Examining the sources of evidence in e-cigarette policy recommendations: A citation network analysis of international public health recommendations

Abstract

Background Public health policies and recommendations aim to be informed by the best available evidence. Evidence underpinning e-cigarettes policy recommendations has been necessarily limited due to the novelty of the technology and the lack of long-term epidemiological studies and trials. Some public health bodies have actively encouraged e-cigarette use whilst others have raised concerns over introducing new health risks and renormalising tobacco smoking. Using citation network analysis we investigated the author conflicts of interest and study funding statements within sources of evidence used by public health bodies when making recommendations about e-cigarette policy. Methods We conducted citation network analysis of public health recommendation documents across four purposively selected diverse jurisdictions: WHO, UK, Australia, and USA. We extracted all citations from 15 public health recommendation documents, with more detailed data collected for influential citations (used in 3+ recommendation documents). We analysed the relationships between the sources of evidence used across jurisdictions using block modelling to determine if similar groups of documents were used across different jurisdictions. We assessed the frequency and nature of conflicts of interest. Results 1700 unique citations were included across the 15 public health recommendation documents, with zero to 923 citations per document (median = 63, IQR = 7.5–132). The evidence base underpinning public health recommendations did not systematically differ across jurisdictions. Of the 1700 citations included, the majority were journal articles (n = 1179). Across 1081 journal articles published between 1998–2018, 200 declared a conflict of interest, 288 contained no mention of conflicts of interest, and 593 declared none. Conflicts of interest were reported with tobacco (3%; n = 37 journal articles of 1081), e-cigarette (7%; n = 72), and pharmaceutical companies (12%; n = 127), with such conflicts present even in the most recent years. There were 53 influential citations, the most common study type was basic science research without human subjects (e.g. examination of aerosols and e-liquids) (n = 18) followed by systematic review (n = 10); with randomised control trial being least common (n = 4). Network analysis identified clusters of highly-cited articles with a higher prevalence of conflicts of interest. Conclusion Public health bodies across different jurisdictions drew upon similar sources of evidence, despite articulating different policy approaches to e-cigarettes. The evidence drawn upon, including the most influential evidence, contained substantial conflicts of interest (including relationships with e-cigarette and tobacco industries). Processes to explicitly manage conflicts of interest arising from the underlying evidence base may be required when developing public health recommendations.