This section describes the methodological steps used in the review: 1) the search of the literature; 2) the selection of relevant studies through application of inclusion and exclusion criteria; 3) quality appraisal of relevant studies; 4) data extraction from higher quality studies; and 5) synthesis of the evidence.
3.1 Literature search Seven electronic databases, abstracting primarily peer-reviewed research journal articles, were searched from their inception until July 2004: MEDLINE (from 1966), EMBASE (from 1980), PsycInfo (from 1887), Sociological Abstracts (from 1963), Safety Science and Risk Abstracts (SSRA, from 1981), EconLit (from 1969) and American Business Inform (ABI, from 1918).5 See Appendix A for details on the range of topics covered by these bibliographic databases. Since the search terms and language of the databases were found to differ significantly, the terms used in the search were customized for each database. A list of the broad terms used in the search can be found in Appendix B, Table B.1.
The search strategy combined two sets of keywords using an "AND" strategy (Appendix B, Figure B.1). The first set of keywords focused on OHSMS terms, while the keywords in the second set included evaluation and OHS effectiveness terms. The terms within each group were combined using an “OR” strategy. The titles, abstracts, case registry, and MeSH subject headings were searched for keywords. The search strategy was not limited in terms of language. In addition, the reference lists of all papers meeting the eligibility criteria and review papers on this topic were manually checked for relevant studies.
3.2 Selection of studies Titles and abstracts of each article were screened by at least two reviewers. Full text articles were retrieved for those studies that appeared to meet the inclusion criteria (Appendix I), and for those in which the information presented in the title, abstract, and key words was insufficient to exclude them. A consensus method was used to resolve any disagreements between the two reviewers on study inclusion. A third reviewer was consulted for those studies in which agreement could not be reached.
3.2.1 Inclusion and exclusion criteria The review team needed some way to recognize an OHSMS intervention in the literature. This was a challenge at the outset of the study in the absence of a definition of an OHSMS.
5 Searches of CCInfoWeb (Canadian Centre for Occupational Health and Safety (CCOHS), from 1900) and Dissertation Abstracts International (DAI, from 1861) were also conducted, but disbanded after the title and abstract screening step because of the large yield found in the seven databases.
Effectiveness of Occupational Health & Safety Management Systems: A Systematic Review