Article type
Year
Abstract
Background: the sensitivity of a systematic or rapid review search strategy can be tested against citations of studies that are deemed to meet the inclusion criteria. Even so, these citations themselves have been shown to be similar to other studies that will eventually be included in the final review; using similarity algorithms to identify ‘similar articles’ in databases like PubMed and Ovid Medline. For example, in a perfect scenario, all citations of included studies in a systematic review would be ‘similar’ to each other. Therefore, if you identified only one then a ‘similar articles’ search could potentially retrieve the citations of all the others.
Objectives: the objective of this study is to report on the development and validation of the Intelligent Related Citation Searching (IRIS) strategy.
Methods: an inter-disciplinary team of methodologists, clinical researchers and librarians developed and validated IRIS using data from Cochrane Reviews published in 2018. We extracted all citations of included studies in these reviews and identified if they were also indexed in PubMed. For feasibility, we picked a random sample of 100 reviews that included at least five citations with PubMed IDs (PMIDs) for analysis.
We developed IRIS for Excel (IrisXl) to interact with PubMed via the Entrez query and database system at the National Center for Biotechnology Information (NCBI). For each included citation, we submitted the PMID to PubMed and requested the PMIDs of all ‘related articles’. We also mapped the retrieved PMIDs into a network diagram for each Cochrane Review. We calculated the percentages of citations that were identified by IRIS and the screening burden to identify the additional studies included in the Cochrane Review.
We used Endnote (X8), Microsoft Excel, and Excel add-ons (NodeXl and IrisXl) for citation management, data management and data analysis.
Results: from the 100 analyzed Cochrane Reviews, we identified 2610 included citations indexed in PubMed. The reviews included 29.2 ± 30.3 citations (range: 6 to 217). For each PubMed ID, IrisXl reported how many citations were retrieved and the number of relationships each retrieved citation had with the included citations. The sensitivity of IRIS was 89.1% ± 13.9% (range: 40% to 100%). Visual inspection of the network diagrams allows identification of study citations that were not related to others for further review.
The screening burden was 41 ± 38 citations screened for every new identified citation that was included in a Cochrane Review. Figure 1 show the steps involved in using IRIS in identifying studies for inclusion.
Conclusions: the use of IRIS may provide an alternative or complementary strategy for identifying articles for inclusion. Further research is being conducted to compare the results with traditional search methods and the comparative screening burden of each approach.
Patient or healthcare consumer involvement: not involved in this methods project
Objectives: the objective of this study is to report on the development and validation of the Intelligent Related Citation Searching (IRIS) strategy.
Methods: an inter-disciplinary team of methodologists, clinical researchers and librarians developed and validated IRIS using data from Cochrane Reviews published in 2018. We extracted all citations of included studies in these reviews and identified if they were also indexed in PubMed. For feasibility, we picked a random sample of 100 reviews that included at least five citations with PubMed IDs (PMIDs) for analysis.
We developed IRIS for Excel (IrisXl) to interact with PubMed via the Entrez query and database system at the National Center for Biotechnology Information (NCBI). For each included citation, we submitted the PMID to PubMed and requested the PMIDs of all ‘related articles’. We also mapped the retrieved PMIDs into a network diagram for each Cochrane Review. We calculated the percentages of citations that were identified by IRIS and the screening burden to identify the additional studies included in the Cochrane Review.
We used Endnote (X8), Microsoft Excel, and Excel add-ons (NodeXl and IrisXl) for citation management, data management and data analysis.
Results: from the 100 analyzed Cochrane Reviews, we identified 2610 included citations indexed in PubMed. The reviews included 29.2 ± 30.3 citations (range: 6 to 217). For each PubMed ID, IrisXl reported how many citations were retrieved and the number of relationships each retrieved citation had with the included citations. The sensitivity of IRIS was 89.1% ± 13.9% (range: 40% to 100%). Visual inspection of the network diagrams allows identification of study citations that were not related to others for further review.
The screening burden was 41 ± 38 citations screened for every new identified citation that was included in a Cochrane Review. Figure 1 show the steps involved in using IRIS in identifying studies for inclusion.
Conclusions: the use of IRIS may provide an alternative or complementary strategy for identifying articles for inclusion. Further research is being conducted to compare the results with traditional search methods and the comparative screening burden of each approach.
Patient or healthcare consumer involvement: not involved in this methods project