|
medpundit |
||
 |
 |
|
Thursday, April 24, 2003Obesity appears to increase the risk of dying from cancer significantly and could play a role in more than 90,000 deaths from the disease each year in the United States, a study by the American Cancer Society released Wednesday reveals. The study, which looked at hundreds of thousands of people over a 16-year period, found "the heaviest men and women in the study had 50-60 percent higher death rates from cancer than normal-weight people," Jeanne Calle, the principal author of the study and director of analytic epidemiology for the American Cancer Society, told United Press International. Based on the results, Calle's team used the current prevalence of obesity in the United States to estimate "as much as 20 percent of cancer deaths in women and 14 percent in men" could be due to being overweight. "This translates to more than 90,000 deaths from cancer each year that could potentially be avoided if we could maintain normal weight throughout adulthood," Calle said. If we’d all just get off our fat butts the world would be such a better place. But the study has some problems that neither it, nor the editorial accompanying it, addresses: -It doesn’t break down its findings by age. Everyone is lumped together by body mass index, a measure of obesity, instead. The authors don’t mention if heavier people tended to be older, a variable that would affect their death rate -It doesn’t tell us the incidence of cancers by body weight, only the number of deaths from cancer. Did fat people get more cancer than skinny people, or less? The authors don’t say. -It doesn’t reveal how many people were included in each weight group. There’s no indication of how many people were skinny, or how many were fat. That does make a difference, especially when expressing results as percentages. If the number of obese people was small compared to those of normal or average weight, then the comparison of death rates may not be valid. Obesity gets the blame for everything these days, but this study doesn’t make it at all clear that it’s responsible for cancer deaths. I have to wonder, though, if California requires restaurants and grocery stores to post warnings about carcinogens in their food, will they read this study and require them to post warnings that their food contains calories? UPDATE: Comments from Educated Guesswork: I believe your criticism of the obesity/cancer study is somewhat off base. (1) While it's true that they don't *break down* their findings by age, they did control for it as a confounder. See the footnotes on p 1631. So, while it's hard to tell when this is significant, it's not correct to suggest that the problem could just be age. This is how the study says it controlled for age: "Age-adjusted death rates were calculated for each category of body-mass index and were directly standardized to the age distribution of the entire male or female study population." This still obscures the role that age could be playing in 1) the incidence of cancer and 2) the incidence of obesity. Most cancers are more common with increasing age - breast cancer, colon cancer, lung cancer, skin cancer, to name a few. Obesity is more prevalent with increasing age, too. It's quite possible that more of the "overweight" group also happened to be older and thus more prone to cancer. (2) You're correct that it doesn't tell us the incidence. So, it could be that being fat just increases your chance of dying from cancer. But that would be pretty bad too. Yes, it would be. But, incidence counts. If fat people are less likely to get cancer in the first place, then perhaps the ones who end up with it have more virulent forms of it, hence greater mortality. Not all cancers are the same. Although the study lumps all breast cancers together, there are subcategories of breast cancer, (and most other cancers, too) that each have different mortality rates. (3) They don't say *directly* how many people are in each weight group, but you can work it out pretty easily from the data in Table 1. Just divide the incidence by the death rate. Clearly, there are a lot more people in the higher group. However, this isn't a problem for the analysis. The Cox proportional hazard technique (which is what they used) works just fine even when the groups are unbalanced. Except that in reality smaller samples make for greater errors. The human organism isn't a precise mechanism. There are so many tiny, uncontrollable variables that can skew results in small populations. If only 393 morbidly obese men died of cancer, extrapolating that number to death rates per 100,000 is going to magnify any error inherent in the small sample. Here's an example that's easy to reproduce. Try taking your pulse. Count it for five seconds and multiply by 12 to get your heart rate in beats per minute. Then, count it for a full sixty seconds. Chances are, they won't be the same. Missing just one beat or counting one additional beat in that five second count will throw your calculated heart rate off from your actual heart rate by 12 beats per minute. And that's with the heart - one of the most mechanically precise organs we have, when it's working right. UPDATE II: Educated Guesswork responds. posted by Sydney on 4/24/2003 08:29:00 AM 0 comments 0 Comments: |
|
