Maureen Rice1, Muhammad Usman Ali1, Donna Fitzpatrick-Lewis2, Meghan Kenny1, Parminder Raina1, Diana Sherifali3. 1. McMaster Evidence Review and Synthesis Centre, McMaster University, 1280 Main St. W., McMaster Innovation Park, Room 207A, Hamilton, Ontario L8S 4K1, Canada; Department of Health Research Methods, Evidence and Impact, Faculty of Health Sciences, McMaster University, Hamilton, Ontario L8L 2X2, Canada. 2. McMaster Evidence Review and Synthesis Centre, McMaster University, 1280 Main St. W., McMaster Innovation Park, Room 207A, Hamilton, Ontario L8S 4K1, Canada; School of Nursing, Faculty of Health Sciences, McMaster University, Health Sciences Centre Room HSC-3N25F, 1280 Main Street West, Hamilton, Ontario L8S 4K1, Canada. 3. McMaster Evidence Review and Synthesis Centre, McMaster University, 1280 Main St. W., McMaster Innovation Park, Room 207A, Hamilton, Ontario L8S 4K1, Canada; School of Nursing, Faculty of Health Sciences, McMaster University, Health Sciences Centre Room HSC-3N25F, 1280 Main Street West, Hamilton, Ontario L8S 4K1, Canada. Electronic address: dsherif@mcmaster.ca.
Abstract
OBJECTIVE: The objective of the study was to test the overall effectiveness of a simplified search strategy (SSS) for updating systematic reviews. STUDY DESIGN AND METHODS: We identified nine systematic reviews undertaken by our research group for which both comprehensive and SSS updates were performed. Three relevant performance measures were estimated, that is, sensitivity, precision, and number needed to read (NNR). RESULTS: The update reference searches for all nine included systematic reviews identified a total of 55,099 citations that were screened resulting in final inclusion of 163 randomized controlled trials. As compared with reference search, the SSS resulted in 8,239 hits and had a median sensitivity of 83.3%, while precision and NNR were 4.5 times better. During analysis, we found that the SSS performed better for clinically focused topics, with a median sensitivity of 100% and precision and NNR 6 times better than for the reference searches. For broader topics, the sensitivity of the SSS was 80% while precision and NNR were 5.4 times better compared with reference search. CONCLUSION: SSS performed well for clinically focused topics and, with a median sensitivity of 100%, could be a viable alternative to a conventional comprehensive search strategy for updating this type of systematic reviews particularly considering the budget constraints and the volume of new literature being published. For broader topics, 80% sensitivity is likely to be considered too low for a systematic review update in most cases, although it might be acceptable if updating a scoping or rapid review.
OBJECTIVE: The objective of the study was to test the overall effectiveness of a simplified search strategy (SSS) for updating systematic reviews. STUDY DESIGN AND METHODS: We identified nine systematic reviews undertaken by our research group for which both comprehensive and SSS updates were performed. Three relevant performance measures were estimated, that is, sensitivity, precision, and number needed to read (NNR). RESULTS: The update reference searches for all nine included systematic reviews identified a total of 55,099 citations that were screened resulting in final inclusion of 163 randomized controlled trials. As compared with reference search, the SSS resulted in 8,239 hits and had a median sensitivity of 83.3%, while precision and NNR were 4.5 times better. During analysis, we found that the SSS performed better for clinically focused topics, with a median sensitivity of 100% and precision and NNR 6 times better than for the reference searches. For broader topics, the sensitivity of the SSS was 80% while precision and NNR were 5.4 times better compared with reference search. CONCLUSION: SSS performed well for clinically focused topics and, with a median sensitivity of 100%, could be a viable alternative to a conventional comprehensive search strategy for updating this type of systematic reviews particularly considering the budget constraints and the volume of new literature being published. For broader topics, 80% sensitivity is likely to be considered too low for a systematic review update in most cases, although it might be acceptable if updating a scoping or rapid review.
Authors: David Metcalfe; Andrew Judge; Daniel C Perry; Belinda Gabbe; Cheryl K Zogg; Matthew L Costa Journal: BMC Musculoskelet Disord Date: 2019-05-17 Impact factor: 2.362