The association between workplace smoking bans and self-perceived, work-related stress among smoking workers

Tobacco use, including passive smoking, is a key risk factor for several illnesses such as cancer, heart attacks, cardiovascular disease, strokes [1] and is responsible for about 6 million deaths each year [2]. Parallel to the increased global awareness of the health risk of environmental tobacco smoke (ETS), public smoking bans, including workplace smoke-free policies, have become widespread in many countries. While these bans are primarily intended to protect non-smokers from the adverse effects of ETS, some empirical evidence suggests that they also encourage smoking cessation behaviors [e.g. [3‐6]]. For example, Moher et al. [7] find consistent evidence that workplace smoke-free policies reduce cigarette consumption during the working day and exposure to ETS at work among non-smokers. In a meta-analysis of 26 studies, Fichtenberg and Glantz [4] find that workplace smoking bans (WSB's) decrease smoking prevalence and daily smoking intensity as well as exposure to ETS.

While there is a large and growing literature regarding the effects of WSB's on smoking behavior, their unintended effects have received little attention. Identifying these effects will be of importance to public policy, particularly since negative effects may limit the effectiveness of WSB's. There is evidence that public smoking bans including WSB's may lead to compensating smoking behavior, such as increased smoking during work breaks [8] and displaced smoking [9, 10]. Adda and Cornaglia [10] find that in some instances, bans may increase exposure to ETS due to smokers switching from smoking in public areas where bans are enforceable to non-enforceable private areas. WSB's may increase tension between non-smoking and smoking workers [11]. In a UK study, Anderson et al. [12] find that WSB's in hospitals make nurses practice some dangerous smoking behaviors like stubbing out cigarettes in bins that contained paper towels and smoking in unsafe places. Moore et al. [13] find that a California smoking restriction in bars may increase exposure to ETS and other adverse effects. Particularly for women, the authors find that compliance with WSB's by smoking outside bars can constitute a threat to their physical safety and public image. In a related study, Adams and Cotti [14] observe an increase in accidents due to drunk driving because individuals drive long distances to bars with no smoking restrictions. WSB's may also lead to the accumulation of smoking trash, like cigarette butts and dead matches at the entrance to a workplace. This could increase the risk of fire and impose a cost on local governments to collect the trash. Also, a congregation of smokers at the entrance of a workplace increases exposure to ETS to people entering and leaving the workplace.

Workplace smoke-free laws may be linked to work-related stress in two ways. First, stress may result from both psychological and physical dependence on nicotine. This could be caused by a reduced level of nicotine in the blood arising from a requirement to restrict smoking during the working day. Studies have identified several symptoms of nicotine withdrawal, all of which may cause higher stress. These include anxiety, tension, depression and difficulty concentrating [15‐17]. Second, smokers with severe self-control problems may be more vulnerable to high work-related stress especially if there is a full smoking restriction in the workplace.

The objective of this paper is to examine whether WSB's are associated with higher self-perceived work-related stress among working smokers, using longitudinal data from the Canadian National Population Health Survey (NPHS).

To examine the relationship between WSB's and self-perceived, work-related stress, a fixed-effects linear regression is used in order to control for measured and unmeasured individual time-invariant confounders. The use of the fixed-effects method is appropriate, given that some crucial unmeasured individual factors (e.g. cognitive abilities) can bias the outcome between WSB's and workplace stress.

A fixed-effect regression model of the following form is estimated:

where i, j and t represent, respectively, an individual, province of residence and time period. The symbol ban represents the three categories of WSB's, × is a vector of additional control variables which include: age, marital status, educational attainment, ethnicity, work status, income, drinking status, health status, number of chronic conditions, provincial and occupational fixed effects. Individual time-invariant characteristics are represented by μ_i and can be correlated with other time-varying covariates in the model. The standard residual term is, ε _ijt which is adjusted for clustering at the individual level. In order to study the heterogeneous effects of workplace smoke-free policies on self-perceived stress, separate analyses are performed for males, females, young adult (age 18-40 years) and those classified as old (age 41-65 years). To check whether the results are sensitive to the inclusion of additional covariates, three different model specifications are estimated. Model 1 is the baseline specification which controls for income level, educational level, marital status, age, work status, ethnicity, drinking status. Model 2 includes the additional covariates: individual health status, number of chronic conditions and provincial fixed effects. In model 3, occupational fixed effects are included, in addition to the model 2 covariates. Also, as a further robustness, a separate analysis is performed for non-smokers. Multivariate analysis is conducted using Stata 11.

The adverse health consequences of environmental tobacco smoke are well established. In response, smoking bans have been progressively used in recent times. Besides protecting non-smokers against environmental tobacco smoke, some studies show that WSB's reduce smoking prevalence and intensity. While there is substantial literature on the effect of this policy in altering smoking behaviors, little is known about its potential unintended consequences. This study uses longitudinal data from the Canadian National Population Health Survey to examine the association between WSB's and work-related stress. We exploit the panel structure of the data set by using a linear fixed-effects model to control for individual time-invariant confounders that may bias the relationship between WSB's and work-related stress. To capture the heterogeneous effect of WSB's, on stress level across different groups of workers, the analysis is stratified by gender and age.

Multivariate analysis reveals that, on average, full or partial WSB's have a positive and statistically significant association with self-perceived work-related stress among smoking workers. These findings are not sensitive to the inclusion of individual's health status, occupational and provincial fixed effects. We also find that these effects vary by gender and age. In particular, WSB's are significantly associated with higher work stress for males and young adults (aged 18-40). We also examined the effect of WSB's on the stress level of nonsmoking workers and no statistically significant effect was found (see Table 3).

These results are supported by some previous studies which find that WSB's have a larger impact on groups with high smoking rates [27‐29]. For example, Heloma and Jaakola [22] find that Finland's national smoke-free workplace laws have a greater impact on males than on females. In a recent survey of 2,103 workers by Croner [30], two thirds of the workers reported that they would not manage their stress very well, or at all, if there were a ban on smoking breaks.

These individual differences in the effect of WSB's on stress level could be explained by the varying degree of nicotine dependence and the smoking rate among individuals. Smokers with low nicotine dependence are less likely to suffer from symptoms of nicotine withdrawal, including tension and mood disturbance [31]. This may help explain why there are gender differences in the results of the present study. Some studies show that females are less nicotine dependent than males [e.g. [32]]. Variations in nicotine dependence have been attributed to some unobserved characteristics, including genetic factors, personality and family background [31].

The results of the current study could imply additional negative effects of higher smoking intensity, given several studies find that work stress is positively associated with more smoking [e.g. [33‐37]]. Also, the health risks and economic costs attributable to job stress are substantial and widely documented [38, 39]. For example, work stress costs U.S. companies over $300 billion annually [40] and has been linked to several adverse health conditions [41].

In addition to increased level of stress among smokers, other unintended consequences of WSB's have been reported in previous studies. WSB's may lead to compensating smoking behaviors [8, 9]. For example, some smokers may increase cigarette consumption before and after work [42], smoke hard during break times by increasing puff frequency and fast smoke each cigarette [8]. Other side effects include smoking in unsafe places and the accumulation of smoking trash like cigarette butts and dead matches at the entrance to a workplace. Also, the congregation of smokers at the entrances and exits of a workplace increases exposure to second-hand smoking.

The current study has some limitations. First, we did not control for individuals' level of nicotine dependence as this information is not available in the data set. Second, the fixed-effects model that we used controls only for individual time-invariant confounders. However, there may be other time-variant unobserved characteristics that are not captured by the current study. Third, it is possible that differences in work-related stress levels among individuals may be due to differences in the type of workplace and not the smoking bans, especially if these bans are constant over time. However, the analysis uses seven occupational dummies to control for unobserved occupation-specific characteristics that may affect stress level. It should be noted that these occupational dummies may not fully capture individual-specific workplace effects, since individual work environments within each occupational classification are likely to be different. Fourth, this study does not stratify the analysis by quantity of cigarettes smoked. Recent studies cast doubt on using the individual quantity of cigarettes smoked as a measure of smoking intensity, where smokers may reduce the quantity of cigarettes smoked but increase the intake of cotinine [43, 44]. The NPHS data set does not have any information on cotinine. Also, there is no uniform definition on the number of cigarettes smoked that can be classified as heavy or light.

Despite these limitations, this paper adds to the literature on the unintended consequences of anti-smoking policies by providing empirical evidence that WSB's may lead to higher self-perceived, workplace-related stress among smoking workers. These results cannot be generalized. Further evidence using data from other countries will be necessary to confirm these findings. The unintended negative side effects of WSB's do not undermine their usefulness in discouraging smoking. However, it is important for policy makers to combine workplace smoke-free policies with other workplace intervention measures, in order to offset the negative side effects. For example, stress management programs [41], individual and group therapy and pharmacological treatment to address nicotine addiction have been shown to be effective intervention strategies [7]. Moreover, it may also be useful to complement WSB with other worksite-based tobacco control intervention measures to promote smoking cessation among workers [45]. The results of this study do not imply that WSB's are the main determinant of self-perceived, work-related stress among smokers but provides suggestive evidence that they may be positively related.