Article Text

Intended and unintended effects of restrictions on the sale of cigarillos to youth: evidence from Canada
1. Hai V Nguyen1,
2. Paul Grootendorst2
1. 1Program in Health Services and Systems Research, Duke-NUS Graduate Medical School Singapore, Singapore
2. 2Leslie Dan Faculty of Pharmacy, University of Toronto, Ontario, Canada
1. Correspondence to Dr Hai V Nguyen, Program in Health Services and Systems Research, Duke-NUS Graduate Medical School Singapore, 8 College, Singapore 169857, Singapore; vanhai.nguyen{at}duke-nus.edu.sg

## Abstract

Background Youth consumption of cigarillos (ie, little cigars) has increased markedly in recent years. In July 2010, the Canadian government banned the sale of flavoured cigarillos and required unflavoured cigarillos to be sold in packs of at least 20 units. This paper assesses changes in young persons’ use of cigarillos and regular cigars, which are potential substitutes, following the policy.

Methods To investigate of the change in cigar smoking following the policy, we constructed a segmented regression model that allowed the policy to change the height and the slope of the trend in the outcome variables. The model was estimated using data from the 2007–2011 Canadian Tobacco Use Monitoring Surveys.

Results We obtained visual and regression-based evidence that use of cigarillos among youth declined following the policy. We also found a small, gradual increase in their use of regular cigars, possibly due to their compensatory switching from cigarillos to regular cigars. Overall, there was a net reduction in cigar use among youth after the intervention.

Interpretation The policy achieved its goal of reducing youth's consumption of cigarillos, but may have an unintended consequence of increasing their use of regular cigars. Policymakers should address the possibility that youth switch to regular cigars in response to restricted access to cigarillos. Possible ways of discouraging this substituting behaviour include extending the ban to cover all flavoured cigars and mandating a minimum pack size for all cigars, or raising taxes on flavoured cigars.

• Public policy
• Prevention
• Cessation
• Packaging and Labelling
View Full Text

## Introduction

Sales of little cigars (also known as small cigars or cigarillos in Canada) have increased substantially in recent years. In Canada, sales jumped from 53 million units in 2001 to an estimated 469 million units in 2008, making them the fastest growing segment of the tobacco market.1 In the USA, between 1997 and 2007, sales of small cigars increased by 240%.2 In Canada, cigarillos are similar to cigarettes in size but wrapped in tobacco leaf, often have a filter and contain no more than 1.4 g of tobacco.1 In the USA, small cigars and cigarillos are different products. The former are the same as little cigars in Canada but the latter are larger and contain about 3 g of nicotine.3

Cigarillos are particularly popular among youth. Canadian children and young people under age 24 accounted for 39% of users of cigarillos, but constituted only 15% of cigarette smokers during 2007–2010.4 Two factors are often suggested to have contributed to their popularity. First, cigarillos are available in different flavours (such as strawberry, watermelon, bubble gum, chocolate and so on) that appeal to youth.2 ,5 A recent study6 showed that 36% of US middle and high school students who are cigar smokers reported using flavoured small cigars. Second, cigarillos could be purchased individually for about $1 each or in ‘kiddie packs’ of five.7 As youth's use of tobacco products is generally responsive to its costs,8 ,9 these small package sizes likely appeal to youth who do not have sufficient funds to purchase a package of 20 cigarettes but who do have funds to buy a single cigarillo. The popularity of cigarillos among youth, which poses significant health risks to users,10–12 has prompted an increasing number of jurisdictions to take measures to limit access to, and the appeal of, cigarillos. These policy interventions include quantity restrictions (eg, sales bans, or minimum package size restrictions) and price increases (eg, minimum price restrictions for cigarillos). In Canada, effective from July 2010 the federal government banned the retail sale of flavoured cigarillos and required unflavoured cigarillos be sold in minimum quantities of 20 units, like normal cigarettes. In March 2013, Singapore passed a similar measure that required cigarillos to be sold in packs of at least 20 units. In the USA, two jurisdictions have regulated small cigars, New York City and Providence, Rhode Island. New York City prohibited the sale of all flavoured tobacco products including small cigars in February 2010. In March 2013, it further restricted access to small cigars by mandating small cigars to be sold in packages of four or more and to carry a minimum price of$10.50 per pack.13 The US Food and Drug Administration banned flavoured cigarettes in 2009 but has not done so for flavoured cigars.14

While policy efforts to regulate use of cigarillos are gathering pace, little is known about the effects of the existing interventions on the use of cigarillos among youth. There are several studies that characterise the use of cigars among youth (eg, King et al6; Schuster et al15), but the literature on the impacts of tobacco control policy on youth cigar use is scant. This may be because most of the existing regulations were only recently introduced, making policy evaluation difficult. An exception is a study by Ringel et al16 that assessed the impacts of cigar prices and tobacco regulations on youth cigar demand by US students grade 6–12. Using the 1999 and 2000 National Youth Tobacco Survey data, they found that youth cigar use is sensitive to cigar prices but not to state tobacco-control regulations. A recent report by the US Government Accountability Office showed (in a descriptive analysis) that there was a shift to large cigars in response to a higher tax imposed on small cigars in 2010 in the USA, with little impact on overall consumption of cigars.17

This study presents empirical evidence on the outcomes of the Canadian ban on the sale of flavoured little cigars and imposition of a minimum pack size of 20 units. Specifically, using segmented regression analysis and Canadian tobacco use survey data, this paper assesses the impact of the policy change on self-reported use of little cigars, use of regular cigars, and overall use of cigars among Canadian youth. Regular cigars were not targeted by the policy and thus flavoured and other types remain available for purchase individually. It is thus possible that youth substituted the unregulated for the regulated cigar varieties. Such compensatory behaviour is regularly observed in tobacco use. For instance, when faced with higher cigarette taxes, smokers will switch from light to regular cigarettes18 or smoke fewer cigarettes, but will draw in more smoke with each puff of the cigarette.19

## Methods

### Data sources

We used individual-level data from the Canadian Tobacco Use Monitoring Survey (CTUMS), a nationally representative survey of tobacco use of Canadians. The CTUMS conducted annually by Statistics Canada employs a cross-sectional stratified survey design. Each CTUMS cycle interviews approximately 20 000 individuals aged 15 years and older (about 1700–1900 persons each month) using computer-assisted telephone. The survey's response rate is about 83%. Survey weights are estimated and placed on each record to represent the number of sampled people that the record represents. In particular, the weighting scheme reflects the sampling design, monthly population totals for each province and the exclusion of people without telephones.20

Our analysis employed annual CTUMS cycles from 2007 to 2011; earlier CTUMS cycles do not distinguish between cigarillos and regular cigars. The CTUMS records the month of interview, allowing us to separate the pre-policy period (January 2007–June 2010) from the post-policy period (July 2010–December 2011).

The CTUMS asks questions about subjects’ recent and lifetime use of different tobacco products, including cigarillos and regular cigars. Use of cigarillos is determined using the questions, ‘Have you ever tried smoking a little cigar or cigarillo?’ and ‘In the past 30 days, did you smoke any little cigars or cigarillos?’ Use of regular cigars is determined using the questions, ‘Have you ever tried smoking a cigar (not including little cigars or cigarillos)?’, and ‘In the past 30 days, did you smoke any cigars (not including little cigars or cigarillos)?’

Using these questions, we constructed several binary outcome variables: ‘Smoked cigarillos over the past 30 days’; ‘Ever smoked cigarillos’; ‘Ever smoked cigarillos but not over the past 30 days’; ‘Smoked regular cigars past 30 days’; ‘Ever smoked regular cigars’; and ‘Ever smoked cigarillos or regular cigars’. These outcomes are intended to capture the effects of the policy on various facets of smoking behaviour. The first outcome measures the change in past 30-day use of cigarillos. The second and third outcomes measure initiation into and 30-day abstinence of cigarillos, respectively. If the policy is effective, the fraction of respondents who have ever tried cigarillos and the fraction of those who tried cigarillos in the past 30 days should decrease, and the fraction of previous users who have not used in the last month should increase. The next two outcomes capture possible substitution from cigarillos to regular cigars. The final outcome measures the net change in the overall prevalence of cigar use.

### Study population

As the goal of the intervention is to reduce the use of cigarillos among youth, our study sample focuses on the CTUMS subjects aged 15–24. This sample from the CTUMS 2007–2011 has just over 46 000 observations.

### Segmented regression model

To investigate the changes in the outcomes following the policy, we constructed a regression model that allows the policy to change the height and the slope of the trend in the outcome variables. The specification of this regression model, often referred to as segmented or piecewise regression, is as follows:

where Yipt are outcomes for individual i in province p in month t; the covariate T is a linear time trend; Policyt is a dummy variable equal to one if the policy is in effect in month t, and equal to 0 otherwise; Policy*T is the interaction term between Policy and the time trend T; β2 and β3 are parameters of interest: β2 estimates the change in the level of the trend post policy, while β3 estimates the change in its slope. We explicitly control for a variety of individual demographic and other characteristics in our regression models. These characteristics, contained in the vector X, include cigarette smoking status, subject age, and indicators for sex, language and education. We include province-specific indicator variables (denoted by the vector ηp) to control for time-invariant differences across provinces. We also include a set of season dummies (µs) to control for possible seasonal effects on youth cigar use (eg, higher rates of use and experimentation in summer months).

The regressions were estimated using STATA Version 12 software. All estimates in the descriptive, graphical and regression analyses were weighted. Robust SEs accounted for survey design and the weighting and were clustered at provincial level to allow for possible within-province correlation. For ease of interpretation in regressions with binary outcomes and interaction terms, linear probability models were used.21

## Results

### Descriptive statistics

Table 1 displays descriptive statistics (means and CIs) for cigar smoking and demographic variables, using data for youth aged 15–24 from the entire study period, that is, January 2007–December 2011. As female and male respondents might exhibit different smoking behaviours, we present the statistics for all respondents and also for male and female subsamples.

Table 1

Descriptive statistics of participants aged 15–24 in Canadian Tobacco Use Monitoring Survey (CTUMS) samples, January 2007 to December 2011

For the whole sample, 39% reported ever smoking cigarillos, while 9% reported using cigarillos in the past 30 days. Fewer young people ever tried regular cigars (28%) and only 4% reported trying regular cigars in the past month. The overall prevalence of all-cigar use is 43% (ever tried) and 10% (tried in the last month). By gender, male individuals are more likely to smoke cigars than female individuals. However, the gender difference is larger for regular cigars (40% vs 16% for ever use, and 7% vs 1% for past 30-day use) than for cigarillos (47% vs 31% for ever use, and 13% vs 5% for past 30-day use). This suggests that female respondents are more attracted to cigarillos than to regular cigars. Current smokers of cigarettes account for 18% while 76% of the sample reported never smoking cigarettes. Most of the sample subjects have finished secondary education or are still studying for it (82%). Forty-nine percent of the sample was female.

Table 2 reports the changes in cigar use after the intervention by age and gender groups. The upper panel reports the means of 30-day use of cigarillos and regular cigars by male and female respondents aged 15–24 for pre-policy and post-policy periods. In this unadjusted comparison, these periods are chosen to be short (ie, 12 months) because using longer periods might give a distorted picture, given a possible secular downward trend in smoking during this period. For cigarillos, past 30-day use prevalence significantly declined following the policy, from 13.7% to 9.3% (p=0.000) for male respondents and from 5.3% to 3.3% (p=0.001) for female respondents. For regular cigar use, however, there was a smaller and insignificant decline for male respondents (5.8% vs 4.9%; p=0.206) and an insignificant increase for female respondents (0.8% vs 0.9%; p=0.673). This indicates that after the intervention the decline in cigarillo use was much stronger than that for regular cigars. Meanwhile, for the adult group aged 25–65 displayed in the lower panel, the changes in 30-day use of cigarillos and regular cigars after the intervention were very small and statistically insignificant.

Table 2

Percentages of Canadian Tobacco Use Monitoring Survey (CTUMS) sample reporting past 30-day use of cigarillos and regular cigars pre and post policy, by age and gender group

### Graphs

We examined the trends in the monthly prevalence of cigarillo use and regular cigar use by subjects aged 15–24 in figures 15. Monthly smoking prevalence is calculated as a fraction of all respondents in a month (accounting for sampling weights) who smoke a certain tobacco product. In each figure, fitted linear regression lines for pre-policy and post-policy periods are overlaid. These lines depict predictions of monthly prevalence from a linear regression of monthly prevalence on a linear monthly time trend. The vertical dashed line in each figure indicates the date when the intervention came into effect (ie, July 2010).

Figure 1

Trend in any use of cigarillos in past 30 days, age 15–24, Canadian Tobacco Use Monitoring Survey (CTUMS) 2007–2011.

Figure 2

Trend in ever use cigarillos but not in past 30 days, age 15–24, Canadian Tobacco Use Monitoring Survey (CTUMS) 2007–2011.

Figure 3

Trend in ever use of cigarillos, age 15–24, Canadian Tobacco Use Monitoring Survey (CTUMS) 2007–2011.

Figure 4

Trend in any use of regular cigars over past 30 days, age 15–24, Canadian Tobacco Use Monitoring Survey (CTUMS) 2007–2011.

Figure 5

Trend in ever use of regular cigars, age 15–24, Canadian Tobacco Use Monitoring Survey (CTUMS) 2007–2011.

Figure 1 shows the trend for the past 30-day use of cigarillos. Although the trend fluctuated across the months (perhaps due to seasonal effects), on average its post-policy segment was more downward sloping and lay significantly lower than the pre-policy segment. This indicates that the prevalence of past 30-day use was substantially lower following the policy. Figure 2 displays the trend in 30-day abstinence of cigarillos. There is an upper shift in the post-policy line, suggesting an increase in abstinence rates post policy. Figure 3 shows the trend in the prevalence of ‘ever use cigarillos’. Although the pre-policy segment trended upwards, the post-policy trend was downward sloping and lay lower than the pre-policy trend. This means that the prevalence of ‘ever used cigarillos’ was also lower after the policy.

Figures 4 and 5 display the trends in past 30-day use and ever use of regular cigars. Both figures display the same pattern: the trend was declining up to the policy adoption at which point the trend flattens. This graphical evidence is consistent with the pre–post comparisons in table 2 and suggests that the decline in use of regular cigars slowed down following the policy.

### Regression results

The upper panel of table 3 reports the estimates from the segmented regressions for the six outcomes using the sample of youth aged 15–24. The first three columns concern cigarillos use. As noted, the segmented regression decomposes the policy effects into the change in the level (β2) and the change in the slope (β3). Column 1 presents estimates from the regression with the ‘past 30-day use’ indicator as outcome. The coefficient (β2) on the level variable is −0.023 and statistically significant. This indicates that past 30-day cigarillo use among young people declined by 2.3 percentage points after the policy was introduced. Compared with the pre-policy prevalence of 10.1%, this represents a decline of 22%. Meanwhile, the coefficient on the slope variable is negative but very small and not statistically significant.

Table 3

Effects of policy on cigar use, by age group

The coefficient estimates from the regression of the 30-day abstinence outcome are displayed in column 2. We obtained a positive and statistically significant coefficient of 0.043 on the level variable, and a small and insignificant coefficient on the slope variable. This indicates that the fraction of users with 30-day abstinence increased abruptly by 4.3 percentage points right after the policy and there was little change in the long term. Column 3 showing the outcome of ‘ever used cigarillos’ displays a similar pattern of results as column 1. The level coefficient of −0.031 indicates that the ‘ever used cigarillos’ prevalence declined by 3.1 percentage points after the policy. This change is statistically significant. Compared with the pre-policy prevalence of ever smoking cigarillos of 39%, this represents a modest decline of 8%. Similar to the previous estimates, the slope coefficient is not significant, suggesting that most of the changes in cigarillo use occurred right after the policy.

Turning to the use of regular cigars, columns 4–5 show that there is no level change: the coefficients (β2) on the level variables in the ‘ever used regular cigars’ and ‘used regular cigars past 30 days’ regressions are not statistically significant. Interestingly, however, the coefficients (β3) on the slope variables are positive and statistically significant. Thus, while the declines in the use of cigarillos concentrated on the level, the changes in the use of regular cigars were centred on the slope. This difference, which is consistent with the graphical evidence in figures 15, suggests that the decline in the use of cigarillos was sudden while the offsetting increases in the use of regular cigars were gradual. Meanwhile, the coefficient (β1) on the pre-policy trend is negative and statistically significant. This indicates that the use of regular cigars has been on a declining trend before the policy, which is consistent with the graphical evidence in figures 4–5. Column 6 presents estimates from the same regression but with ‘ever smoked any type of cigars’ as the outcome. The coefficient estimate of −0.022 on the level variable indicates that the policy significantly reduced the overall prevalence of all-cigar use by 2.2 percentage points.

Other covariates in the regressions have expected signs. For instance, male students are more likely to smoke cigars, both small and regular, than female students. Those who are current cigarette smokers are more likely to smoke cigars. As expected, young people are more likely to smoke cigars in the summer than in other seasons. The full set of regression estimates are reported in the online supplemental material.

### Robustness checks

As a robustness check, we estimated the same models but used a sample of adults aged 25–64. The idea is that, as adults are not major consumers of cigarillos, we would not expect similar and significant changes in the outcomes for this age group. A finding to the contrary could suggest some misspecification of our model of youth smoking. The results of these regressions are displayed in the lower panel of table 3. For the outcomes of ‘ever smoked cigarillos past 30 days’ (column 1) and ‘ever smoked cigarillos’ (column 3), there is little change in the slope. There is a slight decline in the level, but statistically insignificant. In addition, we did not find any significant change in 30-day abstinence in cigarillos (column 2). For past 30-day use of regular cigars (column 4) and any use of regular cigars (column 5), there is almost no change in the slope. Meanwhile, the level coefficients are negative, suggesting that the use of regular cigars by adults even dropped after the policy. This is in contrast to the increases in regular cigar use by youth aged 15–24 shown in the upper panel of table 3. The absence of increases in regular cigar use by adults provides some support for the robustness of the findings from our model of youth smoking.

As another robustness check, we modelled the same outcomes using standard logistic regressions. The pattern of the estimates for the intercepts and slopes (the sign and statistical significance) is similar to those obtained from the linear model. We also estimated the regressions with normal cigarette smoking as the outcome. We found no evidence of higher cigarette smoking after the policy. These results are reported in the online supplemental material.

## Discussion

In this paper, we assessed the changes in the use of cigarillos and regular cigars among youth following the ban on the sale of flavoured cigarillos and the minimum size requirement for unflavoured cigarillo packs in Canada. The evidence suggests that the intervention has encouraged young cigarillo users to discontinue their use; the prevalence of past-30-day cigarillo use dropped by 22%. However, we observed a small, gradual increase in the use of regular cigars among youth. One interpretation of this finding is that some users of flavoured cigarillos switched to regular cigars that, unlike cigarillos, are still available flavoured and are not subject to the packaging requirement. Another complementary factor that likely contributed to the increased use of regular cigars is the response by the cigar industry. It has been reported that following the policy, some cigar manufacturers tried to circumvent the ban by introducing flavoured regular cigars that appear similar to cigarillos, except that they have slightly more tobacco (ie, contain a little over 1.4 g) or have had the filters removed.22 Nevertheless, it should be emphasised that the offsetting increase in regular cigar use was smaller than the reduction in cigarillo smoking, resulting in a net decrease in cigar smoking. Thus, the intervention appears to be beneficial in reducing overall cigar use among youth. This finding is in contrast with the outcome of a tax increase on small cigars in 2010 in the USA. It was reported that tobacco companies introduced ‘large’ cigars (by weight) which were virtually indistinguishable from small cigars, and there was a large shift of consumption from small to large cigars, with little effect on overall consumption.17

Our study has a number of limitations. First, the data we used do not allow us to distinguish the consumption of flavoured cigarillos from that of non-flavoured cigarillos. However, this is unlikely to affect our findings because most young cigar users (for instance, 73% in Florida; 80% in Maryland) smoked flavoured cigars.5 In addition, we could not separate the effect of the ban on the sale of flavoured cigarillos from the effect of the minimum-20-unit-per-package rule. Another limitation is that our research design did not have an external comparison group. Although the segmented model allowed us to control for various determinants of tobacco use at the individual level, there might be other confounding shocks at the national level that affect the use of cigarillos among youth. However, the fact that we observed the policy effect for young people who are the major users of cigarillos—and not for adult users—should alleviate some concern about possible confounding. Also, the abrupt decline in cigarillo use in contrast to the gradual change in regular cigar use, which were shown in the graphical and regression analyses, are consistent with the abrupt nature of the ban. This further lends support to the interpretation that the decline in cigarillo use and the increase in regular cigar use were likely driven by the intervention.

In conclusion, our study demonstrated that the ban on flavoured cigarillos and the minimum pack size requirement were effective in reducing youth's use of cigarillos, but policymakers should address the possibility that such policies induce young cigarillo users to switch to smoking regular cigars. One policy option is to expand the ban to also prohibit flavoured regular cigars and to mandate a minimum pack size (or a minimum price per pack) for all types of cigars. Alberta, a province in Canada, has prohibited all flavoured cigars, and Ontario is considering a similar measure.23 ,24 Another option is to impose a higher tax on flavoured regular cigars. Cigars have been taxed at a uniform rate for flavoured and unflavoured products, and these tax rates have remained unchanged for several years.25 A higher tax on flavoured cigars is likely to discourage price-sensitive children and youth from switching to flavoured regular cigars and still allow adults access to such products.

• A number of jurisdictions have adopted measures to regulate access to cigarillos while several others are deliberating similar policies. However, little is known about the effects of these regulations.

• This paper empirically evaluates Canada's recent restrictions on cigarillo access, which involves banning flavoured cigarillos and requiring unflavoured cigarillos to be sold in packs of a minimum of 20 units.

• We found visual and regression-based evidence that cigarillo use among youth declined after the policy. We also found a small, gradual increase in the use of regular cigars among youth.

• Overall, there was a net decrease in cigar use among youth after the intervention.

View Abstract

• ## Supplementary Data

This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Files in this Data Supplement:

## Footnotes

• Contributors HVN conceptualised the idea, contributed to the study design, conducted the data analysis and contributed to the interpretation of the results and manuscript writing. PG contributed to the study design, contributed to the manuscript writing and interpretation of the results.

• Competing interests None.

• Provenance and peer review Not commissioned; externally peer reviewed.

• Data sharing statement Data for this paper are from Canadian surveys that are available for public use at no charge.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.