Article Text
PDF
Abstract
Background The E-cigarette, or Vaping Product-Use Associated Lung Injury (EVALI) Outbreak of 2019 hospitalised thousands and killed dozens of people in the USA and raised perceptions of the dangers posed to health by electronic cigarettes (e-cigarettes). These illnesses along with continued increases in youth vaping rates lead to the passage of many state and federal laws intended to curtail the sale of flavoured e-cigarettes. Little is known about the impact of these events on US e-cigarette and cigarette retail sales.
Methods Using Nielsen Scantrack sales data from January 2014 to January 2020 for 23 US states, we evaluate the effect of the EVALI outbreak. First-differenced state-panel regressions tracking unit sales of total-level and category-level e-cigarettes and cigarette sales controlling for price, Tobacco 21 policy coverage, product distribution, seasonality, EVALI-attributable deaths, and state-level e-cigarette policies affecting the availability of e-cigarettes (non-tobacco flavoured and total) were employed.
Results Dollar sales of e-cigarettes declined 29% from their pre-EVALI peak by January 2020. Total sales of e-cigarettes declined in response to EVALI deaths and the total e-cigarette sales ban put in place in Massachusetts adopted in its wake. Cigarette sales were largely unchanged by either the direct or indirect policy effects of the EVALI outbreak, except for in Massachusetts, where cigarette sales—particularly those smoked by young people—rose temporarily after a total ban on e-cigarette sales.
Conclusion Sales of e-cigarettes declined in response to the EVALI outbreak and from the most restrictive regulatory policies that were adopted in response, while sales of cigarettes were affected less.
- surveillance and monitoring
- electronic nicotine delivery devices
- public policy
- taxation
Data availability statement
Data may be obtained from a third party and are not publicly available. The data-sharing agreement with Nielsen does not allow us to share the underlying data with third parties.
Statistics from Altmetric.com
Background
Sales of nicotine-containing electronic cigarettes (e-cigarettes) in the USA grew rapidly from their introduction to the market at the beginning of the decade until mid-2019.1–3 August 2019 saw reports that young people who had recently used e-cigarettes were being admitted to intensive care units with pneumonia-like lung injuries.4 5 These hospitalisations were shortly followed by reports of the first-ever deaths from an illness associated with e-cigarette use.6 These reports materialised against a backdrop in which the Food and Drug Administration released early findings from the 2019 National Youth Tobacco Survey in September, which documented an all-time high of 27.5% of high school students reporting past-30-day e-cigarette use, the majority of whom were found to primarily use e-cigarettes flavoured to taste like fruit and mint.7 While e-cigarette use is likely to be less hazardous than smoking for adult smokers who switch completely,8 these high levels of e-cigarette use among US youth pose a significant public health threat. Concerns stem from the evidence demonstrating the risks associated with youth exposure to nicotine, its high addiction potential, and the associations between youth e-cigarette use and future cigarette smoking.9–11 In response to these rising youth prevalence figures and increasing hospitalisations and deaths, federal and state governments issued a series of regulatory policy responses that attempted to control and contain the e-cigarette market, including temporary bans of some or all flavoured e-cigarettes in four states.
In October 2019, the US Centers for Disease Control and Prevention (CDC) named the new syndrome, ‘E-cigarette, or Vaping Product-Use Associated Lung Injury’, or EVALI, although the exact cause of the outbreak remained unclear.12 13 By November 2019, the CDC concluded that EVALI was primarily driven by illicit Tetrahydrocannabinol-containing electronic vaping devices, especially those that contained a dangerous additive, Vitamin E acetate.14 15 Driven to respond to the rise in youth use of flavoured e-cigarettes and the EVALI outbreak, in January 2020, the US Food and Drug Administration (FDA) released new national guidance which gave a higher enforcement priority to refillable e-cigarette cartridges flavoured to taste like anything other than tobacco or menthol under which companies risked FDA enforcement if they did not cease the manufacture, distribution and sale of such products within 30 days.
Previous research suggests that increasing tobacco harm perceptions and tobacco product bans have the potential to shift tobacco product sales and use patterns.1 16–19 The literature seeking to define the economic relationship between cigarettes and e-cigarettes, the two crucial markets in this analysis, is empirically mixed. One study has found cigarettes to be complementary to e-cigarettes,20 some have found indeterminate or mixed results,21–23 and others have found these goods to be economic substitutes.24–27 The latter studies have led economists to express concern that events or policies which reduce the sales of e-cigarettes may increase the sales of combustible cigarettes. The EVALI outbreak was associated with raising harm perceptions of e-cigarettes.28 29 Additional previous research documented shifts in the tobacco product market and tobacco use behaviours following tobacco product bans.16 30–32 Recent work from Friedman documented a short-run rise in cigarette smoking among youth in San Francisco in the wake of that city’s ban on the sale of all flavoured tobacco products,32 while Gammon et al documented large decreases in flavoured product sales with partial substitution towards remaining tobacco flavoured products in the wake of the San Francisco policy change.31 To our knowledge, no study has examined how the EVALI outbreak and contemporaneous policy actions affected the e-cigarette and cigarette markets. In this paper, we explore how these tobacco sales were impacted by the EVALI outbreak as measured by EVALI deaths and state bans of flavoured or all e-cigarettes starting in September 2019 and ending 25 January 2020—the period before the FDA national guidance went into effect.
Methods
We use Scantrack data on sales of e-cigarettes and cigarettes in the USA provided by The Nielsen Company. The dataset contains Universal Product Code-level unit and dollar sales in 4-week period ending first on 1 February 2014 and last on 25 January 2020, in Nielsen-tracked channels (convenience stores, food/drug/mass merchandisers, discount stores and military commissaries), in 23 of the largest US states by population. These states account for 76.4% of cigarette dollar sales, 79.3% of e-cigarette dollar sales in Nielsen channels and 79.8% of the population as of 2018.33 Using data from the Consumer Price Index, we adjusted prices for inflation to January 2017 dollars.34 We aggregated sales of e-cigarettes into four groups: (1) durable hardware (includes parts, empty refills and starter kits), and (2) consumable products, which are defined as products that are intended to be used up and replaced, including disposables devices, prefilled cartridges and e-liquid bottles. We then divided consumable products by flavouring into tobacco, menthol and all other (including mint, fruit and sweet) flavours. Following Liber et al, we use brand information from the National Survey on Drug Use and Health to disaggregate three subcategories of cigarette sales (in addition to total cigarette sales): brands whose users skew younger than the population average (Marlboro Menthol, L&M Menthol, Newport Regular, Camel Menthol and Newport Menthol), brands whose users skew much older than average (Basic Regular, Doral Menthol, Doral Regular, Misty Menthol, Misty Regular, Salem Menthol, Virginia Slims Menthol and Virginia Slims Regular) and age-proportionate brands of those whose user age structure resembles the population of smokers as a whole (includes all other brands).35 We hypothesise that as e-cigarette use is more common among young people in the USA,36 sales of young-skewing cigarette brands should track with substitutional purchasing behaviours.
In multivariate analyses, we employ first-differenced, logged per-capita unit sales as the outcome variable of interest to account for non-stationarity in e-cigarette and cigarette sales, each of which was found to contain unit roots using the Levin-Liu-Chu test. Differencing accounts for this non-stationarity without requiring strong assumptions about the correct parametrization of trends in sales absent EVALI, a particular concern for e-cigarette sales given their relatively steep upward trajectory in the years leading up to EVALI.
To chart the course of the EVALI outbreak, we consider four measures of its direct impact: hospitalisations, media coverage, reported cases and reported deaths. The construction of these measures is documented in the online supplemental material. After fitting the potential measures to e-cigarette sales figure models that are identical on the right-hand side to the results in online supplemental table 1, we chose the best-fitting-measure, deaths, to represent the direct effects of the EVALI outbreak in further analyses.
Supplemental material
We also focus our study on the impact of regulatory policy changes that followed in the wake of the EVALI outbreak, namely, full and partial e-cigarette sales bans imposed by state governments in Washington, Michigan, Oregon and Massachusetts. To evaluate the effect of the EVALI outbreak separately from this rapidly shifting policy environment, we employ several descriptive and multivariate analytical strategies. For figures 1 and 2, we sort states into groups determined by the most strict e-cigarette flavour restriction policy in effect during the study period: No State Policy, where no state-level changes to e-cigarette flavour availability during the study period, State Partial Ban, where only tobacco flavour was available (Washington State (10 October 2019– 10 February 2020), Michigan (18 September 2019–15 October 2019] and Oregon (15 October 2019–17 October 2019)) and Total Ban, where no e-cigarettes of any flavour were available (Massachusetts (24 September 2019–23 December 2019)). After the total ban expired in Massachusetts, it was replaced with a policy banning sales of all non-tobacco flavoured e-cigarettes, a partial ban in our parlance, and is coded accordingly in our data. Our analysis aims to separate the differential effects of these policies from the EVALI outbreak’s direct impact. To achieve this, we first plot the average per-capita total e-cigarette sales split by policy group in figure 1 as well as average per-capita e-cigarette category-level sales split out by policy group in figure 2. To plot average per-capita cigarette sales by policy group in (for Young and Old Brands) (figure 3) and (for Total Sales) in online supplemental figure 2, we used dates before 2019 to calculate state-specific seasonal adjustments to account for the cyclical nature of cigarette sales.
Average per capita sales of e-cigarettes by state e-cigarette policy group (real US$, January 2017–January 2020). Note: Horizontal lines indicate the timing and duration of partial and total state e-cigarette sales bans. State Total Ban=Massachusetts; State Partial Ban=Average of Michigan, Oregon and Washington; No State Policy=Average of Alabama, Arizona, California, Colorado, Florida, Georgia, Illinois, Indiana, Kentucky, Louisiana, Missouri, New Jersey, New York, North Carolina, Ohio, Pennsylvania, Tennessee, Texas and Virginia.
Average per capita category-level sales of e-cigarettes by state e-cigarette policy group (real US$, January 2017–January 2020). Note: State Total Ban=Massachusetts; State Partial Ban=Average of Michigan, Oregon and Washington; No State Policy=Average of Alabama, Arizona, California, Colorado, Florida, Georgia, Illinois, Indiana, Kentucky, Louisiana, Missouri, New Jersey, New York, North Carolina, Ohio, Pennsylvania, Tennessee, Texas and Virginia.
Average per capita cigarette sales, December 2018--January 2020, by e-cigarette policy and brand age category, adjusted for seasonal consumption patterns. Note: State Total Ban=Massachusetts; State Partial Ban=Average of Michigan, Oregon and Washington; indexed per capita cigarette sales No State Policy=Average of Alabama, Arizona, California, Colorado, Florida, Georgia, Illinois, Indiana, Kentucky, Louisiana, Missouri, New Jersey, New York, North Carolina, Ohio, Pennsylvania, Tennessee, Texas and Virginia.
Then, we employ multivariate first-differenced panel regression models predicting change in log-differenced e-cigarette per-capita sales volumes to separate the effect of the EVALI outbreak and the policy changes that followed in its wake. All model covariates are first-differenced to reflect the differenced outcome variable. Non-policy related model covariates include average logged real sales-weighted price for each product category, distribution of Juul pods in fruit and mint flavours—the sales-weighted share of retailers carrying Juul Lab’s Fruit (Mango and Fruit Medley) and Mint-flavoured pods which heavily contributed to a dramatic expansion of the US e-cigarette market—and total distribution points for the category of sales being measured to capture retail availability more generally.1 Policy variables are equal to the proportion of days in each period that a given policy was in effect in a given jurisdiction. For Tobacco 21, this value is further weighted to the share of a given state’s population that is covered by a new Tobacco 21 policy in the latter portion of a 4-week period. All states were coded as being covered by a Tobacco 21 policy after December 2019.37 For sensitivity purposes, each indexed measure of the EVALI outbreak was tested in different versions of the regression (online supplemental table 1), and EVALI deaths were selected as the measure of the outbreak with the best model fit and therefore used for the remainder of the paper. Seasonal adjustments were made using dummies for each of the 13 4-week periods per year as well as heating degree days (HDD), which indicates the share of each period in a given state that was cold enough to necessitate household heating.38 Cigarette sales are cyclical throughout the year, and this pattern is more pronounced in regions with larger temperature changes, which necessitates using a variable to capture this variation. HDD was interacted with the overall average HDDs for each state to allow for unseasonably warm weather to have a differential impact on colder-weather states. The following formula was used for the full model multivariate analysis for e-cigarettes:
We used Bayesian Infomation Criterion (BIC) scores to guide our variable selection, which was minimised to determine the specification for the full model of total sales of e-cigarettes and its category models. The multivariate total cigarette sales and brand group models were identical to the e-cigarette models, except for the removal of 
and 
which were not included because they raised the BIC.
Additional model specifications are included for each market segment in online supplemental tables 3 to 11, building from basic models with the three primary EVALI variables (deaths and state policies) up to the full model. Sensitivity analyses were performed examining the impact of removing Massachusetts from the analytical sample, as the partial e-cigarette ban there represented a loosening of sales restrictions rather than the tightening of what could be sold that the partial ban represented in Michigan, Oregon and Washington.
Results
Descriptive results
E-cigarette sales
Before the peak of the EVALI outbreak, monthly e-cigarette sales by value had exceeded all previous monthly sales in 31 of the previous 33 4-week periods in the USA dating back to February 2017 (see online supplemental figure 1).39 40 The only interruption to that trend was the temporary decrease in sales after JUUL pulled certain fruit-flavoured and sweet-flavoured products from retail stores in November 2018.1 By January 2020, e-cigarette sales in the prior 4-week period were US $295 million or 28.7% below the prior maximum of $416 million achieved in August 2019.
Using average real state-level per-capita data on e-cigarette sales split out by policy group, figure 1 suggests that different state’s e-cigarette markets were not affected equally by the events precipitated by the EVALI outbreak. Massachusetts, with its Total Ban, saw a substantial decline in e-cigarette sales while states that banned all but tobacco-flavoured e-cigarettes lost nearly half their e-cigarette sales. States that did not pass any state-level e-cigarette policies appear to see smaller reductions in the immediate aftermath of EVALI.
Figure 2 shows that Partial Bans in Michigan, Oregon and Washington appear to have had the first-order effect of decreasing all menthol and other flavoured product sales while providing a temporary increase to tobacco-flavoured products. The total ban in Massachusetts cut across all product categories, effectively zeroing them out for 3 months, as sales fell 99% from US$10.7 million in August to US$103 000 in December. Once the temporary Total Ban was replaced with a Partial Ban, only tobacco flavours and hardware sales increased, and they did so at a pace 95% smaller than before the outbreak.
Cigarette sales
Seasonally adjusted cigarette volumes initially declined in Massachusetts through August 2019 possibly due to the implementation of a ban of cigarette sales in all drug stores starting at the beginning of 2019.41 Beginning in the 4 weeks ending 5 October 2019—the first period that included the e-cigarette ban imposed on 24 September—sales volumes for all cigarettes in Massachusetts increased for three consecutive 4-week periods (online supplemental figure 2). Average sales volumes in states with partial bans and states that were not subject to a state ban do not appear to be as visibly affected by the outbreak. The increase in cigarette sales is more pronounced for sales of brands that skew young, while brands that skew old do not show any change in their sales trajectory coinciding with the e-cigarette ban in Massachusetts (figure 3).
Multivariate results
Table 1 displays regression results for first-differenced total and category level logged per capita e-cigarette unit sales as well as total and brand group logged per capita cigarette sales in 23 states over the period of February 2014 through January 2020.
First-differenced regression results for log-differenced total e-cigarette and cigarette sales and by product category and brand group (unadjusted)
E-cigarette sales
The coefficients reported in table 1 represent the effect of a single month, and the first-order (EVALI deaths) and second-order (policy) effects had an additive effect on e-cigarette sales within periods and over time. Peak EVALI deaths were associated with a 10.6% decline in total sales, while the state total ban decreased e-cigarette sales by 234% per month at its maximum value. A state partial ban on flavoured e-cigarette sales has a negative sign but is not statistically significant (p=0.16).
At the category level, sales of hardware and other-flavoured e-cigarettes have a negative, significant association with EVALI deaths. Sales of hardware and tobacco flavour refills were significantly lower under a state total ban when compared with having no restriction in place. A state partial ban was only significantly associated with decreases in sales of menthol-flavoured refills; however, the effect size was very large, at a 240% decline for a full 4-week period of a partial ban. However, given that Massachusetts partial ban followed a full ban—effectively represented an easing of regulatory restrictions rather than the tightening of rules that the partial ban represented in other states—we suspected that the null association between state partial bans and overall non-menthol e-cigarette sales was driven by the inclusion of Massachusetts. In a sensitivity analysis in which Massachusetts is removed from the analytical sample (online supplemental table 12), the findings for the effects of the partial ban on e-cigarette sales change markedly. The partial ban’s effect becomes negative and significant for total e-cigarette sales, hardware, other-flavoured, and (stays negative for) menthol-flavoured sales while becoming positive and significant for tobacco-flavoured e-cigarettes. The effect of price becomes negative and significant for total e-cigarette sales and EVALI deaths becomes positive and significant for menthol-flavoured e-cigarette sales.
Outside of the primary measures of the EVALI outbreak and its consequences, the model found that Tobacco 21 policy coverage was associated with a significant decrease in total and other-flavoured (including fruit and sweet flavours) e-cigarette sales. Tobacco 21 policy coverage was associated with increasing sales of menthol-flavoured e-cigarettes. Price had a significant and negative relationship with sales for hardware, other-flavoured and menthol-flavoured e-cigarettes. Total e-cigarette sales and hardware were higher during cooler weather and unseasonably warm weather was significantly associated with lower hardware sales. Higher distribution was associated with higher sales for total and every e-cigarette sales category. Higher JUUL fruit pod distribution was associated with higher total, hardware and other-flavoured e-cigarette sales as well as with lower menthol and tobacco flavour e-cigarette sales. Higher JUUL mint pod distribution was associated with significantly lower total, hardware and tobacco-flavoured e-cigarette sales.
Cigarette sales
Total cigarette sales and sales for each brand group were negatively associated with EVALI deaths. State partial e-cigarette bans were not significantly associated with total cigarette sales; however, they were associated with significantly lower sales of disproportionately old brands. In Massachusetts, a full 4-week period of a total ban on e-cigarette sales was associated with a significant 4.1% rise in sales of total cigarette sales. This effect seems concentrated among disproportionately young cigarette brands, as those increased by 8.3% per period of a full ban, age-proportionate brands increased a smaller but significant 2.7% and sales of disproportionately old brands were unaffected.
Price had a significant and negative relationship to total cigarettes sales and for each brand group. Greater distribution of products was significantly associated with higher young and old brand cigarette sales. Cooler weather, measured by HDD, was significantly associated with lower cigarette sales and total sales and for each brand group. HDDs had less effect in cold-weather states in the North for total and disproportionately young cigarette brand sales (measured by the HDD and HDD Mean interaction term). Finally, we find that Tobacco 21 policy coverage was associated with a decline in the sales of young brands (p=0.012), but not significantly associated with total sales or sales of another brand group, which is consistent with Liber et al.35
Discussion
Results from this study highlight the impact of EVALI deaths, and the subsequent and varied state bans on the sale of e-cigarettes on the e-cigarette and cigarette markets.
E-cigarette sales
The size and composition of the US e-cigarette market changed substantially in the aftermath of the EVALI outbreak and contemporaneous policy actions. We found that EVALI deaths were associated with a decline in total e-cigarette sales, something that may be attributable to increased perceived risks to the health of those using these products.17–19 Additionally, policies that restricted the availability of flavoured e-cigarettes were further and more strongly predictive of sales declines among the affected product categories (menthol and all other flavours).
Other recent work has found that the e-cigarette market also saw shifts in e-cigarette device types, particularly in the wake of the FDA’s guidance limiting the sales of cartridge-based e-cigarettes in flavours beyond menthol and tobacco; evidence indicates that sales of e-cigarette products that were exempt from FDA’s flavour restrictions such as disposable e-cigarette products, continued to increase.42 43 While disposable product sales are currently lower than other e-cigarette device types covered by the FDA regulations, a significant rise in their sales44 is cause for concern. Disposable e-cigarette products with youth-friendly flavours and price points are widely distributed by convenience stores and other retailers.45 The shifts in flavoured e-cigarette sales documented in this paper as well as the shift to disposable products documented elsewhere reflects how consumers quickly switch preferences as regulatory policies change.42 43 Monitoring the shifts in e-cigarette sales in relation to regulatory actions is key to maximising the efficacy of policy actions to help reduce youth e-cigarette use. Special attention must be paid to future comparisons between states that have enacted permanent bans on the sale of all non-tobacco flavoured e-cigarettes (New York, New Jersey, Massachusetts and California) and those that have not.
Our results also indicate that several distinct tobacco control policies were found to have significant effects on the size of the e-cigarette market in the US during the study period. Flavour restrictions imposed after EVALI had an additive effect on one another, especially in the jurisdictions that imposed temporary partial bans in late 2019 as a result of the EVALI outbreak. In addition, Tobacco 21 policies were associated with a decline in sales of e-cigarette products overall, as well as those products that have been found to be appealing to young people (mint, fruit and sweet flavours).46 To our knowledge, this is the first evidence that Tobacco 21 policies decrease e-cigarette sales. Higher e-cigarette prices, which are directly affected by excise taxes, were significantly associated with lower e-cigarette sales. Together with press reports on the EVALI outbreak and the direct effects of the EVALI outbreak (those not attributable to policy change in reaction to the outbreak), Tobacco 21 policies, flavour restrictions and e-cigarette price increases also contributed to declining e-cigarette sales in the USA.29
Cigarette sales
EVALI deaths were associated with lower cigarette sales, in total and across all brand groups. This pattern of lower sales during the EVALI period only continued for disproportionately old brands during partial e-cigarette sales bans.
In examining the effect of the full ban on e-cigarette sales in Massachusetts, total cigarette sales rose. This effect was concentrated in sales of disproportionately young cigarette brands, which temporarily increased during the period when e-cigarettes were fully banned before declining again when the full ban on e-cigarette sales was repealed and replaced with a partial ban on the sale of flavoured e-cigarettes. In contrast, sales of cigarette brands that skewed older were not significantly affected by the full e-cigarette sales ban in Massachusetts. Given that younger smokers are much more likely to vape than older smokers, the temporary sales increase in cigarette brands smoked by younger people as compared with brands smoked by older people warrants further research attention.36 Our finding of a lack of cigarette sales growth after state partial bans mildly contrasts with some of the findings by Friedman32 who used a population-based health survey to find youth substitution towards cigarettes in the aftermath of a ban on flavoured e-cigarettes in San Francisco (akin to a partial ban) because the share of youth consumption of all cigarettes is small making the findings potentially compatible. Our findings align closer to the more analogous sales data-based study from Gammon et al,31 which found little positive movement in cigarette sales after the San Francisco policy change. Given the remaining unknowns and uncertainties related to the economic relationship between e-cigarettes and cigarettes, more research is needed to explore how a persistent decline in e-cigarette sales would affect cigarette sales.21–26
Limitations
This study has many strengths but is not without limitations. First, the overall effect on the e-cigarette market in states that imposed flavoured e-cigarette sales bans are likely imprecise given that Nielsen captures data from traditional brick-and-mortar retailers but not from vape shops and online stores. Temporary e-cigarette flavour bans may have prompted closures of vape shops and shifted purchasing to online retailers, as initial survey findings from Massachusetts have indicated.47 As such, we cannot confidently predict how failing to account for sales in non-Nielsen channels affected the observed outcomes. Further, the Nielsen state-level data do not cover the entire country, and as small states with more non-Hispanic white and less affluent populations may react differently to the events of the EVALI outbreak and imposed their own flavoured e-cigarette restrictions (Montana and Rhode Island), we cannot speak to whether the results observed above would hold there.
Second, while we do not explicitly account for the presence or level of excise tax in our findings, we do however use a price variable that incorporates excise taxes. We also do not control for the presence of over 230 local tobacco flavour bans of varying comprehensiveness that were in place before the EVALI outbreak or the FDA’s prioritised enforcement policy for cartridge e-cigarettes that began in February 2020.48 However, the presence of these policies, combined with the youth vaping epidemic accelerated the implementation of flavour bans at all levels of government, making the chosen policies of focus a timely and appropriate subject of study.
Additionally, we opted to classify starter kits as hardware when most such kits contain some consumable nicotine liquid given the laborious and time-consuming process of separating the various parts of the device. Although this coding likely led to some level of inflated hardware sales (hardware comprised 15% of total e-cigarette dollar sales and 7% of e-cigarette unit sales) and decreased total unit sales of consumable products, it is unlikely that this variable could have significantly influenced the observed effects of policy or the EVALI outbreak on sales. Also, given the high degree of e-cigarette product heterogeneity and consumption patterns, our price estimates do not reflect standardised units of e-cigarettes; future research is critical to developing validated methods for addressing product variation.
Our analyses of cigarette sales could be subject to additional scrutiny. The categorisation of ‘young’ and ‘old’ cigarette brands has been used in one other study,35 and it remains unclear whether these brand categories reflect differential changes in product usage among consumers by age. Additionally, while an immediate and temporary finding of a likely shift from e-cigarette to cigarettes in Massachusetts is consistent with previous research,21 27 the nature of this finding cannot be determined using market data, and so we are unable to determine whether this aggregate relationship indicates that former smokers returned to smoking, dual users shifted their consumption to cigarettes or never-smokers initiated cigarette use. Future research is needed to determine behavioural patterns among individual consumers. Last, we are unable to determine whether the observed temporary increase in cigarette sales would have persisted over time if the total e-cigarette ban in Massachusetts had remained in place.
Conclusion
Despite the study’s limitations, this study is the first to our knowledge to examine the association between the EVALI outbreak, subsequent e-cigarette bans and nationwide sales shifts in the US e-cigarette and cigarette markets. Findings that indicate partial and full e-cigarette bans more strongly associated with changes in the e-cigarette market, even as the first-order impact of the EVALI outbreak was accounted for, suggesting that e-cigarette sales bans are likely to decrease the size of the e-cigarette market on their own, even in the absence of a similar public health crisis. While sales data are reflective of market-level consumer behaviour, future studies should examine how the EVALI outbreak and e-cigarette bans were associated with individual behaviour change. Moreover, future work ought to examine the effects of the changes described here on e-cigarette use, initiation and cessation as well as its relationship to other tobacco product usage. The EVALI outbreak highlighted the dangers of related to THC vaping products which are federally illegal and operate out of the view of US federal consumer agencies. Refining regulatory and legislative policy interventions to reduce the significant uptake of these products by youth and the health consequences for all users while minimising increased cigarette sales are critical issues to help improve the public’s health.
What this paper adds
Past studies have identified associations between the E-cigarette, or Vaping Product-Use Associated Lung Injury (EVALI) outbreak and the general public’s perception of the health risks of e-cigarettes; however, little is known about the impacts of the EVALI outbreak and subsequent flavour restriction policies on e-cigarette sales in the USA.
This study measures the first-order impact of the EVALI outbreak through EVALI deaths and the second-order policy responses to ban the sale of some or all e-cigarettes and finds that sales of e-cigarettes declined in response to both, while sales of cigarettes were mostly unaffected; Massachusetts’ total ban on e-cigarette sales had only a limited and temporary effect on sales of cigarettes, primarily in those brands used by young people.
The decline in e-cigarette sales in response to e-cigarette flavour bans, along with no consistent corresponding increase in overall cigarette sales, should be considered in further regulating the e-cigarette market.
Data availability statement
Data may be obtained from a third party and are not publicly available. The data-sharing agreement with Nielsen does not allow us to share the underlying data with third parties.
Ethics statements
Patient consent for publication
Ethics approval
This study does not involve human participants.
References
Supplementary materials
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.
Footnotes
Twitter @AlexCLiber, @MeganCDiaz
Contributors ACL and ZC conceived the research question, led the data analysis and drafted the initial manuscript. MCD and ED performed initial data management and helped craft the analytical strategy. DV and BS provided the data and supervised the project. All authors contributed to editing and revising the final manuscript. ACL is responsible as the guarantor of this manuscript.
Disclaimer This article reflects the views of the authors and should not be construed to represent the views or policies of any affiliated institution or entity. The conclusions drawn from the Nielsen data are those of the researcher(s) and do not reflect the views of Nielsen. Nielsen is not responsible for, had no role in and was not involved in analysing and preparing the results reported.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.