its predisposition, so it chose to downplay it. Deep in the report it acknowledges that, “study results for overall health status, cancer outcomes, and hospital-based care, (find that) adults with Medicaid frequently fare no better and sometimes far worse than uninsured patients in their health-related outcomes”…

In fact, the evidence strongly suggests that insurance is not the most important factor in health — income and education are. But RWJ wasn’t paying IOM to come to that conclusion.

Mr. Scandlen had reviewed all of the 139 studies in the report and found that 26 studies had compared uninsured people only to those with private insurance — in other words, they had failed to compare the uninsured to those receiving government health coverage in existing programs, Medicaid and Medicare. He found that 44 studies had compared the health outcomes of Medicaid enrollees to people who had no insurance and more than 3 out of 4 found that those with government-provided healthcare coverage did worse than the uninsured on a range of health treatments and outcomes.

It turns out that the 18,000 death figure was not based on looking at a single medical record or cause of death, it was a mathematical prediction, based on a string of flawed and unsupportable assumptions. He found the “tortured methodology” that came up with the 18,000 deaths figure buried in Appendix D:

First, they rely entirely on a single study that estimated “a higher overall mortality risk for uninsured adults of 25 percent. ”Linda Gorman deconstructs this original study persuasively. She notes the study by Peter Franks begins by looking at people who were uninsured or privately insured in 1971 and then looks at their mortality in 1987. Never mind that this entire population likely went through many spells of being covered or not being covered in the intervening years. IOM then assumes that the incidence of diseases like diabetes, hypertension, breast cancer, and HIV are the same in the uninsured population as they are in the privately insured population. So they multiply the death rate for the insured by 125% and get the “excess mortality” of the uninsured. Voila! 18,000 dead.

As Scandlen went on to explain, the problems with the IOM’s math began by failing to accurately identify who the uninsured are and to recognize that they are a very diverse group, ranging from "young healthy people who don’t want insurance" to "pretty poor people," all with vastly different mortality rates. The uninsured also have different disease rates than the general population, he said, because they tend to be younger (older people are covered under Medicare).

Gorman’s analysis noted that the authors claimed their findings supported government-provided “universal health insurance to reduce both financial barriers to care and the risk of premature mortality.” But, she wrote, “the study itself provided no information about the effects of government coverage on mortality because it excluded “adults with Medicaid, Veterans Administration insurance, or Medicare.”

A 1994 study by researchers at the National Heart, Lung, and Blood Institute at the National Institutes of Health had examined the effects of government coverage on mortality. It reported that people on public programs had higher mortality than either the uninsured or those with private insurance. Mortality most related to income level. Gorman also pointed out that most deaths among younger adults, who make up much of the uninsured, are due to accidents, suicides and assaults (confirmed in Health United States 2007 issued by the Department of Health and Human Services), issues for which health insurance has little affect. Finally, she went step-by-step in describing the math used to arrive at the 18,000 death figure.

Scandlen concluded that one thing that is almost certainly not true about the uninsured is that 18,000 of them die each year simply because they do not have coverage. Nor is there any evidence that government insurance will solve the real reasons behind health disparities, as intuitively correct as it might seem.

Food for thought.

Bookmark and Share