Changes in smoking among restaurant and bar

Health Promotion International, Vol. 23 No. 1 doi:10.1093/heapro/dam041 Advance Access published 18 December, 2007

# The Author (2007). Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]

Changes in smoking among restaurant and bar employees following Norway’s comprehensive smoking ban MARC T. BRAVERMAN1*, LEIF EDVARD AARØ2 and JØRN HETLAND2

SUMMARY Norway implemented a nationwide ban on indoor smoking in June 2004. This study documents the smoking patterns of Norway’s restaurant and bar workers before and after the ban, to determine changes in smoking prevalence and explore which individual and environmental characteristics were related to cessation. A national sample of food service workers was surveyed by telephone or Internet immediately before the ban and at 4 and 11 months post-implementation. Results showed that between baseline measurement and 4 months post-implementation, there were significant declines in prevalence of daily smoking (23.6% points, p , 0.005), daily smoking at work (26.2% points, p , 0.001), number of cigarettes smoked by continuing smokers (21.55, p , 0.001) and number of cigarettes smoked at work by continuing smokers (21.63, p , 0.001). No significant changes occurred in any of these variables between 4 and 11

months post-implementation. Logistic regression analysis revealed that only smokers’ intentions at baseline to quit within 30 days predicted cessation at both follow-up time points. In addition, cessation at 4 months was predicted by lower daily cigarette consumption at baseline, whereas cessation at 11 months was predicted by baseline attitude toward ETS and exposure to ETS as measured at followup. In sum, Norway’s smoking ban was accompanied by a reduction in smoking in the period immediately following the ban, and the reduction was maintained almost a year later. The finding that smoking cessation was consistently associated with smokers’ intentions to quit within 30 days suggests that motivational and support programs could play a significant role in boosting cessation rates. It is recommended that targeted interventions be used to supplement the benefits of a comprehensive ban to achieve tobacco control objectives.

Key words: smoking cessation; tobacco use cessation; smoking bans

INTRODUCTION In June 2004, Norway became the second country to implement a nationwide ban on all indoor smoking, following Ireland by 3 months. The major impetus for the Norwegian legislation, as with most smoking bans (Farrelly et al., 1999; U.S. Department of Health and Human Services, 2006), was the need to protect the health of workers by reducing their exposure to environmental tobacco smoke

(ETS). In achieving this aim, the evidence to date indicates that the ban has been highly successful, as measured by analyses of air quality and a variety of physiological indices including pulmonary function and respiratory symptoms (Eagan et al., 2006; Skogstad et al., 2006). In addition to the primary aim of protecting workers’ health by eliminating ETS, worksite bans may also influence the smoking habits of employees. Evidence indicates that worksite bans often lead to increases in the number of 5

Downloaded from http://heapro.oxfordjournals.org/ by guest on May 31, 2013

1 Department of Human Development and Family Sciences, Oregon State University, 161 Milam Hall, Corvallis, OR 97331, USA and 2Faculty of Psychology, University of Bergen, Norway *Corresponding author. E-mail: [email protected]

6

M. T. Braverman et al.

METHODS Sample Survey respondents consisted of 1525 employees working in restaurants, cafeterias, bars and other eating and drinking establishments in May 2004. The sample was selected through a two-stage process. First, a random sample of businesses was identified, using a public register on which all companies in the Norwegian hospitality industry are listed. These businesses were contacted by telephone. Second, employees from each establishment were identified and invited to participate. To avoid selection bias, the identification of employees to be interviewed followed a procedure involving random selection of the first letter of employee surnames. The overall participation rate was estimated at 53%. Participants were given the choice between being interviewed by telephone (N ¼ 1337) or completing the questionnaire online (N ¼ 188).

Measures At the first survey administration, respondents were asked about background demographic variables, the type of establishment they worked at and their work situations (hours per day, time of day, etc.). In later administrations, they also were asked about whether their work situations had changed. The questions on smoking behavior were identical across all three data collection occasions, with information being collected from all respondents about current smoking status (daily, occasional, nonsmoker) and, for smokers, about the average number of cigarettes smoked daily, smoking habits at work, number of cigarettes smoked daily at work and intentions to quit smoking (within 30 days, 6 months, longer time period, no specific plans). Other questions included respondents’ attitudes at baseline toward the upcoming indoor smoking ban (favor, neutral, oppose) and attitudes toward ETS. A series of questions also asked if they were experiencing respiratory symptoms or problems (coughing in the morning, wheezing, etc.). Two scales were created from these survey items: attitude toward ETS (seven items, Cronbach’s a ¼ 0.81) and respiratory symptoms (five items, Cronbach’s a ¼ 0.77). In addition, respondents were asked


quit attempts, frequency of successful smoking cessation, and reductions in overall cigarette consumption (Glasgow et al., 1997; Chapman et al., 1999; Heloma and Jaakkola, 2003; Levy and Friend, 2003; Bauer et al., 2005; Gallus et al., 2006). There may also be favorable effects on community-level factors that may be associated with cessation, such as prevalence estimates and norms regarding the acceptability of smoking (Albers et al., 2004). There are several reasons why smoking bans might result in reduced smoking levels among employees (Chapman et al., 1999; Levy and Friend, 2003). The increased social attention to tobacco control issues that accompanies the ban may stimulate smokers’ motivation to quit, due to a shift in perceived social norms or other psychosocial processes. Further, there is reduced opportunity to smoke during the day, and smoking may become so logistically awkward that the perceived benefits of quitting can become more salient (Parry et al., 2000). One major category of worksite is the food service industry, including restaurants, bars and other establishments. For several reasons, employees in these businesses comprise a particularly important sector of the workforce on which to focus smoking reduction goals. First, these workers often experience higher ETS exposure in the workplace, even when smokefree policies are present, due to frequent violations (U.S. Department of Health and Human Services, 2006). Second, they have been found to smoke in greater proportions, and at higher levels of consumption, than workers in most other industries (Siegel et al., 2006). Third, a relatively high proportion of restaurant employees are in their 20s or younger, so policies that can promote smoking reduction within this group may have significant benefits far into the future. This article reports on the smoking levels in a national sample of workers in Norway’s restaurants and bars, surveyed 1 month before the ban was implemented and at two time points— 4 and 11 months—thereafter. We examine two primary questions: (i) How did the smoking patterns of food service employees change during the period before and after implementation of the ban, in terms of both smoking prevalence and the number of cigarettes consumed? (ii) What were the demographic, individual and environmental factors that may predict smoking cessation within this group?

Norway’s comprehensive smoking ban

to what degree they experienced smoke from others, both in the work setting and at home or other locations.

Statistical analyses Data were analyzed with SPSS version 15.0. Changes in smoking prevalence across the three data collection points were tested with McNemar’s test for change in proportions. Changes in the mean numbers of cigarettes smoked were tested with repeated measures analyses using SPSS’s general linear modeling procedure. The examination of predictors for smoking cessation was conducted with logistic regression analyses that assessed smoking cessation at each of the two follow-up points, relative to baseline.

RESULTS Descriptive information on sample Table 1 presents descriptive information on the sample at Time 1, including demographic breakdowns and description of the work settings, along with the distribution of smoking status for each subgroup. Slightly, more than half the sample (52.3%) was under 35 years old. The modal level of educational attainment was completion of secondary education (consisting of 13 years of schooling in Norway; 58.0%). Annual income for the majority of the sample (57.1%) fell between 200 000 Norwegian kroner (US$29 120 or E24 300 in May 2004) and 399 000 kroner.

A high proportion of the sample (52.9%) were daily smokers, with another 7.7% smoking occasionally. The percentage of daily smokers was roughly similar across gender but varied across age, with daily smoking being particularly high among 15- to 24-year olds (56.4%). Daily smoking was also high among those whose highest education was lower secondary school (58.4%), and those with an income between 100 000 and 199 000 kroner (60.2%). Across types of establishments, the highest prevalence of daily smoking was found in nightclubs (64.9%) and bars (58.5%). Sample attrition analysis Among the 1525 respondents at T1, 879 (57.6%) remained in the sample at T2, and 579 (38.0%) remained at T3. The most common reason for attrition was that respondents had left their jobs and could not be located. To examine the possibility of sample bias due to this attrition, two sets of analyses were conducted on a broad range of variables, comparing respondents who remained in the sample at each follow-up with those who did not. First, T1 respondents who did participate at T2 (N ¼ 879) were compared with T1 respondents who were lost (N ¼ 646), on the T1 variables of smoking status, smoking status at work, intentions to quit, gender, age, income, type of establishment, time employed at current location, time of day worked, smokers’ average daily cigarette consumption and smokers’ average daily consumption at work. No significant differences (a ¼ 0.05) were found between the groups on any of these comparisons. Second, to test for attrition bias at the second follow-up (T3), respondents who participated at all three data points (N ¼ 579) were compared with those who responded at T1 and T2 but were lost at T3 (N ¼ 300). The groups were compared on the T2 variables of smoking status, smoking status at work, intentions to quit, time of day worked, whether they had changed jobs since T1, smokers’ average daily cigarette consumption and smokers’ average daily consumption at work, as well as the T1 variables of gender, age, income and type of establishment. Significant differences were found between the groups for only two of these variables: age ( p ¼ 0.004), with retained respondents being somewhat older than those lost, and having changed jobs between T1 and T2


Procedures A summary of the study protocol was reviewed and approved by the Regional Committee for Medical Research Ethics for Western Norway, the governing oversight body for human subjects review, located at the University of Bergen, Faculty of Medicine. Data were collected in May 2004 (Time 1 or T1), with follow-up data collection taking place in September/October 2004 (T2) and May 2005 (T3). Attempts were made to recruit all those who participated at baseline for the follow-up interviews. For each participant, the mode of data collection ( phone or online) remained consistent across follow-up waves.

7

8


Table 1: Description of sample at baseline (T1) N (%)a

a

Daily smoker (%)

Occasional smoker (%)

Nonsmoker (%)

1525 (100)

52.9

7.7

39.4

719 (47.1) 806 (52.9)

51.3 54.3

7.8 7.6

40.9 38.1

321 (21.0) 478 (31.3) 403 (26.4) 211 (13.8) 112 (7.3)

56.4 53.6 53.3 48.3 47.3

7.2 10.5 5.5 6.6 7.1

36.4 36.0 41.2 45.0 45.5

36 (2.4) 221 (14.5) 883 (58.0) 383 (25.1)

41.7 58.4 54.4 47.8

11.1 5.4 8.2 7.6

47.2 36.2 37.5 44.6

112 (7.3) 259 (17.0) 871 (57.1) 104 (6.8) 17 (1.1) 7 (0.5) 73 (4.8) 82 (5.4)

45.5 60.2 54.5 45.2 58.8 28.6 37.0 47.6

8.9 7.7 7.6 6.7 11.8 0.0 4.1 11.0

45.5 32.0 37.9 48.1 29.4 71.4 58.9 41.5

166 (10.9) 967 (63.4) 392 (25.7)

51.2 54.0 51.0

8.4 7.3 8.2

40.4 38.7 40.8

222 (14.6) 325 (21.3) 978 (64.1)

49.5 55.1 53.0

5.4 9.5 7.6

45.0 35.4 39.5

467 (30.6) 339 (22.2) 38 (2.5) 501 (32.9) 94 (6.2) 86 (5.6)

50.3 48.7 44.7 58.5 64.9 41.9

9.6 6.2 10.5 7.0 4.3 9.3

40.0 45.1 44.7 34.5 30.9 48.8

Some subcategories do not add up to 1525 due to missing data. Within the three Smoking Status columns, figures in each row add up to 100%.

b

( p ¼ 0.018), with 97% of the retained respondents remaining in the same job versus 94% of the respondents lost. This second result reinforces the assertion that job instability was one factor underlying attrition. These comparisons provide assurance that the attrition across data points was not confounded with variables pertaining to these analyses, and thus did not introduce significant levels of bias. Finally, in further consideration of potential attrition effects, the analyses of changes in smoking prevalence and cigarette consumption

between T1 and T2 are presented for two subsamples: all respondents who participated at T1 and T2 (N ¼ 879), and those who participated only at all three time points (N ¼ 579). The first of these subsamples (which we will refer to as T1 –T2) includes the entire second subsample (T1 –T2 –T3) plus respondents who did not participate at T3. Providing data from both groups helps assess the degree to which the smaller group with more complete longitudinal data (T1 –T2 –T3) might be affected by attrition. As will be seen, the two subsamples provide very


Full sample Gender Male Female Age 15– 24 25– 34 35– 44 45– 54 55– 69 Education Elementary (8 years or less) Lower secondary (9– 10 years) Secondary (11–13 years) Some university Personal income ,100 000 kroner 100 000–199 000 200 000–399 000 400 000–599 000 600 000–799 000 800 000 or more Refused to answer Do not know Hours worked per week 20 or less 21– 40 More than 40 Time of day Day Evening or night Variable Type of establishment Restaurant Cafeteria/Cafe´ Coffee bar Pub/Bar Nightclub/Disco Other

Smoking statusb


similar results for the changes between T1 and T2.

not significant. For occasional smokers, there were no significant changes across any of the three time points. It is noteworthy that all four of the indicators we examined—overall prevalence, prevalence at work, overall daily cigarette consumption and daily consumption at work—display the same pattern: a significant drop (varying between 7.1% for daily smoking prevalence and 20.8% for cigarette consumption at work) between baseline and the first post-legislation measurement 4 months later, followed by a stabilization (i.e. no significant change) over the following 7 months. The two distinct categories of smoking reduction—decline in daily smoking prevalence and reduction in consumption by continuing smokers—were combined to estimate the reduction in overall cigarette consumption (Fichtenberg and Glantz, 2002). We calculated the per capita consumption of cigarettes for respondents in the T1 –T2 – T3 sample, coding daily cigarette consumption as zero for nonsmokers. The mean number of cigarettes consumed per capita was 8.07 at T1, 6.93 at T2 and 7.12 at T3, for an overall decline of 11.8% across the three periods. Who were the smokers who quit? Finally, we examined whether smoking cessation at either 4 or 11 months post-legislation could be predicted from a number of demographic, individual and environmental variables. For that analysis, we created two dichotomous variables reflecting cessation from T1 to T2 and T1 to T3, respectively. The variables were coded 1 for baseline smokers (both daily and occasional) who reported not smoking at all at the respective follow-up and 0 for baseline smokers who reported smoking either daily or occasionally at follow-up. These dichotomous variables were used as dependent variables in separate logistic regressions (Table 4). Only respondents with comprehensive data across all time points (the T1 –T2 –T3 sample) were included in these regression analyses. The predictors for both regressions included gender, age, income and several individual variables measured at baseline: number of cigarettes smoked daily, intentions to quit, attitude toward the upcoming ban (which was set to take effect in June 2004, 1 month after the baseline survey), the seven-item attitude toward


Smoking prevalence and cigarette consumption before and after legislation Table 2 displays changes over time in both general smoking prevalence and smoking prevalence at work. Between T1 and T2, both samples reflect that prevalence of daily smoking declined (24.6%, p , 0.001, in the T1 –T2 sample and 23.6%, p , 0.005, in the T1 –T2 – T3 sample). Concomitant increases occurred in both occasional smoking and nonsmoking. By contrast, between T2 and T3 there was no change in daily smoking prevalence (0.0% change) in the T1 –T2 –T3 sample. A similar pattern was observed for smoking prevalence at work. Both samples reflected that daily smoking prevalence declined between T1 and T2 (26.8%, p , 0.001, in the T1 –T2 sample and 26.2%, p , 0.001, in the T1 –T2 – T3 sample). Between T2 and T3, however, there was no significant change (1.7% increase, NS). Table 3 displays the average numbers of cigarettes smoked by daily and occasional smokers at the three time points, both throughout the day and specifically at work. The table includes only those smokers who did not change their status; thus, for example, daily smokers who later reported being occasional smokers or nonsmokers are excluded from these figures. By reflecting consumption only among smokers whose overall smoking status remained stable, these data provide information that is independent of the changes presented in Table 2, which reflect respondents’ shifts across smoking status categories. For daily smokers, both subsamples show significant reduction in overall daily cigarette consumption between T1 and T2 (16.31 cigarettes to 14.86, p , 0.001, in the T1 –T2 sample, and 16.31 to 14.76, p , 0.001, in the T1 –T2 –T3 sample). Between T2 and T3, there was no significant change (14.76 –14.92, NS). For occasional smokers, there were no significant changes across any of the time periods in number of cigarettes smoked. With regard to the consumption at work, the mean number of cigarettes smoked by daily smokers declined (7.96 cigarettes to 6.32, p , 0.001, in the T1 –T2 sample, and 7.85 –6.22, p , 0.001, in the T1 –T2 –T3 sample), and again the change between T2 and T3 (6.22 –6.44) was

9

10

Sample

All respondents at T1 and T2 (N ¼ 879)

T1: Baseline (%)a

Prevalence at work Smoke at work daily Smoke at work occasionally Do not smoke at work

T1 –T2 change Change (%)

X2 b

T2– T3 change p

Change (%)

X2

b

p

51.8 6.8 41.4 100

47.2 9.1 43.7 100

24.6 2.3 2.3

17.686 ,0.001

46.0 8.8 45.3 100

39.2 12.7 48.0 100

26.8 3.9 2.7

24.201 ,0.001

50.8 6.6 42.7 100

47.2 7.8 45.1 100

47.2 8.6 44.2 100

23.6 1.2 2.4

8.163 ,0.005

0.0 0.8 20.9

0.0

44.7 8.6 46.6 100

38.5 12.8 48.7 100

40.2 11.1 48.7 100

26.2 4.2 2.1

13.611 ,0.001

1.7 21.7 0.0

0.988 NS

General prevallence Daily smokers Occasional smokers Nonsmokers Prevalence at work Smoke at work daily Smoke at work occasionally Do not smoke at work

a

T3: 11 months post-legislation (%)

General prevalence Daily smokers Occasional smokers Nonsmokers

Respondents at all 3 time points (N ¼ 579)

T2: 4 months post-legislation (%)a

Some figures do not sum exactly to 100% because of rounding error. McNemar’s test for change in the proportion of daily smokers.

b

NS


Table 2: General smoking prevalence and prevalence at work, before and after legislation


Table 3: Average daily cigarette consumption by smokers who remained daily or occasional smokers, before and after legislation Sample

T1: Baseline

T2: 4 months post-legislation

T3: 11 months post-legislation

T1 –T2 change Change

All respondents at T1 and T2

a

T2 –T3 change p

Change

Fb

p

16.31 (7.64)a 3.88 (4.34)

14.86 (7.56) 2.56 (3.22)

21.45 21.32

27.629 ,0.001 8.860 0.006

7.96 (5.41) 2.06 (2.45)

6.32 (5.01) 1.84 (2.19)

21.64 20.22

59.409 ,0.001 0.216 NS

16.31 (7.44) 2.94 (3.61)

14.76 (7.09) 2.00 (2.87)

14.92 (7.84) 5.00 (9.22)

21.55 20.94

21.916 ,0.001 4.088 NS

0.16 3.00

0.237 NS 1.569 NS

7.85 (5.21) 1.41 (1.62)

6.22 (4.61) 1.24 (1.39)

6.44 (5.30) 1.53 (1.74)

21.63 20.17

34.011 ,0.001 0.246 NS

0.22 0.29

0.536 NS 0.627 NS

Standard deviation in parentheses. GLM repeated measures test for significance of change in means.

b


Cigarettes smoked daily Daily smokers (N ¼ 390) Occasional smokers (N ¼ 32) Cigarettes smoked daily at work Daily smokers (N ¼ 378) Occasional smokers (N ¼ 32) Respondents at all 3 time points Cigarettes smoked daily Daily smokers (N ¼ 234) Occasional smokers (N ¼ 17) Cigarettes smoked daily at work Daily smokers (N ¼ 225) Occasional smokers (N ¼ 17)

F

b

11


12


Table 4: Logistic regression predicting nonsmoking status at Time 2 and Time 3 Variable

a

T1 (Baseline) to T3 (11 months post-legislation) (N ¼ 319)

N

Odds ratio (95% C.I.)

N

225 270

Ref 1.05 (0.53, 2.11)

144 0.881 175

Ref 1.53 (0.61, 3.84)

0.361

101 171 131 63 29

Ref 1.60 (0.62, 4.12) 0.80 (0.26, 2.46) 1.09 (0.32, 3.74) 1.71 (0.38, 7.80)

58 0.330 107 0.693 93 0.898 41 0.487 20

Ref 1.31 (0.38, 4.59) 0.95 (0.23, 3.89) 0.82 (0.16, 4.25) 3.03 (0.60, 15.34)

0.672 0.942 0.816 0.180

120 335 40 495

Ref 1.77 (0.70, 4.44) 1.45 (0.37, 5.73) 0.925 (0.87, 0.98)

72 0.225 224 0.597 23 0.007 319

Ref 0.61 (0.22, 1.65) 0.63 (0.11, 3.54) 0.99 (0.93, 1.06)

0.329 0.599 0.796

261 84 81 69

Ref 4.24 (1.85, 9.69) 2.22 (0.90, 5.49) 0.39 (0.09, 1.81)

0.001 0.084 0.229

175 48 53 43

Ref 3.44 (1.08, 11.01) 2.26 (0.71, 7.19) 2.07 (0.59, 7.33)

0.037 0.169 0.258

382 113

Ref 0.85 (0.37, 1.98)

0.704

235 84

Ref 1.09 (0.42, 2.88)

0.856

100 162 233 495

Ref 0.66 (0.30, 1.48) 0.43 (0.17, 1.05) 1.04 (0.96, 1.13)

86 0.315 118 0.064 115 0.310 319

Ref 0.64 (0.25, 1.62) 0.15 (0.04, 0.64) 1.14 (1.02, 1.27)

0.343 0.010 0.017

123 188 184 495

Ref 2.29 (0.85, 6.17) 0.68 (0.17, 1.05) 1.01 (0.79, 1.29)

85 0.103 124 0.415 110 0.960 319

Ref 2.13 (0.65, 6.94) 0.34 (0.11, 1.08) 0.95 (0.68, 1.32)

0.210 0.068 0.753

p

Odds ratio (95% C.I.)

p

Exposure to ETS variables: reported at T2 in first model and at T3 in second model.

ETS scale and the five-item respiratory symptoms scale. The predictors also included two environmental variables: exposure to ETS from others’ smoke in the workplace, and exposure to ETS from others’ smoke at home or in other locations. For the model predicting smoking cessation over the 4 months from T1 to T2, the proportion of variance explained (Nagelkerke R 2) was 0.194. Cessation was predicted significantly by only two variables, though those were highly significant: number of cigarettes smoked daily at baseline (odds ratio ¼ 0.925, p ¼ 0.007), and respondents’ intentions to quit smoking within the next 30 days (OR ¼ 4.24, p ¼ 0.001).

For the model predicting cessation over the 11 months from T1 to T3, the overall proportion of variance explained (Nagelkerke R 2) was 0.215. Once again, 30-day quit intentions was a significant predictor (OR ¼ 3.44, p ¼ 0.037), though it was not as strong as in the 4-month prediction. Number of cigarettes smoked daily at baseline did not predict 11-month cessation. However, baseline attitude toward ETS did predict cessation (OR ¼ 1.14, p ¼ 0.017), and moderate or high current exposure to ETS (at the time of follow-up measurement; low values represent high exposure) was also strongly significant (OR ¼ 0.15, p ¼ 0.010). Thus, over both time intervals, 30-day quit intentions emerged


Gender Male Female Age 15– 24 25– 34 35– 44 45– 54 55– 69 Income level Up to 200 000 kroner 200 000 kroner or more Refused or do not know N cigarettes smoked daily at baseline Intentions to quit (at baseline) No plans or do not know Yes, within 30 days Yes, more than 30 days Yes, more than 6 months Exposure to ETS at work (at time of measurement)a No exposure Some exposure Exposure to ETS at home or elsewhere (at time of measurement)a No exposure Some exposure Moderate or high Attitude toward ETS (at baseline) Attitude toward ban (at baseline) Neutral Negative Positive Respiratory symptoms (at baseline)

T1 (Baseline) to T2 (4 months post-legislation) (N ¼ 495)


as the only consistent predictor of cessation over the study period. DISCUSSION

The very high levels of smoking reported by this sample of food service workers underscore the critical importance of working with this population as an audience for tobacco control efforts. As noted in Table 1, 51.3% of males and 54.3% of females in our sample were daily smokers at baseline. By contrast, in a survey conducted 5 months earlier, general smoking prevalence in Norway among adults ages 16–74 was only 26.3% (Statistics Norway, 2006, http:// www.ssb.no/english/subjects/03/01/royk_en/); if those national results are weighted by the age distribution of the present sample, the comparable proportions of daily smokers are estimated as 27.8% among males and 26.1% among females. Thus, smoking prevalence among food service workers stands at roughly twice the national average. Therefore, these results are quite encouraging in determining that consumption among restaurant and bar workers has declined following Norway’s ban. Was the Norwegian ban responsible for the observed smoking reductions? We believe these findings suggest that it was. Because the ban was comprehensive and nationwide, it was not possible to set up a simultaneous comparative design, which could attempt a definitive conclusion. However, the patterns found for all of the prevalence variables are strikingly consistent across time. The stabilization of these variables between the second and third time points indicates that the initial 4-month declines were not simply the result of a general downward trend in smoking. We know of no compelling alternative hypotheses to suggest why all of the smoking variables we measured in this sample would have declined over a brief 4-month period, apart from the ban and its attendant publicity. For example, whereas cross-sectional designs are subject to the potential criticism that selection bias may be operating (Farrelly et al., 1999), e.g. those smokers are leaving these workplaces and being replaced by nonsmokers, we followed the same respondents over the three data points. Therefore, these patterns suggest that the legislation did indeed contribute to the observed reductions in smoking within this population. One important area for further research is whether the reductions accompanying smoking bans extend across income and education brackets, given the strong inverse relationship between smoking prevalence and socioeconomic status (Siahpush et al., 2006).


Our results show that Norway’s implementation of a comprehensive smoking ban in June 2004 was associated with a significant decline in daily smoking among food service workers, as measured in October 2004. In the T1 –T2 –T3 sample, the reduction in daily smoking prevalence over those 4 months was 3.6% points, for a relative decline in daily smoking of 7.1%. Seven months later in May 2005—almost 1 year after the baseline measurements—the percentage of daily smokers remained unchanged, indicating that the decline had been maintained. This pattern of a significant reduction followed by stabilization was consistent in all of the smoking variables we examined: overall smoking status, smoking status at work, daily smokers’ overall cigarette consumption and daily smokers’ cigarette consumption at work. The combined effects of lowered smoking prevalence and continuing smokers’ lowered consumption resulted in an overall decrease in per capita cigarette consumption of 11.8% across the three measurement points. We also found that smoking cessation was predicted at both follow-up measurement points by smokers’ intentions to quit within 30 days, as expressed at baseline. How do these results coincide with findings from other nationwide smoking bans? The nationwide ban in Italy, which took effect in January 2005, was associated with a 7.6% decrease in overall consumption over one year (Gallus et al., 2006). The ban in Ireland was associated with a reduction in smoking prevalence of 3.1% between March 2004 and December 2006, based on 12-month moving averages (Office of Tobacco Control, 2007, http://www.otc.ie/). Finally, Finland’s national worksite smoking ban in 1995 was accompanied by a large drop in daily smoking prevalence from 29.8% to 24.6% over 1 year, a relative reduction of 17%, although 3 years later the rate had risen somewhat to 25.2% (Heloma and Jaakkola, 2003). With the exception of the Finnish study, these reports are generally consistent with the findings reported here for Norway, although comparability is limited because our data focus on a specific sector of the Norwegian workforce.

13

14


examined through chi-square analyses on mode choice ( phone versus online) crossed with smoking and demographic variables measured at T1. No differences were found between the groups on the four major smoking variables or on gender, household income, or length of time at present employment. Online respondents did tend to be younger ( p ¼ 0.001) and more highly educated ( p ¼ 0.016), a common finding in research on Internet use. Second, the study uses self-report data, with no independent validation of smoking status. Given the broad geographic scope of the sample, it was not possible to use multiple measurement strategies to validate self-reported smoking. We do not believe the absence of independent validation compromises the validity of these findings. The smoking behavior questions were carefully developed in accord with recommendations by the World Health Organization and the International Union Against Cancer (UICC), and the questions have been used in previous nationwide health surveys in Norway. All of the above-cited studies of national smoking bans similarly used self-report as the sole measurement strategy, and previous research has confirmed the validity of self-report for assessing smoking behavior in populationbased surveys (e.g. Assaf et al., 2002). In conclusion, these results add to a small but growing research literature on nationally legislated smoking bans and the changes in smoking behavior that may be associated with them. Such bans are singular events and the documentation of smoking prevalence in the time periods immediately before and after their implementation will contribute substantially to our understanding of the relationship between national policy and smoking behavior. These varying studies have begun to present a consistent picture that national, comprehensive indoor smoking restrictions will be accompanied by a reduction in cigarette consumption. Further research should examine the informational, educational and clinical interventions that can accompany these major shifts in national lifestyle, in order to take best advantage of the rich opportunities for achieving tobacco control objectives. FUNDING Directorate of Health and Social Affairs, Division for Tobacco Control, Norway.


With regard to the logistic regression analyses examining the predictors of smoking cessation, the strongest and most consistent predictor of which smokers would quit was the intention to quit smoking within the next 30 days. This finding highlights the fact that psychological preparation is generally a critical component of the cessation process. It also suggests that for many individuals, policy change—as represented here by the smoking ban—cannot be expected to be a fully sufficient impetus to produce behavioral change. The effects of policies can be accentuated by introducing program supports to help smokers quit. For example, it must be noted that despite the strong statistical relationship between 30-day quit intentions and respondents’ behavior, almost 80% of those in this sample who intended to quit within 30 days did not quit. We believe these success rates can be improved by combining smoking bans and other policy initiatives with opportunities for education and support. In Norway, there were three ongoing activities at the time of the ban that may have functioned to support individual smoking cessation efforts: recommendations for physicians and other health professionals to provide cessation counseling, training from the Directorate of Health and Social Affairs for leaders of cessation groups, and a nationwide telephone-based cessation helpline. These efforts could be supplemented with targeted interventions. Restaurants, cafes and bars, which are often the setting for health promotion activities involving patrons, can also be used effectively to reach their workers (Licata et al., 2002). Two methodological points bear discussion. First is the study’s use of two modes of data collection—telephone and online. Consistent with the growing popularity of mixed-mode surveys due to their potential to increase response rates and reduce cost (Dillman, 2000), we believe that providing respondents with the option to participate in their preferred mode helped to maximize their likelihood of remaining in the study across the three data points, thus strengthening the study’s design. We do not have comparative data on whether individual responses were affected by mode choice, but mode differences are generally of most concern when questions are highly sensitive (Groves et al., 2004); we do not consider that to be the case with this population. Initial differences between the two respondent groups were


ACKNOWLEDGMENTS Data collection for this study was sponsored by Norway’s Directorate of Health and Social Affairs, Division for Tobacco Control. We are grateful for valuable advice during the planning phase from Rita Lill Lindbak (Division for Tobacco Control), and Karl Erik Lund and Jostein Rise (Norwegian Institute for Alcohol and Drug Research). We are also grateful to Norman Constantine (University of California, Berkeley) for valuable comments on an earlier draft, and to Sam Vuchinich (Oregon State University) for suggestions on data analysis.

Albers, A. B., Siegel, M., Cheng, D. M., Rigotti, N. A. and Biener, L. (2004) Effects of restaurant and bar smoking regulations on exposure to environmental tobacco smoke among Massachusetts adults. American Journal of Public Health, 94, 1959–1964. Assaf, A. R., Parker, D., Lapane, K. L., McKenney, J. L. and Carleton, R. A. (2002) Are there gender differences in self-reported smoking practices? Correlation with thiocyanate and cotinine levels in smokers and nonsmokers from the Pawtucket Heart Health Program. Journal of Women’s Health, 11, 899–906. Bauer, J. E., Hyland, A., Li, Q., Steger, C. and Cummings, K. M. (2005) A longitudinal assessment of the impact of smoke-free worksite policies on tobacco use. American Journal of Public Health, 95, 1024 – 1029. Chapman, S., Borland, R., Scollo, M., Brownson, R. C., Dominello, A. and Woodward, S. (1999) The impact of smoke-free workplaces on declining cigarette consumption in Australia and the United States. American Journal of Public Health, 89, 1018– 1023. Dillman, D. (2000) Mail and Internet Surveys: The Tailored Design Method, 2nd edition, Wiley, New York. Eagan, T. M. L., Hetland, J. and Aarø, L. E. (2006) Decline in respiratory symptoms in service workers five months after a public smoking ban. Tobacco Control, 15, 242–246. Farrelly, M. C., Evans, W. N. and Sfekas, A. E. (1999) The impact of workplace smoking bans: results from a national survey. Tobacco Control, 8, 272– 277.

Fichtenberg, C. M. and Glantz, S. A. (2002) Effect of smoke-free workplaces on smoking behaviour: systematic review. BMJ, 325, 188– 191. Gallus, S., Zuccaro, P., Colombo, P., Apolone, G., Pacifici, R., Garattini, S., et al. (2006) Effects of new smoking regulations in Italy. Annals of Oncology, 17, 346– 347. Glasgow, R. E., Cummings, K. M. and Hyland, A. (1997) Relationship of worksite smoking policy to changes in employee tobacco use: findings from COMMIT. Tobacco Control, 6, S44– S48. Groves, R. M., Fowler, F. J., Couper, M. P., Lepkowski, J. M., Singer, E. and Tourangeau, R. (2004) Survey Methodology. Wiley, Hoboken, NJ. Heloma, A. and Jaakkola, M. S. (2003) Four-year follow-up of smoke exposure, attitudes and smoking behaviour following enactment of Finland’s national smoke-free work-place law. Addiction, 98, 1111– 1117. Levy, D. T. and Friend, K. B. (2003) The effects of clean indoor air laws: what do we know and what do we need to know? Health Education Research, 18, 592– 609. Licata, M., Gillham, K. and Campbell, E. (2002) Health promotion practices of restaurants and cafes in Australia: changes from 1997 to 2000 using an annual telemarketing intervention. Health Promotion International, 17, 255–262. Parry, O., Platt, S. and Thomson, C. (2000) Out of sight, out of mind: workplace smoking bans and the relocation of smoking at work. Health Promotion International, 15, 125– 133. Siahpush, M., McNeill, A., Borland, R. and Fong, G. T. (2006) Socioeconomic variations in nicotine dependence, self-efficacy, and intention to quit across four countries: findings from the International Tobacco Control (ITC) Four Country Survey. Tobacco Control, 15(Suppl. iii), iii71–iii75. Siegel, M., Barbeau, E. M. and Osinubi, O. Y. (2006) The impact of tobacco use and secondhand smoke on hospitality workers. Clinics in Occupational and Environmental Medicine, 5, 31– 42. Skogstad, M., Kjærheim, K., Fladseth, G., Gjølstad, M., Daae, H. L., Olsen, R., et al. (2006) Cross shift changes in lung function among bar and restaurant workers before and after implementation of a smoking ban. Occupational and Environmental Medicine, 63, 482– 487. U.S. Department of Health and Human Services (2006) The Health Consequences of Involuntary Exposure to Tobacco Smoke: A Report of the Surgeon General. Centers for Disease Control and Prevention; Office on Smoking and Health.


REFERENCES

15