Article type
Year
Abstract
Objective: To determine the appropriateness of rejecting MEDLINE records with a publication type of "randomized controlled trial" when the title and abstract seemingly do not describe a trial.
Introduction: When performing a literature search for a systematic review, authors often divide the search process into two phases: 1) a detailed search strategy of electronic databases using index and free text terms and 2) a manual screening of the titles and abstracts of the electronic search results.
Methods: As part of two systematic reviews of hypertension therapy, 2205 records were retrieved from the MEDLINE database using the Cochrane Optimal Search Strategy for RCTs combined with detailed topic searches. Titles and abstracts of the retrieved records were independently evaluated by two reviewers who were blinded to the publication types assigned by MEDLINE. Of the 823 records assigned the MEDLINE publication type "randomized controlled trial," the reviewers rejected 44 on the grounds that they did not describe RCTs. Full-text copies of 31 of the rejected articles could be obtained for further evaluation.
Results: Twelve of the 31 articles were reports of RCTs. In seven cases neither the title or abstract mentioned a control group, but the methods section in the full text described either a double-blind trial without specific mention of randomization (2) or explicitly described an RCT (5). In two cases a letter to the editor, with a title that seemed to express an opinion, reported the results of an RCT that was otherwise unpublished. Two articles re-analyzed data or corrected erroneous data reported in previously published RCTs. One review article included primary data from an otherwise unpublished RCT.
Discussion: When conducting a literature search as part of a systematic review, all MEDLINE records indexed with the publication type "randomized controlled trial" should be considered RCTs unless an evaluation of the full-text article proves otherwise.
Introduction: When performing a literature search for a systematic review, authors often divide the search process into two phases: 1) a detailed search strategy of electronic databases using index and free text terms and 2) a manual screening of the titles and abstracts of the electronic search results.
Methods: As part of two systematic reviews of hypertension therapy, 2205 records were retrieved from the MEDLINE database using the Cochrane Optimal Search Strategy for RCTs combined with detailed topic searches. Titles and abstracts of the retrieved records were independently evaluated by two reviewers who were blinded to the publication types assigned by MEDLINE. Of the 823 records assigned the MEDLINE publication type "randomized controlled trial," the reviewers rejected 44 on the grounds that they did not describe RCTs. Full-text copies of 31 of the rejected articles could be obtained for further evaluation.
Results: Twelve of the 31 articles were reports of RCTs. In seven cases neither the title or abstract mentioned a control group, but the methods section in the full text described either a double-blind trial without specific mention of randomization (2) or explicitly described an RCT (5). In two cases a letter to the editor, with a title that seemed to express an opinion, reported the results of an RCT that was otherwise unpublished. Two articles re-analyzed data or corrected erroneous data reported in previously published RCTs. One review article included primary data from an otherwise unpublished RCT.
Discussion: When conducting a literature search as part of a systematic review, all MEDLINE records indexed with the publication type "randomized controlled trial" should be considered RCTs unless an evaluation of the full-text article proves otherwise.