Article type
Year
Abstract
Background: Search strategies to identify randomized controlled trials are well established. However, strategies to retrieve diagnostic studies from the literature are less well developed.
Objectives: To develop a sensitive search strategy for MEDLINE to retrieve studies of diagnostic test performance for all imaging modalities.
Methods: A handsearch of radiology journals was conducted for studies of diagnostic imaging test performance using pre-defined inclusion criteria. These articles formed independent derivation and validation sets of records, for evaluating the search strategy. The sensitivity and positive predictive values (PPV) of search terms from the derivation set were used to develop two components of the search strategy. The first component identified any study assessing diagnostic test performance, from the imaging literature; the second identified studies of any imaging modality. The two components were combined in the final search strategy. The sensitivity, specificity and PPVs of the search strategy were calculated with the derivation and validation sets of records. A pilot study was conducted using the strategy in MEDLINE.
Results: Retrieval characteristics for the derivation set were a sensitivity of 92.8% (95%CI 89.5, 95.1), a specificity of 58.5% (95%CI 56.4, 60.5), and a PPV of 25.1% (95%CI 22.7, 27.6). Validation with an independent set of articles achieved a sensitivity of 91.9% (95%CI 87.1, 95.1), a specificity of 52.2% (95%CI 49.2, 55.2), and a PPV of 25.1% (95%CI 22.0, 28.5). Removal of irrelevant publication types further improved retrieval characteristics for specificity (73.0% CI 72.0, 75.7%) and PPV (33.9% CI 30.9, 37.1%). The pilot study achieved a PPV of 29.2% in MEDLINE after removal of non-relevant articles. We found that the volume of imaging literature in MEDLINE had tripled since 1975 using our search strategy.
Conclusions: A sensitive search strategy to identify studies assessing diagnostic performance of imaging tests, for any imaging modality, was developed and validated. The search strategy has adequate retrieval characteristics to develop a register of studies using MEDLINE.
Objectives: To develop a sensitive search strategy for MEDLINE to retrieve studies of diagnostic test performance for all imaging modalities.
Methods: A handsearch of radiology journals was conducted for studies of diagnostic imaging test performance using pre-defined inclusion criteria. These articles formed independent derivation and validation sets of records, for evaluating the search strategy. The sensitivity and positive predictive values (PPV) of search terms from the derivation set were used to develop two components of the search strategy. The first component identified any study assessing diagnostic test performance, from the imaging literature; the second identified studies of any imaging modality. The two components were combined in the final search strategy. The sensitivity, specificity and PPVs of the search strategy were calculated with the derivation and validation sets of records. A pilot study was conducted using the strategy in MEDLINE.
Results: Retrieval characteristics for the derivation set were a sensitivity of 92.8% (95%CI 89.5, 95.1), a specificity of 58.5% (95%CI 56.4, 60.5), and a PPV of 25.1% (95%CI 22.7, 27.6). Validation with an independent set of articles achieved a sensitivity of 91.9% (95%CI 87.1, 95.1), a specificity of 52.2% (95%CI 49.2, 55.2), and a PPV of 25.1% (95%CI 22.0, 28.5). Removal of irrelevant publication types further improved retrieval characteristics for specificity (73.0% CI 72.0, 75.7%) and PPV (33.9% CI 30.9, 37.1%). The pilot study achieved a PPV of 29.2% in MEDLINE after removal of non-relevant articles. We found that the volume of imaging literature in MEDLINE had tripled since 1975 using our search strategy.
Conclusions: A sensitive search strategy to identify studies assessing diagnostic performance of imaging tests, for any imaging modality, was developed and validated. The search strategy has adequate retrieval characteristics to develop a register of studies using MEDLINE.