A systematic review of the effects of e-cigarette use on lung function – npj Primary Care Respiratory Medicine

Our systematic review was conducted according to a protocol developed prior to the launch of the review, which was registered in the PROSPERO register of systematic reviews (CRD42021227121)^ten. This systematic review is reported according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.¹¹.

Search strategy and selection of studies

Using a search strategy (Supplementary Tables 1-5) developed by an experienced health sciences (GG) librarian, we systematically searched the EMBASE, MEDLINE and PsycINFO databases via Ovid, the Cochrane CENTRAL database and the Web of Science Core from 2004 (the year of the first e-cigarette patent) to July 12, 2021. We also conducted a gray literature search by consulting the websites of major government and public health organizations (the World Health Organization, Health Canada, United States Centers for Disease Control and Prevention, United States Food and Drug Administration, Canadian Center on Substance Use and Addiction, European Center for Prevention and Disease Control and the European Public Health Association). Additional articles were identified by manually searching the reference lists of included publications as well as SCOPUS and Google Scholar (first ten pages). Articles were included if they reported quantitative primary data on changes in lung function associated with vaping, defined as the use of any device that works by converting e-liquid into an aerosol using metal coils, in human participants of all ages. Cell and animal studies were excluded. Studies using non-combustion heat devices were also excluded, as they do not meet the above definition of vaping. Eligible studies included randomized controlled trials (RCTs), non-randomised intervention studies (NRSI), and cohort studies; cross-sectional studies and case reports were excluded. We included studies that used non-users of both vaping devices and conventional cigarettes as a comparison group and those that used a before-and-after design in which individuals acted as their own controls. Inclusion was not limited by language or country of publication. Abstracts and conference proceedings were included if enough data could be extracted from these publications.

Search results were downloaded from databases to reference management software (EndNote X9) or added manually (eg, for gray literature results). Duplicates were removed in EndNote and entries were uploaded to Covidence (Veritas Health Innovation, Melbourne, Australia), a systematic review software. Two reviewers (LH and KH) independently screened the titles and abstracts of all identified publications for eligibility. Citations deemed potentially eligible by either reviewer, based on the predefined review inclusion/exclusion criteria (Supplementary Table 6), were extracted for full-text screening and assessed for inclusion. Reasons for exclusion after full-text review were annotated in Covidence and any disagreements were resolved by consensus or by a third reviewer (AH-L.).

Data Extraction

Two independent reviewers (LH and KH) extracted methodological, demographic and outcome data from duplicate included studies; disagreements were detected in Covidence and were resolved by consensus or, if necessary, by a third-party reviewer (AH-L.). Data extracted included study characteristics (first author, journal, year of publication, year(s) of data collection, funding, data source, study design, recruitment strategy, length of follow-up, country origin, sample size); population characteristics (sex, gender, age, race, ethnic origin, socio-economic status, dose/frequency of electronic cigarette use, status of conventional cigarette smoker, consumption of smoked cannabis); and vaping behavior, including type of vaping device used (e.g., disposable e-cigarette versus a pod device such as JUUL), vaping products used (e.g., nicotine-only cartridges versus to THC cartridges exclusively versus dual use nicotine and THC products), and the source of the vaping product (informal information [i.e., friends, family members, or dealers] vs. commercial [i.e., vape shops, stores, dispensaries]).

Initially, the outcomes of primary interest extracted were respiratory signs and symptoms, as they are important to patients and are early signs of respiratory disease. Secondary outcomes included: lung function results; Computed tomography (CT) findings of emphysema, airway remodeling, and loss of small airways; breathing-related quality of life and exercise limitations; the incidence and/or prevalence of respiratory diseases as well as exacerbations of previous respiratory diseases; and the use of health care resources, including outpatient respiratory care, emergency department visits, and hospitalizations. Given the limited number of studies available and the heterogeneity of the data extracted from these studies, no meta-analysis was performed.

Risk of bias

Risk of bias in included publications was assessed independently by two reviewers (LH and KH), and discrepancies were resolved by consensus or, if necessary, by a third reviewer (AH-L.). The risk of bias of the included non-randomized studies (pre-post studies, NRSI with non-vaping reference group, cohort study) was assessed using the ROBINS-I tool (Risk of Bias in Non-randomized Studies —of Interventions).¹². The ROBINS-I tool assesses intervention-specific outcomes for a study across seven domains that assess risk of bias before the intervention, during the intervention, and after the intervention. For each outcome of interest extracted from an included study, the risk of bias in each domain was flagged as ‘low’, ‘moderate’, ‘serious’ or ‘critical’. Included RCTs were assessed using the Cochrane Collaboration’s risk of bias assessment tool (ROB V1)¹³. Similar to ROBINS-I, this tool assesses risk of bias through the assessment of five domains; for each outcome of interest extracted from an included study, the risk of bias for each domain was reported as ‘low risk of bias’, ‘high risk of bias’ or ‘unclear risk of bias’. All eligible publications were included in the qualitative synthesis, regardless of their assessed risk of bias.

Summary of reports

Further information on the research design can be found in the summary of nature research reports linked to this article.