Abstract
Aims
We aimed to investigate how strongly smoking dependence and smoking dependence motives are associated with depressive symptoms among daily smokers and if these associations are independent of measured confounders and shared familial factors.
Design
Cross-sectional individual-based and within-pair analyses.
Setting
Fourth wave of the population-based Finnish Twin Cohort conducted in 2011.
Participants
918 daily smokers born 1945-57 (48% men), mean age 59.5 years including 38 twin pairs discordant for depression.
Measurements
Depressive symptoms were assessed using the Center for Epidemiologic Studies Depression scale with a cut off value ≥20 for depression. Smoking dependence was assessed using the Fagerström Test for Cigarette Dependence (FTCD) and smoking dependence motives with three subscales from the multi-dimensional Brief Wisconsin Inventory of Smoking Dependence Motives (WISDM): Primary Dependence Motives (PDM), Affective Enhancement (AE), and Taste (T). Logistic regressions, using standardized scores of independent variables and adjusted for multiple confounders with correction for sampling as twin pairs, were used in the individual-based analyses. Conditional logistic regression was used to control for shared familial factors in discordant twin pairs.
Findings
Prevalence of depression was 18% (n=163: 61 [14%] in men, n=102 [22%] in women). Higher smoking dependence measured by the FTCD (OR 1.45; 95% CI 1.20, 1.75), and dependence motives measured by the PDM (1.56; 1.30, 1.87) and the AE (1.54; 1.28, 1.85) were associated with higher odds of depression. The associations remained after adjusting for individual confounders, except for neuroticism, which attenuated all associations. FTCD, PDM, and AE showed associations with depression within depression-discordant monozygotic pairs, suggesting an association independent of familial factors.
Conclusions
Depression is associated with smoking dependence and smoking dependence motives related to heavy, automatic use and use to regulate affective states. The associations appear to be confounded or mediated by neuroticism but are independent of shared familial influences.
Keywords: Tobacco smoking, Dependence, Depression, Mental health, Motive, Public Health
Introduction
The association of smoking with depression is widely reported in cross-sectional and longitudinal studies (1-5). However, several other factors, such as living alone and low education, are associated with both smoking and depression (6-8). Furthermore, familial factors increase vulnerability for both phenotypes (9, 10).
Even though the causality and direction of the association between smoking and depression are still unclear (1, 11), several mechanisms underlying the association have been suggested (12, 13). One possible mechanism is the self-medication model, whereby smoking is used to alleviate symptoms, particularly those of negative affect (14), i.e. depressed mood predisposing to smoking initiation and development of nicotine dependence (1). Alternatively, long-term smoking predisposes to the development of depression (15). Notably, the observed associations may be confounded by familial influences, including shared genetic vulnerability (13).
Recent Mendelian randomization studies indicate that smoking is independently associated with depression (11). However, genetic vulnerability for depression also predisposed participants to long-term smoking in the same analyses, albeit with weaker evidence (11). Therefore, genetic confounding in the association between smoking and depression remains a challenge. This makes twin studies powerful tools since they control unmeasured familial confounding in epidemiological studies (16), allowing for deeper understanding of complex associations, and further helps developing prevention and treatment interventions.
Although the association of smoking with depression has been widely studied, less is known about the associations of smoking dependence and smoking dependence motives with depression. For instance, dependence is multi-faceted phenomenon, and it is unknown how depression is associated with global measures of dependence versus specific dependence dimensions. Dependence can be assessed using several measures which have been shown to be related to key dependence criteria (17-21). Smoking dependence is widely assessed using the Fagerström Test for Cigarette Dependence (FTCD) (17, 18), which assesses smokers’ level or severity of dependence by tapping heaviness of smoking and tolerance.
The value of multidimensional measures such as the Nicotine Dependence Syndrome Scale (NDSS) (21) and the Wisconsin Inventory of Smoking Dependence Motives (20), is that they reflect the influence of multiple dependence facets or processes that are associated with classic dependence criteria: relapse vulnerability, nicotine self-administration, and withdrawal magnitude (22-24). In particular, two major WISDM dimensions, i.e. primary (PDM) and secondary (SDM) dependence motives have been meaningfully, differentially associated with FTCD (25). The PDM reflects a pattern of heavy, automatic smoking resulting in a perceived loss of control over use and high levels of craving and is particularly associated with intransigent tobacco use (e.g., relapse). The SDM reflects the instrumental use of smoking to achieve certain purposes (e.g., coping with negative affect, experience of pleasure) and appears to develop with less tobacco exposure than does PDM. Two validated SDM subscales were used in this study; Affective Enhancement (AE) assessing smoking to cope with negative affect and enhance positive emotional states and Taste assessing smoking for taste reward (20, 24, 26). PDM, AE and Taste differ in their relations with key aspects of the phenotypes such as withdrawal severity (27) and are also known to differ based on their genetic associations with smoking (28, 29).
Previously, associations of smoking dependence or smoking dependence motives with depression have been investigated in a limited number of studies using mainly unidimensional measures (1). We are not aware of studies addressing the association between multi-dimensional dependence measures and depression among adults, though depression has been associated with the multi-dimensional NDSS among adolescents (30).
Aims
To investigate how smoking dependence measured by the uni-dimensional FTCD and smoking dependence motives measured by three subscales from the multi-dimensional WISDM, are associated with depressive symptoms in daily smokers.
To investigate if the associations are independent of measured confounders and non-measured shared familial factors including genetic influences.
Methods
Participants
Individual-based data came from the fourth survey of the Finnish Twin Cohort conducted in 2011 (participation rate 72%, n=8,410, 55% (4,657) women, mean age 60 years, born in 1945-57) (31). In 2011, 18% (n=1,475) of the participants reported current daily cigarette smoking. Main analyses were performed for 918 individuals (48% men) following listwise deletion method (Figure 1a). Supplemental analyses were conducted with all participants who had enough information on depressive symptoms (n=1,432 [97%]; 708 [49%] men. Smoking behaviour was assessed similarly in each survey as described in detail elsewhere (32).
Figure 1a.
Flow chart of samples included in the individual-based analyses.
Within-pair data came from responses of 198 same-sexed twin pairs (54% female), where both co-twins were daily smokers; 105 were dizygotic (DZ) and 84 monozygotic (MZ, i.e., genetically identical) pairs (Figure 1b). Nine pairs with uncertain zygosity were excluded from pairwise analyses. Of the 189 pairs, 38 pairs were discordant for depression (18 MZ and 20 DZ) with complete data.
Figure 1b.
Flow chart of twin pairs included in the within-pair analyses.
Dependent variable
The dependent variable was depression (yes/no), assessed with the Center for Epidemiologic Studies Depression (CES-D) scale (33). A total sum score (range from 0 to 60) of current depressive symptoms was calculated from 20 items (scored 0-3) reflecting major facets of depression: depressed mood, feelings of guilt and worthlessness, feelings of helplessness and hopelessness, psychomotor retardation, loss of appetite, and sleep disturbance. CES-D sum distribution was skewed with a floor effect (Figure S1). A cut off value ≥20 was used to classify a participant with clinically relevant level of depressive symptoms (34). Those with depression (CES-D ≥20) were compared to those with no depression (CES-D <20).
Independent variables
We considered two dependence measures; smoking dependence assessed with the Fagerström Test for Cigarette Dependence (FTCD) and smoking dependence motives assessed with three subscales from the multi-dimensional brief WISDM (26): Primary Dependence Motives (PDM), Affective Enhancement (AE), and Taste. PDM was used as a latent profile for the core of smoking dependence and is constructed from four WISDM subscale scores (Automaticity, Craving, Loss of Control, and Tolerance) based on 16 items (four items/subscale). Automaticity reflects tendency to smoking automatically and without conscious awareness or intention, whereas Craving assesses frequency and severity of urges to smoke and smoking in response to cravings. The Loss of Control subscale assesses the perceived loss of volitional control over smoking behaviour. Lastly, Tolerance assesses needing to smoke more to experience desired effects or ability to smoke heavily without adverse consequences. AE (three items) reflects smoking to ameliorate negative internal states or enhance positive feeling states and consists of both negative and positive smoking reinforcement motives from the original WISDM-68 scale whereas Taste (three items) reflects smoking to experience the gustatory sensations associated with smoking (20). While the FTCD attempts to measure dependence largely by assessing its downstream consequences, the WISDM does so by assessing the motivational drives and effects produced by it.
To manage overall participant burden, the current study included 22 items from the total 37 Brief WISDM items. Each statement was rated with a Likert scale ranging from 1 (Not true of me at all) to 7 (Extremely true of me). A value for each subscale (Automaticity, Craving, Loss of Control, Tolerance, Affective Enhancement, and Taste) was calculated as a mean of included items per subscale and the PDM is the mean of the Automaticity, Craving, Loss of Control, and Tolerance subscales (26). The features of these dependence phenotypes (PDM, AE and Taste) have been defined during development of the WISDM and validated with several independent samples (22, 24, 26). The exact full text questions for items included in each subscale have been presented in the Appendix of the article by Smith et al. (2010) (26). The item numbers are as follows: PDM: Automaticity (1,10,14,25); Craving (4,7,23,29); Loss of Control (2,16,21,35); Tolerance (3,28,31,36), and SDM: AE (9,33,37), and Taste (5,15,20).
Participants also completed the global dependence measure FTCD with a sum score of responses to six items that include heaviness of smoking and latency to smoke in the morning (range 0 to 10) (17, 18).
Confounders
Relevant variables associated with depression or smoking were chosen from the literature (6-8, 35) and considered as potential confounders (Table S1). All potential confounders were tested separately for each independent variable (FTCD, PDM, AE, Taste) and depression. To be included as a confounder, a variable had to be associated (p<0.15) with both the depression and a primary independent variable (PDM, AE, FTCD) (36). As shown in Table S1, several confounders were included in the analyses in addition to age and sex. The questionnaires including all variables used in the Finnish Twin Cohort 2011 survey are available online (37). Categorization of confounders, including amount of missing data are shown in Table 1. Validation and origin of each variable are shown in Table S1.
Table 1.
Characteristics of the study variables among all current daily cigarette smokers with sufficient information on their depression status (n=1,432) by CES-D categorization in the Finnish Twin Cohort in 2011.
| Confounder / independent variable |
Depression status | ||
|---|---|---|---|
| Non-depressed CES-D <20 N=1,145 |
Depressed CES-D ≥20 N=287 |
Inferential statistic a |
|
| n (%) | n (%) | p-value | |
| Sex | |||
| Man | 596 (52.1) | 112 (39.0) | 7.90e-05 |
| Woman | 549 (47.9) | 175 (61.0) | |
| Missing | 0 | 0 | |
| Marital status | |||
| Married or living together | 714 (62.4) | 137 (47.7) | 7.43e-05 |
| Single | 160 (14.0) | 54 (18.8) | |
| Divorced or widowed | 263 (23.0) | 95 (33.1) | |
| Missing | 8 (0.7) | 1 (0.4) | |
| Working status | |||
| At work | 618 (54.0) | 79 (27.5) | 1.10e-14 |
| Not at work | 524 (45.8) | 207 (72.1) | |
| Missing | 3 (0.3) | 1 (0.4) | |
| Self-rated health | |||
| Excellent/very good | 688 (60.1) | 75 (26.1) | 7.76e-33 |
| Good | 384 (33.5) | 140 (49.8) | |
| Fair/poor | 63 (5.5) | 70 (24.4) | |
| Missing | 10 (0.9) | 2 (0.7) | |
| Leisure time physical activity a | |||
| Active (>1.5 MET-h/day) | 581 (50.7) | 118 (41.1) | 7.03e-03 |
| Inactive (≤1.5 MET-h/day) | 542 (47.3) | 159 (55.4) | |
| Missing | 22 (1.9) | 10 (3.5) | |
| Number of pass outs due to alcohol in the last year | |||
| Never | 960 (83.8) | 210 (73.2) | 1.07e-06 |
| 1-3 times | 133 (11.6) | 52 (18.1) | |
| At least 4 times | 25 (2.2) | 21 (7.3) | |
| Missing | 27 (2.4) | 4 (1.4) | |
| Sleep length | |||
| 7 – <9 hours | 772 (67.4) | 148 (51.6) | 9.63e-06 |
| <7 hours | 290 (25.3) | 110 (38.3) | |
| ≥9 hours | 78 (6.8) | 26 (9.1) | |
| Missing | 5 (0.4) | 3 (1.1) | |
| Sleep quality | |||
| Well | 323 (28.2) | 31 (10.8) | 6.88e-24 |
| Rather well | 639 (55.8) | 134 (46.7) | |
| Rather poorly | 138 (12.1) | 81 (28.2) | |
| Poorly | 33 (2.9) | 35 (12.2) | |
| Missing | 12 (1.1) | 6 (2.1) | |
| Life satisfaction | |||
| Satisfied | 274 (23.9) | 7 (2.4) | 3.5e-109 |
| Intermediate | 773 (67.5) | 85 (29.6) | |
| Dissatisfied | 98 (8.6) | 194 (67.6) | |
| Missing | 0 (0) | 1 (0.4) | |
| Emotional support | |||
| Strong support | 1,041 (90.9) | 207 (72.1) | 6.37e-17 |
| Some support | 61 (5.3) | 42 (14.6) | |
| Not any support | 37 (3.2) | 37 (12.9) | |
| Missing | 6 (0.5) | 1 (0.4) | |
| Negative life events (weighted sum score) | |||
| Quartile 1 (lowest) | 368 (32.1) | 31 (10.8) | 1.98e-20 |
| Quartile 2 | 186 (16.2) | 31 (10.8) | |
| Quartile 3 | 251 (21.9) | 56 (19.5) | |
| Quartile 4 (Highest) | 306 (26.7) | 148 (51.6) | |
| Missing | 34 (3.0) | 21 (7.3) | |
| Involved in a serious traffic accident in adulthood * | |||
| No | 866 (75.6) | 195 (67.9) | 2.91e-02 |
| Yes | 86 (7.5) | 28 (9.8) | |
| Missing | 193 (16.9) | 64 (22.3) | |
| Injured by physical assault in adulthood * | |||
| No | 823 (71.9) | 152 (53.0) | 1.46e-10 |
| Yes | 144 (12.6) | 77 (26.8) | |
| Missing | 178 (15.6) | 58 (20.2) | |
| Victim of sexual assault in childhood * | |||
| No | 1,028 (89.8) | 216 (75.3) | 2.12e-12 |
| Yes | 19 (1.7) | 24 (8.4) | |
| Missing | 98 (8.6) | 47 (16.4) | |
| Victim of sexual assault in adulthood * | |||
| No | 906 (79.1) | 176 (61.3) | 8.08e-15 |
| Yes | 55 (4.8) | 51 (17.8) | |
| Missing | 184 (16.1) | 60 (20.9) | |
| Victim of or witness to a violent crime in adulthood * | |||
| No | 927 (81.0) | 207 (72.1) | 9.18e-04 |
| Yes | 39 (3.4) | 21 (7.3) | |
| Missing | 179 (15.6) | 59 (20.6) | |
| Parental divorce or separation in childhood * | |||
| No | 958 (83.7) | 221 (77.0) | 3.00e-02 |
| Yes | 86 (7.5) | 30 (10.5) | |
| Missing | 101 (8.8) | 36 (12.5) | |
| Dichotomy of traumatic life events (*) c | |||
| None | 615 (53.7) | 90 (31.4) | 5.46e-10 |
| At least one | 237 (20.7) | 89 (31.0) | |
| Missing info in 1- 5 traumas | 277 (24.2) | 101 (35.2) | |
| Missing info in all 6 traumas | 16 (1.4) | 7 (2.4) | |
| Mean (SD) | Mean (SD) | ||
| Age (years, range 53.3 – 67.3) | 59.5 (3.5) | 59.2 (3.6) | 2.05e-01 |
| Missing, n (%) | 0 (0) | 0 (0) | |
| Education by 1981 (years, range 3 - 16) b | 8.0 (2.5) | 7.6 (2.3) | 2.41e-02 |
| Missing, n (%) | 2 (0) | 1 (0) | |
| Alcohol consumption (drinks [12 g alcohol]/day, range 0 - 22.6) | 1.4 (2.1) | 1.7 (2.9) | 6.85e-01 |
| Missing, n (%) | 2 (0) | 1 (0) | |
| Number of cigarettes (per day, range 0 - 40) | 14.0 (7.4) | 15.5 (8.4) | 3.35e-01 |
| Missing, n (%) | 4 (0) | 2 (1) | |
| Neuroticism (sum score, range 0 - 10) | 3.0 (2.2) | 6.6 (2.1) | 5.0e-108 |
| Missing, n (%) | 30 (3) | 7 (2) | |
| FTCD (sum score, range 0 - 10) | 3.8 (2.3) | 4.6 (2.4) | 4.67e-08 |
| Missing, n (%) | 13 (1) | 1 (0) | |
| WISDM PDM Motives (sum, range 16 - 112) | 55.3 (23.6) | 67.4 (26.8) | 3.00e-13 |
| Missing, n (%) | 39 (3) | 13 (5) | |
| WISDM AE Motives (sum, range 3 - 21) | 10.9 (4.8) | 13.3 (5.1) | 6.43e-14 |
| Missing, n (%) | 14 (1) | 6 (2) | |
| WISDM Taste Motives (sum, range 3 - 21) | 11.6 (4.7) | 12.2 (5.3) | 9.35e-02 |
| Missing, n (%) | 13 (1) | 6 (2) | |
The Center for Epidemiological Studies-Depression (CES-D), The Wisconsin Inventory of Smoking Dependence Motives (WISDM), WISDM Primary Dependence Motives (PDM), WISDM Affective Enhancement Motives (AE), WISDM Taste Motives (T), Fagerström Test for Cigarette Dependence (FTCD), Metabolic Equivalent (MET).
Pearson Chi-square test for categorical variables and two-sample t-test for continuous variables.
The latest information in data from 1981; corresponding information from the 1975 survey was used for those with missing information on their education in 1981.
A dichotomous variable (none vs at least one) of traumatic life events includes those traumatic life events that were related to depression and smoking dependence/motives (*).
Statistical analyses
Missing data
For the main analyses, we imputed some missing data in the CES-D, WISDM and FTCD, but no missing information were allowed in any of the independent variables or confounders. In CES-D, no more than two missing items were allowed, and those values were imputed by the mean of the remaining items. In the WISDM measure, one missing value for each subscale (Automaticity, Craving, Loss of Control, Tolerance, AE or Taste) was allowed, a missing item being imputed as mean of the non-missing values within a subscale. In the FTCD, one missing value was allowed and replaced with a mean of non-missing values. Due to missing values in the confounders, we created a ‘missing’ category for the categorical variables which we used in one part of sensitivity analyses.
Individual-based analyses
Men and women were analysed together as there were no interactions between sex and any independent variable on depression in the Likelihood-ratio tests (p-values >0.3) (38). Descriptive analyses were performed with Pearson Chi-square or two-sample t-test by depression status. The WISDM and FTCD scores were standardized to mean of 0 and SD of 1 for comparisons. Logistic regression (Odds Ratio [OR] with 95% Confidence Intervals [CI]) was used in analysing the associations of PDM, AE, Taste, or FTCD with depression. Basic regression analyses were adjusted for sex and age (Model 1). In Model 2, the effect of a variable of interest (PDM, AE, Taste, or FTCD) was tested after adjusting for sex, age, and a single confounder. To allow comparison, the same confounders were included for all four independent variables. In Model 3, all WISDM subscales (PDM, AE, and Taste) were jointly entered in the model with age and sex to reveal orthogonal relations within the various subscales (25). In addition, two multiple adjusted analyses (Model 4 without neuroticism and Model 5 with neuroticism) were performed including all confounders showing <0.5 correlation with each other. In all analyses, a cluster option was used to control for the effect of sampling twins as pairs on standard errors of the estimates (39).
Within-pair analyses
Conditional logistic regression was used to control for the effect of familial factors in twin pairs concordant for smoking, i.e. both co-twins were daily smokers, but discordant for depression, i.e. one twin was depressed (CES-D ≥20) and the other was not (CES-D<20), with information on FTCD, all WISDM items and relevant confounders (Figure 1b). Analyses were performed MZ/DZ pairs combined and separately for DZ and MZ pairs with a test for heterogeneity between zygosity using the Likelihood-ratio test. Implementing and interpreting within-pair analyses are described in Text S1. Within-pair analyses were adjusted for CPD to control for differences in amount of smoking between co-twins, which might account for some within-pair discrepancies in smoking dependence or dependence motives. Neuroticism was adjusted because of substantial confounding in the individual-based analyses, unlike the other tested confounders.
Sensitivity analyses
Sensitivity analyses are described and reported in the Text S2 and Tables S3-S5.
Stata SE versions 15/16 (StataCorp, College Station, Texas, USA) were used for all the analyses.
Results
Individual-based analyses
Characteristics of studied variables by depression status in the whole sample (n=1,432) are shown in Table 1. Among the 918 daily smokers (48% men), the prevalence of depression (CES-D ≥ 20) was 18 % (n=163: 61 men [14%]; 102 women [22%]) in 2011.
In age- and sex-adjusted analyses, we found statistically significant associations of FTCD, PDM, and AE with depression and the associations remained significant after adjusting for most of the confounders (Table 2). The associations of PDM and AE with depression also remained significant when all three WISDM dimensions were simultaneously included in the model (Model 3). PDM and AE, but not FTCD, were statistically significantly associated with depression when all confounders except neuroticism were simultaneously included in the models (Model 4). The variable confounding the associations most was neuroticism, which strongly attenuated the associations of FTCD, PDM and AE with depression (Models 2 and 5). Associations of each confounder with depression by each dependence measure are shown in the Table S2. Sensitivity analyses demonstrate that findings were robust over different choices for handling missing data (Tables S3-S4).
Table 2.
Associations of standardized values of Fagerström Test for Cigarette Dependence (FTCD) and Brief Wisconsin Inventory of Smoking Dependence Motives (WISDM) measures with depression (CES-D ≥20) among current daily cigarette smokers (n=918 with 163 cases) in the Finnish Twin Cohort. Analyses were performed with logistic regression providing odds ratios (OR) with 95% confidence intervals (CI) including observations without missing values in any variables. In all individual-based analyses, a robust variance estimator was used to adjust for the non-independence of observations within twin pairs. Associations with p-values <0.05 are in bold.
| Individual- based analyses |
Confounders | FTCD | WISDM Dimensions | ||
|---|---|---|---|---|---|
| PDM | AE | Taste | |||
| OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | ||
| Model 1 a | 1.45 (1.20, 1.75) | 1.56 (1.30, 1.87) | 1.54 (1.28, 1.85) | 1.01 (0.85, 1.21) | |
| Model 2 b | Education | 1.44 (1.20, 1.74) | 1.55 (1.29, 1.87) | 1.54 (1.28, 1.85) | 1.02 (0.85, 1.22) |
| Marital status | 1.42 (1.17, 1.72) | 1.54 (1.28, 1.85) | 1.52 (1.26, 1.83) | 1.01 (0.85, 1.20) | |
| Working status | 1.35 (1.11, 1.63) | 1.47 (1.22, 1.80) | 1.49 (1.24, 1.79) | 1.01 (0.85, 1.21) | |
| Self-rated health | 1.19 (0.98, 1.45) | 1.33 (1.10, 1.60) | 1.37 (1.13, 1.66) | 0.99 (0.83, 1.19) | |
| LTPA | 1.42 (1.17, 1.71) | 1.53 (1.27, 1.85) | 1.52 (1.26, 1.82) | 1.01 (0.85, 1.21) | |
| Alcohol consumption | 1.44 (1.19, 1.74) | 1.55 (1.29, 1.86) | 1.53 (1.27, 1.84) | 1.01 (0.84, 1.21) | |
| Pass outs due to alcohol | 1.39 (1.15, 1.68) | 1.50 (1.25, 1.80) | 1.48 (1.23, 1.78) | 0.97 (0.81, 1.17) | |
| Number of cigarettes | 1.51 (1.19, 1.93) | 1.61 (1.29, 2.01) | 1.50 (1.23, 1.83) | 0.95 (0.79, 1.15) | |
| Sleep length | 1.41 (1.17, 1.71) | 1.54 (1.28, 1.85) | 1.50 (1.25, 1.81) | 1.00 (0.84, 1.19) | |
| Sleep quality | 1.41 (1.16, 1.72) | 1.53 (1.27, 1.85) | 1.49 (1.23, 1.80) | 1.03 (0.85, 1.24) | |
| Life satisfaction | 1.19 (0.97, 1.46) | 1.33 (1.09, 1.62) | 1.41 (1.13, 1.76) | 0.93 (0.75, 1.14) | |
| Emotional support | 1.38 (1.14, 1.66) | 1.51 (1.25, 1.81) | 1.51 (1.25, 1.82) | 1.03 (0.85, 1.24) | |
| Negative life events | 1.41 (1.16, 1.16) | 1.51 (1.25, 1.82) | 1.46 (1.21, 1.77) | 0.99 (0.82, 1.20) | |
| Traffic accident in adulthood * | 1.45 (1.20, 1.75) | 1.56 (1.29, 1.87) | 1.53 (1.28, 1.84) | 1.01 (0.85, 1.21) | |
| Physical assault in adulthood * | 1.40 (1.16, 1.69) | 1.51 (1.26, 1.81) | 1.49 (1.24, 1.80) | 1.00 (0.84, 1.20) | |
| Sexual abuse in childhood * | 1.45 (1.20, 1.75) | 1.57 (1.30, 1.88) | 1.51 (1.25, 1.81) | 1.01 (0.84, 1.21) | |
| Sexual abuse in adulthood * | 1.40 (1.15, 1.69) | 1.53 (1.27, 1.84) | 1.52 (1.25, 1.84) | 1.02 (0.85, 1.22) | |
| Victim/witness of violent crime in adulthood * | 1.43 (1.19, 1.72) | 1.54 (1.28, 1.85) | 1.50 (1.25, 1.81) | 1.03 (0.86, 1.23) | |
| Parental divorce in childhood * | 1.44 (1.20, 1.74) | 1.56 (1.30, 1.88) | 1.54 (1.29, 1.85) | 1.01 (0.85, 1.21) | |
| Dichotomy of traumatic life events (none/at least one) ** | 1.39 (1.15, 1.68) | 1.52 (1.26, 1.83) | 1.51 (1.26, 1.82) | 1.01 (0.85, 1.21) | |
| Neuroticism | 1.13 (0.93, 1.39) | 1.12 (0.91, 1.38) | 1.09 (0.88, 1.35) | 0.97 (0.78, 1.20) | |
| Model 3 c | 1.45 (1.09, 1.93) | 1.51 (1.13, 2.03) | 0.66 (0.52, 0.83) | ||
| Model 4 d | 1.09 (0.77, 1.54) | 1.42 (1.06, 1.92) | 1.36 (1.02, 1.82) | 0.87 (0.66, 1.14) | |
| Model 5 e | 0.98 (0.68, 1.40) | 1.16 (0.84, 1.62) | 1.10 (0.81, 1.50) | 0.82 (0.60, 1.12) | |
Fagerström Test for Cigarette Dependence (FTCD); Brief Wisconsin Inventory of Smoking Dependence Motives (WISDM); Primary Dependence Motives (PDM), Affective Enhancement (AE); Taste (T); OR=Odds Ratio; CI=Confidence Interval; n=number of Individuals with Depression in the Analysis by Dependence Motive; LTPA=leisure time physical activity. Depression defined by the Center for Epidemiologic Studies Depression (CES-D) scale with a cut-off value of 20 (Depression CESD ≥20).
Model 1: Individual models for FTCD and each WISDM dimension, adjusted for age and sex (basic model).
Model 2: Individual models for FTCD and each WISDM dimension, adjusted for age, sex and separately for each significant covariate.
Model 3: A multiple adjusted model with all three WISDM dimensions, adjusted for age and sex.
Model 4: Individual multiple adjusted models for FTCD and each WISDM dimension, with all confounders except neuroticism. All traumatic life events * related to depression and smoking dependence/motives were added to the model as a dichotomous variable ** (none vs at least one traumatic life event).
Model 5: Individual multiple adjusted models for FTCD and each WISDM dimension with all confounders (as in the model 4) including also neuroticism.
The Taste dimension behaved differently from the PDM and AE in that it showed a lack of association with depression (Table 2), even before including confounders. Including Taste in the model along with the other smoking dependence motives revealed that higher Taste scores were associated with a reduced depression likelihood.
Within-pair analyses
Altogether there were 38 smoking concordant but depression discordant twin pairs (18 MZ, 20 DZ) (Figures 1b, S2, S3). In the conditional logistic regression analyses, FTCD, PDM, and AE showed statistically significant associations with depression when MZ and DZ pairs were combined (Table 3). The associations of FTCD, PDM, and AE with depression were evident also in MZ pairs with similar effect sizes (Text S1). Adjusting within-pair differences for CPD did not fundamentally change these associations except for PDM in MZ pairs where the association became stronger. When adjusted for neuroticism all associations attenuated.
Table 3.
Within-pair associations (Odds Ratios [OR] with 95% confidence intervals [CI]) of standardized mean values of Smoking Dependence (FTCD) and dimensions from the Brief Wisconsin Inventory Dependence Motives (WISDM) with depression (CES-D ≥20) a among twin pairs discordant for depression, where both co-twins are current daily cigarette smokers with no missing values in any factors of interest or covariates in the Finnish Twin Cohort. Associations are presented by zygosity and for all pairs of known zygosity.
| Within pair analyses | np | FTCD OR (95% CI) b |
WISDM Dimensions | ||
|---|---|---|---|---|---|
| PDM OR (95% CI) b |
AE OR (95% CI) b |
Taste OR (95% CI) b |
|||
| BASIC MODEL c | |||||
| DZ and MZ pairs together | 38 | 1.77 (1.02, 3.08) | 2.30 (1.16, 4.50) | 1.77 (1.08, 2.89) | 1.35 (0.89, 2.06) |
| DZ pairs | 20 | 1.52 (0.82, 2.81) | 1.62 (0.83, 3.16) | 1.47 (0.85, 2.51) | 1.24 (0.75, 2.06) |
| MZ pairs | 18 | 2.55 (0.87, 7.49) | 30.3 (0.88, 1045) | 3.15 (0.96, 10.3) | 1.61 (0.75, 3.44) |
| LR test for MZ/DZ difference (p-value) | p=0.396 | p=0.023 | p=0.208 | p=0.573 | |
| MODEL 1 d | |||||
| DZ and MZ pairs together | 38 | 2.03 (0.91, 4.55) | 2.87 (1.13, 7.32) | 1.81 (0.99, 3.31) | 1.24 (0.78, 1.97) |
| DZ pairs | 20 | 1.45 (0.55, 3.85) | 1.54 (0.59, 3.98) | 1.35 (0.68, 2.69) | 1.09 (0.61, 1.95) |
| MZ pairs | 18 | 3.62 (0.86, 15.3) | 55.6 (1.21, 2552) | 3.55 (0.95, 13.2) | 1.54 (0.69, 3.46) |
| LR test for MZ/DZ difference (p-value) | p=0.448 | p=0.029 | p=0.207 | p=0.489 | |
| MODEL 2 e | |||||
| DZ and MZ pairs together | 38 | 1.14 (0.59, 2.23) | 1.51 (0.70, 3.24) | 1.35 (0.79, 2.31) | 1.16 (0.68, 2.00) |
| DZ pairs | 20 | 1.19 (0.56, 2.54) | 1.21 (0.56, 2.62) | 1.28 (0.70, 2.32) | 1.00 (0.49, 2.03) |
| MZ pairs | 18 | 0.99 (0.21, 4.57) | 23.2 (0.48, 1119) | 1.83 (0.46, 7.25) | 1.43 (0.62, 3.39) |
| LR test for MZ/DZ difference (p-value) | p=0.828 | p=0.094 | p=0.723 | p=0.526 | |
Fagerström Test for Cigarette Dependence (FTCD); Brief Wisconsin Inventory of Smoking Dependence Motives (WISDM); Primary Dependence Motives (PDM), Affective Enhancement (AE); Taste (T); OR=Odds Ratio; CI=Confidence Interval; DZ=Dizygotic; MZ=Monozygotic; np=Number of Twin Pairs Discordant for Depression; LR test=Likelihood-ratio test for difference between MZ and DZ pairs.
Depression defined by the Center for Epidemiologic Studies Depression (CES-D) scale with a cut-off value of 20 (Depression CES-D ≥20).
Conditional logistics regression odds ratios with 95% confidence intervals. Statistically significant associations (p<0.05) are in bold.
Basic model: Age and sex are automatically adjusted in the model (data includes only same-sexed twin pairs).
Model 1: Adjusted with age, sex and number of cigarettes smoked per day.
Model 2: Adjusted with age, sex and neuroticism.
Discussion
Our analyses showed that FTCD, PDM, and AE are associated with depression. Our findings also indicate that in daily smokers, depression is related to core dependence processes, such as craving and automatic smoking arising from a history of heavy smoking, as well as from the instrumental use of smoking for affect regulation. Taste and sensory motivations to smoke are not associated with depression. The associations for AE and PDM but not for FTCD remained after adjusting for multiple confounders except neuroticism, which alone attenuated all associations. Further, based on the MZ pairs, associations of FTCD, PDM, and AE with depression appear to be independent of familial factors. Notably, all within-pair effects were independent of smoking heaviness. Therefore, our analyses add to the evidence of the robust association of two different measures of dependence (FTCD and WISDM) with depression, not explained by smoking quantity .
While FTCD and PDM both reflect heavy smoking (40), in our study, the OR estimates were higher for PDM relative to FTCD in MZ pairs, although with overlapping confidence intervals. This may be because the PDM captures a wider array of information about the “core” features of dependence – e.g., only the PDM measures automaticity, craving, and loss of control. Also, the FTCD may tap unintended environmental features that do not sensitively index dependence per se (e.g., contextual factors that prohibit or permit heavy smoking). Interestingly, all the PDM subscales accounted about for similar amounts of total variability for depression in the sensitivity analyses. However, whether the PDM total value or any of the PDM sub scores alone can be used to examine relations between smoking dependence motives and depression, needs further investigation.
The localization of the effects to PDM and AE indicates that the mechanisms linking depression and dependence may operate by promoting heavy and automatic smoking to suppress cravings and smoking to cope with affective distress, supporting previous findings (14). However, this does not mean that other SDM motives do not play any role. But, amongst SDM scales, the AE motive may be especially relevant due to the need to reduce negative moods associated with depression. Regardless of which SDM subscales might contribute to depression related smoking, their influence may be most apparent early in the development of dependence. Namely, SDM motives appear to influence smoking early in its development (23, 24), but escalating smoking due to SDM motives may then stimulate heavier smoking, which augments PDM.
Our individual-based and within-pair results consistently suggest that depressed persons are not smoking because of taste or taste-reward related to smoking. Smokers, who have taste-related motives for their smoking behaviour, may comprise a specific subgroup of daily smokers behaving differently from others in relation to mood. Rather, depression seems to be related to primary dependence motives and to affect regulation. This finding is also supported by a study in which positive and negative reinforcement subscales used from the original WISDM-68 predicted depression level in smoking college students (41). Vinci et al. (41) also reported that friendship-like attachment to smoking (affiliative attachment) explained significant variance of depression level in college smokers. Unfortunately, we did not include the Affiliative Attachment subscale in our questionnaire in 2011.
Our analyses highlight the role of neuroticism in the association between dependence and depression. Neuroticism is robustly related to depression and anxiety and may account for the comorbidity of many mood disorders (42). There is also a strong genetic correlation between neuroticism and major depression (43). According to our analyses, when neuroticism is used as a confounder, a major portion of the smoking dependence motives-depression relationship is attenuated. This may occur because the statistical control of depression via neuroticism may greatly reduce meaningful information in the depression variable (44) or neuroticism may play a role as a mediator between smoking and depression.
Finally, our analyses yield new evidence about the associations between addictions and depression, since this is one of the first studies using both uni- and multi-dimensional measures of smoking dependence in the study of depression. Although depression is more common among adult women (35, 45) and depressed women are at higher risk for nicotine dependence (46), we did not find sex-interactions in the associations of depression with smoking dependence or smoking dependence motives. Therefore, based on our data, the association between depression and such dependence is not moderated by sex.
Strengths and limitations
This study has several strengths. First, the data are from a population-based cohort with a high participation rate. Second, we controlled for multiple potential confounders. Third, a twin sample enables co-twin control design, which controls for familial confounding on the associations. Thus, the generalization of our findings is high among a population of European ancestry with 53 – 67 years of age.
This study has also weaknesses. First, smoking status is based on self-reports without biochemical verification. Second, we used a cross-sectional design, since the FTCD, WISDM and the CES-D measures were available only in one survey. Therefore, without temporal order, we cannot draw conclusions regarding the direction of a causal pathway between dependence and development or worsening of depression. A longitudinal incidence analysis is needed to test if depression leads to dependence among baseline non-dependent smokers or if dependence leads to depression among baseline non-depressed smokers. Third, as we studied only three WISDM dimensions, associations between other smoking dependence motives and depression need further investigation. Fourth, multiple imputation was not used in all variables to replace missing observations, which might have some influence on the findings. However, zero missingness in the CES-D items in our main analysis sample (n=918: 94%) did not differ significantly from the whole survey sample (n=8,410: 90%) in 2011. Further, sensitivity analyses show that our results are not significantly affected by the number of observations included in the analyses. On the other hand, we cannot rule out a possibility of overfitting our multiple adjusted models. Notably, our sample is restricted to adults with 53 – 67 years of age (47, 48). Further, the observed associations may be underestimated since both dependent smokers and seriously depressed persons likely had higher attrition due to premature death or other reasons (49-51). However, we acknowledge that in the Finnish Twin Cohort, the general participation rates have been relatively high (72 — 89%) compared to many other population-based cohort surveys (47). Finally, although the original surveys included thousands of twin pairs (48), our pairwise analyses include only 18 MZ pairs discordant for depression but concordant for daily smoking and thus lack optimal statistical power resulting in large confidence intervals. However, pairwise analyses are powerful in controlling for familial confounding on the associations and thus our results deepen knowledge of the relationships between dependence and depression.
Conclusion
Depression and smoking dependence are associated with one another independently of shared familial influences, a relationship found with FTCD and two WISDM dimensions, with the latter capturing motives related to heavy, out of control automatic smoking, with high levels of craving, and smoking to regulate affective states. These associations are robust with adjustment for several single or a combination of confounders, except neuroticism which attenuates the association. As for clinical implications, higher dependence and smoking to regulate affect should be considered while helping depressed smokers to quit their habit.
Supplementary Material
Acknowledgements
We acknowledge support by the Academy of Finland (grant #309119 to TK; #308248 and #312073 to JK), Sigrid Juselius Foundation to JK, NHLBI R01 HL109031-06 to MEP. We thank Dr Vilja Eskola for her preliminary work related to the WISDM variables.
Footnotes
Conflict of interest statement: Dr. Korhonen has consulted for Pfizer on nicotine dependence 2011-2017. No other conflict of interest.
Pre-registration of the research question
The primary research question and analysis plan of this study have not been pre-registered on a publicly available platform.
Contributor Information
Maarit Piirtola, Institute for Molecular Medicine Finland (FIMM), University of Helsinki, PO. Box 20, 00014 University of Helsinki, Helsinki, Finland.
Jaakko Kaprio, Institute for Molecular Medicine Finland (FIMM), University of Helsinki, PO. Box 20, 00014 University of Helsinki, Helsinki, Finland; Department of Public Health, University of Helsinki, Po. Box 20, 20014 University of Helsinki, Helsinki, Finland.
Timothy B. Baker, Department of Medicine, University of Wisconsin School of Medicine and Public Health, 1930 Monroe Street, Madison, WI 53711 -2059, United States.
Thomas M. Piasecki, Department of Psychological Sciences, University of Missouri, 210 McAlester Hall, Columbia, MO 65211, United States.
Megan E. Piper, Department of Medicine, University of Wisconsin School of Medicine and Public Health, 1930 Monroe Street, Madison, WI 53711 -2059, United States.
Tellervo Korhonen, Institute for Molecular Medicine Finland (FIMM), University of Helsinki, PO. Box 20, 00014 University of Helsinki, Helsinki, Finland.
References
- 1.Fluharty M, Taylor AE, Grabski M, Munafo MR The Association of Cigarette Smoking With Depression and Anxiety: A Systematic Review, Nicotine Tob Res 2017: 19: 3–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Luger TM, Suls J, Vander Weg MW How robust is the association between smoking and depression in adults? A meta-analysis using linear mixed-effects models, Addict Behav 2014: 39: 1418–1429. [DOI] [PubMed] [Google Scholar]
- 3.Taylor AE, Fluharty ME, Bjørngaard JH, Gabrielsen ME, Skorpen F, Marioni RE, … Munafò MR Investigating the possible causal association of smoking with depression and anxiety using Mendelian randomisation meta-analysis: the CARTA consortium, BMJ Open 2014: 4: e006141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Goodwin RD, Zvolensky MJ, Keyes KM, Hasin DS Mental Disorders and Cigarette Use among Adults in the United States, American Journal on Addictions 2012: 21: 416–423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Korhonen T, Broms U, Varjonen J, Romanov K, Koskenvuo M, Kinnunen T, Kaprio J Smoking behaviour as a predictor of depression among Finnish men and women: a prospective cohort study of adult twins, Psychol Med 2007: 37: 705–715. [DOI] [PubMed] [Google Scholar]
- 6.World Health Organization. Addiction to Nicotine. In: Samet JM & Yoon S-Y, editors. Gender, women, and the tobacco epidemic: World Health Organization; 2010. [Google Scholar]
- 7.Köhler CA, Evangelou E, Stubbs B, Solmi M, Veronese N, Belbasis L, … Carvalho AF Mapping risk factors for depression across the lifespan: An umbrella review of evidence from meta-analyses and Mendelian randomization studies, J Psychiatr Res 2018: 103: 189–207. [DOI] [PubMed] [Google Scholar]
- 8.Kearns NT, Carl E, Stein AT, Vujanovic AA, Zvolensky MJ, Smits JAJ, Powers MB Posttraumatic stress disorder and cigarette smoking: A systematic review, Depress Anxiety 2018: 35: 1056–1072. [DOI] [PubMed] [Google Scholar]
- 9.Liu M, Jiang Y, Wedow R, Li Y, Brazel DM, Chen F, … Vrieze S Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use, Nat Genet 2019: 51: 237–244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Howard DM, Adams MJ, Clarke TK, Hafferty JD, Gibson J, Shirali M, … McIntosh AM Genome-wide meta-analysis of depression identifies 102 independent variants and highlights the importance of the prefrontal brain regions, Nat Neurosci 2019: 22: 343–352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Wootton RE, Richmond RC, Stuijfzand BG, Lawn RB, Sallis HM, Taylor GMJ, … Munafò MR Causal effects of lifetime smoking on risk for depression and schizophrenia: Evidence from a Mendelian randomisation study, 2018, 10.1101/381301/JbioRxiv: 381301. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Mathew AR, Hogarth L, Leventhal AM, Cook JW, Hitsman B Cigarette smoking and depression comorbidity: systematic review and proposed theoretical model, Addiction 2017: 112: 401–412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Royal College of Physicians, Royal College of Psychiatrists. Smoking and mental health. Royal College of Psychiatrists Council Report London; 2013. [Google Scholar]
- 14.Markou A, Kosten TR, Koob GF Neurobiological similarities in depression and drug dependence: a self-medication hypothesis, Neuropsychopharmacology 1998: 18: 135–174. [DOI] [PubMed] [Google Scholar]
- 15.Plurphanswat N, Kaestner R, Rodu B The Effect of Smoking on Mental Health, Am J Health Behav 2017: 41: 471–483. [DOI] [PubMed] [Google Scholar]
- 16.McGue M, Osler M, Christensen K Causal Inference and Observational Research: The Utility of Twins, Perspect Psychol Sci 2010: 5: 546–556. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Heatherton TF, Kozlowski LT, Frecker RC, Fagerström KO The Fagerström Test for Nicotine Dependence: a revision of the Fagerström Tolerance Questionnaire, Br J Addict 1991: 86: 1119–1127. [DOI] [PubMed] [Google Scholar]
- 18.Fagerström K Determinants of tobacco use and renaming the FTND to the Fagerström Test for Cigarette Dependence, Nicotine Tob Res 2012: 14: 75–78. [DOI] [PubMed] [Google Scholar]
- 19.American Psychiatric Association D.-T. F. Diagnostic and statistical manual of mental disorders: DSM-5™ Arlington, VA, US: : American Psychiatric Publishing, Inc.. 2013. [Google Scholar]
- 20.Piper ME, Piasecki TM, Federman EB, Bolt DM, Smith SS, Fiore MC, Baker TB A multiple motives approach to tobacco dependence: the Wisconsin Inventory of Smoking Dependence Motives (WISDM-68), J Consult Clin Psychol 2004: 72: 139–154. [DOI] [PubMed] [Google Scholar]
- 21.Shiffman S, Waters A, Hickcox M The nicotine dependence syndrome scale: a multidimensional measure of nicotine dependence, Nicotine Tob Res 2004: 6: 327–348. [DOI] [PubMed] [Google Scholar]
- 22.Piasecki TM, Piper ME, Baker TB Refining the tobacco dependence phenotype using the Wisconsin Inventory of Smoking Dependence Motives: II. Evidence from a laboratory self-administration assay, J Abnorm Psychol 2010: 119: 513–523. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Piasecki TM, Piper ME, Baker TB, Hunt-Carter EE WISDM primary and secondary dependence motives: associations with self-monitored motives for smoking in two college samples, Drug Alcohol Depend 2011: 114: 207–216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Piper ME, Bolt DM, Kim SY, Japuntich SJ, Smith SS, Niederdeppe J, … Baker TB Refining the tobacco dependence phenotype using the Wisconsin Inventory of Smoking Dependence Motives, J Abnorm Psychol 2008: 117: 747–761. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Piasecki TM, Piper ME, Baker TB Tobacco Dependence: Insights from Investigations of Self-Reported Smoking Motives, Curr Dir Psychol Sci 2010: 19: 395–401. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Smith SS, Piper ME, Bolt DM, Fiore MC, Wetter DW, Cinciripini PM, Baker TB Development of the Brief Wisconsin Inventory of Smoking Dependence Motives, Nicotine Tob Res 2010: 12: 489–499. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Baker TB, Piper ME, Schlam TR, Cook JW, Smith SS, Loh WY, Bolt D Are tobacco dependence and withdrawal related amongst heavy smokers? Relevance to conceptualizations of dependence, J Abnorm Psychol 2012: 121: 909–921. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Chen LS, Baker TB, Grucza R, Wang JC, Johnson EO, Breslau N, … Bierut LJ Dissection of the phenotypic and genotypic associations with nicotinic dependence, Nicotine Tob Res 2012: 14: 425–433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cannon DS, Baker TB, Piper ME, Scholand MB, Lawrence DL, Drayna DT, … Leppert MF Associations between phenylthiocarbamide gene polymorphisms and cigarette smoking, Nicotine Tob Res 2005: 7: 853–858. [DOI] [PubMed] [Google Scholar]
- 30.Dierker L, Rose J, Selya A, Piasecki TM, Hedeker D, Mermelstein R Depression and nicotine dependence from adolescence to young adulthood, Addict Behav 2015: 41: 124–128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kaprio J The Finnish Twin Cohort Study: an update, Twin Res Hum Genet 2013: 16: 157–162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Kaprio J, Koskenvuo M A prospective study of psychological and socioeconomic characteristics, health behavior and morbidity in cigarette smokers prior to quitting compared to persistent smokers and non-smokers, J Clin Epidemiol 1988: 41: 139–150. [DOI] [PubMed] [Google Scholar]
- 33.Radloff LS The CES-D Scale:A Self-Report Depression Scale for Research in the General Population, Applied Psychological Measurement 1977: 1: 385–401. [Google Scholar]
- 34.Vilagut G, Forero CG, Barbaglia G, Alonso J Screening for Depression in the General Population with the Center for Epidemiologic Studies Depression (CES-D): A Systematic Review with Meta-Analysis, PLoS One 2016: 11: e0155431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Bromet E, Andrade LH, Hwang I, Sampson NA, Alonso J, de Girolamo G, … Kessler RC Cross-national epidemiology of DSM-IV major depressive episode, BMC Med 2011: 9: 90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Heinze G, Wallisch C, Dunkler D Variable selection - A review and recommendations for the practicing statistician, Biom J 2018: 60: 431–449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Twin study at University of Helsinki; older cohort. https://wiki.helsinki.fi/display/twineng/Older+cohort. [Google Scholar]
- 38.Heidari S, Babor TF, De Castro P, Tort S, Curno M Sex and Gender Equity in Research: rationale for the SAGER guidelines and recommended use, Research Integrity and Peer Review 2016: 1: 2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Williams RL A note on robust variance estimation for cluster-correlated data, Biometrics 2000: 56: 645–646. [DOI] [PubMed] [Google Scholar]
- 40.Baker TB, Breslau N, Covey L, Shiffman S DSM criteria for tobacco use disorder and tobacco withdrawal: a critique and proposed revisions for DSM-5, Addiction 2012: 107: 263–275. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Vinci C, McVay MM, Copeland AL, Carrigan MH The relationship between depression level and smoking motives in college smokers, Psychol Addict Behav 2012: 26: 162–165. [DOI] [PubMed] [Google Scholar]
- 42.Barlow DH, Sauer-Zavala S, Carl JR, Bullis JR, Ellard KK The Nature, Diagnosis, and Treatment of Neuroticism:Back to the Future, 2014: 2: 344–365. [Google Scholar]
- 43.Nagel M, Jansen PR, Stringer S, Watanabe K, de Leeuw CA, Bryois J, … Posthuma D Meta-analysis of genome-wide association studies for neuroticism in 449,484 individuals identifies novel genetic loci and pathways, Nat Genet 2018: 50: 920–927. [DOI] [PubMed] [Google Scholar]
- 44.Leventhal AM, Zvolensky MJ Anxiety, depression, and cigarette smoking: a transdiagnostic vulnerability framework to understanding emotion-smoking comorbidity, Psychol Bull 2015: 141: 176–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Wesselhoeft R, Pedersen CB, Mortensen PB, Mors O, Bilenberg N Gender-age interaction in incidence rates of childhood emotional disorders, Psychol Med 2015: 45: 829–839. [DOI] [PubMed] [Google Scholar]
- 46.Komiyama M, Yamakage H, Satoh-Asahara N, Ozaki Y, Morimoto T, Shimatsu A, … Hasegawa K Sex differences in nicotine dependency and depressive tendency among smokers, Psychiatry Res 2018: 267: 154–159. [DOI] [PubMed] [Google Scholar]
- 47.Piirtola M, Kaprio J, Waller K, Heikkilä K, Koskenvuo M, Svedberg P, … Ropponen A Leisure-time physical inactivity and association with body mass index: a Finnish Twin Study with a 35-year follow-up, Int J Epidemiol 2017: 46: 116–127. [DOI] [PubMed] [Google Scholar]
- 48.Kaprio J, Bollepalli S, Buchwald J, Iso-Markku P, Korhonen T, Kovanen V, … Waller K The Older Finnish Twin Cohort — 45 Years of Follow-up, Twin Research and Human Genetics, 10.1017/thg.2019.54: 1–15. [DOI] [PubMed] [Google Scholar]
- 49.Tolonen H, Dobson A, Kulathinal S Effect on trend estimates of the difference between survey respondents and non-respondents: results from 27 populations in the WHO MONICA Project, Eur J Epidemiol 2005: 20: 887–898. [DOI] [PubMed] [Google Scholar]
- 50.Karvanen J, Tolonen H, Härkänen T, Jousilahti P, Kuulasmaa K Selection bias was reduced by recontacting nonparticipants, J Clin Epidemiol 2016: 76: 209–217. [DOI] [PubMed] [Google Scholar]
- 51.Lundberg I, Damström Thakker K, Hällström T, Forsell Y Determinants of non-participation, and the effects of non-participation on potential cause-effect relationships, in the PART study on mental disorders, Soc Psychiatry Psychiatr Epidemiol 2005: 40: 475–483. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.