INTRODUCTION
Deprescribing literature has been increasing continuously for the last ten years, as have systematic reviews focusing on deprescribing-related questions. Authors developed various search strategies to retrieve deprescribing literature. To our knowledge, there is still no consensus on the terms to use to exhaustively retrieve deprescribing articles in bibliographic databases (i.e. MEDLINE, Embase, etc. ). The use of carefully selected terms is recommended to ensure the exhaustiveness of a systematic review search strategy1. However controlled vocabulary aimed at identifying deprescribing articles remains unclear.
Search filters are specifically developed to avoid indexing pitfalls due to imprecise controlled vocabulary, therefore improving search effectiveness2. Search filters usually focus on a particular study design or a specific topic3. The performance of a search filter is usually evaluated based on a reference set of relevant articles for the target study design or topic.
We recently developed two deprescribing search filters with maximized sensitivity for MEDLINE (using PubMed interface) and for Embase (using Embase.com interface) with a sensitivity of 92% (95% CI: 83–97) and 91% (95% CI: 82–96) respectively4. The efficiency gained by using these maximized sensitivity filters in systematic reviews search strategies remains unknown. Simultaneously, The US Deprescribing Network (USDeN) developed a deprescribing search strategy that included a deprescribing search filter for MEDLINE using PubMed interface, but its performance has not been evaluated yet5.
The aim of this case study was to implement these three deprescribing search filters in systematic review search strategies and evaluate their performances.