Friday, October 23, 2009

Lies, Darn Lies, and Statistics

I attended a timely and constructive town-hall meeting on health insurance reform on Tuesday night, featuring several of my colleagues from the Department of Economics. I appreciate the time and effort that each of them put into their presentations. I found it interesting that many times, and across many of the various speakers, issues that I have blogged about here --- adverse selection, moral hazard, the historical accident of our employer-based health plans --- were discussed in the presentations.

As in any good debate, of course, and especially among economists, there were a lot of dueling statistics. I realized that I had long had on my “to do” list to go and look up some original source data on topics that I have read about through a lot of secondary discussion. So, I did that, and here are some facts and patterns I found interesting. These questions all address an obvious larger question : “What do we get in the U.S. for our higher per-capita spending on health care?"

1 ) The U.S. adult mortality rates are not impressive covering the entire adult life span. But, there is also a clear pattern that our performance improves with age. Conditional on a male reaching age 50, the U.S. is kind of in the middle of the pack. We still trail countries like Sweden, Spain, the U.K. France, and Canada, but our life expectancy surpasses several other European countries, including Germany, Austria, and Denmark. By age 65 American males surpass their counterparts in the U.K. and Canada, and if you are a 75 year old American male, you have about the longest life expectancy on earth. The position of the U.S. among women in lower, but the pattern is the same. [Source: “International Comparative Study of Mortality Tables for Pension Fund Retirees,” Cass Business School, U.K., data through 2001]. This corresponds to similar arguments from other data sources I have heard in the health care debate.

2 ) One thing the U.S. does relatively well is help people survive cancer. The U.S. is number 1 or 2 in survival rates for a variety of cancers. The other prominent leading country is Japan. [Source for this and the cancer statistics below: Coleman, et al. Lancet Oncology, 2008.]

3 ) Having some type of “universal access” health care system is not a sufficient condition for having bad cancer survival rates. France has an identical survival rate for women from colon cancer. Canada is not far behind on breast cancer. In addition to Japan, Australia ranks well. Jim Cobbe presented an alternative, aggregate cancer statistic that showed the U.S. slightly behind Australia in overall cancer mortality rates.

4 ) What does stand out in the world of universal access is the poor performance of Britain in cancer survival. In all four of the Lancet Oncology categories, the U.K. survival rates were markedly lower (often 15 percentage points lower, or a 15- 20 percent lower survival rate) than those of the U.S..

So what does this tell us about our current system? (By the way, the panelists uniformly agreed that our tying of health care to employment is a bizarre and inefficient historical artifact). One is that one cannot argue that any arbitrary universal access health care system will necessarily make our health outcomes worse than they are now, using mortality figures as a benchmark. Secondly, what we seem to do relatively well in terms of mortality is address health issues of middle and later age. This suggests that what we do less well is address health issues of young adults. Given that young adults are reported to be relatively less covered by health insurance in the U.S., it suggests that our system works best for the approximately 80 - 85 percent of our population (again, dueling statistics) that are insured, and this is consistent with the reports that the majority of Americans are satisfied with their current coverage, but are concerned for those who are uninsured.

Here are some unanswered questions that I still have.

1 ) The U.S. clearly has non-insurance issues that impact young adults. We do have much higher traffic fatality rates, for example, than most of Western Europe. One of my colleagues said that he didn’t think that was enough to explain the data. I wonder if there are any careful statistical controls for things such as traffic accidents and homicides.

2 ) Mortality is mortality, not quality of life. A lot of the discussion at the town hall meeting was over quality of life issues. One of my colleagues reports being able to obtain quasi-emergency glaucoma surgery in the U.K. only because he had U.S. insurance. He reported that he was told that if he had been a U.K. citizen on NHS, he would have been on a waiting list that would have accelerated his blindness by many years. That is a compelling story, but do we have any way of quantifying quality-of-life measures (beyond mortality statistics). The argument of my colleagues supporting more government involvement suggested that the mortality statistics told most of the story (U.S. higher spending = waste). The arguments of my colleagues who were relatively more opposed to government plans suggested that mortality and quality of life would tell different stories.

3 ) Can we explain the wide variety of differences (at least in mortality) of other countries that have some type of universal coverage? I was really surprised at the wide variation among countries like Japan, Australia, France, the U.K. and Denmark. What can we learn about the specific type of universal coverage and its effects?

No comments: