significant level with new onset of major depressive
episodes, dysthymia, manic episodes,
generalized anxiety disorder, panic disorder,
specific phobias, social phobia, and PTSD
(Table 3). Among participants aged 50 years or
older, smoking was only associated with new
onset of manic episodes (Table 3). In addition,
the test for trend in the association of smoking
and dysthymia was statistically significant
among participants aged 50 years or older,
although the overall test was not statistically
significant (Table 3).
These findings corresponded with an overall
decreasing rate of new-onset mood or anxiety
disorders associated with increasing age. A total
of 3181 (14.2%) of the 21 164 participants
aged 18 to 49 years experienced new onset of
mood or anxiety disorders; by contrast, only
1268 (8.9%) of 13 489 participants aged 50
years or older reported a new mood or anxiety
disorder (OR = 1.69; 95% CI = 1.56, 184;
P < .001). We observed a similar pattern for each
individual mood and anxiety disorder, with ORs
ranging from 1.51 for dysthymia to 2.42 for
manic episodes (all, P < 0.01; data not shown).
DISCUSSION
Despite declining rates, smoking remains
prevalent in the United States and other countries.
Almost 1 out of every 5 NESARC
participants reported smoking cigarettes regularly
at some point during the 3-year follow-up.
Thus, smoking remains a major public health
problem. We found a clear association between
regular smoking and new-onset mood and
anxiety disorders. This finding is consistent
with results from other longitudinal studies.
Our use of instrumental variable
models and longitudinal design further supports
a causal link. Instrumental variable analyses
with state cigarette taxes and public
attitudes as instruments did not detect significant
endogeneity for smoking in these models,
suggesting that the coefficients from naive regression
models could be interpreted as consistent
estimators of the effects of smoking on
new-onset mental disorders.
We also found a significant moderating
effect of age on the relationship between
smoking and new onset of mood and anxiety
disorders. The association was significant in
participants aged 18 to 49 years but mainly
nonsignificant in participants aged 50 years or
older. The reasons for the moderating effect of
age on the association of smoking with newonset
mood and anxiety disorders remain unclear.
One possible explanation for this moderating
effect is the existence of a risk window,
during which individuals are especially vulnerable
to exposures.49---51This interpretation is
supported by findings of decreased incidence of
mood and anxiety disorders in older adults in
our study and others52---54 and findings of
strong associations between smoking and new
onset of mood and anxiety disorders in adolescents. Nevertheless, this explanation
remains speculative. Future research should
explore this and other possible explanations, such
as cohort effects and greater mortality among
smokers with mood and anxiety disorders.
Previous research provides few clues to
possible biological mechanisms linking smoking
and mental illness. Agonists of nicotinic
cholinergic receptors (including nicotine itself)
presumably improve cognition and mood.55---56
Indeed, these effects of nicotine are the basis
for the self-medication hypothesis regarding
the association of smoking with mental disorders.
However, it has also been hypothesized
that chronic administration of cholinergic
agents may lead to indirect inhibition of the
nicotinic receptors (functional antagonism) and
hence contribute to increased prevalence of
depression. Furthermore, the mechanisms
linking smoking with mental illness may vary
across individuals according to their genetic
makeup.58 The mechanisms may also vary by
type of mental disorder. In view of the
findings regarding the moderating effect of age,
exploring the potential impact of biological
factors such as hormonal changes associated
with age may help to elucidate the link between
smoking and mental illness.