Meta-analyses in environmental and occupational health. Issue 6 (24th March 2018)

• Record Type: Journal Article
• Authors: Ioannidis, John P A
• Is Part Of: Occupational and environmental medicine. Volume 75:Issue 6(2018)
• Page Start: 443
• Page End: 445
• Publication Date: 2018-03-24
• Subjects: meta-analysis; Medicine, Industrial -- Periodicals; Environmental health -- Periodicals; 616.980305
• DOI: 10.1136/oemed-2016-104128
• Languages: English
• ISSNs: 1351-0711

Abstract

Abstract : Objectives: Meta-analyses are considered generally as the highest level of evidence, but concerns have been voiced about their massive, low-quality production. This paper aimed to evaluate the landscape of meta-analyses in the field of occupational and environmental health and medicine. Methods: Using relevant search terms, all meta-analyses were searched for, but those published in 2015 were assessed for their origin, whether they included randomised trials and individual-level data and whether they had authors from the industry or consultancy firms. Results: PubMed searches (last update February 2017) identified 1251 eligible meta-analyses in this field. There was a rapid increase over time (n=16 published in 1995 vs n=163 published in 2015). Of the 163 eligible meta-analyses published in 2015, 49 were from China, followed at a distance by the USA (n=19). Only 16 considered randomised (intervention) trials and 13 included individual-level data. Only 1 of the 150 meta-analyses had industry authors and none had consultancy firm authors. As an example of conflicting findings, 12 overlapping meta-analyses addressed mobile phones and brain cancer risk and they differed substantially in number of studies included, eligibility criteria and conclusions. Conclusions: There has been a major increase in the publication of meta-analyses in occupational and environmental health over time, with the majority of these studies focusing on observational data, while a commendable fraction used individual-level data. Authorship is still limited largely to academic and non-profit authors. With massive production of meta-analyses, redundancy needs to be anticipated and efforts should be made to safeguard quality and protect from bias.