Article type
Year
Abstract
Background: Up to now no published methodological filter designed for the retrieval of diagnostic test accuracy (DTA) studies has proved sensitive enough for use within a Cochrane DTA systematic review. A recent study tested the hypothesis that normal cross-sectional studies should be treated differently from delayed cross-sectional studies (longitudinal analyses).
Objectives: i) further test the hypothesis that normal cross-sectional studies and delayed cross-sectional studies are generally described differently in the published literature, ii) refine an unpublished methodological filter designed for identification of delayed cross-sectional studies (this will be done in MEDLINE (Ovid)), and iii) develop an equivalent filter for use in PubMed.
Methods: An unpublished filter will be further validated by expansion of the data set. Key elements of reports of potential studies for inclusion within Cochrane DTA reviews that focus on longitudinal diagnosis and prediction will be entered into textual analysis software so as to refine the existing unpublished filter. The filter will then be further tested by a large dataset of reports of potentially relevant studies from existing literature on methodological filters. Finally, a PubMed equivalent filter will be developed.
Results: The results will show the new filterás sensitivity, specificity, precision and accuracy and will be presented at the Cochrane Colloquium in Madrid, October 2011.
Conclusions: DTA studies generally fall into two camps - normal cross-sectional studies (e.g. looking at the accuracy of a new test to see if someone is pregnant or not) and delayed cross-sectional studies (e.g. a new test to see whether someone will develop symptomatic dementia from a 'pre-dementiaá state). The filters published so far treat these studies as one. The literature therefore condemns these filters based on potentially invalid criteria.
Objectives: i) further test the hypothesis that normal cross-sectional studies and delayed cross-sectional studies are generally described differently in the published literature, ii) refine an unpublished methodological filter designed for identification of delayed cross-sectional studies (this will be done in MEDLINE (Ovid)), and iii) develop an equivalent filter for use in PubMed.
Methods: An unpublished filter will be further validated by expansion of the data set. Key elements of reports of potential studies for inclusion within Cochrane DTA reviews that focus on longitudinal diagnosis and prediction will be entered into textual analysis software so as to refine the existing unpublished filter. The filter will then be further tested by a large dataset of reports of potentially relevant studies from existing literature on methodological filters. Finally, a PubMed equivalent filter will be developed.
Results: The results will show the new filterás sensitivity, specificity, precision and accuracy and will be presented at the Cochrane Colloquium in Madrid, October 2011.
Conclusions: DTA studies generally fall into two camps - normal cross-sectional studies (e.g. looking at the accuracy of a new test to see if someone is pregnant or not) and delayed cross-sectional studies (e.g. a new test to see whether someone will develop symptomatic dementia from a 'pre-dementiaá state). The filters published so far treat these studies as one. The literature therefore condemns these filters based on potentially invalid criteria.