Article Text
Abstract
A consistent approach using standardised items to assess e-cigarette use in both youth and adult populations will aid cross-survey and cross-national comparisons of the effect of e-cigarette (and tobacco) policies and improve our understanding of the population health impact of e-cigarette use. Focusing on adult behaviour, we propose a set of e-cigarette use items, discuss their utility and potential adaptation, and highlight e-cigarette constructs that researchers should avoid without further item development. Reliable and valid items will strengthen the emerging science and inform knowledge synthesis for policy-making. Building on informal discussions at a series of international meetings of 65 experts from 15 countries, the authors provide recommendations for assessing e-cigarette use behaviour, relative perceived harm, device type, presence of nicotine, flavours and reasons for use. We recommend items assessing eight core constructs: e-cigarette ever use, frequency of use and former daily use; relative perceived harm; device type; primary flavour preference; presence of nicotine; and primary reason for use. These items should be standardised or minimally adapted for the policy context and target population. Researchers should be prepared to update items as e-cigarette device characteristics change. A minimum set of e-cigarette items is proposed to encourage consensus around items to allow for cross-survey and cross-jurisdictional comparisons of e-cigarette use behaviour. These proposed items are a starting point. We recognise room for continued improvement, and welcome input from e-cigarette users and scientific colleagues.
- Electronic nicotine delivery devices
- Public policy
- Surveillance and monitoring
Statistics from Altmetric.com
Introduction
E-cigarette use has grown in many high-income and middle-income countries,1–12 resulting in a rapidly evolving e-cigarette marketplace. As e-cigarette use is still a relatively new behaviour, researchers have taken a variety of approaches to measuring use, often adapting cigarette smoking items to assess e-cigarette use. The lack of a consistent approach to assessing e-cigarette use is a barrier to knowledge synthesis13 14 and to conducting meaningful cross-national comparisons of the effect of e-cigarette policies on population tobacco use patterns. It has been recommended that monitoring, evaluation and research use standardised approaches and definitions of e-cigarette use for trial, occasional and regular users, and among youth and adult populations.13 15
As evidenced by at least 139 countries’ adoption16 17 of the Global Adult Tobacco Survey (GATS) and Global Youth Tobacco Survey, researchers, governments and funders are aware of the power of common items for understanding the effect of policy on behaviour. The following suggested core e-cigarette items are the result of a Robert Wood Johnson Foundation-funded (RWJF) project (‘Harvesting Global Learning on Alternative Nicotine Delivery Systems (ANDS) to Inform U.S. Policy Action, Policy Research, and Surveillance’) that brought together researchers and government representatives to identify existing needs to support cross-national e-cigarette research and learning. While no formal Delphi method was employed, the following recommendations are based on input from the 65 individuals from 15 countries included in the RWJF meeting series, as well as the authors’ own experiences developing questions, analysing responses, and/or interpreting findings for the International Tobacco Control (ITC) Four Country Study,18 Smoking Toolkit Study (STS),19Population Assessment of Tobacco and Health (PATH) Study,20 Online Panel Survey in Great Britain,21–23 Truth Initiative Young Adult Cohort Study,24 National Health Interview Survey3 and National Youth Tobacco Study25 26 surveys, providing unique insight into the strengths and limitations of various e-cigarette items.
Focusing on adults, the purpose of this paper is to propose an efficient set of e-cigarette use items to enable accurate cross-jurisdictional comparisons of e-cigarette use behaviour and to allow systematic evaluation of the effects of policy on e-cigarette and tobacco product use. While they still need to undergo systematic evaluation, we hope that these proposed items will promote open dialogue and further development of rigorous items for national and subnational e-cigarette surveillance research.
Assessing e-cigarette use
There are several general issues that need to be considered when developing a survey with e-cigarette items. These include the survey’s target population, the policy setting, and the mix of tobacco products and e-cigarette devices available to the target population. It is also important to take into account e-cigarette terminology and to accurately differentiate between e-cigarettes, other emerging products and traditional tobacco products. We describe these issues here before introducing a core set of recommended items.
E-cigarette terminology
E-cigarettes are known by a variety of names, with terms varying by region, age group, tobacco use status or reason for use.27 28 Terms that have been used include electronic cigarettes, e-cigarettes, electronic nicotine delivery systems (ENDS), alternative nicotine delivery systems (ANDS), electronic vapour products, e-cigars, e-pipes, e-hookahs, e-shishas, personal vaporisers, vape pens and hookah pens. The meaning of these terms is not standardised, and the same term may be employed to refer to different subtypes of devices. The researcher-generated terms ‘ENDS’ and ‘ANDS’ inaccurately imply that these devices always contain nicotine. These are academic terms and should be avoided in public-facing documents and presentations.28 Currently, it is likely that the most universally understood terms are ‘electronic cigarettes,’ ‘e-cigarettes,’ or the phrase ‘e-cigarettes or other vaping devices.’
A helpful way to introduce the relevant terminology in surveys is to include a ‘preamble,’ or a brief introduction at the start of the e-cigarette section. For example, the Wave 1 survey instrument for the PATH Study used this preamble:
‘The next questions are about e-cigarettes. Some e-cigarettes can be bought as one-time, disposable products, while others can be bought as reusable kits with a cartridge or tank system. Some people refill their own e-cigarettes with nicotine fluid, sometimes called ‘e- liquid.’ Disposable e-cigarettes, e-cigarette cartridges and e-liquid come in many different flavours and nicotine concentrations. Some common brands include Fin, NJOY, Blu, e-Go and Vuse.’
This preamble was developed for use in the USA in 2014, and has been updated in each PATH Study survey wave. Researchers should be aware that introductory text such as this preamble will need to be modified as products change and the public develops familiarity with e-cigarettes. In markets where ‘heat-not-burn’ products, such as iQos, have been introduced, the preamble could note that respondents should not consider these products when answering e-cigarette items. We strongly suggest pilot research to assess the appropriate e-cigarette terms in surveys, and if possible, we suggest including both the standard (eg, e-cigarette) and colloquial (eg, vaping device) names. Future research should include regular cognitive testing of terminology used to identify e-cigarette native terms used by the survey’s target population. Surveys with appropriate modes may consider using pictures of devices. These pictures should also be cognitively tested and updated as e-cigarettes evolve in the target population’s setting.
Differentiating e-cigarettes from cigarettes and new emerging products
An additional challenge in e-cigarette survey item development is clearly differentiating cigarette items from e-cigarette items. As these devices do not produce smoke, it is not appropriate to refer to ‘smoking’ or ‘smoker’ when describing e-cigarette use, nor are these terms generally employed among established e-cigarette users.29 The scientifically accurate term for e-cigarette emissions is ‘aerosol’; however, the popular term for e-cigarette emissions understood by the public is ‘vapor'. E-cigarette use behaviour should be described as ‘use’ or ‘vaping.’ We recommend differentiating e-cigarette use from ‘smoking’ when smoking is first mentioned in the survey. For example, the 2016 Online Panel Survey in Great Britain21–23 used the following text at the beginning of its tobacco use section: ‘When we refer to cigarettes, pipes, cigars, or other tobacco products, we are not referring to electronic cigarettes or vaping devices (because these do not contain tobacco).’ In the case of vaping devices that could be used for nicotine or cannabis consumption, researchers could consider an additional item about the substance most commonly consumed with the device, which could then be used as a basis for skip patterns or form fills. We also strongly recommend that heat-not-burn products be assessed separately from e-cigarette products. As heat-not-burn products continue to spread within and beyond the European Union and Japan, the research community will need to seriously consider how to assess use of these products so that they are differentiated from traditional combusted tobacco and vaping products.
Recommended e-cigarette items
Table 1 presents a minimum set of e-cigarette items that, in the experience of the authors, are essential to assessing the role of policy on e-cigarette and tobacco use behaviour. The items cover eight constructs: ever use, frequency of use, former daily use, relative perceived harm, device type, presence of nicotine, flavour preference and reasons for use. The first two constructs, ever use and frequency of use, are further identified as minimum core e-cigarette items when survey space is limited. It should be noted that this minimum set of items is insufficient for surveys of tobacco users or vapers.
Recommended minimum core items to assess e-cigarette use in national surveys
E-cigarette ever use: Ever use of e-cigarettes captures initiation or trial. While this construct is useful for quantifying the proportion of initiates in a population and constructing skip patterns within surveys, it is minimally useful in analyses, as most ever use is limited to one or two instances.3 15 Researchers should use caution when employing this item as a measure of exposure to e-cigarettes.30 Prior research has employed ever use as a measure of e-cigarette exposure among adult smokers;31–33 however, this weak measure yields uninterpretable estimates of the effect of e-cigarette use on smoking. It should be noted that ‘ever e-cigarette use’ (ie, trial) is different than ‘former daily use,’ which we present below.
We recommend assessing e-cigarette ever use with an item from the ITC Four Country Survey.34 This item should be asked of all survey respondents and allow a ‘Don’t know’ response. Additionally, e-cigarette ever use should be asked on its own rather than as part of a list of tobacco products, as the list approach is likely to underestimate use.3
Frequency of e-cigarette use: It is still not known what levels of e-cigarette use are relevant to behavioural and health outcomes. Frequency of e-cigarette use is commonly assessed by asking the participant about the number of days he or she has used an e-cigarette in the past 30 days. However, due to the transience of e-cigarette use in some populations (ie, smokers or young adults), we do not recommend this approach for estimating frequency of use for two reasons. First, this item encourages equating any level of use in the past 30 days with ‘current’ use, conflating recent initiates or experimental users who may be unlikely to progress to daily use with current, established e-cigarette users.10 12 15 35 Second, this item is most useful in combination with a subsequent item assessing the length of time the use pattern has endured, which increases the number of items in our core set of items.
Our proposed item, adapted from the ITC Four Country Survey,34 allows for flexibility in defining a meaningful level of e-cigarette use. In addition to surveillance surveys, this item is also appropriate to assess within-person changes in cohort studies and could be used alone when only one or two questions on e-cigarettes are possible due to space restrictions. Researchers should include parallel items assessing frequency of e-cigarette use and cigarette smoking so that co-use of these products (ie, ‘dual use’) can be compared.
Former daily use: Assessing patterns of former daily use is important for understanding the impact of e-cigarette use on uptake or reduction of smoking, as well as e-cigarette related health outcomes. PATH Waves 1–2 and the ITC Surveys ask whether respondents who do not currently use e-cigarettes have ever used e-cigarettes ‘fairly regularly.’ Rather than leaving the definition of ‘fairly regularly’ to respondents, we suggest asking about at least daily use over a month or more, which would indicate that the user had vaped for an extended period and may be relevant for behavioural outcomes.
Relative perceived harm: Common theories of health behaviour posit that harm perceptions influence tobacco use behaviour, with lower perceived harm encouraging higher levels of experimentation and current use. PATH, ITC, STS, GATS and the Truth Initiative Young Adult Cohort ask about absolute or relative perceived harm. We suggest assessing perceived harm relative to cigarettes (rather than absolute perceived harm) among all survey respondents due to their common use as smoking cessation or harm reduction tools.36 37 Using an item adapted from PATH and the ITC Four Country Survey, we suggest assessing relative perceived harm to understand how tobacco use prevention and health communication campaigns, as well as media coverage, affect the perceptions of non-, former and current tobacco users, and how these perceptions affect e-cigarette use.
Device type: E-cigarettes are a diverse product class and must not be treated as a single product. With the wide variation in design, content, function, nicotine delivery, price and availability of these products, different types of e-cigarette devices may have different behavioural and health effects. A growing body of work suggests that device characteristics such as nicotine content and type of battery are correlated with e-cigarette use behaviours and may affect smoking cessation.22 38–41 Given the diversity of the products, it is unsurprising that surveys vary widely in their approach to capturing device type. Some surveys (eg, PATH Wave 2) split questions about devices into two parts: a first question about the size/shape of the device, and a second question about whether the device is disposable, uses prefilled cartridges or is refillable with liquids. Some studies (eg, PATH Waves 1–2 and ITC) also use pictures of e-cigarette types.
The wording and response options for our suggested device type item are driven by battery size, which has been shown to affect nicotine delivery42 43 and smoking cessation.38 Our proposed response options identify four mutually exclusive types of devices (table 1). For analyses, these items may be collapsed into groups 1 and 2, likely to have less powerful batteries (often called ‘cigalikes’), and groups 3 and 4, likely to have more powerful batteries, often called ‘second generation’ and ‘mods'. Devices with larger batteries are normally refillable with e-liquid (eg, a liquid containing some mix of propylene glycol, glycerine, water, flavouring, impurities and often nicotine), which may be associated with a risk of unintentional poisoning and is a relevant data point in estimating population harms.44 45 While we have found this approach useful in understanding device characteristics in the UK and USA, items assessing device characteristics will need to be adapted according to availability of e-cigarettes in different jurisdictions and the evolution of the devices. The utility of this approach may diminish as devices with more powerful batteries and prefilled cartridges or sealed tanks are made available.
Nicotine content: Similar to understanding device characteristics, assessing e-cigarette nicotine content has no direct parallel cigarette survey item. Few of the first national and international surveys of e-cigarettes asked if the device contained nicotine. As countries banned nicotine-containing e-cigarettes, more surveys asked about whether devices used contained nicotine.12 46–49 These items often ask about nicotine concentration by per cent, milligrams per millilitre (mg/mL) or by an ordinal descriptive term (eg, ‘low,’ ‘medium’ and ‘high’). Each of these approaches, however, has serious drawbacks. Asking about nicotine concentration by per cent or mg/mL is difficult for inexperienced users, yielding a number of ‘don’t know’ responses (eg, 12% ‘don’t know’ in a recent Action on Smoking and Health Survey).50 51 Using terms that correspond to manufacturers’ descriptions (eg, ‘low’) is also problematic because these labels do not necessarily capture similar ranges of nicotine concentrations across brands or jurisdictions.
Our proposed nicotine content item requires some respondent knowledge and is similar to an item used in PATH Wave 2. Our item asks about ‘the vaping device you use most often’ because sophisticated users may use multiple nicotine concentrations, employing different strengths of nicotine in different situations or over time. It may be possible to collect more reliable information on e-liquid nicotine concentration among experienced users; however, the ultimate amount of nicotine delivered to the user depends on the device, the nicotine concentration and the user’s experience with e-cigarettes, among other variables.42 52–55 Assessing nicotine fluid concentration is of limited application until we have refined items to accurately assess device characteristics such as battery wattage and coil resistance. In jurisdictions where certain nicotine concentrations are banned, it may be useful to adapt our suggested item to assess use of the banned e-liquid nicotine concentrations.
Flavours: Most e-cigarettes, even those that taste like traditional cigarettes, are flavoured because their main constituents (eg, nicotine, propylene glycol) have little flavour. However, truly unflavoured e-liquids (eg, those that contain only propylene glycol, glycerin, water and nicotine) are also available. Most existing surveys of e-cigarette use ask about flavours, but their approach differs. PATH Waves 1–2 and ITC ask about flavours that are available in cigarettes (eg, traditional tobacco, menthol or mint), as well as several other flavour categories (eg, chocolate, fruit, clove or spice, alcoholic drink, dessert). While understanding the prevalence of different e-cigarette flavour preferences may shed light on the behavioural and public health impact of flavour use, this is a difficult behaviour to accurately assess. First, respondents may find describing their preferred flavour using a list of generic terms challenging if their preferred flavour fits into multiple categories. For example, is ‘piña colada’ an alcoholic drink or a fruit flavour? Second, like nicotine concentration, e-cigarette users may vape a variety of e-cigarette flavours. Respondents to the PATH Wave 2 and the 2016 ITC Survey were provided a list of individual flavours and asked to ‘select all that apply’ to describe their flavour use in the past 30 days. With this approach, however, it is unclear whether the respondents are describing one preferred e-liquid flavour or a range of preferred flavours.
Ultimately, the flavour response options should be dependent on the current situation in the target population’s jurisdiction and the purpose of the research. In the USA, for example, menthol cigarettes are legal and prevalent, but other flavoured cigarettes are banned. Thus, it makes sense to ask about menthol e-liquid use separately from other flavours. In different policy contexts, it may make sense to ask about other flavours. Our recommended item focuses on the most common flavour because some users may consume multiple flavours in a day or week. The proposed response options avoid the problem of multiple categorisation of a flavour and decrease response burden. While switching between flavours is an important construct that should be assessed in surveys with a large sample of daily vapers, this item is of limited use in a general population survey in settings where daily e-cigarette use is uncommon, which describes nearly all current settings.
Reasons for use: Given the opportunity, e-cigarette users will nominate multiple reasons for e-cigarette use.36 56 While allowing respondents to choose multiple reasons for use reflects complex motivations for the behaviour, it has limited utility for understanding the role of e-cigarette use and behavioural intention in e-cigarette and tobacco use behaviour. It could be argued that qualitative research may be more appropriate for in-depth explorations of reasons for use. However, where survey space allows, a single question on the main reason why e-cigarettes or vaping products are/were used may be relevant for policy and practice. If the survey mode allows, researchers may also consider asking respondents to rank their reasons for use, which would still allow for comparisons of top reasons across jurisdictions. If the purpose of a survey is to measure the effectiveness of e-cigarettes for smoking cessation, we recommend including ‘e-cigarette or vaping device’ in a list of questions that assess what approach, support or aids were used during a specific attempt to stop smoking (eg, in the last attempt).57
E-cigarette items of limited utility
In addition to our eight recommended items, we highlight three constructs which we believe are of limited utility in most jurisdictions: e-cigarette awareness, e-cigarette or e-liquid quantity of consumption, and e-cigarette or e-liquid brands.
E-cigarette awareness: Until recently, most national surveys asked about awareness of e-cigarettes. In 2014 and 2015, the US National Adult Tobacco Survey and STS in England did not assess awareness because previous surveys had shown e-cigarette awareness was near universal (93% in the UK as early as 201258 and 86.4% in the USA in 201359). We recommend dropping the awareness item in jurisdictions where awareness has reached saturation.
E-cigarette or e-liquid quantity of consumption: One complex issue in e-cigarette research is evaluating how much e-liquid users consume. Research suggests that frequency of e-cigarette use is relevant to smoking cessation effectiveness.22 23 60 Many surveys follow approaches similar to those assessing heaviness of smoking among cigarette smokers,61 asking about consumption of cartridges or disposable e-cigarettes per day or number of e-cigarette puffs per day. For users of refillable e-cigarettes, items ask how long it takes to use a specific amount of e-liquid (ie, 10 mL) or the size (in mL) of the last bottle purchased and how long it usually lasts. Interestingly, this is similar to methods that have been developed for assessing cannabis consumption.62 PATH, ITC and STS all ask questions about the daily quantity of e-cigarette use. PATH and ITC ask about daily consumption in product units, and how long one’s last purchase of liquid will last (ITC), while STS asks about number of times per day the e-cigarette is used. However, e-liquid bottles and e-cigarette refillable reservoirs are of varying sizes, so time to depletion is of limited utility without reliable information about the respondent’s device. Additionally, e-liquid consumption as a function of puffs per day will vary by the user’s puff topography and device settings. Similar to frequency of use, the field is in its infancy and we are only beginning to accurately measure and understand how heaviness of use/daily quantity may predict public health outcomes. Without item testing, we recommend including these items with caution and ask that researchers share their lessons learnt and publish formative work to advance the field.
E-cigarette and e-liquid brands: It is common practice in surveys of smoking behaviour to include questions about the respondent’s preferred brand; this practice has been applied to brand varieties of devices and e-liquid. Assessing e-cigarette device brands is challenging, as there are thousands of varieties, and it is unclear how brand loyal e-cigarette users are to device and e-liquid makers. Experienced e-cigarette users may have more than one device, or may combine components from multiple brands. Casual e-cigarette users may not know the brand of their device or e-liquid. Despite these challenges, brand is a worthwhile construct for understanding the effect of marketing on e-cigarette use behaviour.
Conclusion
Using the combined experience of an international group of researchers, we have proposed a minimum set of e-cigarette items to encourage consensus around items and allow for cross-jurisdictional comparisons and surveillance of e-cigarette use. These proposed items are meant to open a dialogue on meaningful items for national e-cigarette surveillance and should be updated as measurement of e-cigarette use behaviour evolves. We recognise that there is room for continued improvement of these items, and we welcome input from e-cigarette users and academic/public health colleagues. We also encourage discussion of how common definitions of e-cigarette use and consistency in reporting of results could advance the field. Additionally, this paper focuses on items for surveys and studies with adults only; future recommendations are needed for youth surveys, although some of the same items are applicable to youth. Standardised, reliable and valid surveillance items will speed knowledge synthesis both within and across countries, will place patterns and reasons for e-cigarette use in the context of the emerging complexity of poly-tobacco/nicotine product use, and will better inform policy making and regulation and the overall public health impact of e-cigarettes and related products.63
What this paper adds
Jurisdictions have taken different approaches to regulating e-cigarette devices and e-cigarette use. These different approaches present an opportunity to evaluate the effect of e-cigarette policies and regulation on e-cigarette and tobacco product use. However, for cross-jurisdictional comparisons to be useful, approaches to assessing e-cigarette use must be similar. The recommended set of eight e-cigarette measures for surveillance includes two core items to distinguish ever use from more frequent use and six items to assess former use, relative perceived harm, primary device type, primary flavour preference, nicotine content and primary reason for use.
Acknowledgments
The authors thank the Robert Wood Johnson Foundation for funding ‘Harvesting Global Learning on Alternative Nicotine Delivery Systems (ANDS) to Inform U.S. Policy Action, Policy Research, and Surveillance,’ during which the idea for this manuscript was formed. The authors also thank the authors of relevant surveys whose work has contributed to the survey items recommended in this manuscript.
References
Footnotes
Contributors All authors conceived the manuscript topic. JLP wrote the first draft of the paper and led the writing team. SCH, LSB, LB, AMG, ACV, AM, DBA and JEC contributed to manuscript writing. JEC and DBA were project leads on ’Harvesting Global Learning on Alternative Nicotine Delivery Systems (ANDS) to Inform U.S. Policy Action, Policy Research, and Surveillance'.
Funding LSB is funded by a Cancer Research UK (CRUK)/BUPA Foundation Cancer Prevention Fellowship (C52999/A19748). LSB, SCH and AM are members of the UK Centre for Tobacco and Alcohol Studies, a UK Clinical Research Collaboration Public Health Research: Centre of Excellence with funding from the Medical Research Council, British Heart Foundation, Cancer Research UK, Economic and Social Research Council and the National Institute for Health Research under the auspices of the UK Clinical Research Collaboration (MR/K023195/1). Effort on this manuscript for AMG, ACV and DBA was also supported by Truth Initiative. JLP’s time was supported by the National Institute On Drug Abuse of the National Institutes of Health under Award Number K01DA037950.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Correction notice This paper has been amended since it was published Online First. Owing to a scripting error, some of the publisher names in the references were replaced with ’BMJ Publishing Group'. This only affected the full text version, not the PDF. We have since corrected these errors and the correct publishers have been inserted into the references.