Article Text

Download PDFPDF

Heterogeneity in the measurement and reporting of outcomes in studies of electronic cigarette use in adolescents: a systematic analysis of observational studies
  1. Carlos Echevarria1,
  2. Ian P Sinha2
  1. 1Respiratory Department, North Tyneside General Hospital, North Shields, UK
  2. 2Respiratory Department, Alder Hey Children's Hospital, Liverpool, UK
  1. Correspondence to Dr Carlos Echevarria, Respiratory Department, North Tyneside General Hospital, North Shields NE29 8NH, UK; CarlosEchevarria{at}doctors.org.uk

Abstract

Objective To examine consistency between cross-sectional studies of conventional and electronic cigarette use among adolescents in terms of the measurement, analysis and reporting of parameters.

Design A systematic analysis of cross-sectional studies of conventional and electronic cigarette use in adolescents, to identify measured and reported parameters.

Data sources Studies examining use of electronic and conventional cigarette use in adolescents were identified by searching the SCOPUS database in August 2015.

Study selection The selection criteria for studies were: cross-sectional studies, in English, on e-cigarette use in adolescents. Two reviewers independently selected relevant studies from the search. 60 abstracts were identified, from which 31 papers were eligible for review (23 unique studies).

Data extraction Measured and reported parameters were identified and tabulated. These included the prevalence of cigarette and/ or electronic cigarette use, and the definitions of terms. Data were extracted independently by two reviewers.

Data synthesis With regards basic parameters of ‘ever’ or ‘current’ use of electronic or conventional cigarettes, there were 31 unique measured parameters across 23 studies. Of 16/23 studies in which authors collected information on dual current use, prevalence was reported in 11/16, with six different definitions of ‘dual use’.

Conclusions There are substantial differences in measurement and reporting of parameters across observational studies of electronic and conventional cigarette use in adolescents. These studies are at risk of reporting bias, and results are difficult to interpret. A core outcome set that should be measured and reported in all observational studies is required, using structured consensus techniques.

  • Addiction
  • Electronic nicotine delivery devices
  • Nicotine
View Full Text

Statistics from Altmetric.com

Background

There is debate, worldwide, around the benefits and harms of electronic cigarettes (e-cigarettes).1–5 One uncertainty is whether e-cigarette use among adolescents results in conventional cigarette use.6

Observational studies may be the most appropriate study design to explore this issue, as randomised controlled trials face ethical and legal obstacles. Three longitudinal cohort studies and numerous cross-sectional studies, from various countries, have examined the use of e-cigarettes and conventional cigarettes in adolescents. One longitudinal cohort study in 694 people aged 16–24 years reported that in smoking-naïve individuals, those that used e-cigarettes were more likely to progress to cigarette smoking, although absolute numbers were small (11/16, 68.8% vs 128/678, 18.9%).7

A second longitudinal study of 2530 tobacco-naïve 14-year-olds showed similar findings: at 6 months the proportion of those who had tried combustible tobacco was 182/ 2247 (8.1%) in never e-cigarettes users compared to 67/218 (30.7%) in ever e-cigarette users. Combustible tobacco was defined as the ever use of conventional cigarettes, cigars, and/ or hookah at any time. Numbers were smaller for conventional cigarette use: 68/2267 (3.0%) vs 21/217 (9.7%).8 This would be consistent with a ‘gateway effect’ towards conventional cigarette use, but the authors correctly state that this does not show causation. For example, ‘ever use’ may be a surrogate measure of experimentation which could account for the risk of future cigarette ever use. A third longitudinal survey of 2338 adolescents from Hawaii showed that transition from never-use to smoking was associated with e-cigarette use, even when other covariates such as age, ethnicity and rebelliousness were included in the multivariate model.9

There are many more cross-sectional studies than longitudinal studies, and it seems likely this trend will continue. Although they also cannot prove causation, multiple cross-sectional studies may provide useful information on associations and trends in e-cigarette and conventional cigarette use. For example, an increase in the number of e-cigarette users and conventional cigarette smokers with time would be consistent with the gateway effect; an increase in e-cigarette use with a reduction in conventional cigarette smoking would be inconsistent with this phenomenon. Currently, the interpretation of these studies appears to vary. Some feel that the results reassure against a gateway effect,10 but others urge caution in light of the results.2 ,8 Therefore, these observational studies require transparent and consistent reporting to aid their interpretation.

However, within these studies, researchers can measure various parameters, and analyse and present them in different ways (box 1). There is currently no guidance around these methodological decisions, and no core set of results that should be reported in all studies. This may lead to reporting bias and non-uniformity between studies, which makes them more difficult to appraise, interpret and synthesise. We aimed to examine consistency between studies in terms of the measurement, analysis and reporting of parameters.

Box 1

Potential parameters that could be reported in studies of the use of electronic cigarettes and conventional cigarettes

‘Level 1’ results—proportion of adolescent, across the whole sample, who:

  •  Have never used e-cigarettes

  •  Have ever used e-cigarettes

  •  Currently use e-cigarettes

  •  Have never used conventional cigarettes

  •  Have ever used conventional cigarettes

  •  Currently use conventional cigarettes

‘Level 2’ results—proportion of adolescents, across the whole sample, who use electronic and conventional cigarettes in the following combinations:

  •  Use both products currently

  •  Have used both products ever

  •  Have never used either product

  •  Never conventional cigarettes, ever e-cigarettes

  •  Never conventional cigarettes, current e-cigarettes

  •  Ever conventional cigarettes, never e-cigarettes

  •  Ever conventional cigarettes, current e-cigarettes

  •  Current conventional cigarettes, never e-cigarettes

  •  Current conventional cigarettes, ever e-cigarettes

‘Level 3’ results—use of one product stratified by use of the other

  •  Proportion of adolescents who never smoked conventional cigarettes who have never used e-cigarettes

  •  Proportion of adolescents who never smoked conventional cigarettes who have ever used e-cigarettes

  •  Proportion of adolescents who never smoked conventional cigarettes who currently use e-cigarettes

  •  Proportion of adolescents who have ever smoked conventional cigarettes who have never used e-cigarettes

  •  Proportion of adolescents who have ever smoked conventional cigarettes who have ever used e-cigarettes

  •  Proportion of adolescents who have ever smoked conventional cigarettes who currently use e-cigarettes

  •  Proportion of adolescents who currently smoke conventional cigarettes who have never used e-cigarettes

  •  Proportion of adolescents who currently smoke conventional cigarettes who have ever used e-cigarettes

  •  Proportion of adolescents who currently smoke conventional cigarettes who currently use e-cigarettes

  •  Proportion of adolescents who have never used e-cigarettes who have never smoked conventional cigarettes

  •  Proportion of adolescents who have never used e-cigarettes who have ever smoked conventional cigarettes

  •  Proportion of adolescents who have never used e-cigarettes who currently smoke conventional cigarettes

  •  Proportion of adolescents who have ever used e-cigarettes who have never smoked conventional cigarettes

  •  Proportion of adolescents who have ever used e-cigarettes who have ever smoked conventional cigarettes

  •  Proportion of adolescents who have ever used e-cigarettes who currently smoke conventional cigarettes

  •  Proportion of adolescents who currently use e-cigarettes who have never smoked conventional cigarettes

  •  Proportion of adolescents who currently use e-cigarettes who have ever smoked conventional cigarettes

  •  Proportion of adolescents who currently use e-cigarettes who currently smoke conventional cigarettes

Methods

We conducted a systematic analysis of cross sectional studies in which children and adolescents (younger than 18 years) were asked about e-cigarette use. We included studies published, in full, in English. We excluded studies in adolescents and adults, unless data for each age group were reported separately. Studies were identified by searching SCOPUS (a platform for searching Medline and other databases), in August 2015, using the search strategy (TITLE-ABS-KEY (electronic cigarette OR e cigarette OR e-cigarette) AND (adolescen* OR teenage* OR child*)). Both authors (CE and IPS) screened the titles and abstracts for eligibility. All potential articles were reviewed and summarised in separate documents by each authors, and were compared. Disagreements were re-examined, and resolved by discussion.

Two reviewers (CE and IPS) independently extracted the following data from each study:

  • Geographic setting, number of participants, and source of funding

  • Whether the following parameters were measured:

    • ever (lifetime) use of e-cigarettes

    • ever (lifetime) use of conventional cigarettes

    • current use of e-cigarettes

    • current use of conventional cigarettes

  • Whether the following results were reported:

    • ‘Level 1’ results in box 1, as a proportion of the whole study population:

      • Prevalence of ever (or never) use of e-cigarette

      • Prevalence of ever (or never) use of conventional cigarette

      • Prevalence of current use of e-cigarette

      • Prevalence of current use of conventional cigarette

    • ‘Level 2’ results in box 1, as a proportion of the whole study population:

      • Prevalence of dual ever use of e-cigarette and conventional cigarette

      • Prevalence of dual current use of e-cigarette and conventional cigarette

      • The proportion of adolescents who had never used either e-cigarette or conventional cigarettes

      • Prevalence of the remaining ‘Level 2’ (box 1) combinations of never, ever, and current use of e-cigarette and conventional cigarettes: ever e-cigarette, never conventional cigarette; current e-cigarette, never conventional cigarettes; never e-cigarette, ever conventional cigarette; current e-cigarette, ever conventional cigarette; never e-cigarette, current conventional cigarette; ever e-cigarette, current conventional cigarette

    • ‘Level 3’ results in box 1 (The frequency with which studies reported the proportion of adolescents using conventional cigarettes stratified by e-cigarette status, and vice versa)

  • Definitions used for current and ever use of e-cigarettes and conventional cigarettes and dual use.

Results

Of 60 abstracts screened, 23 individual studies (31 papers) were deemed eligible for the review.11–41 Eleven were conducted in North America, eight in Europe, three in Asia and one in Australia. The median number of participants was 3127 (IQR=1320 to 16 193). Participants’ ages were reported as: (A) middle and high school age,14 ,18 ,19 high school age only20; (B) aged 9–18+,17 10–16,33 11–18,15 12–18,29 13–18,32 ‘predominantly 14 to 15’,39 14–17,28 14–18,24 15,25 15–19,26 16–17,12 and 17–25 years37 ;and (C) a mean age (SD) of 12.2 (0.9) and 15.6 (1.2) for middle and high school students,31 14.0 (0.87) and 16.5 (0.77) for middle and high school students,22 12.4 (0.5),34 12.5 (0.6),27 14.5 (2.0),23 14.6 (0.7),40 14.6 (1.9),38 ∼14.7 (0.1),16 15.1 (2.1),41 16.2,36 and 16.4 (1.4).11 All studies were funded by non-industry sources.

Taking into account possible ever or current use of e-cigarettes or conventional cigarettes there were 31 unique measured parameters across the 23 studies. The parameters reported in each study are shown in table 1.

Table 1

Reported parameters in all studies

The prevalence of e-cigarette and conventional cigarette use (regardless of use of the other product) was reported in most of the studies in which it was measured. Ever use of e-cigarettes was measured in 19/23 studies and reported in 18 of these, current use of e-cigarettes was measured in 16/23 studies and reported in all 16 of these, ever use of conventional cigarettes was measured in 21/23 studies and reported in all 21 of these, and current use of conventional cigarettes was measured in 20/23 studies and reported in 18 of these.

In 17/23 studies, the authors collected sufficient information to calculate the prevalence of dual ever use, which was reported or easily calculated in 7/17, or dual never use, which was reported or easily calculated in 5/17. In 16/23 studies, there would have been sufficient information collected for the authors to calculate the prevalence of dual current use, and it was reported (or could be easily derived from other results) in 11/16.

Of the 23 included studies, 20 reported some results relating to use of one product when stratified by the other (‘level 3’ in table 1). Across these 20 studies, there were 18 unique combinations of results that were reported. Of the 18 possible results that could have been reported, each was reported in at least one study. The most commonly reported result was the proportion of e-cigarette users.

Definitions

The greatest variation in reported values was for ‘dual use’. While this term implies concurrent use of e-cigarettes or conventional cigarettes, it was commonly defined as the use of both, at least once, in the last month. Reported figures varied from 1% to 50% depending on definition and analysis. ‘Dual use’ was calculated and reported in six different ways, as a percentage or proportion of: the whole sample who used both conventional cigarettes and e-cigarettes currently (11/23) or ever (1/23); the whole sample who were ever used e-cigarettes and daily conventional cigarette smokers (1/23); current e-cigarette users who were current conventional cigarette smokers (2/23); ever e-cigarette users who had ever tried conventional cigarettes (1/23); and current smokers who are current e-cigarette users (1/23).

In some studies ‘ever’ use of a product implied that it had ever been utilised, whereas in others it was considered mutually exclusive from ‘current use’ (eg, an ‘ever’ user had used the product in the past, but not in the last month). The term ‘ever smoker’ usually described adolescents who had had even one puff of a cigarette,13–15 ,17–23 ,26 ,27 ,30–40 but was also used to describe adolescents who had smoked at least 100 cigarettes. The terminology between studies varied around the same definition. In one study, adolescents who had smoked more than 100 cigarettes but did not currently smoke were called ‘ever smokers’,24 but in other studies they were called ‘former smokers’.16 ,37 ,41 ‘Former smoker’ was also used to describe someone who used to smoke but has quit,19 and someone who has smoked at least one puff, but not in the past 30 days.32 ‘Current use’ was most commonly defined as someone who used a product in the last month, regardless of frequency.11–21 ,23 ,24 ,26 ,30–32 Other terms we identified included ‘puffers’,24 ‘experimental users’,24 ,29 ,33 ,36 ,38 ‘infrequent users’,39 ‘regular users’, ‘occasional users’,21 and ‘heavy users’ or ‘daily users’.24 ,25 ,29 ,38

Many studies assumed that ‘e-cigarette’ use is synonymous with nicotine use, although not all e-cigarettes contain nicotine. Only three studies explicitly asked users if they thought that their e-cigarette contained nicotine, which is of importance when considering the gateway effect.29 ,30 ,36

Discussion

In this systematic analysis of cross sectional studies of e-cigarette and conventional cigarette use in adolescents, we identified variation in measurement, definition, and reporting of parameters around the use of these products. In only one study were all results reported, appearing in a simple 3 by 3 table with absolute values so all percentages could be easily calculated.12 We have adapted this table and included it as part of our reporting recommendations (table 2).

Table 2

Hypothetical example to show data reporting, with the minimum required information in bold and large font

This non-uniformity has implications for people and agencies trying to establish whether the use of e-cigarettes is changing, and whether e-cigarettes may entice adolescents into nicotine addiction and cigarette smoking. Many parameters can be measured, each of which has some relevance, but it is not clear which is most important. For example, it is not immediately obvious which is the best way to describe dual use of e-cigarettes and conventional cigarettes. Two studies12 ,24 presented three possible combinations of results that reflect dual use: the prevalence of this behaviour in the whole sample (2% in one study, 4% in the other), the proportion of current conventional cigarette smokers who have used e-cigarettes in the last month (11% in one study, 34.8% in the other), or the proportion of current e-cigarette users who concurrently use tobacco (62% in one study, 50% in the other). Meaningful interpretation or comparison between studies around ‘dual current use’ is not possible unless all studies report results in a uniform manner.

‘Dual current use’ suggests a person is smoking conventional cigarettes and using e-cigarettes on a daily basis, while in most studies of adolescents this can refer to someone who has had one cigarette and one puff of an e-cigarette within the last month. It is preferable to use descriptive terms that are unambiguous, such as ‘daily user’ or ‘use within the last month’.

The definitions of ‘ever use’ and ‘current use’ not only vary between studies of adolescents, but have different meanings to adult patients. For example, adult smokers who had one cigarette in the last month may not consider themselves a current user; current use would imply daily smoking.

Some authors may have analysed more results than they reported, and could have chosen which results to report after reviewing the results. Selective reporting of outcomes in this way is commonplace in clinical trials, and is likely to be of importance in studies employing other research methods. A Cochrane review has highlighted that outcome reporting bias is prevalent in studies of interventions for preventing smoking in young people.42

We focused on e-cigarettes and conventional cigarettes, but have not examined reporting of other alternative tobacco or nicotine products. If a systematic review that aims to collate study results were planned, we would recommend that study authors be contacted for non-published results given the likelihood of selective reporting of outcomes.

We recommend the development of a core set of variables by an international collaboration of researchers, stakeholders and policymakers, using structured consensus techniques. These variables should be measured and reported in all observational studies examining the use of conventional cigarettes and e-cigarettes in adolescents. Core outcome sets such as this have been developed for clinical trials in the paediatric field.43 The potential advantages are that they may increase the likelihood that important outcomes are measured, promote uniformity across studies and reduce selective outcome reporting.44 This should also include agreed definitions around never, ever, current and dual use of these products. Given the number of potential combinations of parameters, we recommend that this core outcome set should include a summary of why these results are important, and what they may indicate, to enable those reading observational studies in the future to better contextualise the results. In the meantime, given the marked heterogeneity in reporting that is already present, we make the following brief recommendations for future publications:

  • Terms should be descriptive and unambiguous; for example, ‘daily use’ or ‘use in the last month’. If terms such as ‘ever use’ are used they should be clearly defined and easy to locate within the paper. For ease of data interpretation, categories such as ‘current use’ and ‘ever use’ should be mutually exclusive.

  • Absolute values of collected data should be clearly reported in all studies as shown in table 2 (the bold and large font items represent the minimum required information). The total, row, and/ or column percentages (non-bold) can be included. For studies that are presenting and comparing data on multiple nicotine products, figures/ tables may differ from the example. However if data on conventional cigarettes and e-cigarette data is available, it should appear in the article or supplemental data as shown in table 2.

  • If quoting ‘dual use’ percentages within the written text, absolute data should be provided as a proportion of the entire population prior to any reported data on subpopulations. When reporting information on subpopulations it should be clear if these data were selected a priori.

Conclusions

There is marked variation across observational studies that examine the use of e-cigarettes and conventional cigarettes in adolescents. This makes it difficult to synthesise, appraise, and interpret studies, and renders them prone to selective reporting bias. A core outcome set that should be measured, and reported, in all observational studies addressing this topic is warranted.

What this paper adds

  • Cross-sectional studies of e-cigarette and cigarette use vary considerably, in the parameters reported and the definitions of terms, which makes it difficult to interpret what the results mean.

  • There is a need for a core outcome set that should be measured and reported in observational studies that examine the use of e-cigarette and conventional cigarettes in adolescents, with clear and consistent definitions of terms.

References

View Abstract

Footnotes

  • Contributors For the submitted study, CE and IPS were responsible for the conception, design, analysis and interpretation of data. IPS wrote the original draft, and both contributed to re-drafts and approved of the final version.

  • Competing interests None declared.

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

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.