Case study on comprehensive searching in the area of overcrowding in emergency departments

Tags: Poster
Friesen C, Hartling L, Bond K, Ospina M, Rowe B

Background: Comprehensive searching, a crucial step in the systematic review process, aims to maximize sensitivity through a thorough and unbiased process. Ideally, it should retrieve all relevant articles including those identified by content experts. A key component of comprehensive searching is the use of controlled vocabularies (i.e. subject headings from electronic databases) and keywords, the combination of which can improve the sensitivity of searches, especially for searches of broad topic areas. We recently completed a systematic review of quality measures used to document overcrowding in the emergency department (ED) and interventions to reduce ED overcrowding, topics that are poorly indexed in electronic databases. Following a comprehensive search, we used two databases developed by content experts to check the search’s accuracy.

Objectives: The objectives of this case study are to compare the accuracy of 1) electronic database searches versus expert databases, and 2) using subject headings alone versus subject headings with keywords for selected databases.

Methods: The reference standard includes the total number of all unique articles identified through the electronic searches of 19 databases, the expert databases, and grey literature searches. Review team members determined the relevancy of articles according to established criteria. Accuracy will be determined by calculating sensitivity (the proportion of all relevant articles retrieved); specificity (the proportion of all irrelevant articles not retrieved); and precision (the proportion of relevant articles). These measures will be calculated and compared for 1) the electronic database searches, 2) the expert databases, and 3) key database searches with and without keywords.

Results: Results describing the accuracy of electronic database searches versus expert databases and using subject headings with and without keywords will be available at the Cochrane Colloquium in Melbourne, Australia.

Conclusions: These results will demonstrate the value of comprehensive searching of electronic databases versus expert databases in the area of ED overcrowding. The results will also illustrate the relative value of using keywords in addition to subject headings. We hope to contribute to the understanding of different approaches to developing comprehensive searches for systematic reviews, particularly in broad topic areas where indexing is poor.