Article type
Year
Abstract
Introduction: Randomised controlled trials are not always feasible when evaluating interventions within the scope of Cochrane Effective Practice and Organization of Care Review Group (EPOC). We therefore include interrupted time series (ITS) and controlled before and after (CBA) study designs in our reviews. While the Cochrane optimal search strategy for RCTs has been developed to identify randomised and clinical controlled trials in Medline its ability to identify ITS and CBA studies is unknown.
Objective: To determine the sensitivity and precision of the existing Cochrane Medline optimal strategy for identifying ITS and CBA designs and to develop the strategy to improve the sensitivity.
Methods: A gold standard set of studies, within the scope of EPOC in terms of type of intervention and study design was identified by handsearching Medical Care (1969-95), BMJ (1992-4) and full text searching from the Ovid Biomedical Core Collection of all original and miscellaneous articles from Annals of Internal Medicine, BMJ, JAMA and Lancet (1995-6). The Cochrane Medline strategy was run on this set of journals for the specified years to determine the sensitivity and precision in identifying our gold standard. An examination of frequency of occurrence of MeSH terms and textwords in Medline records for studies which were not identified was undertaken and the most frequently occurring identified. Their sensitivity and precision will be calculated against the gold standard and then used to develop the existing Cochrane strategy.
Results: Preliminary analysis on the hand searched journals only suggest that the Cochrane Medline optimal strategy identifies 82.9% of our studies including 93.2% of RCT or CCT studies, 70.2% of CBA and 55.5% of ITS studies with a precision of 7.0%. Inclusion of additional search terms increased the sensitivity to 96.4%, identifying 100% RCT or CCT, 83.3% CBA and 93% ITS studies, while precision dropped to 6.0%.
Discussion: Preliminary results suggest that the Cochrane Medline optimal strategy identifies a substantial proportion of ITS and CBA studies but that the strategy can be adapted to increase sensitivity with a small decrease in precision.
Objective: To determine the sensitivity and precision of the existing Cochrane Medline optimal strategy for identifying ITS and CBA designs and to develop the strategy to improve the sensitivity.
Methods: A gold standard set of studies, within the scope of EPOC in terms of type of intervention and study design was identified by handsearching Medical Care (1969-95), BMJ (1992-4) and full text searching from the Ovid Biomedical Core Collection of all original and miscellaneous articles from Annals of Internal Medicine, BMJ, JAMA and Lancet (1995-6). The Cochrane Medline strategy was run on this set of journals for the specified years to determine the sensitivity and precision in identifying our gold standard. An examination of frequency of occurrence of MeSH terms and textwords in Medline records for studies which were not identified was undertaken and the most frequently occurring identified. Their sensitivity and precision will be calculated against the gold standard and then used to develop the existing Cochrane strategy.
Results: Preliminary analysis on the hand searched journals only suggest that the Cochrane Medline optimal strategy identifies 82.9% of our studies including 93.2% of RCT or CCT studies, 70.2% of CBA and 55.5% of ITS studies with a precision of 7.0%. Inclusion of additional search terms increased the sensitivity to 96.4%, identifying 100% RCT or CCT, 83.3% CBA and 93% ITS studies, while precision dropped to 6.0%.
Discussion: Preliminary results suggest that the Cochrane Medline optimal strategy identifies a substantial proportion of ITS and CBA studies but that the strategy can be adapted to increase sensitivity with a small decrease in precision.