Mind the gap: identifying what is missed when searching only the broad scope with clinical queries

The PubMed Clinical Study Category filters are subdivided into "Broad" and "Narrow" versions that are designed to maximize either sensitivity or specificity by using two different sets of keywords and Medical Subject Headings (MeSH). A searcher might assume that all items retrieved by Narrow would also be found by Broad, but there are occasions when some [Filter name]/Narrow citations are missed when using [Filter name]/Broad alone. This study quantifies the size of this effect.

For each of the five Clinical Study Categories, PubMed was searched for citations matching the query Filter/Narrow NOT Filter/Broad. This number was compared with that for Filter/Broad to compute the number of Narrow citations missed per 1,000 Broad. This process was repeated for the MeSH terms for "Medicine" and "Diseases," as well as for a set of individual test searches.

The Clinical Study Category filters for Etiology, Clinical Prediction Guides, Diagnosis, and Prognosis all showed notable numbers of Filter/Narrow citations that were missed when searching Filter/Broad alone. This was particularly true for Prognosis, where a searcher could easily miss one Prognosis/Narrow citation for every ten Prognosis/Broad citations retrieved.

Users of the Clinical Study Category filters (except for Therapy) should consider combining Filter/Narrow together with Filter/Broad in their search strategy. This is particularly true when using Prognosis/Broad, as otherwise there is a substantial risk of missing potentially relevant citations.