Article type
Abstract
"Background: There is a growing use of the GRADE system in Systematic Reviews (SR) of epidemiological studies (prevalence or incidence). However, there is no standardization for the application of its criteria in this type of SR, so the evaluation of their certainty may be incorrect.
Objectives: To describe the use of the GRADE system and its domains (risk of bias, imprecision, lack of direction, inconsistency, publication bias, and confounding factors) in SRs of epidemiological studies.
Methods: We searched in PubMed, Scopus, Web of Science, and Embase from January 2017 to December 2021. We included SRs that evaluated the prevalence-incidence of some clinical diseases and that declared the use of the GRADE system. We describe the methodological characteristics of the elaboration of SR and the use of each domain according to the GRADE handbook.
Results: We analyzed 96 SR, of which 40 were in the community population. The median number of databases consulted was 5 (interquartile range [IQR]: 3-6). 66.7% of the included reviews registered a protocol. A statistical analysis was performed on 84.4% of the reviews. Nineteen studies used the GRADE system to assess the risk of bias. Regarding the use of GRADE, 93.5% of reviews clearly described the population and outcome, 58.4% presented a table of summary of findings. The clear description of the domain to assess the certainty of the evidence was variable (study design: 64.9%; risk of bias: 79.2%; inconsistency: 83.1%; indirectness: 72.7%; imprecision: 79.2%; publication bias: 58.4%; confounding factor: 10.4%; large effect size 13.0%; dose-response gradient: 9.1%).
Conclusion: There is significant variability and inconsistencies concerning how the GRADE system and its domains were used, reflecting a lack of consensus and knowledge about the certainty rating of the evidence. The inappropriate use of the GRADE system can limit the validity of the results in decision-making.
"
Objectives: To describe the use of the GRADE system and its domains (risk of bias, imprecision, lack of direction, inconsistency, publication bias, and confounding factors) in SRs of epidemiological studies.
Methods: We searched in PubMed, Scopus, Web of Science, and Embase from January 2017 to December 2021. We included SRs that evaluated the prevalence-incidence of some clinical diseases and that declared the use of the GRADE system. We describe the methodological characteristics of the elaboration of SR and the use of each domain according to the GRADE handbook.
Results: We analyzed 96 SR, of which 40 were in the community population. The median number of databases consulted was 5 (interquartile range [IQR]: 3-6). 66.7% of the included reviews registered a protocol. A statistical analysis was performed on 84.4% of the reviews. Nineteen studies used the GRADE system to assess the risk of bias. Regarding the use of GRADE, 93.5% of reviews clearly described the population and outcome, 58.4% presented a table of summary of findings. The clear description of the domain to assess the certainty of the evidence was variable (study design: 64.9%; risk of bias: 79.2%; inconsistency: 83.1%; indirectness: 72.7%; imprecision: 79.2%; publication bias: 58.4%; confounding factor: 10.4%; large effect size 13.0%; dose-response gradient: 9.1%).
Conclusion: There is significant variability and inconsistencies concerning how the GRADE system and its domains were used, reflecting a lack of consensus and knowledge about the certainty rating of the evidence. The inappropriate use of the GRADE system can limit the validity of the results in decision-making.
"