8.22.2007

Numbers Game

MSNBC is reporting on a pair of studies proclaiming the safety of weight loss surgery. It doesn't take long to see through the games being played with the studies, though.

First up, these "long-term" studies only extended for 7 to 10 years. If similiar procedures done without the intention of weight loss hadn't been shown to severely reduce life expectancy, this might be significant. Alternately, if we assume fat people only have a decade to live, this would be significant. If the study didn't also show that 94% of the fat people were likely to make it through that decade.

Which brings us to the second issue. About 94% of fatties didn't die. 95% of surgery patients didn't die. Wow. What a difference. I did round up from 93.67% for the fat people, though, so I maybe I minimized the difference. So, for every 100 people, the difference was slightly more than 1 person.

If they were honest, of course. See, the article doesn't mention whether or not the study controlled for inconvenient data. Like, deaths FROM the surgery. Its the kind of thing that could very easily have been quietly "controlled for" and thus the physical risks of the surgery may be completely disregarded. "If you only count the people who lived, the results are quite good," style reasoning.

The study also only mentions percentages in terms of relative risk. Meaning, the risk for surgery relative to the risk without surgery which they pronounce as decreased by 30-40%. They don't mention that the risk without the surgery wasn't high to begin with. Otherwise, it relies on "this many people died!" reporting which is basically meaningless in context. It mentioned a second study, for instance, only with the raw data. No doubt because it would show that in this study a 96% of fat people didn't die. Again, the difference was about 1%, with a little more than 97% of surgery patients living. If they didn't control, of course. They then quickly slid into pronouncing relative risks for specific diseases without any context to understand what the risks really were. When it's time to note that surgery patients had increased risks of death from accident, suicides (!), and other causes no relative risk is mention. Though they do note how this "puzzled" the scientists.

Which brings us to yet another problem with all this. See, data that didn't serve the preordained purpose was "puzzling". No conclusions or pronouncements that WLS drives people to kill themselves. No, it just puzzles. But if the same kind of data falls in favor of WLS, then it means something. No puzzlement at all, why such conclusive data should justify MORE weight loss surgery. Its this kind of selective interpretation that plagues fat science and is why fat activists are right to be very skeptical whenever the news wire mentions some horrible new thing blamed on fatness. There really is no reason to think we're getting the whole story. These studies report the numbers they want and the conclusions they were looking for. If it doesn't fit, it gets controlled for or it dismissed as puzzling. To steal from a fellow fat blog, one is reminded of the saying attributed to Disraeli: "There are three kinds of lies: lies, damned lies, and statistics."

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.