Tuesday, February 19, 2013

When evidence is not evidence

From Baby Boomers Sicker Than Parents’ Generation, Study Finds by Nicole Ostrow.

This study illustrates the temptation to overreach on conclusions and the dangers of not comparing like-to-like.
The study, among the first to compare the generations, shows that baby boomers aren’t as healthy and active as most would believe, said Dana E. King, the lead author. They become sicker earlier in life than the previous generation, are more limited in what they can do at work and are more likely to need the use of a cane or walker, the research found.
But is that really what they found? When comparing two things, you have to normalize the extraneous in order to be able to discern those elements which are pertinently different. So how do you normalize two different generations? Are they really identical?

In this particular case, one can imagine several confounding issues. The life expectancy is dramatically higher for those born after 1946 than those born before, i.e. a greater percentage of each cohort is surviving. That would seem to imply that the average health for the longer lived group will necessarily be worse. To put it differently, if you have health conditions that in earlier times would have carried you off but with modern medicine (both the quality of medicine being exercised as well as the quantity) you survive, it seems reasonable to assume that you will survive in some diminished capacity. You would not have been part of the health average with the earlier cohort but you will be with the second.

The consequence of this difference in longevity/survivability is that if all things else were equal and everyone lived exactly the same lifestyle, you would still expect the later generation to demonstrate worse average health than the earlier generation. I see nowhere in the report where this intergenerational difference in longevity is controlled for.

Without that control/normalizing of populations being compared, what exactly can the study tell us? Not much.

Even more telling are the recommendations of the lead author, a medical doctor and health researcher. Astonishingly the recommendations are 1) "There needs to be a new emphasis and continued attention to programs to improve healthy lifestyles in this age group" and 2) "more studies are needed."

So really, absent any normalizing between the two populations, all we know is what we already know - there is an opportunity for people to improve their health by eating better and exercising more. It does not appear that the study does provide any new evidence at all. All it appears to provide is an opportunity for self-interested advocates to advocate for actions that they already supported. An example of conformation bias, perhaps?

No comments:

Post a Comment