Precision of healthcare systematic review searches in a cross-sectional sample
Margaret Sampson, MLIS, PhD, Children's Hospital of Eastern Ontario Research Institute, 401 Smyth Road, Ottawa, Ontario, Canada K1H 8L1.
In systematic reviews, search precision is generally traded off against the desire to retrieve all relevant studies; however, there is no published evidence on typical precision values. The objective of this study is to establish typical values for the precision of systematic review searches in healthcare.
From an existing cross-sectional sample of 300 MEDLINE-indexed systematic reviews, those that reported the flow of bibliographic records through the review process (n = 109) were examined. Where the ratio of the number of included studies and the number of unique retrievals could be determined, overall and median precision of the search was calculated. Subgroup analyses were conducted by review type (treatment/prevention, diagnosis/prognosis, epidemiology, other), eligible study designs, number of databases searched and for updates of existing systematic reviews.
Precision could be calculated for 94 systematic reviews. The median [interquartile range] precision was 0.029 [0.013, 0.081] with a range of 0.007–0.358. In this sample, precision did not differ significantly in any of the subgroups examined.
Search precision of approximately 3% was typical in this cross-section of health related systematic reviews. This finding is useful for systematic review teams to gauge review resource needs and for information specialists in evaluating their searches. Copyright © 2011 John Wiley & Sons, Ltd.