Article Text

Download PDFPDF

Cigarette and cigar sales in Hawaii before and after implementation of a Tobacco 21 Law
  1. Rebecca Glover-Kudon1,
  2. Doris G Gammon2,
  3. Todd Rogers2,
  4. Ellen M Coats2,
  5. Brett Loomis2,
  6. Lila Johnson3,
  7. MaryBeth Welton1,
  8. René Lavinghouze1
  1. 1 Office on Smoking and Health, Centers for Disease Control and Prevention, Atlanta, Georgia, USA
  2. 2 Center for Health Policy Science and Tobacco Research, RTI International, Research Triangle Park, North Carolina, USA
  3. 3 Tobacco Prevention and Education Program, Hawaii State Department of Health, Honolulu, Hawaii, USA
  1. Correspondence to Dr Rebecca Glover-Kudon, Office on Smoking and Health, Centers for Disease Control and Prevention, Atlanta, GA 30341, USA; rgloverkudon{at}


Introduction On 1 January 2016, Hawaii raised the minimum legal age for tobacco access from 18 to 21 years (‘Tobacco 21 (T21)’) statewide, with no special population exemptions. We assessed the impact of Hawaii’s T21 policy on sales of cigarettes and large cigars/cigarillos in civilian food stores, including menthol/flavoured product sales share.

Methods Cigarette and large cigar/cigarillo sales and menthol/flavoured sales share were assessed in Hawaii, California (implemented T21 in June 2016 with a military exemption), and the US mainland using the only Nielsen data consistently available for each geographical area. Approximate monthly sales data from large-scale food stores with sales greater than US$2 million/year covered June 2012 to February 2017. Segmented regression analyses estimated changes in sales from prepolicy to postpolicy implementation periods.

Results Following T21 in Hawaii, average monthly cigarette unit sales dropped significantly (−4.4%, p<0.01) coupled with a significant decrease in menthol market share (−0.8, p<0.01). This combination of effects was not observed in comparison areas. Unit sales of large cigars/cigarillos decreased significantly in each region following T21 implementation. T21 policies in Hawaii and California showed no association with flavoured/menthol cigar sales share, but there was a significant increase in flavoured/menthol cigar sales share in the USA (7.1%, p<0.01) relative to Hawaii’s implementation date, suggesting T21 may have attenuated an otherwise upward trend.

Conclusions As part of a comprehensive approach to prevent or delay tobacco use initiation, T21 laws may help to reduce sales of cigarette and large cigar products most preferred by US youth and young adults.

  • surveillance and monitoring
  • public policy
  • prevention

Statistics from

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.


Tobacco use is the leading cause of preventable disease and death in the USA.1 Adolescence and young adulthood is a critical time for intervention to prevent tobacco use initiation and dependence.2 Most smokers report trying their first cigarette before 18 years of age, and progression to daily smoking typically occurs during young adulthood.2 To prevent youth initiation, increasing the minimum legal age (MLA) to purchase tobacco products from 18 to 21 years (‘Tobacco 21 (T21)’ policies) is an emergent tobacco control strategy that is gaining momentum at the state and local levels.3

On 1 January 2016, Hawaii began enforcing the first statewide law that raised the MLA from 18 to 21 years of age.4 As of November 2019, 17 other US states (Arkansas, California, Connecticut, Delaware, Illinois, Maine, Maryland, Massachusetts, New Jersey, New York, Ohio, Oregon, Texas, Utah, Vermont, Virginia and Washington), 1 US territory (Guam) and over 500 cities or counties have passed similar legislation.5 At a national level, the estimated impact of removing 18–20 years old cigarette purchasers from the market is a 2% reduction in total retail sales.6 The estimated public health impact, however, is projected to be far greater.

The Institute of Medicine (IOM) concluded that a nationwide MLA of at least 21 years would substantially reduce population-level smoking prevalence.7 Further, the IOM report demonstrated that a nationwide T21 policy over time would prevent 249 000 premature deaths in the USA and 4.2 million lost-life years among Americans born in the past decade.7 A key lever to this trajectory is preventing or delaying tobacco use initiation among adolescents and young adults. According to the IOM, the predicted effect of an MLA of 21 years is largest among 15–17 year olds who would be placed at a greater social distance from older peer sources of tobacco.7

Hawaii’s T21 law prohibits possessing, consuming or purchasing tobacco products in any form if under 21 years of age; and selling or furnishing tobacco products to anyone under 21 years. Because the T21 law contains no exemptions for special populations and because military bases in Hawaii are voluntarily complying with state law,8 9 everyone under 21 years of age is subject to Hawaii’s T21 law. California’s T21 law10 (effective 9 June 2016), in contrast, contains a military exemption and is not currently observed by military installations and some tribal lands.11 This contrast provides an opportunity to examine the initial outcomes on tobacco sales of a universal statewide T21 law (Hawaii) relative to a comparison location with a similar tobacco control policy environment but a T21 law with exemptions (California), and against the broader policy landscape of the contiguous USA.

In the USA, adolescents (12–17 years of age) and young adults (18–25 years of age) are heavy consumers of mentholated products,12–14 influenced by marketing practices that target youth, females, persons of colour and sexual minorities.14–16 Compared with one in three adult smokers over 25 years of age who smoke menthol (vs non-menthol) cigarettes, the percentages of adolescent smokers and young adult smokers who smoke menthol cigarettes are 56.7% and 45.0%, respectively.13 In these younger populations, while non-mentholated cigarette use is declining, menthol cigarette use is stable or increasing.13 A majority of youth and young adults who have ever used a tobacco product report initiating tobacco use with a flavoured product,14 which can increase the likelihood of sustained tobacco use and compromise efforts to quit.14 15 Flavour enjoyment and ease of use relative to the product’s non-flavoured equivalent are cited as reasons for product selection.14 Flavoured cigars, which are driving growth in the overall cigar market,17 also are popular, especially among younger people.12 Sometimes used in combination with marijuana (eg, blunts),12 18 flavoured (vs non-flavoured) cigars are preferred to provide sweetness, reduce tobacco’s bitterness and mask marijuana’s characteristic odour when burned.19

Limited evaluation studies to date suggest that communities with T21 policies experience greater declines in youth smoking and reduced cigarette purchases among current smokers than comparison communities.20 It is acknowledged, however, that time and effective enforcement are needed to realise T21’s full potential.21–23 At the time this study was conducted, only Hawaii and California had sufficient statewide implementation experience to assess early outcomes of T21, such as changes in tobacco product sales.24 25 It is also important to note that Nielsen data were only available from large food stores in Hawaii (vs other types of retail outlets), which delimited the analytic scope. Therefore, this study had two aims. First, we primarily sought to assess the association of T21 on unit sales of cigarettes and cigars as well as the share of observed unit sales for each product that was flavoured (including menthol) in large food stores in Hawaii. As a secondary objective, we sought to differentiate implementation of T21 in various settings and account for secular trends; we compared sales trends in Hawaii with those of California (T21 policy with special population exemptions) and similar outlets on the US mainland operating under various policy conditions.



We acquired Nielsen Company (Nielsen) Universal Product Code (UPC) level data for cigarette and cigar sales occurring in civilian large-scale (>US$2 million sales/year) food stores (eg, Foodland Super Market, Safeway) in 4-week aggregates from 27 May 2012 to 25 February 2017. The analysis focused on sales of cigarettes and ‘cigars’, comprised of large cigars and cigarillos (the latter dominating food store sales in this product category). Sales were assessed for the Oahu, Hawaii Designated Market Area (DMA) (henceforth, ‘Hawaii’), the aggregate of the four California DMAs: Los Angeles, San Diego, San Francisco and Sacramento (henceforth, ‘California’), and the total mainland US market (henceforth, 'USA'), which excludes Hawaii and Alaska. DMAs are Nielsen-defined geographical areas that are a collection of counties centred around a metropolitan area. Counties included in the Hawaii and California DMAs represent approximately 69% and 84% of the total population of each state, respectively. The analysis focused on food store sales exclusively because these were the only civilian sales projections available for Hawaii from Nielsen.


The data were cleaned and categorised using UPC-level (ie, bar code) characteristics, such as brand, common name (eg, cigarillo), flavour (eg, menthol) and pack size (eg, 20 sticks). Unit sales were standardised so that each cigarette unit represents a single pack of 20 sticks and each cigar unit represents a single cigar. Each tobacco product was coded as non-flavoured or flavoured. Flavours include characterising flavours such as fruit, alcohol or candy, as well as menthol, mint and wintergreen. All flavour descriptors that referenced tobacco or something general like ‘original’ or ‘robust’ were coded as non-flavoured. Also analysed were the proportion of total cigarette sales that are menthol (referred to as the sales share of menthol cigarettes), and the proportion of total cigar sales that are flavoured/menthol (referred to as the sales share of flavoured/menthol cigars), calculated by geographical area.

We plotted trends in total unit sales and the sales share of menthol cigarettes and flavoured/menthol cigars in Hawaii, noting the date in which Hawaii’s T21 policy was implemented, to visually observe changes in sales around the time of T21 implementation. Following methods similar to those used elsewhere,26 27 we assessed changes in the level of total unit sales and flavoured/menthol sales share from prepolicy to postpolicy using a segmented regression analysis. Separate regression analyses were conducted for each outcome and geographical area, with models centred at the data period containing the date of policy implementation. The date of implementation for Hawaii T21 (1 January 2016) was used to assess changes in Hawaii and US sales, while the date of implementation for California T21 (9 June 2016) was used to assess changes in California. The dependent variables were unit sales and the sales share of flavoured/menthol products. Independent variables included (1) separate time trends for the prepolicy and postpolicy periods, allowing for estimation of the slope before and after policy implementation and (2) indicators for the prepolicy and postpolicy periods, which enable estimation of the level of sales or share just before and at the time of implementation. Models were fit using Stata V.14.228 and controlled for first-order autocorrelation and heteroscedasticity to evaluate coincidence of changes in sales with T21 implementation. For each geographical area, model estimated sales or share from each regression were used to calculate the average monthly sales or share for the prepolicy and postpolicy periods. Postregression t-tests were conducted to test for significant differences (p<0.05) in average monthly estimated unit sales or flavoured/menthol share of sales in the prepolicy and postpolicy periods. Change in average monthly unit sales and flavoured/menthol share between prepolicy and postpolicy implementation reflects 47 prepolicy and 15 postpolicy periods for the Hawaii and US models and 52 prepolicy and 10 postpolicy implementation periods for the California model. Model specification did not allow for statistical comparison across geographical areas.



A comparison of average monthly unit sales between the prepolicy and postpolicy periods revealed a significant decrease in cigarette sales in Hawaii following implementation of T21 (−4.4%, p<0.01) (table 1). Significant reductions in average monthly unit sales of cigarettes also were observed in the mainland USA over the same periods (−10.6%, p<0.01) and in California across its prepolicy and postpolicy periods (−11.7%, p<0.01). Average menthol cigarette sales share also decreased significantly in Hawaii from prepolicy to postpolicy periods (−0.8%, p<0.01), while the menthol cigarette sales share increased across the same time periods in USA (1.7%, p<0.01). Average menthol cigarette sales share increased significantly in California following its T21 implementation (3.6%, p<0.01).

Table 1

Regression results for the effect of Tobacco 21 policies on average monthly unit sales of cigarettes and menthol sales share, 2012–2016

Large cigars and cigarillos

Average monthly unit sales of cigars decreased significantly in Hawaii following implementation of T21 (−12.1%, p<0.01) from the prepolicy to postpolicy period (table 2). Similar changes were observed for the USA and California. T21 policies had no statistically significant effect on average flavoured/menthol cigar sales share in Hawaii and California; however, there was a significant increase in the flavoured/menthol cigar sales share in the USA relative to the Hawaii policy implementation date (7.1%, p<0.01).

Table 2

Regression results for the effect of Tobacco 21 policies on average monthly unit sales of large cigars/cigarillos and flavoured/menthol sales share, 2012–2016


This study examined changes in unit sales and flavoured/menthol product share for cigarettes and large cigars/cigarillos following statewide implementation of a T21 policy. In Hawaii, we observed a decrease in average monthly unit sales for cigarettes coupled with a decrease in menthol share. This combination of effects was not observed in California or the USA. In relative terms, this pattern suggests the effect of T21 might vary in different implementation contexts. Given that the use of menthol cigarettes is higher among smokers <25 years of age (vs older smokers),29 it is possible that the sales share reduction in menthol products is attributable to fewer retail sales to youth and young adults.

T21 laws also had an observable effect on sales of large cigars/cigarillos. Average unit sales decreased significantly in both Hawaii and California following T21 policy implementation. While similar to decreases observed in the USA, the magnitude of change was larger in the two states with T21 policies and larger still in Hawaii, which had no special population exemptions. Although the statistical significance of differences across regions is unknown, results suggest that exemptions may blunt the full effect of statewide T21 policies. Notably, while average flavoured/menthol sales share for cigars/cigarillos did not change with statistical significance in Hawaii or California following T21 policy implementation, it increased in the USA, which suggests that T21 policies may have attenuated an otherwise upward trend.

Prior national-level research demonstrated that young adults (aged 18–24 years) are more likely to be cigar smokers than their older counterparts (aged 25–44 years).30 For adults and youth alike, cigar smokers are more likely to identify as black than other race–ethnicities,30 31 reflecting cigar advertising trends that target young adults and African-Americans.32 While the prevalence of current tobacco use among youth is lower in Hawaii compared with the USA,33 a larger percentage of Native Hawaiian, Pacific Islander and non-Hispanic Black high school students use tobacco than other race–ethnicities.31 For high school students overall, nationally, past 30-day cigar use (8.6%) rivals that of cigarettes (9.3%).34

Taken together, these results indicate that T21 policies may decrease sales of some tobacco products, especially cigars and flavoured/menthol products that are disproportionately consumed by youth, young adults and persons of colour. Given patterns observed before and after implementation of statewide T21 laws in varied policy conditions, results suggest that exemptions may blunt the potential impact of statewide T21 policies.

For both Hawaii and California, intense activity occurred in the 1–2 months preceding the T21 law going into effect and continued in earnest during the first year of implementation (LJ, 2017).25 Prior to the date T21 took effect, Hawaii’s Tobacco Prevention and Education Programme staff and coalition partners, for example, communicated with military officials about voluntary on-base compliance; distributed toolkits with new signage, window decals, and flyers to tobacco retailers; and began an education campaign with materials that targeted the general public, grocers, and tourists/retail customers whose primary language is Japanese, Chinese or Korean. During the planning period in Hawaii, enforcement authorities agreed to emphasise public education and delay imposing fines and citations for possession for the first 3 months of T21 implementation. In addition, Hawaii’s Department of Health worked with the US Substance Abuse and Mental Health Services Administration to revise protocols for compliance checks on underage sales to reflect the increased MLA. California’s multipronged information campaign is detailed elsewhere.25

Statewide T21 policies are in their implementation infancy; at present, research on the impact of these policies continues to emerge. Analysis of sales data serves as a barometer for immediate outcomes of T21 policies driven by age-related access restrictions; however, continued surveillance will be important to examine longer term behavioural change. Time may be a major catalyst for achieving T21 policy goals21; disruption of the older-to-younger peer social supply chain requires time to take hold, and changes in social norms that will ultimately reinforce tobacco prevention and control outcomes may take time to evolve.

While it is encouraging to observe marketplace changes associated with statewide T21 implementation, results are likely fractional to what could be achieved with more comprehensive tobacco prevention and control measures. T21 is regarded as one of several recommended strategies aimed at preventing or delaying tobacco use initiation among youth and young adults including: increasing tobacco product prices, limiting retail outlet density and proximity to schools, and restricting sales of flavoured tobacco products (including menthol varieties).35 It is important to note that, in at least one jurisdiction, critics of menthol product sales restrictions made unsubstantiated claims in their opposition that T21 policies were a more effective alternative to reducing youth initiation.36 The present study makes no claims of comparative effectiveness. Rather, it examines T21 as an additional component of a comprehensive suite of tobacco prevention and control strategies. To maximise full benefits of T21 policies, education and enforcement are also critical.7 As with other tobacco prevention and control measures,37 38 states and localities enacting T21 policies will vary in their ability to fully and consistently implement, monitor and enforce the law. Uneven implementation and enforcement may yield suboptimal results.23 Before enacted laws take effect, jurisdictions could consider collaborating with stakeholders, tobacco control partners and relevant authorities to develop and implement plans to create public awareness, support compliance and operationalise enforcement.


This study is subject to several limitations. First, Nielsen scanner data on retail tobacco sales in Hawaii were only available for large food stores, which may not be representative of all tobacco product purchases. Convenience stores are the most common retail outlet type for cigarette sales in the contiguous USA,39 including cigarette sales to underage persons aged 15–17 years.40 Second, retail scanner data cannot distinguish sales by age of purchaser and so cannot isolate these findings to adults aged 18–20 years (ie, those most directly affected by the increased MLA). Third, US comparison data could not be disaggregated to exclude sales in California and other cities/counties with T21 policies; however, the California-specific data enable a non-confounded comparison. Fourth, the present study may not have accounted for other state-level or local-level tobacco control policy changes implemented during or adjacent to the study period. For example, authors are aware of a large tax increase on cigarettes in California that went into effect after the study period as well as the legalisation of recreational marijuana, which is sometimes added to cigar blunts. It is possible that product hoarding occurred in anticipation of these other policy changes, which could affect the estimated association of T21 policies and cigarette and cigar sales. Despite these limitations, the data presented here are the most complete picture to date of cigarette and cigar sales in Hawaii and California before and after the adoption of their T21 policies.


Evidence about the effectiveness of statewide T21 policies is limited, and this study adds to the science base underlying efforts to expand the reach of T21 policies. Relative to comparison areas, sales data from large food stores suggest that implementation of a universal T21 law in Hawaii is associated with a decline in the market share of menthol cigarettes in the state. Further, results suggest T21 policies in varied settings may reduce flavoured/menthol cigar consumption. Given youth and young adults’ preference for flavoured/mentholated tobacco products, results suggest that T21 laws have the potential to reduce tobacco use among this population; future studies of consumer behaviour could further assess this topic. Additional surveillance efforts are also important to understand the full impact of T21 laws; studies of economic impact may be assessed alongside behavioural research on tobacco use patterns. As part of a comprehensive approach to prevent or delay tobacco use initiation, T21 laws may help to reduce sales of various tobacco products, including those most preferred by youth and young adults in the USA.

What this paper adds

  • In 2016, Hawaii began enforcing the first statewide Tobacco 21 law with voluntary compliance by US military installations. Later the same year, California implemented its Tobacco 21 law with military and tribal land exemptions.

  • Given that statewide Tobacco 21 laws are still in their infancy, evidence about their effectiveness in varied policy contexts is limited.

  • Following Tobacco 21, average monthly cigarette unit sales in large-scale food stores in Hawaii dropped along with menthol market share. Decreases in unit sales of large cigars/cigarillos in food stores were larger in Tobacco 21 states than in the USA, particularly in Hawaii. Tobacco 21 was not associated with flavoured/menthol cigar sales share; however, compared with sales in the USA, results suggest that Tobacco 21 attenuated an otherwise upward trend in these products.



  • Contributors All coauthors have made substantial contributions to the creation of the manuscript and take responsibility for it. RG-K, DGG, TR, EMC, BL and RL conceptualised the study. DGG, TR, EMC and BL conducted data analysis and interpreted results. RG-K, DGG, TR, EMC, BL, LJ and MW drafted portions of the manuscript. RG-K, DGG, TR, EMC, BL, RL, LJ and MW critically reviewed and/or revised the manuscript during peer review. RG-K was responsible for overall content and manuscript submission.

  • Funding Funding support was provided by the US Centers for Disease Control and Prevention, Office on Smoking and Health.

  • Disclaimer The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention or RTI International.

  • Competing interests None declared.

  • Patient consent for publication Not required.

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

  • Data availability statement No data are available.