Is the highly sensitive search worth the effort?

Article type
Authors
Parkhill A, Clavisi O, Bragge P, Gruen R
Abstract
Background: Evidence mapping is a method of determining the important questions in a broad topic area and finding relevant research evidence. Search strategies are an integral component of both evidence mapping and systematic reviewing. Due to the breadth of evidence mapping, traditional highly sensitive systematic review search strategies have not proven feasible from a time and resource perspective, have unmanageably large yields and a low signal to noise ratio. We have therefore developed an alternative evidence mapping search method. Objectives: To compare highly sensitive systematic review search strategies (HS) with specific evidence mapping (EM) search strings on the following parameters: yield, sensitivity, time requirements and resource use. Methods: HS and EM search results were compared by trialing the two search strings in MEDLINE for four sample questions in spinal cord injury and traumatic brain injury. Each search strategy was checked to monitor if it retrieved the MEDLINE available citations which had been included in our reports for the two conditions. For each strategy, we measured and compared: total number of retrieved references, percentage of relevant citations, hit rate of included references, and time spent in developing the search string, conducting the search and reviewing the references. Results: In comparison with HS searches, our EM search strategies: retrieved fewer references which were of a proportionately higher relevance overall, found all the included references, took more time to develop and took less time to both run and review. Conclusions: HS Boolean search strings are the current and most familiar search method but may not be as time-effective for evidence mapping as newer methods.