Journal of Psychiatric Research 45 (2011) 143e149

Contents lists available at ScienceDirect

Journal of Psychiatric Research

journal homepage: www.elsevier.com/locate/psychires

Tobacco smoking as a risk factor for depression. A 26-year population-based

follow-up study
Trine Flensborg-Madsen a, *, Mikael Bay von Scholten a, Esben Meulengracht Flachs b,
Erik Lykke Mortensen c, Eva Prescott d, Janne Schurmann Tolstrup a
a
Research Programme on Lifestyle and Health, National Institute of Public Health, University of Southern Denmark, Øster Farimagsgade 5, 1399 København K, Denmark
b
Research Programme on Register-based Research, National Institute of Public Health, University of Southern Denmark, Denmark
c
Department of Environmental Health, Institute of Public Health, and Center for Healthy Aging, University of Copenhagen, Denmark
d
Department of Cardiology, Bispebjerg University Hospital, Copenhagen, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: Background: A key question regarding the documented association between smoking and depression is
Received 4 January 2010 whether it reflects a causal influence of smoking on depression; however, only a limited number of
Received in revised form longitudinal studies exist in the literature, all of which have relatively short time frames. The purpose
5 May 2010
was to prospectively assess the risk of depression according to daily tobacco consumption in a Danish
Accepted 11 June 2010
longitudinal study.
Methods: A prospective cohort study, the Copenhagen City Heart Study (n ¼ 18,146), was analyzed with
Keywords:
up to 26 years of follow-up. It contains three updated data sets on tobacco consumption and potential
Depression
Tobacco smoking
confounding factors. The study population was linked to Danish hospital registers to detect registrations
Epidemiology with depression. Individuals with depression before baseline were excluded.
Cohort study Results: Women smoking more than 10 g of tobacco per day were at significantly increased risk of
depression compared to women who did not smoke. The adjusted risk of depression among women
smoking 11e20 g per day was 1.74 (CI:1.33e2.27) and 2.17 (CI:1.45e3.26) among women smoking more
than 20 g per day. For men, there was an increased risk of depression for those smoking more than 20 g
per day (HR ¼ 1.90; CI:1.05e3.44). All tests for linear trend were significant (all p < 0.01). The estimates
remained significant in sensitivity analyses aiming to eliminate reverse causation, and in analyses based
on a reduced sample without individuals with chronic diseases or psychiatric disorders other than
depression.
Conclusion: The study suggests that smoking is associated with increased risk of developing depression.
This underlines the potentially harmful consequences of smoking for mental health and supports efforts
to prevent and stop smoking.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction
By the year 2020, unipolar depression is estimated to be the
second largest contributor to the disease burden in the world, while
tobacco use is estimated to be the largest single health problem in
terms of mortality and disability (Lopez and Murray, 1998). It is
becoming increasingly clear that an association between smoking
and depression exists, as depression is over represented among
smokers (Haines et al., 1980) and smoking is over represented
among individuals with depression (Chang and Chiang, 2009;
* Corresponding author. Tel.: þ45 39 20 77 77/þ45 35 32 63 59(Direct); fax: þ45
3920 8010.
E-mail address: tfm@niph.dk (T. Flensborg-Madsen).
Fucito and Juliano, 2009). A key question regarding the associa-
tion between smoking and depression is whether it reflects a causal
influence of tobacco on depression. Only a limited number of
longitudinal studies in the existing literature have investigated this,
and in addition, most of these have used time frames of less than 5
years (Breslau et al., 1998; Goodman and Capitman, 2000; Wu and
Anthony, 1999), which may not be sufficient to demonstrate the
effects of tobacco on depression. Among the few studies using
longer time frames, 10 or 11 years were used (Klungsoyr et al.,
2006; Pasco et al., 2008), but these studies did not collect poten-
tial changes in smoking habits during follow-up, and results were
not stratified according to sex, which could be of importance due to
the increased incidence of depression among women (Alonso et al.,
2004; Kessler et al., 2003) in addition to sex-differences in
smoking-patterns (Preston and Wang, 2006).

0022-3956/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jpsychires.2010.06.006
144 T. Flensborg-Madsen et al. / Journal of Psychiatric Research 45 (2011) 143e149
In this epidemiological study we prospectively assessed the risk
of being registered with depression according to daily tobacco
consumption in a Danish longitudinal study with linkage to Danish
hospital registers over a period of 26 years. Information on lifestyle
factors was evaluated three times during the follow-up period.
Updated measures of both smoking and a range of possible con-
founding factors were used, and several sensitivity analyses were
performed in order to avoid bias due to reverse causation, chronic
diseases, or other psychiatric disorders. In addition, our data
material was large enough to stratify the study population by sex.
2. Materials and methods
2.1. Study population
Data from the Copenhagen City Heart Study (CCHS) were used.
The CCHS is an ongoing study conducted in the Danish population
and initiated in 1976, when a random sample of men and women
above 20 years of age and living in the Copenhagen area was invited
to participate. The sample was randomly drawn from the Central
Population Register, by use of the unique personal identification
number, and invited by letter to participate in a physical exami-
nation and answer self-administered questionnaires in the years
1976e78. The number of participants in the baseline study was
14,223, with a participation rate of 74%. This examination was fol-
lowed by two follow-up examinations in the years 1981e83 (CCHS-
II) (number of participants 12,698; response rate 70%) and 1991e94
(CCHS-III) (number of participants 10,135; participation rate 61%).
Both follow-up examinations were supplemented with younger
participants in order to keep the population large and representa-
tive. Detailed descriptions of the study have been published else-
where (Appleyard et al., 1989; Schnohr et al., 2001).
2.2. Assessment of tobacco smoking
Information on amount and type of tobacco smoking was
obtained from the CCHS-IeIII where participants were asked to
describe their smoking habits. In all three waves, CCHS-IeIII, the
average daily consumption of cigarettes, cheroots, cigars, and pipe-
tobacco (weekly consumption of 50 g-packs) was summed to the
total tobacco consumption based on the assumption that the
tobacco content of cigarettes, cheroots, and cigars are 1, 3, and 5 g
respectively (Prescott et al., 1998). Participants were divided into
the following groups:
1) Never smoker
2) Former smoker
3) Smoke 1e10 g of tobacco per day
4) Smoke 11e20 g of tobacco per day
5) Smoke more than 20 g of tobacco per day
In supplementary analyses, former smokers were investigated
and divided into the following groups, according to their status at
each round of CCHS: Former-smokers for 5 years or more; former
smokers for 1e5 years; former smokers for less than 1 year.
2.3. Assessment of depression by linkage to national registers
All individuals invited to the CCHS IeIII were followed by
linkage with Danish registers using the unique personal identifi-
cation number. According to Danish regulations, participants do not
have to give informed consent for the use of the identification
number for linkage with national registers. The Danish Hospital
Discharge Register (Jurgensen et al., 1986) contains information on
dates of hospital admissions and discharge diagnoses from Danish
hospitals since 1976 and the Danish Psychiatric Central Register
(Munk-Jorgensen and Mortensen, 1997) contains information on
dates of hospital admissions and discharge diagnoses from Danish
psychiatric hospitals since 1969. Diagnoses in the registers are
classified according to the World Health Organization’s Interna-
tional Classification of Diseases (ICD) using ICD-8 until 1994, and
ICD-10 from 1994 and onwards (ICD-9 was never used in Denmark).
Individuals registered with an ICD diagnosis of depression in
either the Danish Hospital Discharge Register or the Danish
Psychiatric Central Register were considered to have a depression
at the given time. In this study depression comprised the following
diagnoses: ICD-8: (296.09; 298.09; 300.49; 296.29; 296.89; and
296.99) and ICD-10: (F32 and F33). Only first-time registrations
were used and individuals registered with depression before their
entry into the study (n ¼ 274) were eliminated from the analyses.
2.4. Assessment of potential confounding factors
A number of covariates were considered potential confounders
in the association between smoking and depression: Alcohol
consumption per week (0 drinks, 1e7 drinks, 8e14 drinks, 15e21
drinks, or >21 drinks); Educational level (less than 8 years, 8e10
years, and more than 10 years); Income (three monthly income
groups: lowest, middle, and highest); Marital status (married or
unmarried); Children living at home (none or at least one); Physical
activity in leisure time (almost completely physically passive or light
physical activity <2 h per week, light physical activity 2 to 4 h per
week, exhausting physical activity or regular hard training >4 h per
week); Smoking-related diseases; an individual registered with an
ICD of smoking-related diseases in the Danish Hospital Discharge
Register was considered to have the disease at the time of regis-
tration. Smoking-related diseases comprised the following diag-
noses: Diseases of the circulatory system (ICD-8:393e458, ICD-
10:I00e99), chronic lower respiratory diseases (ICD-8:490e493,
ICD-10:J40e47), or malign neoplasms in respiratory organs and
intrathoracic organs (ICD-8:161e162, ICD-10:C32-34). Smoking-
related diseases were adjusted for as a time-dependent covariate
with the status for each individual changing with first registration.
2.5. Statistical analyses
Characteristics of the study population are shown for subjects
who participated in the second wave of the study in 1981e83 (CCHS-
II) (Tables 1 and 2). The analyses used to estimate risks of depression
were Cox proportional hazard regression (Collett, 2008), and age
was used as the underlying time variable. Subjects were followed
from their date of entry, to the date of the first registration of
depression, death, disappearance, emigration, or until end of follow-
up (January 2002) e whichever occurred first. Information on time
of death was obtained from the Danish Causes of Death Register (Juel
and Helweg-Larsen, 1999). The assumption of proportional hazards
was tested for smoking variables and no violations were detected.
Although no significant interaction was found between sex and
smoking (p ¼ 0.60), all estimates were stratified according to sex due
to possible interaction with other covariates.
Smoking and the potential confounding variables were
measured repeatedly over time in 1976e78, 1981e83, and
1991e1993. The risk of depression was assessed in-between wave
increments based on variables derived from the preceding ques-
tionnaire. Technically, this means that several observations were
analyzed for each individual who was characterized anew in each of
the consecutive waves and that information from all observation
intervals was pooled as if the information recorded at each interval
was a new observation. In case of missing data the last observation
was carried forward. Results of Cox proportional hazard regression
 T. Flensborg-Madsen et al. / Journal of Psychiatric Research 45 (2011) 143e149 145

Table 1
Characteristics of women according to smoking habits in the second round of the Copenhagen City Heart Study (CCHS-II) 1981e83 (N ¼ 6999).

Never smokers Former smoker 1e10 g of daily 11e20 g of daily >20 g of daily *p-value
tobacco smoking tobacco smoking tobacco smoking

N % N % N % N % N %
Participants 1981 28.3 1284 18.3 1635 23.4 1740 24.9 359 5.1
Average age (years) 57.8 e 58.3 e 56.2 e 52.8 e 55.9 e **<0.0001

Alcohol intake
0 drinks/week 994 50.4 506 39.5 716 43.9 664 38.3 137 38.9 <0.0001
1e7 drinks/week 742 37.6 527 41.1 617 37.9 640 37.0 100 28.4
8e14 drinks/week 160 8.1 175 13.7 208 12.7 271 15.6 53 15.1
15e21 drinks/week 42 2.1 50 3.9 63 3.9 85 4.9 27 7.7
>21 drinks/week 36 1.8 23 1.8 26 1.6 72 4.2 35 9.9

Education
<8 years 909 46.0 533 41.5 799 48.9 838 48.2 163 45.4 0.0002
8e10 years 787 39.8 513 40.0 617 37.7 670 38.5 151 42.1
>10 years 281 14.2 237 18.5 219 13.4 232 13.3 45 12.5

Income
Low 790 41.5 477 38.5 636 40.8 525 31.0 135 39.2 <0.0001
Medium 743 39.1 543 43.8 633 40.6 822 48.5 154 44.8
High 369 19.4 220 17.7 290 18.6 347 20.5 55 16.0

No. of children living at home

None 1600 80.8 1036 80.7 1328 81.3 1315 75.6 302 84.1 <0.0001
1 381 19.2 247 19.3 305 18.7 425 24.4 57 15.9

Marital status
Married 1069 54.0 710 55.4 877 53.7 990 57.0 156 43.4 0.0001
Unmarried 909 46.0 572 44.6 755 46.3 748 43.0 203 56.6

Physical activity during leisure time

Low level 336 17.0 195 15.2 300 18.4 327 18.8 105 29.3 <0.0001
Moderat level 1037 52.4 681 53.0 843 51.6 947 54.4 172 47.9
High level 607 30.6 408 31.8 489 30.0 466 26.8 82 22.8

*X2-test, ** F-test.

Table 2
Characteristics of men according to smoking habits in the second round of the Copenhagen City Heart Study (CCHS-II) 1981e83 (N ¼ 5663).

Never smokers Former smoker 1e10 g of daily 11e20 g of daily >20 g of daily *p-value
tobacco smoking tobacco smoking tobacco smoking

N % N % N % N % N %
Participants 51 11.5 1390 24.6 938 16.6 1731 30.6 953 16.8
Average age 50.3 e 59.1 e 57.3 e 54.8 e 54.3 e **<0.0001

Alcohol intake
0 drinks/week 122 18.9 254 18.4 174 18.7 222 13.0 116 12.2 <0.0001
1e7 drinks/week 232 35.9 423 30.6 300 32.3 481 28.1 184 19.4
8e14 drinks/week 147 22.8 328 23.7 232 25.0 346 20.3 203 21.4
15e21 drinks/week 71 11.0 170 12.3 124 13.4 256 15.0 140 14.8
>21 drinks/week 74 11.4 208 15.0 98 10.6 403 23.6 305 32.2

Education
<8 years 206 31.7 616 44.4 443 47.3 839 48.6 431 45.2 <0.0001
8e10 years 248 38.2 522 37.6 316 33.7 621 35.9 352 36.9
>10 years 195 30.1 250 18.0 178 19.0 268 15.5 170 17.9

Income
Low 147 22.8 377 27.3 306 32.9 437 25.5 165 17.4 <0.0001
Medium 310 48.1 647 46.8 419 45.1 883 51.6 498 52.6
High 187 29.1 357 25.9 205 22.0 393 22.9 283 30.0

No. of children living at home

None 461 70.9 1122 80.7 752 80.3 1333 77.0 719 75.4 <0.0001
1 189 29.1 268 19.3 185 29.7 398 23.0 234 24.6

Marital status
Married 436 67.0 1072 77.1 684 73.0 1255 72.6 680 71.4 <0.0001
Unmarried 215 33.0 318 22.9 253 27.0 473 27.4 272 28.6

Physical activity during leisure time

Low level 71 10.9 199 14.3 134 14.3 293 16.9 219 23.0 <0.0001
Moderat level 267 41.1 574 41.3 404 43.1 774 44.8 393 41.2
High level 312 48.0 617 44.4 399 42.6 662 38.3 341 35.8

*X2-test, ** F-test.
146 T. Flensborg-Madsen et al. / Journal of Psychiatric Research 45 (2011) 143e149

are shown in Tables 3 and 4 in which the association between
smoking and depression is adjusted for age in addition to: a)
alcohol and education; b) all lifestyle covariates; c) change-in-
estimate 5% (in which variables were selected by backward elimi-
nation, removing at each stage the one variable for which removal
caused the smallest change in hazard ratio (HR), provided that the
change in HR was less than 5%); d) Smoking-related diseases as
a time-dependent variable. By assigning the median value to each
group of smokers and treating these as a single continuous variable,
tests of linear trend were conducted among smokers using never-
smokers as the reference (former smokers were not included in this
analysis). Supplementary analyses were conducted to investigate
the subgroups of former smokers.
Sensitivity analyses were conducted in order to investigate bias
due to existing, but unregistered, depression at the time of the
examinations (Table 5). Firstly, all analyses were repeated using
a “time-lag” of three years meaning that three years follow-up time
after each examination (CCHS-IeIII) was eliminated. Secondly, all
analyses were repeated on a restricted sample from the CCHS-II
including only individuals that were very satisfied with life (this
variable was not included in the main analyses because of the
possible correlation between life satisfaction and depression and
because the question was only present in the CCHS-II). In addition,
sensitivity analyses were conducted in order to evaluate the effect
of chronic diseases upon the association between smoking and
depression by excluding individuals who within the past three
years before time of answering the questionnaire were registered
with stroke, acute myocardial infarction, angina pectoris, other
cardiovascular diseases and other diagnoses implying diseases of
a chronic character. The latter includes infectious, endocrinological,
nervous system, chronic lung, gastrointestinal, alcohol-related,
urological, and muscular diseases. To evaluate the effect of previous
or existing co-morbid psychiatric disorders, sensitivity analyses
were conducted in which individuals registered with a psychiatric
diagnosis other than depression before baseline were eliminated.
All analyses were performed using SAS software package SAS 9.1.
3. Results
Of the 23,189 individuals invited to participate in the CCHS-IeIII,
18,146 individuals completed least one of the three questionnaires
and of these, 622 participants (3.4%) developed depression during
follow-up.
was only the case for 12% of the men. Of the men, 17% smoked more
than 20 g per day, which was only the case for 5% of the women
(Tables 1 and 2). The average age of never-smoking women was
generally higher than that of smoking women, while the average
age of never-smoking men generally was lower than that of
smokers. For both sexes there was a tendency towards higher
alcohol intake with higher daily tobacco consumption. For women,
especially the proportion of non-married individuals and individ-
uals with a low level of physical activity was higher among
smokers. For men, especially the proportion of individuals with
a low level of physical activity was higher among smokers (Tables 1
and 2). Cigarettes was the main source of tobacco smoking for both
women (84%) and men (66%), followed by cheroots, cigars, and
pipe-tobacco respectively (data not shown).
3.2. Smoking and risk of depression
Compared to women who did not smoke, the adjusted risk of
being registered with depression among women smoking 11e20 g
per day was 1.74 (95% CI: 1.33e2.27) and it was 2.17 (95% CI:
1.45e3.26) among women smoking more than 20 g per day.
Women who were former smokers had a decreased risk of
depression (HR ¼ 0.71 (95% CI: 0.51e0.99). Adjustment for a variety
of confounders in different models did not change the estimates
notably and all tests of linear trend were significant (p < 0.0001)
(Table 3).
Among men, the adjusted risk of being registered with
depression was 1.66 (95% CI: 0.96e2.88) for those smoking 11e20 g
per day while it was 1.90 (95% CI: 1.05e3.44) for those smoking
more than 20 g per day compared to never-smokers. Former
smokers had approximately the same risk of depression as never-
smokers, and adjustment for confounding factors did not change
any estimates notably. Tests of linear trend were all significant
(p < 0.009) (Table 4).
Analyzing both sexes together showed hazard ratios (change-in-
estimate) of 0.80 (95% CI: 0.61e1.06), 0.90 (95% CI: 0.69e1.19), 1.69
(95% CI: 1.33e2.14) and 1.97 (95% CI: 1.44e2.70) respectively for the
four smoking categories (data not shown). A further analysis of
subgroups of former smokers according to years since smoking
cessation did not show tendencies of a dose-dependent relation-
ship for either women or men (data not shown).
3.3. Sensitivity analyses

3.1. Descriptive results Neither a time window of three years, exclusion of those with
chronic disease, or exclusion of those with psychiatric disorders
In the CCHS-II, women were more likely to be never-smokers before baseline changed the estimates notably (Table 5). Thus, the
than men. Thus, 28% of the women were never-smokers while this risk among women was still significantly increased among those

Table 3
Hazard ratios for depression among women, according to tobacco smoking in the Copenhagen City Heart Study (CCHS).

Tobacco smoking WOMEN

N(cases) Age-adjusted Adjusteda Adjusted b

Adjusted c
Adjusted d

Hazard 95% CI Hazard 95% CI Hazard 95% CI Hazard 95% CI Hazard 95% CI
ratios ratios ratios ratios ratios
Never smoker 2661 (36) 1 e 1 e 1 e 1 e 1 e
Former smoker 1429 (15) 0.73 0.53e1.01 0.71 0.51e0.99 0.75 0.52e1.07 0.71 0.51e0.99 0.69 0.50e0.95
1e10 g* 2357 (28) 0.93 0.68e1.25 0.91 0.67e1.23 0.90 0.64e1.28 0.91 0.67e1.23 0.87 0.65e1.18
11e20 g* 2522 (64) 1.70 1.31e2.22 1.74 1.33e2.27 1.72 1.27e2.33 1.74 1.33e2.27 1.53 1.17e1.99
>20 g* 510 (14) 2.11 1.41e3.14 2.17 1.45e3.26 2.33 1.49e3.65 2.17 1.45e3.26 1.87 1.25e2.79
P for trend <0.0001 <0.0001 <0.0001 <0.0001 0.0001

* Daily.
a
Adjusted for age, alcohol & education.
b
Adjusted for all covariates (age, alcohol, education, income, no. of children living at home, marital status & physical activity during leisure time).
c
Adjusted according to Change-In-Estimate; age, alcohol & education.
d
Adjusted for age & smoking-related diseases.
 T. Flensborg-Madsen et al. / Journal of Psychiatric Research 45 (2011) 143e149 147

Table 4
Hazard ratios for depression among men, according to tobacco smoking in the Copenhagen City Heart Study (CCHS).

Tobacco smoking MEN

N (cases) Age-adjusted Adjusteda Adjustedb Adjustedc Adjustedd

Hazard 95% CI Hazard 95% CI Hazard 95% CI Hazard 95% CI Hazard 95% CI
ratios ratios ratios ratios ratios
Never smoker 1077 (7) 1 e 1 e 1 e 1 e 1 e
Former smoker 1609 (11) 0.92 0.52e1.62 1.00 0.56e1.79 1.00 0.53e1.91 1.00 0.56e1.79 0.89 0.50e1.57
1e10 g* 1385 (13) 0.82 0.43e1.55 0.89 0.46e1.71 0.65 0.29e1.44 0.89 0.46e1.71 0.81 0.43e1.53
11e20 g* 2704 (30) 1.51 0.89e2.56 1.66 0.96e2.88 1.72 0.93e3.19 1.66 0.96e2.88 1.44 0.85e2.45
>20 g* 1560 (17) 1.80 1.03e3.18 1.90 1.05e3.44 2.14 1.11e4.13 1.90 1.05e3.44 1.75 0.99e3.09
P for trend 0.0065 0.0081 0.0037 0.0081 0.0092

* Daily.
a
Adjusted for age, alcohol & education.
b
Adjusted for all covariates (age, alcohol, education, income, no. of children living at home, marital status & physical activity during leisure time).
c
Adjusted according to Change-In-Estimate; age, alcohol & education.
d
Adjusted for age & smoking-related diseases.

smoking more than 10 g per day, and there was still a tendency of
higher risk of depression among men smoking more than 20 g of
tobacco per day. Including only a selected sample of individuals
from the CCHS-II, who were very satisfied with their life, changed
the estimates notably for both women and men as they became
insignificant.
4. Discussion
Our results suggest that individuals who smoke are at increased
risk of being hospitalized with depression, and that the risk
increases with increasing tobacco consumption. Adjusting our
results for various confounding factors did not alter the overall
picture; neither did sensitivity analyses aiming to account for
reverse causation, chronic diseases, or psychiatric disorders.
Other studies, using cross-sectional data, have reported
increased odds of depression in smokers (Hamalainen et al., 2001;
Pasco et al., 2008), and prospective studies, although they are few,
support the findings in the present study of smoking as a risk factor
of depression. In a population based longitudinal Norwegian study
(11 years), a dose-dependent relationship between smoking and
depression was found, with the risk of the heaviest smokers (>20
cigarettes per day) being more than fourfold elevated compared to
the risk of those who had never smoked (Klungsoyr et al., 2006);
a population based retrospective Australian study (10 years) found
almost a doubling of the risk for developing major depressive
disorder among heavy smokers (>20 cigarettes per day) (Pasco
et al., 2008). Both studies used personal interviews to identify
depression.
There are several ways in which the positive association
between smoking and development of depression can be
explained, all of which are most probably relevant. One explanation
is that smoking is actually an important risk factor in the causal
network leading to development of depression. Nicotine use may
increase vulnerability to depression as nicotine influences several
neurochemical systems, such as acetylcholine and catecholamine
systems (Pomerleau and Pomerleau, 1984), which may play an
etiologic role in depression (Carmody et al., 2007). Also, smoking-
induced oxidative stress could explain the causal effect. Tobacco
smoke generates free radicals, causing oxidation of proteins and
other tissue damage (Ozguner et al., 2005), and depression has
been characterized by elevated markers of oxidative stress which
is positively correlated with the severity of the depression
(Khanzode et al., 2003; Yanik et al., 2004). A study of rats found
that exposure to nicotine during adolescence, but not during
adulthood, leads to a depression-like state manifested in
decreased sensitivity to natural reward, enhanced sensitivity to
stress e and anxiety-eliciting situations later in life (Iniguez et al.,
2009). In addition, some longitudinal cohort studies found that the
causality of smoking and depression mainly go in the direction
from smoking to depression, as depression did not significantly
increase the risk of smoking, while conversely smoking increased
the risk of depression (Goodman and Capitman, 2000; Wu and
Anthony, 1999). A second explanation is that the results are due
to reverse causation meaning that individuals registered with
depression had symptoms already when they answered the
questionnaire and that depressed people are more likely to start
smoking. This could be explained by the fact that smoking can
produce an elevation in mood and sense of well-being (Carmody,
1989; Pomerleau and Pomerleau, 1984), and that depressed people
therefore may self-medicate with cigarettes (Hughes, 1988;
Markou et al., 1998). Longitudinal studies have shown this
inverse relationship, in which the presence of depression
increased the risk of smoking progression (Breslau et al., 1993;
Breslau, 1995; Fergusson et al., 2003). However, in the present
study, the sensitivity analyses trying to minimize reverse causa-
tion did not alter the conclusions notably, except for the fact that
the estimates became insignificant when analyses were conducted
on a selected subsample of only satisfied CCHS-III-individuals who
were very satisfied with their lives e and this may simply be
explained by lack of statistical power (Table 5). A third explanation
is that smoking dependence and depression are manifestations of
the same underlying causes. Genetic variation in certain brain
neurotransmitter systems, such as for example the dopaminergic
system may influence both the probability of smoking, by affecting
the self-reinforcing properties of nicotine (Di Chiara and Bassareo,
2007), and the probability of depression (Malhi and Berk, 2007).
The efficacy of bupropion in both helping people quit tobacco and
in the treatment of depression (Foley et al., 2006) may indicate
some commonality between the two on a neurochemical level.
Alternatively, genetic factors may influence personality traits that
may affect the risk of both smoking (Munafo et al., 2007; Elkins
et al., 2006) and depression (Hirschfeld et al., 1983; Carmody
et al., 2007; Jylha et al., 2010, 2009).
There are a number of strengths and potential weaknesses in
our study. The advantages are the prospective design, the large
study population sample, register information on depression, and
several updated measures of smoking and lifestyle covariates.
Obtaining data on smoking prospectively in several follow-ups
meant that data on smoking could be updated with each person
being characterized anew in each of the following examinations.
Thus, we had a unique opportunity to investigate the risk of
admission to a hospital with a diagnosis of depression in relation to
smoking in the previous period.
148 T. Flensborg-Madsen et al. / Journal of Psychiatric Research 45 (2011) 143e149

The advantage of assessing depression from registers is the ease by

0.51e1.73
0.34e1.42

1.03e3.63
0.77e2.51
which the study population can be followed continuously.

95% CI
However, register data can also induce several biases, especially
No psychiatric
misclassification if for example individuals fulfilling criteria for

e
disorder d

7869 (62)
depression were not diagnosed at a Danish hospital; or if individ-

Hazard

0.0039
uals not fulfilling criteria for depression were diagnosed as such.

ratios

0.94
0.69

1.94
1.39
Although all Danish residents have equal access to hospitals and all

1
treatments are free of charge, it is likely that only a fraction of all

0.56e1.86
0.50e1.88

1.00e3.40
0.95e2.97
cases of depression are diagnosed at hospitals - this may particu-
c
No chonic diseases

95% CI

larly be the case for individuals with more co-morbid conditions. In

addition, the use of registers means than the study is based on the
e

diagnoses of clinicians at somatic e and psychiatric hospitals. The

8088 (75)

Hazard

0.0242 conclusions of this study may therefore not be generalizable to all

ratios

1.02
0.97

1.84
1.68

individuals with depression symptoms but may only be used to

1

describe the risk of being admitted to a hospital with a diagnosis of

Only individuals who were very satisfied with their life was included. Only measured from data from the second round of the Copenhagen City Heart Study (1981e83). depression; thus, the depression outcome variable in this study is
0.27e1.98
0.18e1.98
0.42e2.99
0.18e2.37

assumed to have a limited sensitivity and a high specificity.

95% CI

Patients may be affected by depressive symptoms for longer or

Very satisfied

e
individuals b

shorter periods before they are admitted, so in addition to possible

2574 (41)

differences in clinical diagnostic practice, register-based research in

Hazard

0.8449
ratios

depression is also subject to the potential noise and bias that this
0.74
0.60
1.13
0.66
1

period may induce.

Finally, as in all observational studies, there may be unrecog-
0.49e1.86
0.28e1.39
0.73e2.63
0.91e3.59

nised confounding. Such factors predisposing to both smoking and

95% CI

depression could be personality traits such as neuroticism,

a
Time-window

perceived stress, social network, family history of depression, and

7830 (45)

genetics. These factors were not included in the analyses as data

Hazard

0.0156

were not available. Among women, former smokers had

ratios
MEN

0.96
0.62
1.38
1.81

a decreased risk of depression. This suggests that strong mental

1

characteristics of individuals who stop smoking or other lifestyle

0.53e1.05
0.67e1.27
1.25e2.24
1.10e2.86

factors (that would go along with a tobacco-stop) could also be

95% CI

confounding factors in the association between smoking and

No psychiatric

depression. Previous research supports this hypothesis by linking

e
9049 (131)

inability to stop smoking and relapse to smoking with depressive

disorder d

Individuals diagnosed with chronic diseases within three years before study entry has been excluded.
Hazard

symptoms (Covey et al., 1990; Glassman et al., 1988, 1990; Yanik

0.0003
ratios

0.75
0.92
1.68
1.77

et al., 2004).
1

Our study has methodological limitations, but it suggests that

smoking is an important risk factor for depression. The results were
0.46e0.93
0.62e1.17
1.26e2.19
1.43e3.30
c
No chonic diseases

Individuals diagnosed with a psychiatric disorder before study entry has been excluded.

robust to adjustment for a number of confounding factors, and in

95% CI

sensitivity analyses conducted to evaluate the possibility of reverse

e

causation and the potential influence of somatic diseases and of

9288 (154)

<0.0001

existing psychiatric disorders. Thus, the findings underline the

Hazard
Hazard ratios for depression according to tobacco smoking - sensitivity analyses.

ratios

potential harmful consequences of smoking on mental health and

0.66
0.85
1.66
2.17
1

raise the possibility that depression should be included among the

tobacco attributable diseases.
0.25e1.03
0.43e1.41
0.75e2.35
0.74e4.25
95% CI

Contributors
Very satisfied

e
individuals b

3109 (88)

TFM contributed to the conception and design of the study, the

Hazard

0.1919

analysis and interpretation of data, and wrote the first draft of

ratios

0.51
0.78
1.32
1.77

the manuscript. MBS, EMF, ELM, EP and JST contributed to the

1

conception and design of the study, interpretation of data and to

0.49e1.06
0.68e1.37
1.15e2.18
1.26e3.37

critically revising the paper. MBS performed the statistical analyses.

95% CI

JST is the guarantor.

a
Time-window

e
9179 (102)

Role of funding source

WOMEN

Hazard

0.0004

Three year time window.

ratios

0.72
0.97
1.58
2.06

Funding for this study was provided by a research grant from the
1

IMK Almene Fond, grant number: 30206e224. The IMK Almene

Fond had no further role in study design; in the collection; analysis
Tobacco smoking

Former smoker
Never smoker

and interpretation of data; in the writing of the report; and in the

P for trend

decision to submit the paper for publication.

11e20 g*
N (Cases)

1e10 g*

>20 g*
Table 5

* Daily.

Conflict of interest
b
c
d
a

None.
 T. Flensborg-Madsen et al. / Journal of Psychiatric Research 45 (2011) 143e149
Acknowledgements
We thank the steering committee of the Copenhagen City Heart
Study for kindly providing the data necessary for our analyses, and
we thank the IMK Almene Fond for funding this project. Thanks are
due to Vibeke Munk, MA, and Anders Enevold Christensen, Ph.d.,
for help with the manuscript and critical comments.
References
Alonso J, Angermeyer MC, Bernert S, Bruffaerts R, Brugha TS, Bryson H, et al.
Prevalence of mental disorders in Europe: results from the European study of
the epidemiology of mental disorders (ESEMeD) project. Acta Psychiatrica
Scandinavica 2004;420:21e7. Supplementum.
Appleyard M, Hansen AT, Schnohr P, Jensen GN. The Copenhagen City Heart study.
Osterbroundersogelsen. A book of tables with data from the first examination
(1976e78) and a five year follow-up (1981e83). The Copenhagen City Heart
study group. Scandinavian Journal of Social Medicine 1989;41:1e160.
Supplement.
Breslau N. Psychiatric comorbidity of smoking and nicotine dependence. Behavior
Genetics 1995;25:95e101.
Breslau N, Kilbey MM, Andreski P. Nicotine dependence and major depression. New
evidence from a prospective investigation. Archives of General Psychiatry
1993;50:31e5.
Breslau N, Peterson EL, Schultz LR, Chilcoat HD, Andreski P. Major depression and
stages of smoking. A longitudinal investigation. Archives of General Psychiatry
1998;55:161e6.
Carmody TP. Affect regulation, nicotine addiction, and smoking cessation. Journal of
Psychoactive Drugs 1989;21:331e42.
Carmody TP, Vieten C, Astin JA. Negative affect, emotional acceptance, and smoking
cessation. Journal of Psychoactive Drugs 2007;39:499e508.
Chang HH, Chiang TL. Depressive symptoms, smoking, and cigarette price elasticity:
results from a population-based survey in Taiwan. International Journal of
Public Health 2009;54:421e6.
Collett D. Modelling Survival data in Medical research. United States of America:
Chapman & Hall/CRC Press LLC; 2008.
Covey LS, Glassman AH, Stetner F. Depression and depressive symptoms in smoking
cessation. Comprehensive Psychiatry 1990;31:350e4.
Di Chiara G, Bassareo V. Reward system and addiction: what dopamine does and
doesn’t do. Current Opinion in Pharmacology 2007;7:69e76.
Elkins IJ, King SM, McGue M, Iacono WG. Personality traits and the development of
nicotine, alcohol, and illicit drug disorders: prospective links from adolescence
to young adulthood. Journal of Abnormal Psychology 2006;115:26e39.
Fergusson DM, Goodwin RD, Horwood LJ. Major depression and cigarette smoking:
results of a 21-year longitudinal study. Psychological Medicine 2003;33:
1357e67.
Foley KF, DeSanty KP, Kast RE. Bupropion: pharmacology and therapeutic applica-
tions. Expert Review of Neurotherapeutics 2006;6:1249e65.
Fucito LM, Juliano LM. Depression moderates smoking behavior in response to a sad
mood induction. Psychology of Addictive Behaviors 2009;23:546e51.
Glassman AH, Helzer JE, Covey LS, Cottler LB, Stetner F, Tipp JE, et al. Smoking,
smoking cessation, and major depression. Journal of the American Medical
Association 1990;264:1546e9.
Glassman AH, Stetner F, Walsh BT, Raizman PS, Fleiss JL, Cooper TB, et al. Heavy
smokers, smoking cessation, and clonidine. Results of a double-blind,
randomized trial. Journal of the American Medical Association 1988;259:
2863e6.
Goodman E, Capitman J. Depressive symptoms and cigarette smoking among teens.
Pediatrics 2000;106:748e55.
Haines AP, Imeson JD, Meade TW. Psychoneurotic profiles of smokers and non-
smokers. British Medical Journal 1980;280:1422.
149
Hamalainen J, Kaprio J, Isometsa E, Heikkinen M, Poikolainen K, Lindeman S, et al.
Cigarette smoking, alcohol intoxication and major depressive episode in
a representative population sample. Journal of Epidemiology and Community
Health 2001;55:573e6.
Hirschfeld RM, Klerman GL, Clayton PJ, Keller MB. Personality and depression.
Empirical findings. Archives of General Psychiatry 1983;40:993e8.
Hughes JR. Clonidine, depression, and smoking cessation. Journal of the American
Medical Association 1988;259:2901e2.
Iniguez SD, Warren BL, Parise EM, Alcantara LF, Schuh B, Maffeo ML, et al. Nicotine
exposure during adolescence induces a depression-like state in adulthood.
Neuropsychopharmacology 2009;34:1609e24.
Juel K, Helweg-Larsen K. The Danish registers of causes of death. Danish Medical
Bulletin 1999;46:354e7.
Jurgensen HJ, Frolund C, Gustafsen J, Mosbech H, Guldhammer B, Mosbech J.
Registration of diagnoses in the Danish national Registry of patients. Methods
of Information in Medicine 1986;25:158e64.
Jylha P, Mantere O, Melartin T, Suominen K, Vuorilehto M, Arvilommi P, et al.
Differences in neuroticism and extraversion between patients with bipolar I or
II and general population subjects or major depressive disorder patients.
Journal of Affective Disorders; 2010.
Jylha P, Melartin T, Rytsala H, Isometsa E. Neuroticism, introversion, and major
depressive disorderetraits, states, or scars? Depression and Anxiety
2009;26:325e34.
Kessler RC, Berglund P, Demler O, Jin R, Koretz D, Merikangas KR, et al. The
epidemiology of major depressive disorder: results from the National Comor-
bidity Survey Replication (NCS-R). Journal of the American Medical Association
2003;289:3095e105.
Khanzode SD, Dakhale GN, Khanzode SS, Saoji A, Palasodkar R. Oxidative damage
and major depression: the potential antioxidant action of selective serotonin re-
uptake inhibitors. Redox Report 2003;8:365e70.
Klungsoyr O, Nygard JF, Sorensen T, Sandanger I. Cigarette smoking and incidence of
first depressive episode: an 11-year, population-based follow-up study. Amer-
ican Journal of Epidemiology 2006;163:421e32.
Lopez AD, Murray CC. The global burden of disease, 1990e2020. Nature Medicine
1998;4:1241e3.
Malhi GS, Berk M. Does dopamine dysfunction drive depression? Acta Psychiatrica
Scandinavica; 2007:116e24. Supplementum.
Markou A, Kosten TR, Koob GF. Neurobiological similarities in depression and drug
dependence: a self-medication hypothesis. Neuropsychopharmacology
1998;18:135e74.
Munafo MR, Zetteler JI, Clark TG. Personality and smoking status: a meta-analysis.
Nicotine Tob Res 2007;9:405e13.
Munk-Jorgensen P, Mortensen PB. The Danish psychiatric Central register. Dan Med
Bull 1997;44:82e4.
Ozguner F, Koyu A, Cesur G. Active smoking causes oxidative stress and decreases
blood melatonin levels. Toxicology and Industrial Health 2005;21:21e6.
Pasco JA, Williams LJ, Jacka FN, Ng F, Henry MJ, Nicholson GC, et al. Tobacco smoking
as a risk factor for major depressive disorder: population-based study. British
Journal of Psychiatry 2008;193:322e6.
Pomerleau OF, Pomerleau CS. Neuroregulators and the reinforcement of smoking:
towards a biobehavioral explanation. Neuroscience and Biobehavioral Reviews
1984;8:503e13.
Prescott E, Osler M, Hein HO, Borch-Johnsen K, Lange P, Schnohr P, et al. Gender and
smoking-related risk of lung cancer. The Copenhagen Center for Prospective
Population Studies. Epidemiology 1998;9:79e83.
Preston SH, Wang H. Sex mortality differences in the United States: the role of
cohort smoking patterns. Demography 2006;43:631e46.
Schnohr P, Jensen G, Lange P, Scharling H, Appleyard M. The Copenhagen City Heart
study e Østerbroundersøgelsen. Tables with data from the third examination
1991e94. European Heart Journal; 2001:H1e83. Supplement.
Wu LT, Anthony JC. Tobacco smoking and depressed mood in late childhood and
early adolescence. American Journal of Public Health 1999;89:1837e40.
Yanik M, Erel O, Kati M. The relationship between potency of oxidative stress and
severity of depression. Acta Neuropsychiatrica 2004;16:200e3.