Public Health (2003) 117, 383–388
The effect of work environment and heavy smoking on the social inequalities in smoking cessation Karen Albertsen*, Harald Hannerz, Vilhelm Borg, Hermann Burr ´ 105, Department of Psychology and Sociology, National Institute of Occupational Health, Lersø Parkalle Copenhagen DK-2100, Denmark Received 12 August 2002; received in revised form 20 January 2003; accepted 13 March 2003
KEYWORDS Social class; Socioeconomic class; Work environment; Smoking; Cessation
Summary Objectives. To investigate social differences with regard to smoking cessation in Denmark. Methods. Social differences in smoking cessation were estimated from 3606 smokers from the Danish National Work Environment Cohort Study in 1990, 1995 and 2000. The differences were investigated using heavy smoking and four work environment factors as explanatory variables. Results. The odds ratio for cessation was more than twice as high in social class I compared with social class V. When heavy smoking was controlled, this explained 28% of social differences, the work– environment factors alone explained 36% of social differences, and together the factors explained 55% of social differences. Conclusions. A large proportion of the social differences in smoking cessation could be explained by differences in work –environment exposures and smoking intensity. Q 2003 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved.
Introduction Smoking prevalence depends on social class and occupation. The highest proportion of smokers is found in the lowest social classes and in occupations with low educational requirements.1,2 The proportion of smokers that quit smoking is also dependent on social class and occupation.3 – 5 It has been shown that smoking cessation is most common among smokers in the upper social classes.3,4,6 The success of an attempt to quit smoking is influenced by the degree of addiction as well as by norms and attitudes towards smoking. Since smoking may be used to cope with hardships and discomforts in *Corresponding author. Tel.: þ45-3916-5466; fax: þ45-39165201. E-mail address:
[email protected]
life,7 – 10 the environment and the person’s material resources also influence the success rate. Degree of addiction is probably highly dependent on smoking intensity, and there may be social differences in this respect, with a larger proportion of heavy smokers among the smokers in the lowest social classes. It is known that attitudes towards tobacco depend on educational background,11 and there are, by definition, economic and work environment differences between social classes. Hence, there are several possible explanations for a social gradient in the rate of smoking cessation. In a previous study, we investigated the effects of work environment factors on the probability of smoking cessation over a 5-year period.12 The study analysed a logistic regression model comprising measures of base-year, gender, age group, social class, heavy smoking, environmental tobacco
0033-3506/03/$ - see front matter Q 2003 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/S0033-3506(03)00103-3
384
smoke, working hours, chemical exposure, cold, noise, physical strain, conflicts at work, job insecurity, possibilities for development, social support, decision latitude, workload, concentration and responsibility at work. Of these variables, age, base-year, heavy smoking (. 14 g tobacco/day), physical strain, noise, workload and responsibility at work were found to be significantly associated with smoking cessation after 5 years. The probability of smoking cessation was lower for heavy smokers compared with other daily smokers, for the age groups between 30 and 49 years compared with those between 18 and 29 years, for those exposed to noise for a moderate period of time compared with a short exposure, and for those with high compared with low levels of physical strain. The probability of cessation was higher in 1995 – 2000 compared with 1990 – 1995, indicating an increased cessation rate during this period. Moderate levels of responsibility were more predictive than low levels, and against the hypothesis, high levels of workload were more predictive of cessation than low levels. In accordance with previous studies of gender differences in predictors of smoking cessation,9,13, 14 no gender differences were found when heavy smoking was included in the analysis. The aim of the present study is to investigate social differences with regard to smoking cessation in Denmark. It was explored how much of the social gradient in cessation could be explained by: (1) smoking intensity, (2) work environment factors, and (3) smoking intensity and work environment factors.
Materials and methods The Danish National Work environment Cohort Study is an open cohort study conducted at the National Institute of Occupational Health.15 – 17 In 1990, a random sample aged 18 – 59 years was drawn from the Central Population Register of Denmark. In 1995 and 2000, the sample was contacted again and new participants aged 18 – 22 years and immigrants to Denmark were added to the cohort. The database comprises data from 5940 employees in 1990, 5575 employees in 1995 and 6024 employees and self-employed people in 2000, all interviewed by telephone. The response rate of all contacted people was 90% in 1990, 80% in 1995 and 75% in 2000. Of the people who answered in 1990, 86% answered again in 1995, and of the people who answered in 1995, 82% answered again in 2000. In the present study, the analyses were based on answers from all smokers in 1990, who were
K. Albertsen et al.
interviewed again in 1995, and all smokers in 1995, who were interviewed again in 2000; a total of 3606 observations. We estimated social differences in smoking cessation, and investigated the effect of these differences when controlling for the factors that showed statistical significance with cessation in the previous analyses:12 (1) smoking intensity (heavy smoking), (2) work environment factors, and (3) work environment factors and smoking intensity.
Measures Social class The respondents were classified into five groups according to employment grade, job title and education.
† Social class I: executive managers and/or academics. † Social class II: middle managers and/or 3 – 4 years of further education. † Social class III: other white-collar workers. † Social class IV: skilled blue-collar workers. † Social class V: semi-skilled or unskilled workers. People receiving education were classified separately. Smoking ‘Do you smoke?’ (yes/did smoke before, but not anymore/have never smoked). ‘Heavy smoking’ was registered if a person smoked at least 15 g tobacco/day (equivalent to 15 cigarettes), whether it was from cigarettes, pipe or cigars. ‘Cessation’ was registered if persons reported that they were a smoker in the first interview and stated that they no longer smoked 5 years later. Among the 3606 observations, there were 606 cases of cessation. Age 18 – 29 years/30 – 39/40 – 49/ $ 50 Base year 1990 or 1995 Workload ‘Is your workload so heavy that you haven’t got time for talking or thinking about anything else than work?’ Responsibility ‘Do you perform work tasks that—if you make mistakes—can be a risk for the health of others or yourself or can cost a lot of money?’
The effect of work environment and heavy smoking on the social inequalities in smoking cessation
385
Statistical analyses
Noise ‘Are you exposed to noise so high that you have to raise your voice to talk with others?’ Physical strain Physical strain was measured by a scale with three items: ‘Is your work so physically strenuous that you breathe faster’, ‘Does your work cause you to work with your hands raised to shoulder height or higher?’, and ‘Does your work require that you squat or kneel?’.15 (Appendix D, p. 356 for scoring). Response categories for all of the work environment variables were: almost all the time/about three-quarters of the time/about half of the time/about one-quarter of the time/seldom/ never. The distribution of answers was explored in preparation for division into tertiles: low, moderate or high.
Four logistic regression models were used to evaluate the social gradient in smoking cessation. In the first model, social class, age and base-year were used as predictors for cessation over 5 years. In the second model, heavy smoking was included in the analysis. In the third model, heavy smoking was excluded while the four work environment factors were included. In the fourth model, both heavy smoking and work environment factors were included. Each of social classes I – IV was compared with social class V, and odds ratios (OR) for cessation were calculated. The contribution to the explanation of the social differences in smoking cessation was measured by the decrease in the OR exceeding 1.0, when comparing the first model with each of the other models, i.e. ‘percent of gradient explained’ ¼ 100(OR1 2 ORx)/(OR1 2 1) where OR1 is the OR obtained in model 1 and ORx is the OR obtained in model x. All analyses were performed by use of SAS statistical software.
Characteristics of sample Fifty-seven percent of the observations are from the 1990 –1995 cohort and 43% are from the 1995 – 2000 cohort; 56% male (22% between 18 and 29 years, 31% between 30 and 39 years, 30% between 40 and 49 years, and 17% above 50 years). Table 1 shows the distribution of each of the predictor variables within the social classes and the percentage of cessation in each class.
Table 1 Distribution of exposure/outcome levels (percentages) for each social class (I–V). Parameter
Level
I
II
III
IV
V
Noise
Low 89.2 69.4 77.7 49.6 48.2 Medium 7.4 19.8 13.2 28.9 26.4 High 3.4 10.8 9.1 21.5 25.4
Physical strain
Low 92.0 72.1 67.5 38.9 56.1 Medium 6.9 24.2 26.6 41.4 32.7 High 1.1 3.7 5.9 19.7 11.2
Workload
Low 49.0 53.7 66.4 75.1 79.8 Medium 19.8 21.2 17.9 15.6 11.4 High 31.2 25.1 15.7 9.3 8.8
Responsibility
Low 41.1 48.6 50.7 34.7 51.7 Medium 21.5 18.5 21.7 19.9 15.8 High 37.4 32.9 27.6 45.4 32.5
Heavy smoker
No Yes
48.9 48.0 47.7 33.3 36.6 51.1 52.0 52.3 66.7 63.4
Smoking cessation No Yes
76.2 81.8 82.6 86.6 86.4 23.8 18.2 17.4 13.4 13.6
Results Table 2 shows OR and 95% confidence intervals (CI) for cessation after 5 years for each of the social classes, using social class V as the reference group. In addition, it is shown how many of the differences between the classes are explained by each of the models. It appears from model 1 that the OR for cessation is more than twice as high in social class I as in social class V. Heavy smoking was controlled in model 2 and this explains 28% (weighted mean) of the differences between the social classes. The work environment factors were controlled in model 3 and this explains 36% of the differences. In model 4, where heavy smoking and work environment factors were combined, these factors explain 55% of the differences. Thus, although heavy smoking and work environment may be associated, both aspects contribute independently to the explanation of the social gradient in smoking cessation.
Discussion The present study confirmed earlier research which found that smoking-cessation rates depend on social class, with a higher proportion of quitters among smokers in the upper classes.3 In addition, it showed that a large proportion of the social differences in smoking cessation could be
386
K. Albertsen et al.
Table 2 Odds ratio (OR) and 95% confidence interval (CI) for smoking cessation by social class, according to logistic regression models with and without control for work environment factors and heavy smoking.
Social class I Social class II Social class III Social class IV Social class V Weighted mean P value a b c d
n
Model 1a OR (95%CI)
Model 2b OR (95%CI)
Model 3c OR (95%CI)
Model 4d OR (95%CI)
Gradient explained (%) 1– 2 1–3 1–4
377 558 1207 429 870
2.11 (1.50– 2.98) 1.54 (1.11–2.13) 1.50 (1.14–1.97) 1.11 (0.77–1.61) 1.00
1.87 (1.32 –2.65) 1.35 (0.97 –1.88) 1.32 (1.00 –1.74) 1.11 (0.77 –1.61) 1.00
1.73 (1.24 –2.43) 1.37 (1.01 –1.87) 1.27 (0.98 –1.64) 1.10 (0.78 –1.55) 1.00
1.56 (1.10–2.19) 1.22 (0.89–1.67) 1.15 (0.88–1.49) 1.10 (0.78–1.57) 1.00
22 35 36 2 – 28
0.0005
0.0216
0.0457
0.1563
34 31 46 16 – 36
50 59 70 9 – 55
Model 1: social class, age and base-year. Model 2: model 1 þ heavy smoking. Model 3: model 1 þ work environment factors: noise, physical demands, workload and responsibility. Model 4: model 1 þ heavy smoking þ work environment factors: noise, physical demands, workload and responsibility.
explained by social differences in work environment exposures and smoking intensity. The strengths of the study are that it was prospective, that the individuals were sampled at random, that the sample was large, that missing values among participants were few, and that the participation rates were relatively high. Due to potential changes in the work environment as well as the social status of the investigated individuals, the main weakness of the study is the long period that elapsed between the measurements of risk factors and outcome. Thirty-five percent of people who were employed in both 1990 and 1995 changed to a different social class between the two surveys. Since people in the highest social class could move down but not up, while people in the lowest social class could move up but not down, it is evident that the difference in smoking cessation between these classes was likely to be conservatively underestimated. The estimated effect of the work environment factors may have been less significant, as 33% belonged to a different category of physical strain in 1995 compared with 1990, 28% belonged to a different noise category, 42% belonged to a different workload category and 43% belonged to a different category of responsibility. We used the dichotomization ‘heavy smoker’ and ‘light smoker’ as a rough measure of strength of habit. Since strength of habit would depend not only on current smoking but also on the past, and since we did not have any data pertaining to the smoking history of the individuals, we did not find it meaningful to divide the smokers into more detailed categories than these. Social differences in smoking habits might, however, also exist within the two smoking categories, with a resultant residual confounding when used as
a control variable. Another objection that may be raised is that self-reported smoking status was not confirmed with a biomedical marker of tobacco consumption. Self-reported smoking has, however, been validated several times and appears to be a satisfactory way to classify smokers in epidemiological surveys.4,18 In the face of the drawbacks mentioned above, it may be wise to regard the results of the present study as conservative estimates rather than precise numbers. The observed patterns are, however, very clear and they correspond with the general notion of social gradients in health behaviour.4,19 – 21 People in social class I were more likely to quit smoking than people in social class II, who in turn were more likely to quit than those in class III, etc. according to each of the four models. As expected, controlling for work environment factors had the largest effect on the difference between social class V (unskilled blue-collar workers) and social class III (lower grade whitecollar workers). In Denmark, these groups are similar with regard to level of education and income, but differ significantly with regard to work environment exposures. A large number of studies have shown that biological, cultural and social factors associated with low social class are predictors of increased risk of adverse health and, especially, cardiovascular disease.20,21 The connections and possible processes taking place between behaviours such as smoking and characteristics of the social environment are not so well documented. In the present study, differences in work environment explained more than one-third of the differences in cessation rate between social classes over a 5year period. The specific impact and possible mechanisms behind the work environment factors
The effect of work environment and heavy smoking on the social inequalities in smoking cessation
that predicted cessation in this study are discussed elsewhere,12 but overall the results pointed at cultural factors associated with a specific work environment, at work-related stress and at work-related resources as contributors to the social gradient in cessation. In addition to the factors included in this study, other factors might also contribute to the explanation of the social gradient in smoking cessation. Social capital as expressed in social participation has been shown to be associated with cessation in a recent study.6 Workplace smoking policy and smoking bans might also play a role.22 – 24 Susceptibility to general antismoking education, self-efficacy regarding smoking cessation, locus of control and neuroticism might constitute person-related factors that vary between the social classes.11 The relative contributions and possible additive or interactional mechanisms between these factors and the work environment factors should be explored in future studies. In addition to the effect of the work environment factors, the results showed that the likelihood of cessation was seriously reduced for heavy smokers. In a number of previous studies, smoking rate and smoking intensity among smokers have been shown to be associated with high levels of stress at work.9, 10,25,26 Efforts to minimize the social gradient in cessation could therefore also be aimed at reducing stress-producing work environments, at increasing work-related resources and individual coping resources, and at changing workplace smoking cultures. Interventions in the work environment might contribute to a reduction in socio-economic inequalities in health, not only directly by its influence on health, but also indirectly by its influence on lifestyle.
Acknowledgements This research was supported by a grant from: The Service Centre of the Danish Working Environment Council, grant no 2000-08. Computer Programmer Ebbe Villadsen from the National Institute of Occupational Health kindly supported us with advice regarding use of the database.
References [1] Berman BA, Read LL, Marcus AC, Gritz ER. Nurses enrolled in a stop smoking program: the role of occupational stress. J Women Health 1992;1:41—6.
387
[2] Nelson DE, Emont SL, Brackbill RM, Cameron LL, Peddicord J, Fiore MC. Cigarette smoking prevalence by occupation in the United States. A comparison between 1978 to 1980 and 1987 to 1990. J Occupat Med 1994;36: 516—25. [3] Lundberg O, Rosen B, Rosen M. Who stopped smoking? Results from a panel survey of living conditions in Sweden. Soc Sci Med 1991;32:619—22. [4] Osler M. Social class and health behaviour in Danish adults: a longitudinal study. Public Health 1993;107: 251—60. [5] Covey LS, Zang EA, Wynder EL. Cigarette smoking and occupational status: 1977—1990. Am J Public Health 1992; 82:1230—4. [6] Lindstrom M, Hanson BS, Ostergren PO, Berglund G. Socioeconomic differences in smoking cessation: the role of social participation. Scand J Public Health 2000;28: 200—8. [7] McMahon SD, Jason LA, Salina D. Stress, coping, and appraisal in a smoking cessation intervention. Anxiety Stress Coping 1994;161—71. [8] Serxner S, Catalano R, Dooley D, Mishra S. Tobacco use: selection, stress, or culture? J Occupat Med 1991;33: 1035—9. [9] Shields M. Long working hours and health. Health Rep 1999; 11:33—48. [10] Steptoe A, Wardle J, Lipsey Z, et al. A longitudinal study of work load and variations in psychological well-being, cortisol, smoking, and alcohol consumption. Ann J Behav Med 1998;20:84—91. [11] Stronks K, van de Mheen HD, Looman CW, Mackenbach JP. Cultural, material, and psychosocial correlates of the socioeconomic gradient in smoking behavior among adults. Prev Med 1997;26:754—66. [12] Albertsen K, Hannerz H, Borg V. Work environment and smoking cessation over a five-year period. Scandinavian Journal of Public Health 2003 accepted for publication. [13] Gritz ER, Thompson B, Emmons K, Ockene JK, McLerran DF, Nielsen IR. Gender differences among smokers and quitters in the Working Well Trial. Prev Med 1998;27: 553—6. [14] Hennrikus DJ, Jeffery RW, Lando HA. Occasional smoking in a Minnesota working population. Am J Public Health 1996; 86:1260—6. [15] Borg V, Burr H. Danske Lønmodtageres Arbejdsmiljø og Helbred 1990—95. Copenhagen: Arbejdsmiljøinstituttet; 1997. [16] Borg V, Kristensen TS. Social class and self-rated health: can the gradient be explained by differences in life style or work environment? Soc Sci Med 2000; 51:1019—30. [17] Borg V, Kristensen TS, Burr H. Work environment and changes in self-rated health: a five year follow-up study. Stress Med 2000;16:37—47. [18] Patrick DL, Cheadle A, Thompson DC, Diehr P, Koepsell T, Kinne S. The validity of self-reported smoking: a review and meta-analysis. Am J Public Health 1994;84: 1086—93. [19] Aro S, Hasan J. Occupational class, psychosocial stress and morbidity. Ann Clin Res 1987;19:62—8. [20] Brunner E, Shipley MJ, Blane D, Smith GD, Marmot MG. When does cardiovascular risk start? Past and present socioeconomic circumstances and risk factors in adulthood. J Epidemiol Commun Health 1999;53: 757—64. [21] Engstrom G, Tyden P, Berglund G, Hansen O, Hedblad B, Janzon L. Incidence of myocardial infarction in
388
women. A cohort study of risk factors and modifiers of effect. J Epidemiol Commun Health 2000;54: 104—7. [22] Jeffery RW, Kelder SH, Forster JL, French SA. Restrictive smoking policies in the workplace: effects on smoking prevalence and cigarette consumption. Prev Med 1994;23: 78—82. [23] Longo DR, Brownson RC, Johnson JC, et al. Hospital smoking bans and employee smoking behavior: results of a national survey. JAMA 1996;275:1252—7.
K. Albertsen et al.
[24] Eriksen MP, Gottlieb NH. A review of the health impact of smoking control at the workplace. Am J Health Prom 1998; 13:83—104. [25] Steptoe A, Wardle J, Pollard TM, Canaan L, Davies GJ. Stress, social support and health-related behavior: a study of smoking, alcohol consumption and physical exercise. J Psychosom Res 1996;41:171—80. [26] Rose JE, Ananda S, Jarvik ME. Cigarette smoking during anxiety-provoking and monotonous tasks. Addict Behav 1983;8:353—9.