Article type
Year
Abstract
Background:
Updating systematic reviews (SRs) may generate new and very important information, but the process of updating the SR is time-consuming and may inflate larger type I errors. And it is still unclear whether the precision of outcomes is improved during the update process.
Objectives:
To investigate the precision of outcomes between updated SRs and original SRs.
Methods:
We conducted comprehensive electronic searches in PubMed and Embase.com on March 3, 2019, and included 30 updated SRs and 30 original SRs. We calculated the ratio of effect size and the difference of mean differences or standardized mean differences to estimate the change of outcomes. Review Manager 5.3 software was adopted to create forest plots showing comparable outcomes and standard error was used to compare the precision of the outcomes.
Results:
The average update time was 56.0 months and incorporating new trials (23 SRs, 76.7%) was the main reason for the update. Compared with original SRs, 24 (80.0%) updated SRs included more randomized controlled trials, and 22 (73.3%) updated SRs involved a larger number of patients. The newly included randomized controlled trials (RCTs) more often clarified the blinding of participants and personnel compared with previously RCTs included in original SRs (P
Updating systematic reviews (SRs) may generate new and very important information, but the process of updating the SR is time-consuming and may inflate larger type I errors. And it is still unclear whether the precision of outcomes is improved during the update process.
Objectives:
To investigate the precision of outcomes between updated SRs and original SRs.
Methods:
We conducted comprehensive electronic searches in PubMed and Embase.com on March 3, 2019, and included 30 updated SRs and 30 original SRs. We calculated the ratio of effect size and the difference of mean differences or standardized mean differences to estimate the change of outcomes. Review Manager 5.3 software was adopted to create forest plots showing comparable outcomes and standard error was used to compare the precision of the outcomes.
Results:
The average update time was 56.0 months and incorporating new trials (23 SRs, 76.7%) was the main reason for the update. Compared with original SRs, 24 (80.0%) updated SRs included more randomized controlled trials, and 22 (73.3%) updated SRs involved a larger number of patients. The newly included randomized controlled trials (RCTs) more often clarified the blinding of participants and personnel compared with previously RCTs included in original SRs (P