Article type
Abstract
Background
Searching for studies to include in a systematic review uses a standardized approach that has changed little over the years. This approach typically involves searching bibliographic databases using Boolean logic to combine index terms and search words to create a search string, and supplementary methods such as checking reference lists. Alternate approaches have been proposed in an effort to reduce searching time and effort, but the performance of modifications to or replacements of the standard approach has not been systematically examined.
Objectives
To determine the comparative performance of standard and alternative searching approaches used to find studies for inclusion in systematic reviews
Methods
We searched PubMed, Library Information Science and Technology Abstracts (LISTA), and the SureINFO website and conducted backward and forward citation searches for studies comparing the standard search approach with an alternate approach (that is intended to be a replacement for or modification of the standard approach) in ≥5 searches and reporting recall (number of relevant studies found), precision (number of irrelevant studies found), or time to conduct the search.
Results
We screened 2268 records and included 28 studies. Four studies evaluate the following strategies as replacements for the standard approach: 1) searching Google Scholar, 2) using citation searching, 3) using PubMed’s similar articles feature, and 4) restricting the number of terms combined with a filter. Ten studies evaluate modifications to the standard approach, including 1) using a study design filter, 2) using seed studies to design the search, 3) focusing index terms, 4) using a single word for each concept, 5) using automation for search translation, 6) contacting manufacturers, 7) checking reference lists, 8) using frequency of words in title/abstracts, 9) searching the title field only, and 10) using subject term subheadings. Just over half (15/29) of the included studies are considered to be at unclear or high risk of bias.
Conclusions
Analysis of the comparative performance of the search approaches is underway. This review will provide critical information to enable evidence informed decisions regarding how to most efficiently search for studies for inclusion in systematic reviews.
Searching for studies to include in a systematic review uses a standardized approach that has changed little over the years. This approach typically involves searching bibliographic databases using Boolean logic to combine index terms and search words to create a search string, and supplementary methods such as checking reference lists. Alternate approaches have been proposed in an effort to reduce searching time and effort, but the performance of modifications to or replacements of the standard approach has not been systematically examined.
Objectives
To determine the comparative performance of standard and alternative searching approaches used to find studies for inclusion in systematic reviews
Methods
We searched PubMed, Library Information Science and Technology Abstracts (LISTA), and the SureINFO website and conducted backward and forward citation searches for studies comparing the standard search approach with an alternate approach (that is intended to be a replacement for or modification of the standard approach) in ≥5 searches and reporting recall (number of relevant studies found), precision (number of irrelevant studies found), or time to conduct the search.
Results
We screened 2268 records and included 28 studies. Four studies evaluate the following strategies as replacements for the standard approach: 1) searching Google Scholar, 2) using citation searching, 3) using PubMed’s similar articles feature, and 4) restricting the number of terms combined with a filter. Ten studies evaluate modifications to the standard approach, including 1) using a study design filter, 2) using seed studies to design the search, 3) focusing index terms, 4) using a single word for each concept, 5) using automation for search translation, 6) contacting manufacturers, 7) checking reference lists, 8) using frequency of words in title/abstracts, 9) searching the title field only, and 10) using subject term subheadings. Just over half (15/29) of the included studies are considered to be at unclear or high risk of bias.
Conclusions
Analysis of the comparative performance of the search approaches is underway. This review will provide critical information to enable evidence informed decisions regarding how to most efficiently search for studies for inclusion in systematic reviews.