Thursday, January 15, 2009

New Study Advises that Practitioners Emphasize Effects of Smoking on Obesity; However, Study Doesn't Find Such Effects

A new study published in the just-released February issue of the American Journal of Public Health examines the relationship between adolescent smoking on obesity in early adulthood (see: Saarni SE, et al. Association of smoking in adolescence with abdominal obesity in adulthood: a follow-up study of 5 birth cohorts of Finnish twins. Am J Public Health 2009; 99:348-354).

This study was a prospective cohort study involving five different birth cohorts of Finnish twins, with a total sample size of more than 4,000. They were followed for approximately 11 years, between the ages of 16 and 27. The risk of abdominal obesity in young adulthood was related to whether the individual was a smoker in adolescence, while controlling for a number of potential confounding variables, including diet, physical activity, socioeconomic status, parental body mass index, and individual body mass index.

The study reports that without the confounding variables, there was a significant elevation of the odds ratio for obesity (OR = 1.77; 95% confidence interval, 1.39-2.26). After controlling for the potential confounders, there was no significant elevation of the odds ratio for obesity (OR = 1.34; 95% confidence interval, 0.95-1.88).

The study acknowledges that it failed to find a statistically significant relationship between smoking and obesity and that this may be explained by the hypothesis that it is other factors - not smoking itself - that lead smokers to tend to have a higher incidence of obesity. As the authors state: "The increase in the risk of abdominal obesity or overweight among the women was no longer statistically significant in the fully adjusted models. This may have been caused by insufficient statistical power. Another possible interpretation of our findings is that adolescent smoking is a proxy for other factors or is part of a cluster of factors predisposing to later abdominal obesity."

The authors conclude that: "Given the greater risk of overweight and abdominal obesity among girls who smoked daily and the fact that adolescent smoking is often associated with preoccupation with weight, emphasizing the deleterious effect of smoking on abdominal fat accumulation could be effective in smoking prevention among young women."

The Rest of the Story

Wait one second here. The study fails to find a significant relationship between adolescent smoking and obesity in young adulthood after confounding variables - such as diet, physical activity, parental BMI, and individual BMI - are accounted for. The study acknowledges that one interpretation of the findings is that smoking may simply be a proxy for other factors (such as other health behaviors) that actually are influencing obesity. But then the study concludes by advising readers that we should be telling the public (emphasizing to the public, no less) that one of the deleterious effects of smoking is to cause obesity later in life.

Do you mean to tell me that even though the study cannot conclude that smoking causes obesity, we should tell the public that smoking causes obesity because it might lead to a decrease in smoking initiation?

This is definitely an example of viewing the ends as justifying the means. Because reducing smoking initiation is a good thing, it doesn't matter if we lie to people about the effects of smoking or give them information which is not scientifically supported.

What's most odd is that this conflicting information appears in the very same paper. This suggests that the authors of the study are not concerned about the validity of the information that they are recommending be disseminated to the public.

When the study controlled for all the confounders, including body mass index at age 24, the association between smoking and obesity was no longer statistically significant. This suggests to me that the reason why smoking at an early age is associated with obesity later in life is most likely that smokers tend to have a constellation of poor health habits, including poorer diet and less physical activity. It is most likely that these factors, and not the smoking itself, is what is causing them to become obese.

This doesn't mean that smoking is not associated with obesity; it just means that the relationship is not a causal one. I think what this study really points to is the fact that smoking is just one of a constellation of health behaviors that tend to coincide, including smoking, poor diet, and inadequate physical activity. It is these other behaviors which are directly related to obesity. Smoking just happens to be associated with obesity because it is related to these other causal factors. But once you adequately control for these other factors, smoking is no longer significantly associated with obesity.

In fact, there is not a lot of biological plausibility for such an effect. The paper does not in fact outline any potential mechanisms by which smoking could cause obesity. There is some evidence that smoking may help keep a little bit of weight off, but not so much for the reverse. I think, again, that the most likely explanation here is not that smoking is causing obesity, but that smokers also tend to have poorer behaviors in other areas which lead to the obesity.

This research is still important, because it reinforces that smoking fulfills a need for teens. Teens who are less self-confident and lack self-esteem are more likely to smoke. But they are also more likely to engage in other unhealthy behaviors. This research suggests that public health practitioners may not want to treat smoking, diet, and physical activity separately, but that there may be a unifying cause for poor health behavior in general, and that perhaps we should try to focus more on the underlying reasons for youths to engage in poor health behaviors. There is too much compartmentalization in how we deal with adolescent health issues, and perhaps we need to take a more holistic approach.

But the one thing the paper does not demonstrate is that smoking during adolescence causes obesity or has deleterious efects on abdominal fat accumulation. The authors of the study themselves admit this. The authors themselves point out a plausible alternative explanation for the study findings - that "adolescent smoking is a proxy for other factors ... predisposing to later abdominal obesity." However, this does not stop them from recommending that public health practitioners should be "emphasizing the deleterious effect of smoking on abdominal fat accumulation...".

They could have made that recommendation without conducting the study. In fact, they would actually have been more justified in making that recommendation prior to conducting the study, because at that point there was reasonable speculation about the hypothesis. The study, however, suggests that the hypothesis is incorrect and it removes any justification for telling the public that smoking causes obesity.

As we saw with the thirdhand smoke issue, however, science doesn't appear to be necessary in tobacco control. Anti-smoking groups and advocates are communicating information to the public without any scientific evidence to back it up. Apparently, the attitude is that since this is all for a good cause, it is acceptable to lie or deceive the public.

Needless to say, I disagree. I think that as public health researchers and advocates, we have an ethical responsibility to communicate accurate information to the public and to be able to back up our communications with solid scientific evidence.

As I'm finding out, that's not happening in tobacco control.

No comments: