Development and Validation of a Search Filter to Identify Research on Bisexuality

Locating health research on bisexual and pansexual (bisexual+) populations is challenging, as the data are usually collated with data for additional sexual minority and gender identities. To improve the findability of health research on bisexual+ populations, this study sought to develop a sensitivity-maximizing PubMed search filter for bisexuality. Using the relative recall method for the development and validation of search filters, we used PubMed, CitationChaser, and Covidence to search for and screen studies. To be included, studies had to report bisexual/pansexual-specific data and be MEDLINE-indexed with a PubMed Identifier (PMID). Of 291 eligible records, 252 had PMIDs; these records constituted the gold standard set used to develop the search filters. Combinations of search terms were tested against the gold standard set. Sensitivity and number needed to read (NNR) were calculated for each combination. Two search filters are presented. The sensitivity-maximizing search retrieved 100% of the gold standard set, with an NNR of 129.85. The optimized search retrieved all but one of the gold standard articles (99.60%), with an NNR of 74.88. Two PubMed search filters are presented for bisexual+ populations. These filters were validated using the relative recall method against a gold standard set derived from citing and cited references of systematic reviews on bisexual+ health. Use of the sensitivity-maximizing search filter is recommended for exhaustive searches, while the optimized filter is considered more appropriate for nonexhaustive searches.