Article type
Year
Abstract
Background: Within Cochrane Occupational Health Field, a possible (29.3.2004) new Cochrane entity, we have started our own specialiced trials register. In order to be able to gather all relevant occupational health (OH) related intervention studies (including interrupted time series and befor-after studies) from Medline we have started to explore the best predicting search terms and their combinations to be included in a search string.
Objectives: To create an optimal search strategy for Medline for OH intervention studies.
Methods: First we created a classification system for OH intervention studies. This classification serves both as the inclusion criteria for this study and enables us later to tag the original articles in the speacialised trials register by their target-of-intervention (reducing exposure, affecting worker behaviour, reducing work related disease and disability), and the type of their study method (controlled trial, interrupted time series, before-after trial). To test the diagnostic properties of the electronic search strategy we will construct a gold standard of core articles by hand searching several categories of medical journals published in 2001 and 2002 for articles that fulfil our predefined inclusion criteria. All original articles published in 2000 and 2001 in these journals will form the total population. We expect that this search will produce about 100 articles to be marked as the Gold Standard. We will transpose the search word findings from the searches to SPSS. We will use logistic regression analysis to find out which search words will predict the Gold Standard best independently (best ORs). With further modelling with the best predictive terms we aim at best possible set of search terms to be included in a search string. The new search strategy for Medline will be tested in other sets of journals.
Results: At present we are hand searching, two searches independently, the determined journals for gold standard. The results of the regression analysis and the proposed search string will be presented at the Ottawa Colloquium.
Objectives: To create an optimal search strategy for Medline for OH intervention studies.
Methods: First we created a classification system for OH intervention studies. This classification serves both as the inclusion criteria for this study and enables us later to tag the original articles in the speacialised trials register by their target-of-intervention (reducing exposure, affecting worker behaviour, reducing work related disease and disability), and the type of their study method (controlled trial, interrupted time series, before-after trial). To test the diagnostic properties of the electronic search strategy we will construct a gold standard of core articles by hand searching several categories of medical journals published in 2001 and 2002 for articles that fulfil our predefined inclusion criteria. All original articles published in 2000 and 2001 in these journals will form the total population. We expect that this search will produce about 100 articles to be marked as the Gold Standard. We will transpose the search word findings from the searches to SPSS. We will use logistic regression analysis to find out which search words will predict the Gold Standard best independently (best ORs). With further modelling with the best predictive terms we aim at best possible set of search terms to be included in a search string. The new search strategy for Medline will be tested in other sets of journals.
Results: At present we are hand searching, two searches independently, the determined journals for gold standard. The results of the regression analysis and the proposed search string will be presented at the Ottawa Colloquium.