PREVENTIVE
Changes
MEDICINE
19, 314-322 (I%@)
in Acceptance of Workplace Smoking Bans following Their Implementation: A Prospective Study
RON BORLAND,
PH.D., *,’ NEVILLE OWEN, PH.D.,~ DAVID SIMON CHAPMAN, PH.D.S
HILL,
PH.D.,*
AND
*Centre for Behavioural Research in Cancer, Anti-Cancer Council of Victoria, I Rathdowne Street, Carlton South, 3053, Victoria, Australia: tDepartment of Community Medicine, University of Adelaide, Adelaide 5000, S.A., Australia; and SDepartment of Community Medicine, University of Sydney, Westmead Hospital, Westmead 2145, NSW Australia Recent cross-sectional and retrospective studies suggest that the attitudes of employees toward workplace smoking bans may become more positive after the bans have been introduced. Using a prospective design, we found that the attitudes of nonsmokers, ex-smokers, and smokers became more favorable over the 6 months following the introduction of smoking bans to Australian Government Offtces. The magnitude of changes in approval ratings was greatest for smokers, but smokers were also more likely to disapprove of the bans before and after their introduction than were nonsmokers and ex-smokers. Among smokers, there was a strong relationship between the extent to which they saw themselves inconvenienced by the bans and the extent to which they disapproved of them. Overall, the bans were accepted and viewed in a positive light, but there was nevertheless a significant minority of smokers who remained disgruntled with them. Some ways in which the concerns of this subgroup of smokers may be addressed are considered. o IWO Academic PESS, IW.
INTRODUCTION
The accumulating evidence of the harmful effects of environmental tobacco smoke on nonsmokers (l-4) and increasing public concern about passive smoking, are leading to increased pressure to restrict smoking in public places and in the workplace (5-7). At present there are relatively few data on how those who are affected by these restrictions actually perceive them, and what effect the bans have on the attitudes of smokers and nonsmokers. Hill’s survey of opinions (8) found high levels of community approval of some restrictions on smoking in restaurants and work settings. Studies on the impact of workplace smoking bans (9-12) have also reported positive attitudes toward restrictions on smoking in the workplace; two studies (9, 12) reported retrospective data suggesting that approval had increased after restrictions were introduced. Increases in acceptance may mean that anticipated negative consequences may not have come about or that they may have been less severe than expected. No published studies using a prospective design which show changes in acceptance of workplace smoking restrictions after they have been introduced could be found. In 1986, the Australian Public Service Board announced that there would be a ban on smoking on all public service premises to take effect about 18 months from the initial announcement. Over that period departments, and lower level organi’ To whom reprint requests should be addressed. 314 0091-7435/90 $3.00 Copyrieht 0 1990 by Academic Press, Inc. AII rights of reproduction in my form reserved.
SMOKING
BANS
315
zational units within departments, were expected to develop frameworks for implementation. This process was conducted with staff input, either directly or through staff unions. Some departmental units phased in bans ranging from partial to total, others chose to wait until the date of the mandated ban. All departments provided the opportunity for staff affected by the ban to obtain assistance before and after the implementation of the ban. This assistance included smoking cessation and smoking regulation courses in the workplace, time off and/or some financial aid to attend outside programs, talks, and self-help literature. There appears to have been considerable diversity in the extent to which such services were actively promoted and in the extent to which staff availed themselves of the services. There was no formal position regarding sanctions, and different departments adopted different positions. As far as we are aware, at least at the time of our follow-up survey, no punitive sanctions were in place beyond an offtcial rebuke. Over 4,000 Australian public service employees in three cities were surveyed in the month before the mandated introduction of a total ban on smoking in the workplace. A high level of acceptance of the bans was found (13). To determine changes in acceptance following the implementation of the ban, staff in the same work areas were resurveyed 5-6 months after the bans were implemented. This provided an opportunity to use a prospective study design to examine whether there were changes in acceptance. Of particular interest was whether the extent to which the bans were successfully enforced would affect attitudes and whether attitudes and attitude change would vary as a function of workers’ smoking status. There is evidence from research on attitude change that attitudes are likely to shift to become more consistent with behavior (14-16). Thus, being unable to smoke at work, and not seeing others smoke may result in a more positive attitude toward restrictions on smoking behavior. It was predicted that acceptance of the workplace smoking bans would increase following their implementation and that the increase would be independent of whether a statf member did, or did not, smoke. METHODS Sample
All respondents were employees of the Australian Public Service and came from selected work areas in six departments spread across three cities and 44 locations. The departments were selected to provide a range of central offtce staff housed in multistory buildings, and staff of smaller units, housed in suburban areas. The workforce surveyed was predominantly white collar. There were 4,215 respondents to the initial survey, which took place 2-4 weeks before the workplace smoking bans were formally introduced. A questionnaire was distributed in the selected work areas to all employees who were at work on the day of the survey or were expected to be at work the next day. There was a 79% return rate for the initial survey. The follow-up survey was administered 6 months after the initial survey and was completed by 3,360 respondents; 2,113 could be matched to the initial survey and 1,247 either were new or their surveys could not be matched. The reduction
316
BORLAND
ET AL.
in sample size from the initial to the follow-up surveys was mainly due to narrower criteria for inclusion in the follow-up survey as opposed to the initial survey. Of those who could not be matched, only those who were at work on the day were surveyed. Also, one large workplace had reduced its staff level significantly at the location that was surveyed initially. Overall, only 51% of the initial survey respondents were matched at follow-up. This was attributable mainly to 26.3% who did not include their names on the initial questionnaire or who provided names that could neither be deciphered nor traced to attempt recontact. Another 11.5% had left their jobs or were on extended leave. This left 2,623 whom we attempted to recontact. A total of 2,169 matched surveys were returned (a return rate of 83%). However, 56 of these were not usable, leaving a final sample of 2,113. The mean age of the initial sample was 32.8 years. Fifty-three percent were male; 25% were smokers; 33% had tertiary education, and another 39% had either completed secondary school and/or acquired certificate or trade qualifications. The sample was therefore better educated than the general community. Among the smokers, average consumption on working days was 15.3 cigarettes. In line with the distribution of smoking in the community (17), there were more smokers among younger respondents and among those with lower educational qualifications. The proportions of male and female smokers were similar. Questionnaire The initial questionnaire dealt with: l Awareness of the forthcoming mandated ban on smoking at work, of any existing smoking restrictions, and of how existing restrictions had been brought about. 0 A series of questions concerning beliefs and opinions about smoking (18), particularly at work l Questions on age, sex, education level, and current smoking status. l Six items concerning attitudes toward the ban. Items dealt with approval of the ban; approval of the way the ban was introduced; perceptions of rights infringed; quality of own and other’s work environment resulting from the bans; and anger about the bans. Responses to the items were on a 5-point Likert scale from “strongly agree” to “strongly disagree.” Each item was scored such that the response indicating strong approval of the ban was scored + 2, a neutral response was scored zero, and strong disapproval was scored -2. The combination of scores from individual items to form a single score provides a more reliable assessment of attitudes, where it can be shown that the items form a unidimensional scale (19). For the present data, maximum likelihood factor analysis indicated a one-factor solution which accounted for 63.8% of the variance. All items loaded highly on this factor and in the predicted direction, indicating that it was valid to combine the items into a single scale. An attitude score, indicating level of acceptance of the ban based on the total score for the six items, was calculated for each respondent. A score of zero indicated a neutral position, while a score of more than zero indicated some degree of acceptance. The possible range of scores
SMOKING
BANS
317
was from - 12 to + 12. The scale had a high level of internal consistency (Cronbath’s (Y = 0.91). The follow-up questionnaire contained many of the same questions, including the six-item attitude scale. It also included two additional items in the same response format as the attitude scale: “On the whole the ban has been a good thing,” and “The ban on smoking at work has inconvenienced me.” RESULTS
Preliminary analyses revealed that those who completed both surveys were less likely to have been smokers when they completed the initial survey (23.0%), compared with 28.2% for those who completed the initial survey only (x2 = 15.8, df = 2, P C 0.001). In the follow-up survey, the respondents who answered both surveys were again less likely to be smoking (21.8%), compared with 26.3% (x2 = 8.87, df = 2, P < 0.05). This represents a decline in smoking prevalence of 1.2% in the matched sample and 1.9% in the unmatched sample. There were more males in the follow-up sample (57%) compared with the single survey samples (51%), but there was no difference in mean age or in overall daily cigarette consumption among the smokers. Reported compliance with the ban was high at the time of the follow-up with 73.9% of respondents reporting that they were aware of compliance in areas in which smoking was banned, and only 3.3% reporting regular violations. The remaining 22.8% reported occasional violations. The reporting of violations was related to attitudes toward the ban (F = 6.1, df = 2, 2,037, P C 0.01). Post hoc comparisons using Scheffe’s procedure showed that those who reported occasional violations were significantly more accepting than those who reported regular violations, with those who reported no violations in between. There was no relationship between reported violations and changes in acceptance; the pattern reflected a preexisting difference in attitudes. Attitude toward the Ban The sum of respondents’ scores on the six-item attitude scale was used to assess acceptance of the ban. Before the ban was implemented, most respondents approved (average level = +5.5,95% CI = 5.3,5.7), with nonsmokers (+7.8,95% CI = 7.6, 8.0) being more approving than ex-smokers (+6.8,95% CI = 6.4, 7.1) and both being far more approving than current smokers (-0.2, 95% CI = 0.6, 0.2). There was no relationship between age or sex and acceptance of the ban; however, those with higher education levels were significantly more accepting than those with lower levels of education (tertiary education, mean = 7.0; completed secondary, 5.6; partial secondary, 3.8). Those respondents who completed both surveys and were matched were more likely to approve of the ban. Results from the two groups are reported separately (see Table 1). Levels of acceptance of the ban increased substantially from the initial to the follow-up surveys for matched respondents. This included an increase from 83 to 87% in the percentage of respondents who supported the ban overall (scores on the attitude scale of >O). There were also significantly higher levels of approval when those who completed the initial survey only were com-
318
BORLAND
ET AL.
TABLE 1 ATTITUDES TOWARD THE WORKPLACE SMOKING BAN BEFORE AND AFTER ITS IMPLEMENTATION
Mean attitudes score with 95% confidence intervals Follow-up (both surveys) (N = 1,979) Before ban After ban Difference
Single survey
+6.16 (5.91, 6.41) +6.82 (6.59, 7.04) +0.66 (0.51, 0.81)*
+4.87 (4.60, 5.14) n = 1,879 +6.05 (5.14, 6.36) n = 1,208 + 1.18 (0.77, 1.59)**
* Within-subjects effect. ** Between-subjects effect.
pared with those who completed the follow-up survey only and those who could not be matched with an initial survey. Table 1 shows that among those who were unmatched, the change in acceptance is at least as great as that for the matched group. Exploration of effects of smoking status on attitude change was limited to the matched sample. As might be expected, smokers were less likely to approve of the ban, but interestingly, as can be seen in Table 2, there was an interaction between smoking status and time of survey which analysis of variance showed to be highly significant (F = 9.88, df = 2, 1,931; P < 0.0001); smokers increased their level of approval more than did nonsmokers following implementation of the bans. When smoking status was controlled, there were no relationships between changes in acceptance and age, sex, or educational level. The differences as a function of educational status before the ban persisted after the ban. In response to the question in the second survey, “overall, the ban was a good thing,” 95% of never-smokers, 90% of ex-smokers, and 57% of smokers agreed or agreed strongly. This variable was significantly correlated with the follow-up attitude scale (r = 0.81, P < O.OOOl),as might be expected. In response to the question about inconvenience caused by the ban, a bare majority (53%) of smokers agreed that the ban was an inconvenience to them compared with 2% of never-smokers and 3% of ex-smokers. To investigate the relationship between reported inconvenience and attitude to the ban among smokers, a one-way analysis of variance was performed using data TABLE 2 CHANGES IN LEVELS OF APPROVAL OF THE BAN AS A FUNCTION OF SMOKING STATUS (MATCHED SAMPLE ONLY)
Mean attitude score (with 95% CIs) Survey
Never-smokers (n = 1,086)
Ex-smokers (n = 433)
Smokers* (n = 451)
Before ban After Difference
8.16 (7.93, 8.40) 8.55 (8.34, 8.76) 0.39 (0.56, 0.22)
7.09 (6.65, 7.54) 7.85 (7.44, 8.25) 0.75 (0.41, 1.09)
0.56 (-0.01, 1.11) 1.76 (1.23, 2.29) 1.20 (0.82, 1.58)
* Smoking status at the initial survey.
319
SMOKING BANS
on attitude toward the ban at follow-up, compared across the five levels of the inconvenience variable. As can be seen from Table 3, there was a highly significant effect, with those who agreed strongly that they were inconvenienced by the ban being against the ban, those unsure or merely agreeing were neutral, while those who disagreed or disagreed strongly about being inconvenienced were in favor of the ban. To explore the extent to which smokers inconvenienced by the ban might disapprove or approve of it overall, the attitude variable was collapsed into two categories: a score SO represented overall negative or neutral attitude, while a score >O represented a positive attitude. From this analysis, it was found that 3% of those smokers who reported themselves to be inconvenienced by the ban generally approved of it. This level of support is confirmed by the finding that 41% of smokers who reported themselves to be inconvenienced thought that the ban was a good thing overall. DISCUSSION There was a clear increase in levels of approval of the workplace smoking bans following their implementation, and smokers appeared to have increased their levels of approval more than did nonsmokers. Further, many of those who reported being inconvenienced by the ban considered it to be a good thing overall. These results suggest that following the implementation of the ban, it became an accepted part of the work environment. The increases in acceptance suggest that these workers saw the merits of this innovation and were able to adjust to any inconvenience that it caused them. A sizeable proportion of smokers were willing to accept what they regarded as an inconvenience, because they recognized the overall benefits of the ban. One-third of the smokers reported that they were inconvenienced and that they disapproved of the ban. This subgroup may require extra attention when worksite smoking restrictions are introduced. Opportunities to discuss the impact of the bans on them, access to nonprescription or prescription nicotine gum, or other support in coming to terms with the ban could be considered. The findings of high compliance is suggestive that at least where staff are involved in choosing the manner of phasing in restrictions, the vast majority of statf can and do comply after a mandated deadline. The diversity in the mode of TABLE 3 ATTITUDE OF SMOKERS TO THE BAN FOLLOWING ITS IMPLEMENTATION AS A FUNCTION OF THE EXTENT TO WHICH THEY BELIEVED THAT THE BAN HAD INCONVENIENCED THEM (MATCHED SAMPLE ONLY)
Inconvenienced Strongly agree (n = 76) Agree (n = 175) Can’t say (n = 27) Disagree (n = 146) Strongly disagree (n = 52)
Mean attitude score
95% confidence interval
-4.46 +0.65 -0.26 +4.57 +6.92
(-5.60, -3.32) (-0.02, 1.33) (- 1.53, 1.01) (3.94, 4.90) (5.32, 8.52)
320
BORLAND
ET AL.
phasing in of restrictions prior to the mandated ban suggests that it is the consultation, rather than the specific approach that results from it, that is important. This study extends the generality of findings about workplace smoking bans from the health sector (primarily hospitals) in the United States where all previous studies had been conducted (9-12) to white collar public sector employees, in another industrialized country. Smokers were somewhat less likely to complete both surveys. One possible explanation is that some disgruntled smokers refused to complete the second survey, which could have shown lower overall acceptance of the ban had their data been included. We believe this scenario is unlikely since the Time 2 (followup) mean acceptance in the unmatched sample was high, and higher than their Time 1 acceptance. The drop in the percentage of smokers in the unmatched group from pre- to postban was 1.9% compared with 1.2% for the matched group. Part of the differential in responding as a function of smoking status may have been due to the higher absenteeism that commonly occurs among smokers (18), making them relatively less likely to be at work on any given day. Even if we assume the most biasing possibility-that there was selective attrition between surveys of smokers with the utmost negative attitudes-the increase in acceptance cannot be explained away. Even if 2% of the unmatched sample (all smokers) threw their surveys away, and assuming that all of these would have responded with a maximally negative score ( - 12), the average level of acceptance at follow-up would only drop from 6.05 to 5.70 (a reduction of 0.35). This would reduce the change from Time 1 to Time 2 for the unmatched groups to 0.83, a difference that is still numerically greater than the change in the matched sample. The existence of a shift toward greater acceptance following the ban does appear to occur for all subgroups identified and is unlikely to be due to any sampling artifact. However, it should be noted that the actual levels of acceptance of the ban for the matched sample is not representative, as its difference from the unmatched survey groups shows; the true level of acceptance presumably lies somewhere between these two estimates. From a theoretical point of view, the results are of particular interest. They provide non-laboratory support for theories that postulate a drive, or at least a tendency, to bring attitudes into line with behavior, regardless of whether the behavior is voluntarily performed. Just as attitudes may determine volitional behavior (16), it seems that enforced behavior change is followed by attitude change (14, 15). The general significance of the support provided by this study for such theories may be judged by its having occurred in a field setting and being focused upon an issue with which all respondents were familiar and which was of demonstrable importance to them. The small reduction in the percentage of smokers at follow-up is likely to reflect a true change in smoking prevalence as self-reports of smoking status have been shown to be reliable in Australian samples (20). The reduction is at the higher end of that which might be expected for spontaneous cessation based on published data on changes in smoking prevalence in Australia (17). Some smoking cessation might be due to the ban, but in the absence of a control group this cannot be demonstrated (21).
SMOKING BANS
321
CONCLUSION Although levels of acceptance of the bans were quite high, it is clear that there remained a proportion of smokers who were disgruntled. While smokers were offered assistance to stop or regulate their smoking habit before the introduction of the ban, it might be that they need and would be willing to accept extra help in facing the reality of extended periods of time without cigarettes. Perhaps an approach in these circumstances might be to offer more extensive worksite-wide health promotion programs (22). Such programs offer personal assistance in such areas as smoking cessation, dietary improvement, and exercise. But they also have mass-communication components, which promote broader goals of health improvement. Efforts to promote a broader and more well-informed understanding of health issues in workplaces where smoking bans are introduced may help create a more positive perspective among those whose attitudes are more likely to be, and remain, negative. For example, a recent trial in the United States (23) offered a combination of self-help materials, personal advice, and monetary incentives to smokers in a large worksite. They also conducted an internal “media” campaign to better educate smokers and persuade them to attempt cessation when the services were ultimately offered. This program resulted in rates of smoking cessation that were higher than those in a control group. The introduction of workplace smoking bans, particularly given the fact that the majority of workers see them in a positive light, does provide a context, or at least a background, in which broader educational efforts and more specific health promotion programs might also be offered to workers. However, our findings suggest that the bans themselves have a positive effect on the attitudes of both smokers and nonsmokers, and their long-term benefits and contributions to public health should not be underrated. ACKNOWLEDGMENTS This study was funded by the Anti-Cancer Council of Victoria and by a grant from the Anti-Cancer Foundation of the Universities of South Australia. Thanks go to Trudy De Luise, Barbara Murphy, and Elvira Spano for assistance in conducting the survey and to Penny Schofield for data analysis.
REFERENCES 1. Armstrong BK. Commentary: Passive smoking and lung cancer. Community Health Stud 1987; 11:6&s. 2. Peto J, Doll R. Passive smoking. Br .I Cancer 1986; 54381-383. 3. United States Surgeon General. Smoking and Health: A Report of the Surgeon General. Washington: United States Department of Health, Education and Welfare, Public Health Service, 1979. 4. United States Surgeon General. The Health Consequences of Involuntary Smoking. U.S. Department of Human Services and Health, 1986. 5. Clutterbuck D. Persuading employees to break the smoking habit. Znt Management 1981; 36:2729. 6. National Health and Medical Research Council. Effects of Passive Smoking on Health. Report of NH and MRC Working Party on the Effects of Passive Smoking on Health. Canberra: Australian Government Publishing Service, 1987. 7. Woodward A, McMichael AJ. Smoking at work. Aust N Z J Occup Health Safety 1987; 3(6):578 583.
322
BORLAND
ET AL.
8. Hill D. Public opinion about smoking in restaurants and work. Med J Aust 1986; 145:657-658. 9. Andrews JL. Reducing smoking in the hospital: An effective model program. Chest 1983; 84:206209. 10. Beiner L, Abrams DB, Follick MJ, Dean L. A comparative evaluation of a restrictive smoking policy in a general hospital. Am J Public Health 1989; 79192-195. 11. Petersen LR, Helgerson SC, Gibbons CM, Calhoun CR, Ciacco KH, Pitchford KC. Employee smoking behaviour changes and attitudes following a restrictive policy on worksite smoking in a large company. Public Health Rep 1988; 103:115-120. 12. Rosenstock IM, Stergachis A, Heany C. Evaluation of smoking prohibition policy in a health maintenance organisation. Am J Public Health 1986; 76~10%1015. 13. Borland R, Hill DJ, Owen N, Chapman S. Acceptance of workplace bans on smoking in Commonwealth Government Offices. Med J Aust 151:525-528. 14. Bern DJ. Self-perception theory. In: Berkowitz L, Ed. Recent Advances in Experimental Social Psychology. New York: Academic Press, 1972. 15. Festinger L. A Theory of Cognitive Dissonance. New York: Harper and Row, 1957. 16. Fishbein M, Ajzen I. Belief, Attitude, Intention and Behaviour: An Introduction to Theory and Research. Boston: Addison-Wesley, 1975. 17. Hill D, White V, Gray N. Measures of tobacco smoking in Australia 1974-1986 by means of a standard method. Med J Aust 1988; 149:1&12. 18. Smith GC, Athanansou JA, Reid CC, Ng TKW, Ferguson DA. Sickness absence, respiratory impairment and smoking in industry: An Australian study. Med J Aust 1981; 1:235-237. 19. Anastasi A. Psychological Testing, 4th ed. New York: MacMillan, 1976. 20. Dwyer T, Pierce JP, Hannan CD, Burke N. Evaluation of the Sydney “Quit for Life” antismoking campaign. Part 2. Changes in smoking prevalence. Med J Aust 1986; 144~344-347. 21. Borland R, Chapman S, Owen N, Hill D. Effects of workplace smoking bans on cigarette consumption. Am J Public Health 1990; in press. 22. Fielding JE. Health promotion and disease prevention at the worksite. Ann Rev Public Health 1984; 5~237-266. 23. Windsor RA, Lowe JB, Bartlett E. The effectiveness of a worksite self-help smoking cessation program: A randomised trial. J Behav Med 1988; 11:407-421. 24. Velicer WF, Di Clemente CC, Prochaska JO, Brandenburg N. Decisional balance measure for assessing and predicting smoking status. J Pers Sot Psycho1 1985; 881279-1289.