Article type
Abstract
Background
A COVID-19 living network meta-analysis (NMA) addressing drug prophylaxis and treatment has informed the development of several World Health Organization (WHO) practice recommendations. Conducting the NMAs, however, was time-consuming at different stages of systematic review, including literature search and selection and data abstraction, and increased the complexity of the statistical analysis, the certainty of evidence rating, and the interpretation of the results.
Objectives
Using the data from the COVID-19 living NMA, we aimed to determine what proportion of outcomes estimates of effect and certainty of evidence differ importantly when using pairwise meta-analysis versus NMA, and to explore the potential factors associated with the important difference when they exist.
Methods
We evaluated for each comparison and outcome that informed the WHO recommendations (1) whether there is an important difference between the pairwise meta-analysis and the NMA in the effect estimates and (2) whether there is an important difference between the pairwise meta-analysis and the NMA in the certainty of evidence. For calculating the proportions of comparisons and outcomes, we accounted for outcomes belonging to the same comparison by using a generalized linear mixed model. To explore potential scenarios in which the NMA is importantly different from the pairwise meta-analysis, we prespecified scenarios in which we think the NMA might be importantly different from the pairwise meta-analysis based on effect estimate or certainty of evidence (eg, the sample size or the number of events in pairwise meta-analysis does not meet the optimal information size) and used a generalized linear mixed model for analysis.
Results
This is a work in progress. We will present the proportion of comparison and outcomes for which the NMA is importantly different from the pairwise meta-analysis and discuss the scenarios in which the important changes in effect estimates or certainty of evidence eventually changed the direction or strength of recommendations.
Conclusions
The findings from the case of the COVID-19 living NMA that informed the WHO recommendations might indicate the scenarios in which NMA will be importantly different from pairwise meta-analysis and will be worth conducting.
A COVID-19 living network meta-analysis (NMA) addressing drug prophylaxis and treatment has informed the development of several World Health Organization (WHO) practice recommendations. Conducting the NMAs, however, was time-consuming at different stages of systematic review, including literature search and selection and data abstraction, and increased the complexity of the statistical analysis, the certainty of evidence rating, and the interpretation of the results.
Objectives
Using the data from the COVID-19 living NMA, we aimed to determine what proportion of outcomes estimates of effect and certainty of evidence differ importantly when using pairwise meta-analysis versus NMA, and to explore the potential factors associated with the important difference when they exist.
Methods
We evaluated for each comparison and outcome that informed the WHO recommendations (1) whether there is an important difference between the pairwise meta-analysis and the NMA in the effect estimates and (2) whether there is an important difference between the pairwise meta-analysis and the NMA in the certainty of evidence. For calculating the proportions of comparisons and outcomes, we accounted for outcomes belonging to the same comparison by using a generalized linear mixed model. To explore potential scenarios in which the NMA is importantly different from the pairwise meta-analysis, we prespecified scenarios in which we think the NMA might be importantly different from the pairwise meta-analysis based on effect estimate or certainty of evidence (eg, the sample size or the number of events in pairwise meta-analysis does not meet the optimal information size) and used a generalized linear mixed model for analysis.
Results
This is a work in progress. We will present the proportion of comparison and outcomes for which the NMA is importantly different from the pairwise meta-analysis and discuss the scenarios in which the important changes in effect estimates or certainty of evidence eventually changed the direction or strength of recommendations.
Conclusions
The findings from the case of the COVID-19 living NMA that informed the WHO recommendations might indicate the scenarios in which NMA will be importantly different from pairwise meta-analysis and will be worth conducting.