Article Text

Download PDFPDF

Smoking uptake is associated with increased psychological distress: results of a national longitudinal study
  1. Kristie N Carter1,
  2. Frederieke S van der Deen1,
  3. Nick Wilson2,
  4. Tony Blakely1
  1. 1Health Inequalities Research Programme, Department of Public Health, University of Otago, Wellington, New Zealand
  2. 2Department of Public Health, University of Otago, Wellington, New Zealand
  1. Correspondence to Kristie Carter, Health Inequalities Research Programme, University of Otago, Wellington, PO Box 7343, Wellington 6242, New Zealand; kristie.carter{at}otago.ac.nz

Abstract

Background There is evidence that smoking is associated with poorer mental health. However, the underlying mechanisms for this remain unclear. We used longitudinal data to assess whether smoking uptake, or failed quit attempts, are associated with increased psychological distress.

Methods Data were used from Waves 3 (2004/05), 5 (2006/07) and 7 (2008/09) of the longitudinal New Zealand Survey of Family, Income and Employment. Fixed-effects linear regression analyses were performed to model the impact of changes in smoking status and quit status (exposure variables) on changes in psychological distress (Kessler 10 (K10)).

Results After adjusting for time-varying demographic and socioeconomic covariates, smoking uptake was associated with an increase in psychological distress (K10: 0.22, 95% CI 0.01 to 0.43). The associations around quitting and distress were in the expected directions, but were not statistically significant. That is, smokers who successfully quit between waves had no meaningful change in psychological distress (K10: −0.05, 95% CI −0.34 to 0.23), whereas those who tried but failed to quit, experienced an increase in psychological distress (K10: 0.18, 95% CI −0.05 to 0.40).

Conclusions The findings provide some support for a modest association between smoking uptake and a subsequent increase in psychological distress, but more research is needed before such information is considered for inclusion in public health messages.

  • Cessation
  • Smoking Caused Disease
  • Socioeconomic status

Statistics from Altmetric.com

Introduction

There is growing evidence that current smokers are more likely than non-smokers to report poorer mental health (eg, depression, depressive symptoms, mental illness, mental disorders and psychological distress).1–5 Similarly, people with poorer mental health are more likely to smoke.6–10 Existing longitudinal studies on smoking and mental health suggest that it is smoking that contributes to adverse mental health outcomes rather than vice versa.11–15

In the four studies that investigated the causal association between smoking and depression, strong evidence was found for the causal pathway from smoking to depression or depressive symptoms (using dichotomous disease classification), with limited evidence for depression being an antecedent of subsequent smoking.12–15 However, it is important to examine the impact of smoking uptake on changes in mental health using a continuous measure of mental health (or distress), rather than dichotomous disease classification, to identify small (but important) changes on the spectrum of mental health over time. Therefore, building on previous research findings,12–15 we proposed the following research question: ‘What is the effect of smoking uptake on psychological distress?’, hypothesising that moving from non-smoking into smoking will lead to an increase in psychological distress (a continuous measure of mental health).

The possible mechanisms that may underlie the pathway from smoking to changes in mental health are still unclear. Nicotine changes neurotransmitter activity in the brain and this may be the mechanism for an increased risk of depression,12 ,16 ,17 but psychological mechanisms may also be relevant. For example, it is only recently that the role of quit attempts within the relationship between smoking and poorer mental health has been considered.1 ,3 ,18 Cross-sectional studies which have considered the role of quit attempts have found high levels of psychological distress among smokers compared with non-smokers and current smokers who tried to quit smoking but failed, compared with ex-smokers or never-smokers.1 ,3 ,18 In a cross-sectional analysis, we found that current smokers who tried to quit smoking in the last 12 months were over 70% more likely to report high to very high levels of psychological distress than long-term ex-smokers.18 In addition, it has been suggested that levels of psychological distress decrease as time since quitting increases, suggesting a mental health benefit of successful quitting.3 Another possible explanation for the association between smoking and mental health may therefore be the increased distress in current smokers who try but fail to quit smoking. Such an increase in distress after a relapse may be explained by psychological mechanisms like feelings of guilt and shame, low self-efficacy and lost self-control.11 ,19–21 Therefore, failed quit attempts may partly explain the apparent relationship between current smoking and poorer mental health. Yet, a recent longitudinal study reported no evidence of an association between quitting smoking and depression or anxiety questions, in a small prospective multinational cohort of smokers.22 Nevertheless, the group of actual successful quitters who were still abstinent after 6 months was, however, small, limiting the statistical power of the analysis, making it impossible to explore if there were any differences between the mental health outcomes of successful quitters and non-quitters. Therefore, we examined this issue further with a second research question: ‘What is the effect of quit attempts on changes in psychological distress?’. We hypothesised that current smokers who tried to quit but failed would show an increase in psychological distress, whereas those who successfully quit would show a decrease in psychological distress.

Methods

Study design

This study used longitudinal data from Waves 3 (2004/05), 5 (2006/07) and 7 (2008/09) from the Survey of Family, Income and Employment (SoFIE; Wave 1–7 data release, V.2), which was conducted from 2002 to 2010 in New Zealand.23 In short, SoFIE is a nationally representative longitudinal survey of the usually resident population living in private households in New Zealand. During annual face-to-face interviews, information was obtained on individual and family factors, such as labour market activity, education, marital status and household income. In Waves 3, 5 and 7 of SoFIE, a detailed health module was included with questions on health-related quality of life, psychological distress, chronic disease and health behaviours, including smoking and alcohol use. The initial SoFIE baseline sample comprised approximately 11 500 responding private households (response rate of 77%) with over 22 000 adults responding in Wave 1 which reduced to just over 19 000 adults in Wave 3, about 18 000 adults in Wave 5 and over 16 000 adults in Wave 7. A total of 17 140 respondents were included in the longitudinal analyses (response rate=77% of Wave 1 respondents). Attrition was higher in Māori or ‘Other ethnicity’ respondents, as well as those with low income or with poorer health status, were more likely to drop out over the seven waves of the study.

Exposure measures

Smoking status

Responses to two questions about cigarette smoking were used to determine the respondents’ smoking status at each wave: never-smoker, ex-smoker or current smoker. A person was classified as never-smoker if he or she had never smoked one or more cigarettes a day on a regular basis; as ex-smoker if he or she reported not regularly smoking (at the time of the interview) but having ‘ever been a regular smoker of one or more cigarettes a day’; and finally as current smoker if he/she reported to ‘regularly smoke one or more tobacco cigarettes per day’. Based on this information, a two-level smoking status variable was created: (i) non-smokers (never-smokers and ex-smokers grouped together) and (ii) current smokers. Non-smokers were used as the reference group, so the fixed-effects regression modelled smoking uptake between waves.

Quit status

The quit status of current smokers and ex-smokers was derived by asking current smokers: ‘In the last 12 months, have you tried to stop smoking altogether?’. Consequently, a three-level quit status variable was created: (i) current smoker who had not tried to quit smoking, (ii) current smoker who had tried to quit smoking but failed and (iii) successful quitter (ie, ex-smokers who had quit in the last 2 years). Current smokers who had not tried to quit were used as the reference group, so the fixed-effects regression modelled quit attempts (successful or unsuccessful) between waves compared with current smokers who did not try to quit.

Outcome measure

The main outcome of interest was changes in the level of non-specific psychological distress. This was measured by the Kessler Psychological Distress Scale (K10),24–26 which is a 10-item questionnaire with questions regarding negative emotional states (eg, feeling nervous, hopeless or worthless) in the 4 weeks preceding the survey. An example of a K10 question is: ‘During the past 4 weeks, about how often did you feel downhearted and depressed?’, with answers on a Likert scale from ‘none of the time’ to ‘all of the time’. The K10 scores range from 10 to 50 and were grouped into four levels for descriptive purposes: low (10–15), moderate (16–21), high (22–29) and very high (30+).24 ,27 Psychological distress was treated as continuous in the regression analyses.

Other variables

In the fixed-effect regression analyses, we controlled for the following time-varying variables: social marital status (partnered (reference group) vs not partnered), labour force status (employed (reference group) vs not employed), and NZiDep which is a measure of individual deprivation for New Zealand using eight questions including ‘In the past 12 months have you been forced to buy cheaper food so that they could pay for other necessities or have you received help in the form of clothes or money from a community organisation?’ (this was coded as: no deprivation factors (least deprived, reference group) vs one or two deprivation factors, and three or more deprivations factors (most deprived)).28

Statistical analysis

All analyses were conducted using individual unit data from Waves 3, 5 and 7 of SoFIE in SAS 8.2 (in the Statistics NZ data laboratory in Wellington). All numbers of participants presented in this paper are rounded to the nearest multiple of five, with a minimum value of five, as per Statistics New Zealand confidentiality protocol. Tabular analyses are conducted on a non-balanced panel (n=17 140) adult original sample members who responded in Waves 3, 5 and/or 7 (about 15 100 responded in all three waves, with an additional 2040 in either Waves 3 and 5 or Waves 5 and 7). The first longitudinal analysis, of smoking status, included all people with non-missing psychological distress and smoking data (n=17 140; approximately 47 000 observations across the three waves). The second analysis, of quitting smoking, included respondents who were current or ex-smokers at Wave 3 (n=7950; approximately 22 000 observations).

To explore changes in smoking and quit status over time, transition tables of smoking status and quit status summing transitions from Wave 3 to Wave 5, with transitions from Wave 5 to Wave 7 were used. Fixed-effects linear regression models were used to investigate the longitudinal association between: (i) two-level smoking status (exposure) and psychological distress (outcome) and (ii) three-level quit status (exposure) and psychological distress (outcome). The fixed-effects linear regression explores the effect of a change in smoking status (from non-smoking to smoking), or a change in quit status (from current smoker who did not try to quit into current smoker who tried to quit but failed (failed quit attempt) or ex-smoker (successful quit attempt)) on change in psychological distress score between Waves 3, 5 and 7. Model 1 presents the crude model. The fully adjusted model adjusted for time-varying social marital status (partnered, not partnered) and socioeconomic variables (labour force status and NZiDep).

Results

Of the 17 140 respondents included in the longitudinal analyses at Wave 3, 52% were never-smokers, 28% ex-smokers (successful quitters), 7% current smokers who had tried to quit in the past year, and 11% current smokers who had not tried to quit (table 1). Cross-sectionally (at Wave 3), the mean K10 score was 13.4 with a SD of 4.6. High and very high levels of psychological distress were more common in current smokers (11%; mean=14.6, SD=5.7) than in ex-smokers (5%; mean=13.3, SD=4.5) or never-smokers (5%; mean=13.0, SD=4.1). There were much higher levels of psychological distress in current smokers who tried to quit smoking (15%), compared with current smokers who did not try to quit (9%). Furthermore, females, sole parents, not-partnered respondents, respondents who were not employed but actively looking for work, most deprived individuals, and respondents who rated their own health as poor were more likely to report high to very high levels of psychological distress (table 1).

Table 1

Demographic and socioeconomic variables at Wave 3 by Kessler 10 Group (levels of psychological distress)

Turning to the time-varying results over the study period, 660 respondents started smoking and 810 successfully quit smoking. table 2 presents transition matrices of smoking status (A) and quit status (B) over time. There was relative stability in smoking status between waves with 76% of current smokers remaining current smokers in the next wave (2 years later). However, of those current smokers, 13% became ex-smokers (successfully quit) by the next wave. Table 2A also shows that 5% of ex-smokers relapsed and reported being current smokers in the next waves, and 1% of never-smokers start smoking by the next wave. Table 2B presents transitions in the quit status in respondents who were current or ex-smokers in Wave 3. Of those respondents who were current smokers but tried to quit in the past 12 months, 16% had successfully quit (become ex-smokers) by the next wave. Of those respondents who were current smokers but had not tried to quit in the past 12 months, 19% were still current smokers but had tried to quit smoking, and 12% had successfully quit (became ex-smokers). Table 2 also highlights the small number of non-sense transitions (changes from current or ex-smoker to never-smoker, about 1% of all observations) which are common to survey data with large samples. 

Table 2

Transition probability matrices for A: smoking status (n=17 140) and B: quit status (n=7950—current and ex-smokers at Wave 3)

The pooled regression (fully adjusted) model (not controlling for within-individual correlations) estimated a highly significant and strong effect of smoking on K10 (0.65, 95% CI 0.55 to 0.76). However, this analysis does not control for time-invariant confounding. The fixed-effects linear regression analysis of smoking status and psychological distress suggested that respondents who took up smoking between waves experienced a small (but significant) increase in psychological distress (Model 1: increase in K10 score: 0.25, 95% CI 0.04 to 0.47), which persisted in the fully adjusted model (Model 2: K10: 0.22, 95% CI 0.01 to 0.43) (table 3). The smoking estimate of psychological distress was half that of other significant life events in the fully adjusted model, such as changes in marital status or changes in labour force status.

Table 3

Fixed-effects linear regression model, with change in smoking status as the exposure variable and psychological distress (K10) as the outcome variable (n=17 140)

The pooled regression (fully adjusted) model (not controlling for within-individual correlations) estimated a strong association of quitting smoking on K10 (ex-smoker: −0.33, 95% CI −0.47 to −0.18; current smoker but tried to quit: 0.63, 95% CI 0.42 to 0.84). However, this does not control for time-invariant confounding. The fixed-effects linear regression analysis (Table 4) of quit status and psychological distress (in respondents who were current or ex-smokers at Wave 3, n=7950) indicated that ex-smokers (current smokers who successfully quit between waves) experienced a small (but non-significant) decrease in psychological distress (Model 1: K10: −0.06, 95% CI −0.34 to 0.23), whereas, current smokers who tried but failed to quit smoking experienced an increase in psychological distress (Model 1: K10: 0.22, 95% CI −0.01 to 0.44). However, this increase in psychological distress reduced by 18% (ie, from 0.22 to 0.18) with the CI including the null after adjusting for the time-varying confounding factors (Model 2: K10: 0.18, 95% CI −0.05 to 0.40). The fully adjusted models suggest that becoming non-partnered, unemployed or experiencing an increase in individual deprivation was associated with increased psychological distress over time (tables 3 and 4).

Table 4

Fixed-effects linear regression model, with quit status as exposure variable and psychological distress (K10) as the outcome variable (n=7950—current and ex-smokers at Wave 3)

Discussion

Main findings

Based on the growing evidence for a causal pathway from smoking to depression, we hypothesized that smoking uptake would increase psychological distress over time. This hypothesis was supported in the present study, albeit with the increase not being particularly large in terms of K10 scores (though half the size of associated with change in marital status and labour force status). The increase in psychological distress with smoking uptake of 0.22 (95% CI 0.01 to 0.43) equates to about 4%–5% of the population SD in K10 scores, and is small in terms of the ‘clinical’ groupings of K10 scores (low (10–15), moderate (16–21) and high (22–29) levels of psychological distress). We also hypothesized that current smokers who tried to quit smoking but failed between waves, would experience an increase in psychological distress, whereas, those who successfully quit would experience a decrease in distress. This hypothesis was partly supported in the present study, but the impact was not statistically significant in the fully adjusted model. Indeed, the findings in this study suggest the importance of factors such as NZiDep on psychological distress (individuals experiencing an increase in deprivation resulted in an increase in psychological distress).

Strengths and limitations

The strengths of the present study were the use of longitudinal data, the large sample size, the ability to adjust for time-varying demographic and socioeconomic variables, and the use of psychological distress as a continuous measure of mental health. We also used fixed-effects linear regression modelling which controls for all observed and unobserved time-invariant confounding by modelling changes within individuals over time, and ignoring ‘between-individual’ differences.29

It is important to recognise that conclusions concerning the findings in the present study rely on some of the underlying assumptions of the fixed-effects method. The fixed-effects linear regression model is driven by change in the exposure variable.29 Thus, information on SoFIE respondents who did not change with regard to their smoking or quit status over time, do not provide information on the smoking estimate. It is well known that smoking cessation can be a difficult process with multiple quit attempts often preceding long-term abstinence,30 and it is therefore especially likely that multiple failed quit attempts will lead to an increase in psychological distress in current smokers. The mean-centred estimate that is produced by the fixed-effects linear regression method may be sensitive to unobserved time-varying confounding (eg, changes in feelings of mastery; other life stressors that cause both smoking uptake and psychological distress), and health selection (eg, the reverse pathway from psychological distress to smoking status or quit status),31 which, consequently, might have biased the estimates that were found in the present study. However, with respect to the latter, previous research found only a weak (if any) association for the pathway from depression to smoking.12–15 The structure of the SoFIE questions made it not possible to model the reverse pathway, as smoking status and quit attempts were measured over the past 12 months, and the K10 questions were measured over the previous 4 weeks. Therefore, the data has a natural time lag between the smoking variables and K10.

The smoking status variable used in this analysis grouped never-smokers and ex-smokers together as non-smokers. Therefore, the effect of increased psychological distress may hide differential effects of smoking uptake, compared with relapse, and future research should aim to separate out these effects. The K10 scale was used as a measure of psychological distress in this study, whereas other longitudinal studies on smoking and mental health tend to use more clinical measures of mental health or disorders (which are also sometimes limited by their dichotomous nature). The K10 was originally developed to screen for serious mental illness in the general population,24–26 and high levels of K10 are predictive of anxiety and depressive disorders.24 ,27 Although, the cross-sectional distribution of the K10 is skewed (towards lower scores; mean K10 score at Wave 3 was 13.4, SD 4.6), the distribution of the change in K10 scores between waves was normal.

Although the original SoFIE study population (Wave 1) was a nationally representative sample of New Zealand households, the health module data was only collected in Waves 3, 5 and 7. It has been found that younger people of lower socioeconomic status are more likely to drop out of SoFIE.23 This sample attrition may therefore have led to selection bias, and reduced the generalisability of the present study results. However, unless the dropout rates were jointly distributed by smoking status, quit status and psychological distress, the effect of attrition on the exposure–outcome relationship is likely to be minimal.32 We compared the Wave 3 health information on respondents who dropped out (n=2045) by Wave 7 with those who responded in all three health modules (n=15 100), and found no major differences in the distribution of the smoking and K10 variables in those who missed a wave in the study.

Despite the findings of this study, the evidence base for health authorities being able to state in public health messages that ‘smoking causes stress’ is still insufficient. While health authorities have many other scientifically stronger messages around tobacco-related harm (that can be used in health warnings and mass media campaigns), further research on the topic of smoking and stress remains important. This is because it has implications for addressing public misperceptions around smoking and mental health (eg, beliefs around smoking relieving stress) and also for optimising quitting support. That latter can potentially involve how health workers and others (family and friends) can best support quit attempts, but also how best to mobilise any distress from a failed quit attempt into planning for a new quit attempt.

Conclusions

The findings provide some support for a modest association between smoking uptake and a subsequent increase in psychological distress, but more research is needed before such information is considered for inclusion in public health messages.

What is already known

  • There is growing evidence for a causal pathway from smoking to poorer mental health, but there is still much uncertainty about the mechanisms.

What this study adds

  • More specific evidence for a small increase in psychological distress after smoking uptake (in a longitudinal study with multiple waves).

  • While there are possible relationships between psychological distress and both successful quitting and failed quit attempts, these were not at a statistically significant level in this relatively large study.

Acknowledgments

We thank the SoFIE survey participants for their assistance. Ethics approval was obtained for the SoFIE Health module from the University of Otago Ethics Committee, Wellington. This work was undertaken as part of a Masters Dissertation for a Masters in Health and Society (FSvdD) from Wageningen University, The Netherlands, and we thank the university staff involved in supporting this.

References

Footnotes

  • Contributors KNC is the principal investigator of the SoFIE-Health project. She developed the research questions, led the analyses and final drafting of the paper. FSvdD undertook this research towards a Masters of Health & Society, Wageningen University, The Netherlands. She helped develop the research questions, led the literature review and first draft of the paper. NW helped in the interpretation of the analyses and critically revised drafts of the paper. TB helped in developing the analysis plan and interpretation of the analyses for the paper and critically revised drafts of the paper. All authors approved the final version of the paper.

  • Funding SoFIE-Health is primarily funded by the Health Research Council of New Zealand as part of the Health Inequalities Research Programme (Grant 08/048).

  • Disclaimer Statistics New Zealand Security Statement: access to the data used in this study was provided by Statistics New Zealand in a secure environment designed to give effect to the confidentiality provisions of the Statistics Act, 1975. The results in this study, and any errors contained therein, are those of the author, not Statistics New Zealand.

  • Competing interests None.

  • Ethics approval University of Otago Ethics.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Data from the longitudinal Survey of Family, Income and Employment is available from Statistics New Zealand www.stats.govt.nz.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.