Article type
Abstract
Background:
In 2023 the World Health Organization requested a rapid search update to identify any new studies published since the last search of a systematic review conducted in 2020 on the effectiveness of school food and nutrition policies. The timeline for the update was short and efficient ways to speed up the process without compromising on quality were needed.
Objectives:
To share an approach used to identify which databases to search in a systematic review update.
Methods:
To determine which of the 11 original databases to search for the update, we searched the original results from each database in an EndNote file to identify which databases had picked up each of the studies included in the review. We built a matrix of this information in Excel, based on which we then determined which databases had identified most of the included studies as well as unique studies. This information then helped us determine which databases to focus on in the search update. We then searched the additional databases (Scopus, Web of Science, Cochrane Central Register of Controlled Trials and Global Index Medicus) except Open Grey and ClinicalTrials.gov to determine how many additional records needed to be screened.
Results:
Of the 86 included records, most were identified by six of the 11 original databases: PubMed (n=58), Epistemonikos (n=37), and CINAHL, Africa-Wide and ERIC databases searched through the EbscoHost platform (n=33). These were also where most (20 of 25) unique records i.e. only identified in one database, were identified. Although many records were also identified in Web of Science (n=24), only one was unique. The five databases selected for the search update were those which included most studies as well as the databases where the unique studies were identified. The update retrieved 4146 records and searching the additional databases would have required screening an additional 1514 records, an increase of 36.5%.
Conclusion:
Identifying which databases to search for a rapid update may speed up the screening process and prevent the screening of many irrelevant records.
There were no public and/or consumer involvement in this project.
In 2023 the World Health Organization requested a rapid search update to identify any new studies published since the last search of a systematic review conducted in 2020 on the effectiveness of school food and nutrition policies. The timeline for the update was short and efficient ways to speed up the process without compromising on quality were needed.
Objectives:
To share an approach used to identify which databases to search in a systematic review update.
Methods:
To determine which of the 11 original databases to search for the update, we searched the original results from each database in an EndNote file to identify which databases had picked up each of the studies included in the review. We built a matrix of this information in Excel, based on which we then determined which databases had identified most of the included studies as well as unique studies. This information then helped us determine which databases to focus on in the search update. We then searched the additional databases (Scopus, Web of Science, Cochrane Central Register of Controlled Trials and Global Index Medicus) except Open Grey and ClinicalTrials.gov to determine how many additional records needed to be screened.
Results:
Of the 86 included records, most were identified by six of the 11 original databases: PubMed (n=58), Epistemonikos (n=37), and CINAHL, Africa-Wide and ERIC databases searched through the EbscoHost platform (n=33). These were also where most (20 of 25) unique records i.e. only identified in one database, were identified. Although many records were also identified in Web of Science (n=24), only one was unique. The five databases selected for the search update were those which included most studies as well as the databases where the unique studies were identified. The update retrieved 4146 records and searching the additional databases would have required screening an additional 1514 records, an increase of 36.5%.
Conclusion:
Identifying which databases to search for a rapid update may speed up the screening process and prevent the screening of many irrelevant records.
There were no public and/or consumer involvement in this project.