Article type
Year
Abstract
Background: The search in network meta-analysis is very critical, as network meta-analysis aims to rank the benefits (and harms) of interventions based on all available randomized controlled trials (RCTs).
Objectives: So it is necessary to assess the search status in network meta-analyses.
Methods: Published network meta-analyses were retrieved by searching databases (Pubmed, the Cochrane library, Embase, and ISI Web of Knowledge) and hand-searching other sources (Google engine, HTA websites, references lists). Two independent reviewers conducted search, select studies, abstracted data. Statistical analyses were conducted using SPSS version 15.0 for Windows.
Results: 104 network meta-analyses were included. The searches in network meta-analysis were not comprehensive, although 92.6% searched databases and hand-searched one or more other resources. The median number of databases was 3 (IQR 2–4). Medline, Embase, and CENTRAL are the most common used databases. However, 7.4% used the included studies in specific systematic reviews and hand-searching other resources and unpublished data were not well conducted. Although most network meta-analyses (70.4%) stated search terms, their reporting for search was poor and inadequate, as search strategy was reported in 24.1% network meta-analyses using simple format and the restrictions and details for search were not well stated.
Conclusions: The searches in network meta-analysis were not comprehensive, and the reporting for search was poor and inadequate. The best way for retrieving studies for network meta analysis should be using of two or more databases, cross-checking the reference lists of previous systematic reviews and hand searching of other resources by two independent searcher and/or information professionals. And the reporting guidelines for the searches of network meta-analysis should be developed, and at least include items related to search resources, search terms, search restrictions, search date and the searchers.
Objectives: So it is necessary to assess the search status in network meta-analyses.
Methods: Published network meta-analyses were retrieved by searching databases (Pubmed, the Cochrane library, Embase, and ISI Web of Knowledge) and hand-searching other sources (Google engine, HTA websites, references lists). Two independent reviewers conducted search, select studies, abstracted data. Statistical analyses were conducted using SPSS version 15.0 for Windows.
Results: 104 network meta-analyses were included. The searches in network meta-analysis were not comprehensive, although 92.6% searched databases and hand-searched one or more other resources. The median number of databases was 3 (IQR 2–4). Medline, Embase, and CENTRAL are the most common used databases. However, 7.4% used the included studies in specific systematic reviews and hand-searching other resources and unpublished data were not well conducted. Although most network meta-analyses (70.4%) stated search terms, their reporting for search was poor and inadequate, as search strategy was reported in 24.1% network meta-analyses using simple format and the restrictions and details for search were not well stated.
Conclusions: The searches in network meta-analysis were not comprehensive, and the reporting for search was poor and inadequate. The best way for retrieving studies for network meta analysis should be using of two or more databases, cross-checking the reference lists of previous systematic reviews and hand searching of other resources by two independent searcher and/or information professionals. And the reporting guidelines for the searches of network meta-analysis should be developed, and at least include items related to search resources, search terms, search restrictions, search date and the searchers.