Map of life expectancy at birth from Global Education Project.

Tuesday, July 31, 2007

Invidious Comparisons

While it's tolerably straightforward to compare the states on indicators of health care quality and access, as the Commonwealth Fund has done, trying to sort out the causes of disparities in actual health outcomes is far more vexing. (To be honest, people will bicker quite a bit about the health care quality measures, and they could be a lot more comprehensive and nuanced, but some of them are quite robust and they seem to correlate strongly with the more questionable ones.)

C. Corax asked about breast cancer rates in Utah. Actually they are on the low side. If you enjoy playing with numbers (and who doesn't?), the National Cancer Institute let's you generate your own tables on state cancer incidence and death rates, along with county-by-county comparisons, maps, historical trend charts -- it's a veritable McDonald's Playplace for the inquiring mind. Utah happens to have a high quality cancer registry, and so we know that its age-adjusted rate of breast cancer in 2004 was 113.7/100,000 women. Alabama had the lowest rate, and Washington state the highest -- a lot higher, at 136.6. Massachusetts, BTW, was pretty high too, at 130.1.

So why is this? We don't know all the reasons for variations in breast cancer incidence, but the most powerful factor we do know about is childbearing history. Having kids early and often is beneficial, and that's something women in Utah and Alabama are likely to do. People want to blame breast cancer on environmental factors like pesticides and radiation, and diet, and maybe that has a little bit to do with it, but as far as anybody has been able to tell so far, not very much. There's certainly good evidence for some important environmental factors, likely something about the diet, based on international comparisons, but we haven't figured out what they are yet. E.g., Japanese women have a low incidence, but when they immigrate to the U.S., in two generations their rate is similar to the U.S. average. Soy? Who knows.

But then I checked the kind of cancer that is most likely to be caused by exposure to radioactive fallout, thyroid cancer, and guess what? Utah is indeed number one -- 13/100,000, compared with the U.S. average of 8.6 and the lowest incidence state, Oklahoma, at 5.1. Is that because of bomb testing? Dunno. (Nevada isn't on the list because they don't have a tumor registry, sorry.) Pennsylvania, New Jersey, Connecticut, Massachusetts and other states you wouldn't necessarily predict are also on the high side, but that doesn't rule out that nuke testing is related to the high rate in Utah -- something else could be going on in the northeast. Remember, though, that thyroid cancer is pretty uncommon so we aren't talking a huge number of cases and the confidence intervals for the top four or five states overlap.

And finally, for all cancers combined, Utah has almost the lowest rate, in a statistical tie with New Mexico, at 405.9. We have a pretty good idea why -- a low rate of smoking and drinking in Utah. I don't know enough about New Mexico to say for sure, but I suspect they have a low rate of smoking there also. The state with the highest overall incidence of cancer is Maine, and Massachusetts is right up there as well in third place. But there are likely some paradoxical factors at work here. States with better health insurance coverage and better health care -- and Maine and Massachusetts are number one in that department -- have higher incidences in part because more cancers are diagnosed, because more people are screened for colon, prostate and breast cancer. In other states, some people have these cancers but die of other causes before they are discovered.

Life is easy, but epidemiology is hard.