Tuesday, June 10, 2008

New Study Concludes that Smoking Causes Impairment in Memory among Middle Aged Adults

A study published in this week's issue of the Archives of Internal Medicine concludes that smoking causes memory loss among middle-aged adults (see: Sabia S, Marmot M, Dufouil C, Singh-Manoux A. Smoking history and cognitive function in middle age from the Whitehall II study. Archives of Internal Medicine 2008; 168:1165-1173).

Data came from the Whitehall II study of over 5,000 British civil servants who were interviewed at baseline in 1985-88 and followed prospectively for approximately 12-17 years. Cognitive function, including memory, reasoning, vocabulary, and semantic and phonetic fluency was assessed in middle-age, when most of the participants were between the ages of 47 and 72.

Compared to never smokers, current smokers were significantly more likely to have memory impairment (odds ratio = 1.5) at follow-up. This effect remained (odds ratio = 1.4) significant after controlling for other health behaviors and health measures.

Before controlling for other health behaviors and health measures, current smoking was associated with a significant decrement in reasoning, vocabulary, and phonemic and semantic fluency. However, all of these effects disappeared after controlling for the other health behaviors and health measures (which included socioeconomic status, education, marital status, physical activity, nutrition, alcohol use, and other medical problems, such as stroke, high blood pressure, heart disease, high cholesterol and diabetes).

Long-term ex-smokers were found to have significantly increased memory (odds ratio = 0.8), as well as significantly increased vocabulary, phonemic fluency, and semantic fluency, even after controlling for other health behaviors and health conditions.

The study concludes that "smoking in middle age is associated with memory deficit" and that "long-term ex-smokers are less likely to have cognitive deficits in memory, vocabulary, and verbal fluency."

In a Health Day news article about the study, the lead author was quoted as concluding: "Our results suggest that smoking had an adverse effect on cognitive function."

In addition, a scientific consultant to the American Lung Association was quoted as stating that the study either shows that smoking makes people stupid or that stupid people choose to smoke. According to the article: "Dr. Norman H. Edelman, scientific consultant to the American Lung Association, said: 'This study is generally a confirmation of previous work. But there is a fundamental question: Are they stupid because they smoke or do they smoke because they are stupid?'"

The Rest of the Story

Before concluding that this study demonstrates an effect of smoking on memory in middle age, I think it is important to consider two major limitations of the study.

First, the study finds significant impairment of other cognitive functions in current smokers - reasoning, vocabulary, and verbal fluency - which disappears after controlling for other health behaviors and health conditions. This demonstrates that there is substantial confounding in the study. Although the results for memory remained significant after controlling for the other variables in the model, the presence of strong confounding suggests that there may be other factors, not measured in the study, which could explain the relationship between smoking and lower memory function scores.

The second limitation, which is more troubling, is the study's finding that ex-smokers actually performed better in memory, vocabulary, and verbal fluency tests. Clearly, one would not conclude that smoking for a short period of time increases cognitive ability in these areas. Instead, what these results demonstrate is that there is a strong selection effect present in these data. Those smokers who make a decision to quit and do so successfully represent a very different population from smokers who continue to smoke. And the factors which separate these two populations are associated with precisely the outcome variables measured in the study: cognitive function.

The authors are quick to dismiss their finding that ex-smokers had better cognitive function by arguing that they must have had other health behaviors that led to better cognitive function. The idea that the smoking caused the increased cognitive function is not entertained. However, for the finding that current smokers had impaired memory, the study does not similarly dismiss the finding by arguing that they must have also had other health behaviors that led to worse cognitive functioning.

The problem is that this is a very biased interpretation of these data. If one is going to invoke other unmeasured behavioral factors to explain the positive findings for ex-smokers, then one must also be willing to entertain the hypothesis that these unmeasured factors also explain the negative findings for current smokers.

Once one acknowledges that there are unmeasured factors which are affecting the study results, I don't think one can conclude that those unmeasured factors are only affecting the positive results for ex-smokers, but not the negative results for current smokers.

In fact, what I believe is going on here is most likely a selection effect. What the study is doing is examining two different populations of ever smokers: one which makes the decision to quit and does so successfully and the other which either decides not to quit or tries to quit unsuccessfully. These populations are quite different and most importantly, they are likely to differ on precisely the variables that would be expected to affect cognitive function. The reported results for long-term ex-smokers confirm this hypothesis. Because this is a plausible alternative explanation for the study findings regarding the observed impairment in memory among current smokers, the validity of the study's conclusion is thrown into question.

That ends the epidemiology portion of this commentary. But I cannot conclude without saying something about Dr. Edelman's comment that these results either show that smoking makes people stupid or that it is stupid people who decide to smoke.

First of all, even if the study conclusion is valid, it does not demonstrate that smoking makes people stupid. Having a measurable decline in memory functioning does not equate to being "stupid." Using such a term is not only unscientific and inaccurate, but it is degrading to smokers and frankly, offensive.

Second, the study does not show that it is stupid people who start to smoke. What it likely shows is that those who continue to smoke have other behavioral or other factors that are associated with deficits in memory. This doesn't mean that those people are stupid. We do know that smoking is associated with lower levels of education, in general. But again, that doesn't equate to people being "stupid." Once again, I find Dr. Edelman's comment to be unscientific, inaccurate, degrading, and offensive.

No comments: