Article type
Abstract
Background: When making decisions, it is important to incorporate the preferences of patients regarding their options and, more specifically, the relative importance patients place on health outcomes. There is a growing interest on this topic, however, there is little guidance on how to retrieve efficiently research evidence for its assessment in systematic reviews, clinical guidelines or other type of decisions.
Objectives: To develop and validate a search filter for the retrieval of studies in this topic in MEDLINE (using Pubmed)
Methods: We constructed a 'gold standard' performing a handsearch of high-volume journals, and a random sample of MEDLINE articles. Eligibility and classification of relevant articles was performed in duplicate; discrepancies were solved by consensus. We performed term-by-term searches in MEDLINE (via PubMed) using an initial list of terms developed by our group, and other terms provided by experts in the field. We used a diagnostic test accuracy assessment framework to calculate sensitivity, specificity, precision, and accuracy of each term comparing the search results from the database with the articles included in the '“gold standard'. We combined the search terms in multiple permutations to identify the combination with the best sensibility, specificity and balance within specificity and sensibility. We validated these filters using a subset and recalculated sensitivity, sensibility, precision and accuracy.
Results: We are currently conducting the term-by-term searches. We will present the characteristics of the gold standard and the filters with their corresponding sensitivity, specificity, precision, and accuracy at the Summit.
Conclusions: The proposed filter will fill an important gap in research, facilitating the retrieval of studies on this topic. This, in turn, will facilitate the development of systematic reviews, and the incorporation of this type of evidence when developing recommendations and other type of decisions.
Objectives: To develop and validate a search filter for the retrieval of studies in this topic in MEDLINE (using Pubmed)
Methods: We constructed a 'gold standard' performing a handsearch of high-volume journals, and a random sample of MEDLINE articles. Eligibility and classification of relevant articles was performed in duplicate; discrepancies were solved by consensus. We performed term-by-term searches in MEDLINE (via PubMed) using an initial list of terms developed by our group, and other terms provided by experts in the field. We used a diagnostic test accuracy assessment framework to calculate sensitivity, specificity, precision, and accuracy of each term comparing the search results from the database with the articles included in the '“gold standard'. We combined the search terms in multiple permutations to identify the combination with the best sensibility, specificity and balance within specificity and sensibility. We validated these filters using a subset and recalculated sensitivity, sensibility, precision and accuracy.
Results: We are currently conducting the term-by-term searches. We will present the characteristics of the gold standard and the filters with their corresponding sensitivity, specificity, precision, and accuracy at the Summit.
Conclusions: The proposed filter will fill an important gap in research, facilitating the retrieval of studies on this topic. This, in turn, will facilitate the development of systematic reviews, and the incorporation of this type of evidence when developing recommendations and other type of decisions.