As of the first half of 2005, 20 percent of Canadians and 14 percent of British Columbians currently smoked, with British Columbia (BC) being the province with the lowest prevalence.1 The prevalence rate for BC declined by 58 percent in the last 20 years, the largest decline in the country.2,3 Although the decline in prevalence is impressive, tobacco smoking remains the leading cause of preventable deaths in Canada, with estimates of preventable deaths attributable to tobacco smoking ranging from 25 percent to 50 percent in Canada,4 and those for BC at around 5,700 annually.5
Governments in developed countries have been taking a proactive approach to managing public health issues and undertaken interventions designed to change unhealthy behaviors. Research evidence is mounting that public health interventions that focus on smoking prevention or cessation can have a positive impact on healthy behavior,6–20 and that they are cost-effective,18,21 although impacts are not always been significant for all targeted groups,10,12 or persistent,13 and measured impacts vary.20
In light of an ongoing concern regarding the impact of smoking on population health, the BC Ministry of Health, with joint funding from Health Canada, recently undertook an initiative to develop a two-part smoking cessation mass media campaign that would target 20- to 30 year-old blue collar BC smokers, the demographic group with the highest smoking prevalence. The campaign comprised two segments, each lasting approximately 4 weeks. The first segment of the campaign was conducted in early 2005 and the second segment in early 2006. The objectives of the campaign were to inform the target population of the ministry's smoking cessation support program and encourage them to quit. The support program is available through an Internet Web site (www.quitnow.ca). The focus of this study is on the 2005 segment of the campaign.
The 2005 campaign ran between late February and early March, and consisted of a television and radio ad campaign aired over a period of approximately 4 weeks, together with a poster campaign. The costs of media buys amounted to $338,564 Canadian dollars or around 10 cents per BC resident aged 15 years or older, a relatively small amount for a mass media campaign.22 Campaign messages focused on the various short-term and long-term health benefits of cessation, and conversely the short-term and long-term costs of smoking, and are relevant to all smokers and to nonsmokers at risk of taking up smoking. The television ad was aired in selected regions representing less than 20 percent of the BC population, providing opportunities to compare geographic areas that did and did not see the television ad, an analysis conducted by Gagné.23 The radio ad was aired and the posters posted across the province. The text for the radio ad, the main component of the 2005 campaign, consisted of the following:
You already know that every time you have a cigarette you're doing your body serious harm. But did you know that just two months after quitting your lung capacity increases by as much as 30%? Coughing, fatigue, and shortness of breath also decrease. By one year, your chances of suffering a fatal heart attack significantly lowers. And after just five years of quitting smoking, the chance of getting lung cancer is reduced by as much as half. Cigarettes are hurting you, but you can get better. Log on to quitnow.ca today.24
Gagné further describes the campaign and notes that several complementary media (television) campaigns were in force before, during, and possibly shortly after the 2005 BC Ministry of Health Campaign.23 Most complementary ads ran just before the 2005 BC campaign ads began, in the annual period during which smokers are most likely to quit. BC ads would have provided added reinforcements to recent quitters who may have needed it. Thirty-four percent of a sample of recent smokers in the target population who were interviewed during or just after the campaign and were provided a brief description of the ads stated that they recalled recently seeing or hearing at least one of these ads. Although this level of recall may be good for a campaign of such short duration, 20 percent of recent smokers interviewed before the campaign began also stated that they recalled seeing or hearing at least one of the campaign ads, suggesting that prompted recall overstated exposure.23 However, respondents interviewed before the 2005 BC campaign began likely confused complementary ads with 2005 BC campaign ads.
The 2005 BC campaign can be contrasted to other campaigns. For example, the Florida Pilot Program on Tobacco Control, a comprehensive tobacco control program, was initially funded in February 1998. By June 1999, more than $80 million had been spent on this multifaceted program in a state with a population of 15 to 16 million people at the time. By May 1999, 92 percent of youths aged 12 to 17 were able to describe a media message from that campaign, and by 2000, current cigarette use had dropped by 40 percent from its 1998 level of 18.5 percent to 11.1 percent among middle school students and by 17.5 percent from its 1998 level of 27.4 percent to 22.6 percent among high school students.7 Another comprehensive tobacco control initiative was undertaken in California starting in 1989. The initiative was well funded with a tobacco excise taxes over several years. A quasi-experimental approach that compares trends in California with trends in the rest of the United States before and during the initiative finds large impacts for the first 4 years of the initiative (a 52% increase in the annual rate of decline in per capita consumption of cigarettes between 1989 and 1993), but smaller or no impacts in later years (between 1994 and 1996), which may be attributable to lack of innovation, decreases in cigarette prices, increased industry counter measures, and reduced funding for the initiative.13 Although the 2005 BC campaign is not an exhaustive representation of tobacco control activities in BC, the 2005 BC campaign was a much smaller mass media intervention than those of the Florida or California programs. It focused on cessation that can be less successful than prevention19 and on a high-risk population, two significant hurdles.
This article evaluates the impact of the 2005 BC campaign on short-term smoking prevalence and quantity smoked. The effect on prevalence and quantity of cigarettes smoked is estimated for the overall population and for the target demographic. Given the nature and scope of the 2005 BC campaign, it is reasonable to expect limited impacts for this intervention. Other reasons why the campaign may not yield the intended impact on the target demographic are investigated and solutions discussed.
The analysis is conducted with two sets of cross-sectional data. The Canadian Tobacco Monitoring Use Survey (CTUMS) has been conducted semiannually in two waves since 1999, and data were available at the time of writing up to wave 1 of 2005. Surveys are administered using random digit dialing between February and June (wave 1) and between July and December (wave 2). Month of interview is provided in the CTUMS data, allowing the data to be partitioned between a before and an after the campaign (started) period. February 2005 has been omitted from the analysis because it spans both periods. The target population for CTUMS includes persons 15 years and older except full-time residents of institutions and residents of the Yukon, Northwest Territories, and Nunavut. Samples are geographically stratified with a census metropolitan area stratum and a non–census metropolitan area stratum within each province. The sample is designed to overrepresent the population aged 15 to 24. Each observation in the sample has an associated sample weight.25
The British Columbia Tobacco Behaviour and Attitudes Survey (TBAS) is a cross-sectional survey conducted monthly by BC STATS (British Columbia Ministry of Labour and Citizens' Services), and is part of the larger Community Health, Education, and Social Services Survey. Data collection for the TBAS began in September 2004. The target population for Community Health, Education, and Social Services Survey is all persons 15 years or older living in BC, except First Nations residents living on reserves and full-time residents of institutions. The sample is geographically stratified and the sample selected using random digit dialing techniques. Each observation in the sample has an associated sample weight.26 The sample used for this analysis includes monthly cycles from September 2004 to September 2005 for some analyses and from September 2004 to April 2005 for others.
The main research question is whether the BC campaign had an impact on short-term smoking behavior, and is examined using CTUMS data. Smoking behavior is assessed in the analysis by looking at prevalence and at quantities of cigarettes smoked. Prevalence is estimated on the basis of the proportion of survey respondents who answered “every day” or “occasionally” to the question, “At the present time, do you smoke cigarettes every day, occasionally or not at all?” Current smokers were also asked how many cigarettes they had smoked in each of the previous 7 days. The total number of cigarettes smoked in the previous 7 days was used to derive an average daily quantity of cigarettes smoked. Nonsmokers are assigned a quantity of zero.
Estimating changes in prevalence or quantity smoked is difficult when a relatively small proportion of the population smokes, as changes in the behavior of small numbers are difficult to identify statistically, unless these changes are considerable. Furthermore, trends in prevalence or quantities smoked need be taken into consideration in answering the research question as there may already have been a underlying downward trend in smoking prevalence because of national and other ongoing antismoking activities and because of demographic changes: a comparison of prevalence or quantities smoked just before and just after the campaign may simply reflect that trend. If 20 percent of the population smokes at baseline and the trend is for a 1 percent annual decline in that proportion, then a 0.2 percent additional decline may be acceptable from a cost-benefit point of view and may affect many people, but it may not be statistically perceptible, even with a “large” sample: 0.2 percent of 1,000 is equal to 2, hardly a sufficient change for statistical significance. If the underlying population is 5 million, a difference of two smokers in a sample of 1,000 represents 10,000 people and could be considered a major accomplishment, although the change will not be perceptible in the sample. Finally, it is possible that national trends suffered a shock in the period under study because of changes in national advertising pertaining to smoking or for some other reasons. For these reasons, estimation strategies exploit differences in trends and current changes between BC and the rest of Canada. Because the overall CTUMS sample is relatively large, this quasi-experimental approach to answering the research question can partially assist in resolving small sample issues that arise from the estimation of small but important changes. Exploitation of a natural or quasi-experiment in this manner has been widely used in the literature to examine the impact of antismoking interventions.8,11,27–32 The drawback of this quasi-experiment approach is that there is insufficient information to control for similar activities in other Canadian jurisdictions. The comparison then is between the impact of overall activities in BC and that of overall activities in the rest of Canada.
Smoking prevalence and average daily quantity of cigarettes smoked are first estimated and compared for the immediate precampaign (June 2004–December 2004) and postcampaign (March 2005–June 2005) periods for BC and the rest of Canada. The recent changes are then informally compared with the longer term trends between 1999 and 2004. The postcampaign period refers to the period after which the campaign was either in progress or completed and for which CTUMS data were currently available. February 2005 data were excluded from the analysis as the campaign began during that month and some survey respondents for February 2005 could have been interviewed before and some after the campaign began. CTUMS data were not collected in January, and only the first half of 2005 was available at the time of writing, leaving observations from March 2005 to June 2005 for the postcampaign period. In this part of the analysis, which is reported in Tables 1 and 2, the following (null) hypotheses are formally tested:
- There is no difference between precampaign and postcampaign periods for prevalence or quantities of cigarettes smoked in BC.
- There is no difference between precampaign and postcampaign periods for prevalence or quantities of cigarettes smoked in the rest of Canada.
- Precampaign and postcampaign differences are the same for BC and the rest of Canada.
- There has been no trend in prevalence or quantities of cigarettes smoked for the 1999–2004 period in BC.
- There has been no trend in prevalence or quantities of cigarettes smoked for the 1999–2004 period in the rest of Canada.
- There is no difference between BC and the rest of Canada for the prevalence or quantities of cigarettes smoked trends for the 1999–2004 period.
Hypotheses 1 to 6 are tested using ordinary least squares (OLS). All estimates are weighted by the sample weights included in the data. For prevalence, OLS translates into a linear probability model, which suffers from heteroskedasticity,33 but the weighted estimates use a robust estimator of variance, which compensate for this problem.34
Next, to formally test whether there has been a statistically significant recent divergence in trends between BC and the rest of Canada, the following OLS equation is estimated (subscripts representing individuals have been omitted for simplicity):
where S is equal to 1 if an individual is a smoker and it is equal to 0 otherwise for prevalence estimates and is equal to the average number of daily cigarettes smoked for quantity smoked estimates. ROC is an indicator variable equal to 1 if the individual resides outside of BC and 0 otherwise. BC is an indicator variable equal to 1 if the individual resides in BC and 0 otherwise. The variable t represents the half-year in which the observation was collected with July to December 2000 set at 0 and each preceding half-year decremented by 1 and each subsequent half year incremented by 1. The variable t is multiplied by BC and ROC to arrive at tBC and tROC. Coefficients associated with these variables represent the semiannual change in the dependent variable for BC and the rest of Canada, respectively. The variable y 2005 is equal to 1 if the observation is from 2005 and 0 otherwise. The variable y 2005 is multiplied by BC and ROC to arrive at y 2005BC and y 2005ROC. Coefficients associated with these variables represent the 2004–2005 divergence from trend for BC and the rest of Canada, respectively. Estimates of β1 to β6 are shown in Tables 3 and 4 with their associated t values, and a label that characterizes what they represent.
Equation (1) allows a formal test of the following hypotheses:
- 7. There is no divergence in prevalence or quantities of cigarettes smoked trends in the rest of Canada between 2004 and 2005 compared with 1999–2004 (H0: β5 = 0).
- 8. There is no divergence in prevalence or quantities of cigarettes smoked trends in BC between 2004 and 2005 compared with 1999–2004 (H0: β6 = 0).
- 9. The divergence in prevalence or quantities of cigarettes smoked trends between 2004 and 2005 compared with 1999–2004 is the same for BC as it is for the rest of Canada (H0: β5 = β6).
Equation (1) is estimated for the entire CTUMS sample (except February 2005), for the target subgroup (individuals aged 20–34 without a university degree), and for nontarget subgroup, allowing a test of the hypotheses 7 to 9 for the full sample and for the subgroups. The question of whether the campaign affected smoking can thus be examined for the entire population, for the subgroups targeted by the campaign, and for the subgroups not targeted by the campaign.
The question of why the target population is doing poorly is then examined by looking ways in which the target population is more likely to be exposed to cigarette smoking than the nontarget population. Likelihood of exposure is discussed in the context of peer groups, workplace smoking incidence, and leisure activities.
Except for tabulations used to generate Figure 1, OLS with the Huber-White robust estimator for variance is used to produce estimates of trends and quantity smoked. Probit and logistic regressions are used along with OLS to estimate the impact of workplace restrictions on prevalence. All estimates control for sample weights.
British Columbia and rest of Canada recent smoking trends
Figure 1 illustrates smoking prevalence in BC and the rest of Canada by year since 1999, using CTUMS data. Data for 2005 are restricted to the first half of the year. Although the trend has generally been for a consistent decline from one year to the next, data for the first half of 2005 suggest that BC fared relatively better for 2005 than did the rest of Canada. In particular, smoking prevalence increased in the first half of 2005 compared with 2004 in the rest of Canada, whereas it declined for BC, possibly at a rate higher than the average. Series for the average number of cigarettes smoked per day display an almost identical pattern to the ones in Figure 1 and are not shown here. With BC potentially faring better after the campaign than the rest of Canada, the trend suggests that the media campaign may indeed have been effective.
Table 1 shows smoking prevalence rates in BC and the rest of Canada, shortly before (July 2004–December 2004) and during and after (March 2005–June 2005) the campaign, along with average changes in prevalence over the period between 1999 and 2004, using CTUMS data. Table 2 shows the average number of cigarettes smoked per day in BC and the rest of Canada, shortly before and during and after the campaign, along with average changes in daily quantity smoked over the period between 1999 and 2004, also using CTUMS data. Hypotheses 1 to 6 outlined in the “Methods” section are identified by their numbers in Tables 1 and 2 beside their P values.
Results given in Table 1 suggest that smoking prevalence in BC did not decline significantly after the mass media campaign. Reported prevalence fell 0.2 percentage points in BC between the last half of 2004 and March to June of 2005. For the rest of Canada, however, reported prevalence increased by 1.8 of a percentage point over the same period. However, neither of these estimates is statistically significant and the null hypothesis that the change in BC was equal to the change in the rest of Canada cannot be rejected (P = .44).
The bottom part of Table 1 shows estimates of long-term trends in smoking prevalence for BC and the rest of Canada. Estimates indicate that the annual percentage point decline in prevalence is 0.9 for BC and 1.1 for the rest of Canada, although these estimates are not statistically different (P = .51).
Table 2 uses the same analytical approach as Table 1, but focuses on average daily number of cigarettes smoked per capita (all respondents) and per smoker. The top panel of the table shows that for all BC respondents, average number of cigarettes smoked per day declined by 0.22 over the period before and after the campaign. Outside of BC, there was an average increase of 0.38 cigarettes smoked per day. However, neither of these estimates is statistically significant, and the hypothesis that the BC change equals the change in the rest of Canada cannot be rejected (P = .13).
The bottom panel of Table 2 focuses on the change in the amount smoked by smokers. BC smokers' daily consumption of cigarettes declined by 1.18 over the period before and after the campaign. Outside of BC, smokers increased their consumption by 0.74 cigarettes. However, neither of these estimates is statistically significant, and the hypothesis that the BC change equals the change in the rest of Canada cannot be rejected (P = .20).
Tables 1 and 2, along with Figure 1, suggest that while the 2004–2005 changes in smoking for BC and Canada are not statistically different from one another, recent period changes deviate from past trends. Tests of these hypotheses (hypotheses 7–9) are conducted by estimating Equation (1) and are reported in Tables 3 and 4 for prevalence and average number of cigarettes smoked per day, respectively.
In Tables 3 and 4, coefficients for the null hypotheses (7–9) that were rejected at the 10 percent level of significance are shaded along with their associated P values. Table 3 results indicate that while there is a statistically significant deviation from trend in 2004–2005 toward increasing prevalence in the rest of Canada (0.023; P = .08), there is no such deviation in BC (0.003; P = .86), although the results suggest that there may have been a similar upward trend for the campaign's target demographic (0.051 and 0.053). In other words, while the BC campaign may not have resulted in decreased prevalence over and above the 1999–2004 downward trend, it may have prevented a reversal of the downward trend in prevalence, at least for the nontarget population. The target population appears to have done worse between 2004 and 2005 than previously in both of BC and the rest of Canada, but the changes are not statistically significant and hypotheses 7 and 8 cannot be rejected for these subpopulation.
The top panel of column 1 in Table 4 indicates that the 1999–2004 downward trend in the average daily quantity of cigarettes smoked overall was reversed between 2004 and 2005 for all CTUMS respondents residing outside of BC (0.65; P < .01), but not for BC residents (−0.10; P = .72). The estimates also indicate that for smokers only, this reversal was large and significant in the target population (2.97; P < .01), but not in the nontarget population (0.17; P = .85). On the other hand, in BC, the nontarget population of smokers significantly reduced its average daily consumption from 2004 to 2005 over and above the 1999–2004 trend (−2.23; P = .10). This short-term reduction in quantity smoked is likely to eventually translate into cessation, as quantity smoked is a predictor of cessation.35–37 Finally, the hypothesis that the deviation from trend in average daily quantity of cigarettes smoked is the same for BC as for the rest of Canada (hypothesis 9) can be rejected at the 10 percent level for the overall population of smokers (P = .09), for the overall population (P = .04), and for the overall nontarget population (P = .07), and in each of these cases, BC did better than the rest of Canada.
To summarize, CTUMS cross-sectional data suggest that the 2005 campaign has had a positive impact on the nontarget population, but less so, if at all, on the target population. The impacts are marginally statistically significant, but if point estimates are reliable and changes sustained, given prior trends, the impacts would not only be more significant but also quite large. Results however do suggest that Canada may be experiencing a resurgence in smoking and that the BC campaign may simply have prevented this resurgence. Public health agencies therefore need to remain vigilant if further declines in smoking are desired and increases averted, and while short-term campaigns may affect short-term responses, a more comprehensive and sustained approach to public health management may be required. Questions also do remain as to why campaign effects are not clearly identified for the target population, and target population challenges are examined below.
An examination of prevalence
Young smokers face particular challenges in their efforts to quit smoking. Smoking is addictive, and conquering an addiction is more likely to be successful when exposure to the addictive substance is limited than when it is not. Exposure can occur at home, in the workplace, or with friends (peer effects). There is research evidence that household bans are associated with lower prevalence for adolescents,38 and that workplace bans predict reduced prevalence.20,38–40 Moreover, there is considerable evidence of peer effects.41–45 If workplace bans are less likely and/or prevalence is higher for the target demographic, then exposure is likely to be a greater problem for these groups, thereby negatively affecting the probability of cessation. Given that the target demographic was selected on the basis that its prevalence rate is higher than for other groups, then peer effects are likely an issue for that demographic. It is also possible that the target demographic is less likely to be employed where there are workplace bans, which is explored in the following.
TBAS working respondents who answered to the survey between September 2004 and April 2005 were asked, “At your place of work, is smoking restricted completely, allowed in designated areas, restricted only in certain areas or not restricted at all?” Table 5 shows smoking prevalence rates (column 1) along with incidence of (a) complete restrictions (column 3) and (b) complete restrictions or designated (smoking) areas (DSA) (column 5) for the full TBAS sample, and for individuals without university degrees who are 20 to 34 years of age. Prevalence is estimated using observations between September 2004 and September 2005. Incidence of workplace restrictions is estimated using observations on workers between September 2004 and April 2005, as the workplace restriction question was not asked after April 2005 and does not apply to nonworkers. Results in Table 5 indicate that prevalence for the targeted demographic (shown here as aged 20–34 without a university degree) is much higher than for the overall population (26% vs 18% for the full sample). This suggests that peer effects are likely to be an important issue for the target demographic groups.
Table 5 also shows that complete restrictions, and complete restrictions or DSA, are much less likely to be in place for young workers without a degree (32% complete restrictions and 71% complete restrictions or DSA) than for the overall population (49% complete restrictions and 78% complete restrictions or DSA). This suggests that exposure to smoking in the workplace may also be an issue for young workers without a degree. In particular, absence of complete workplace restrictions makes it more difficult for someone who wants to quit smoking to do so, because they are exposed to others smoking at work and have the opportunity to join them.
To examine the impact of workplace smoking restrictions in BC, Table 6 shows linear probability (OLS) estimates, probit estimates, and odds ratios (logistic estimates) for smoking prevalence during the periods from September 2004 to April 2005, using the TBAS data.
The reference person is a nonworker aged 25 to 64 without a university degree, and the linear estimate of prevalence for that demographic is 21.9 percent. Seniors have the lowest age group prevalence rate (10.1% less than those aged 25–64), followed by teenagers (7.0% less than those aged 25–64). Those aged 20 to 24 have the highest prevalence rate (5.5% more than those aged 25–64). Those with university degrees have a prevalence rate 8.5 percent lower than those without. Workers who work where there are limited or no smoking restrictions have a prevalence rate 5.1 percent higher than nonworkers. Workers where smoking is completely restricted have a prevalence rate 11.9 percent lower than that of other workers, and 6.8 percent (5.1–11.9) lower than nonworkers, while those working where smoking is restricted to DSAs have a prevalence rate equal to that of workers with little or no smoking restrictions in the workplace (1.5; P = .62). Workplace smoking restrictions have been shown elsewhere to have a negative effect on prevalence and quantity smoked, and a positive impact on cessation and efforts to quit,38–40,46 even after controlling for selection bias.39 Evans et al39 find that workplace bans predict a 5 percent lower prevalence, but their definition of bans is broad and includes both full restrictions and DSAs. For this broad category of workplace bans, we find similar effects, but based on the results shown in Table 6, none of these effects can be attributed to DSAs in the TBAS sample: they are simply an average of large and null effects.
The evidence presented in Tables 5 and 6 supports the hypothesis that prevalence in the target population may be difficult to manage because smokers in these populations are more likely to be exposed to smoking by others. In particular, since the target group belongs to a higher prevalence population, quitting may be more difficult because peers are more likely to smoke. Furthermore, quitting may be more difficult because smoking is more likely to be allowed at work. Finally, while no evidence is presented regarding this mediator, it is highly likely that the target population spend more time in bars, where smoking is often allowed in designated areas.
The results of this research indicate that the 2005 BC Smoking Cessation Mass Media Campaign may have contributed to avert an increase in smoking prevalence and quantities of cigarettes smoked in the target population similar to what was observed in the rest of Canada and to a decrease in quantities of cigarettes smoked in the nontarget population. The lack of a stronger impact on the target population may be the result of national trends toward more smoking in that demographic, of a short duration campaign, or may be because campaign messages take longer to be translated into action in high-risk groups. It has been argued in this research and elsewhere that high-risk groups face a variety of challenges in their efforts to quit smoking. Although the campaign may have resulted in an increase in the resolve to quit smoking for some members of the high-risk group, the challenges they face in their efforts to quit smoking may be such that more active or prolonged public health approaches are required to affect these populations. A recent meta-analysis of health campaigns found that health campaigns with enforcement mechanisms were more effective than campaigns without.47 The BC campaign may have been more effective if it was accompanied with enforcement mechanisms such as workplace smoking bans. On the other hand, approaches like the 2005 BC campaign may be sufficient to reach low-risk groups that are already have lower likelihoods of exposure. Other measures that could be undertaken along with a media campaign include the provision of free cessation aids, which have been shown to increase the probability of successful quitting,48–50 and to increase interest in contacting the cessation support program.48,49 Finally, and more generally, recent trends indicate that it may be necessary for BC and other Canadian jurisdictions to increase their tobacco control efforts and spending if additional progress in smoking prevention and cessation is to be achieved.
Findings from this study suggest that the 2005 BC Ministry of Health's Mass Media Smoking Cessation Campaign has had positive impacts on short-term smoking cessation. However, findings suggest that short-run trends improved for the nontarget population, but not for the target population. On the other hand, the campaign may have averted a recent national trend toward the consumption of more cigarettes by smokers in the target demographic.
Poor showings for the target population may be explained by increased exposure to smoking for this group. It is therefore important to take into consideration the hurdles that high-risk groups face in quitting smoking if mass media campaigns are to work for these groups. Initiatives such as full workplace smoking bans and the provision of free cessations aids may improve the effectiveness of smoking cessation programs that target high-risk groups, and longer and generally more comprehensive interventions should be considered.
1. Tobacco Control
Programme, Health Canada. Supplementary tables, CTUMS Wave 1, 2005.
Available at: http://www.hc-sc.gc.ca/hl-vs/pubs/tobac-tabac/ctums-esutc-2005/index_e.html
. Accessed January 21, 2006.
2. Smoking in Canada.
Available at: http://www.smoke-free.ca/factsheets/pdf/prevalence.pdf
. Accessed January 21, 2006.
3. Gilmore J. Report on Smoking in Canada: 1985 to 2001
. Ottawa, Ontario: Statistics Canada; 2002. Statistics Canada Catalogue No. 82F0077XIE, No. 1.
4. Ghadirian P. Sleeping With a Killer: A Report from Health Canada's Tobacco Control Programme
. Ottawa, Ontario: Health Canada.
5. Bridge J, Turpin B. The Cost of Smoking in British Columbia and the Economics of Tobacco Control
. Vancouver, British Columbia: Health Canada; 2004.
6. Backinger CL, Fagan P, Matthews E, Grana R. Adolescent and young adult tobacco prevention and cessation: current status and future directions. Tob Control.
7. Bauer UE, Johnson TM, Hopkins RS, Brooks RG. Changes in youth cigarette use and intentions following implementation of a tobacco control
program: findings from the Florida Youth Tobacco Survey, 1998–2000. JAMA
8. Flynn BS, Worden JK, Secker-Walker RH, et al. Mass media and school interventions for cigarette smoking prevention: effects 2 years after completion. Am J Public Health.
9. Hu T-W, Sung H-Y, Keeler TE. Reducing cigarette consumption in California: tobacco taxes vs. an anti-smoking media campaign. Am J Public Health.
10. Levy DT, Friend K. Gauging the effects of mass media policies: what do we need to know? J Public Health Manag Pract.
11. McAlister A, Morrison TC, Hu S, et al. Media and community campaign effects on adult tobacco use in Texas. J Health Commun.
12. Pechmann C, Reibling ET. Anti-smoking advertising campaigns targeting youth: case studies from USA and Canada. Tob Control.
13. Pierce JP, Gilpin EA, Emery S, et al. Has the California tobacco control
program reduced smoking? JAMA.
14. Pisinger C, Vestbo J, Borch-Johnsen K, Jorgensen T. Smoking reduction intervention in a large population-based study. The Inter99 study. Prev Med.
15. Skara S, Sussman S. A review of 25 long-term adolescent tobacco and other drug use prevention program evaluations. Prev Med.
16. Sly DF, Heald GR, Ray S. The Florida “ Truth” Anti-Tobacco Media evaluation: design, first year results, and implications for planning future state media evaluations. Tob Control.
17. Sussman S. Effects of sixty six adolescent tobacco use cessation trials and seventeen prospective studies of self-initiated quitting. Tob Induced Dis.
18. Tengs TO, Osgood ND, Chen LL. The cost-effectiveness of intensive national school-based anti-tobacco education: results from the tobacco policy model. Prev Med.
19. Wakefield M, Flay B, Nichter M, Giovino G. Effects of anti-smoking advertising on youth smoking: a review. J Health Commun.
20. Levy DT, Chaloupka FJ, Gitchell J. The effects of tobacco control
policies on smoking rates: a tobacco control
scorecard. J Public Health Manag Pract.
21. Secker-Walker RH, Worden JK, Holland RR, Flynn BS, Detsky AS. A mass media programme to prevent smoking among adolescents: costs and cost effectiveness. Tob Control
22. Gagné L. Analysis of the Impacts of the British Columbia 2005/06 Tobacco Cessation Mass Media Campaign
. Victoria, British Columbia, Canada: University of Victoria; 2006.
23. Gagné L. The 2005 British Columbia Smoking Cessation Mass Media Campaign and Short Term Changes in Smokers Attitudes
. Victoria, British Columbia: University of Victoria; 2006.
24. BC Ministry of Health. You Can Get Better
. Victoria, British Columbia: BC Ministry of Health; 2005.
25. Microdata User Guide: CTUMS Cycle 1, February–June 2005
. Ottawa, Ontario: Statistics Canada; 2005.
26. CHESS SPSS Public Use Data File User Guide
. Victoria, British Columbia: BC STATS; 2005.
27. Farrelly MC, Davis KC, Haviland ML, Healton CG, Messeri P. Evidence of a dose-response relationship between “truth” antismoking ads and youth smoking prevalence. Am J Public Health.
28. Flynn BS, Worden JK, Secker-Walker RH, Badger GJ, Geller BM, Costanza MC. Prevention of cigarette smoking through mass media intervention and school programs. Am J Public Health.
29. Manley M, Pierce JP, Gilpin EA, Rosbrook B, Berry C, Wun M. Impact of the American Stop Smoking Intervention Study on cigarette consumption. Tob Control.
30. McAlister A, Gumina T, Urjanheimo E-L, et al. Promoting smoking cessation in Russian Karelia: a 1-year community-based program with quasi-experimental evaluation. Health Promot Int.
31. McVey D, Stapleton J. Can anti-smoking television advertising affect smoking behaviour? Controlled trial of the Health Education Authority for England's anti-smoking TV campaign. Tob Control.
32. White V, Tan N, Wakefield M, Hill D. Do adult focused anti-smoking campaigns have an impact on adolescents? The case of the Australian National Tobacco Campaign. Tob Control.
33. Wooldridge JM. Econometric Analysis of Cross Section and Panel Data
. Cambridge, Mass: MIT Press; 2002.
34. StataCorp. Stata Statistical Software.
Release 9. College Station, Tex: StataCorp LP; 2005.
35. Hymowitz N, Cummings KM, Hyland A, Lynn WR, Pechacek TF, Hartwell TD. Predictors of smoking cessation in a cohort of adult smokers followed for five years. Tob Control
36. Monso E, Campbell J, Tonnesen P, Gustavsson G, Morera J. Sociodemographic predictors of success in smoking interventions. Tob Control.
37. Osler M, Prescott E. Psychosocial, behavioral, and health determinants of successful smoking cessation: a longitudinal study of Danish adults. Tob Control.
38. Farkas AJ, Gilpin EA, White MM, Pierce JP. Association between household and workplace smoking restrictions and adolescent smoking. JAMA.
39. Evans WN, Farrelly MC, Montgomery E. Do workplace smoking bans reduce smoking? Am Econ Rev.
40. Glasgow RE, Cummings KM, Hyland A. Relationship of worksite smoking policy on changes in employee tobacco use: findings from COMMIT. Tob Control
41. Alexander C, Piazza M, Mekos D, Valente T. Peers, schools, and adolescent cigarette smoking. J Adolesc Health.
42. Krauth BV. Peer effects and selection effects on smoking among Canadian youth. Can J Econ.
43. Leatherdale ST, Cameron R, Brown SK, McDonald PW. Senior student smoking at school, student characteristics and smoking onset among junior students: a multilevel analysis. Prev Med.
44. Norton EC, Lindrooth RC, Ennett ST. Controlling for the endogeneity of peer substance use on adolescent alcohol and tobacco use. Health Econ.
45. Powell LM, Tauras J, Ross H. The importance of peer effects, cigarette prices and tobacco control
policies for youth smoking behavior. J Health Econ.
46. Tauras J. The Transition to Smoking Cessation: Evidence From Multiple Failure Duration Analysis
. Cambridge, Mass: National Bureau of Economic Research; 1999. NBER Working Paper No. 7412.
47. Snyder LB, Hamilton MA, Mitchell EW, Kiwanuka-Tondo J, Fleming-Milici F, Proctor D. A meta-analysis of the effect of mediated health communication campaigns on behavior change in the United States. J Health Commun.
48. Bauer JE, Carlin-Menter SM, Celestino PB, Hyland A, Cummings KM. Giving away free nicotine medications and a cigarette substitute (Better Quit) to promote calls to a quitline. J Public Health Manag Pract.
49. Cummings KM, Fix B, Celestino PB, Carlin-Menter SM, O'Connor R, Hyland A. Reach, efficacy, and cost-effectiveness of free nicotine medication giveaway programs. J Public Health Manag Pract.
50. Silagy C, Lancaster T, Stead L, Do M, Fowler G. Nicotine replacement therapy for smoking cessation. Cochrane Database Syst Rev.