It's nice reaching a point in your career when you can say something on a website that gets you a personal email reply from Steven D. Levitt, author of Freakonomics.
Steve has a controversial post at the Freakonomics blog: The editor of the Journal of Clinical Psychiatry needs to have his head examined. The post references a recent research report that suggests that smokers may have worse outcomes after traumatic events than non-smokers. Levitt appears to go after some low-hanging fruit in his criticism of the press release, and a portion of his criticism only holds given the information in the press release, but not in the actual article. Citing the original article, the research design appears fairly sound, although not perfect, just as no epidemiologic study is craziness-free.
A part of freakonomics is taking causal pathways assumed true based on correlational evidence and examining the plausibility of the opposite direction. Levitt and Dubner are making quite a popular career (in addition to an impressive academic career, in Levitt's case) of questioning assumptions that certainly deserve questioning. In this case, Levitt suggests that the folks who smoke are probably the ones that are helped the most by smoking in coping. He therefore finds no evidence to suggest that smoking cessation might be beneficial, and reasons that it might even be harmful.
Levitt's points make sense, but they assume that there's no significant difference in the biology of stress regulation between folks with psychiatric trauma symptoms and those who don't have them. Smoking may be a coping mechanism with calming properties in the short run, a potential effect in the long run of further dysregulating the HPA axis could plausibly account for the smoking causing worse symptoms in the smoking group.
Levitt's argument, in whole, seems to be one of biologic plausibility. It is more plausible, given the knowledge of psychology of the average well-educated person, that people who were worse off (or in this case, going to be worse off, as I would argue that effective controls are in place at the first assessment of smoking) were more likely to smoke.
But the biology of trauma, stress, and anxiety, a field outside the knowledge pool of even the average medical student, suggests that PTSD-like symptoms could certainly be modulated by smoking if smoking, a biologically stressful event in itself, modulates the HPA axis in an unhealthy way.
So in summary,
A) Press releases of medical research, in a paraphrase of one commenter, do little more than let you know which author you should be looking for on PubMed.
B) Freakonomics methodology is valuable, but it may be limited when examining fields for which specialist knowledge may be necessary. Biologic plausibility is an important aspect of epidemiologic research.
C) Steven Levitt is gracious in his emails.
D) I'm a total nerd, because I get all excited about getting an email from a University of Chicago economics professor.
