Searching Methods for Rapid Evidence Reviews During Public Health Emergencies

Objectives: As Covid-19-related evidence surged early in the pandemic, numerous Covid-specific databases and registries with varying features and functions emerged. Our team established a Rapid Evidence Service to conduct rapid reviews to inform the public health COVID-19 response. The service involved developing novel searching strategies to synthesize and disseminate the most up-to-date evidence available on COVID-19. Here, we outline our lessons learned in searching for evidence related to the COVID-19 pandemic and provide recommendations for researchers and database developers when responding to current and future public health emergencies.

Methods: Evidence sources, including databases and registries, used to inform our rapid reviews were evaluated based on three domains: topics and specialties included (i.e., infection prevention and control, disease characteristics, surveillance and epidemiology, equity, policy, and mental health and substance use); level of evidence of indexed citations (i.e., systematic review, single studies, expert opinion); and specific usability and functionality (i.e., advanced search functionality, search filters, citation exports, and citation sorting). These domains were assessed to determine usability and ability to reduce time and effort. To ensure comprehensiveness of our search strategy, our rapid review methods included general databases, specialized databases, and grey literature searching methods.

Results: We evaluated a total of 14 evidence sources, including 10 databases and four registries. Of these, three databases and one registry had been newly created in response to the pandemic and indexed only COVID-19 literature, while seven databases and three registries had been adapted from previously existing resources. Content and types of evidence included in each source varied. All databases included at least some advanced search functionality; repositories had limited advanced search functionality. We continue to evaluate the content and functionality of each evidence source and alter our search strategies accordingly.

Conclusions: Features and functionality associated with reducing time and effort for conducting rapid reviews included advanced searching features, filters, sorting and export functions. Controlled vocabulary allowed us to efficiently identify the most relevant evidence in a timely manner; the ability to sort and filter evidence by title and abstract allowed us to reduce the search volume; and export functions allowed us to deduplicate results across databases and allowed for transparent screening methods. By understanding the functionality and parameters of each database, reviews can be completed and disseminated in a timely manner for rapid response to public health emergencies. Our experiences can help inform future research strategies and the development of new evidence sources and enhancement of previously existing evidence sources.