Article type
Year
Abstract
Background: Despite the increasing information overload and the great advances in the methodology of systematic reviews, evidence gaps do exist and impose barriers to well-informed decision making. In such cases, further studies need to be designed to boost existing evidence and narrow the evidence-practice gap. When there are multiple competing interventions for a healthcare problem, network meta-analysis (NMA) can be used to guide the design of new studies.
Objectives: Our objective is to provide a general framework for using NMA evidence in planning future studies.
Methods: The targeted parameter is the precision of the results obtained from NMA: the precision of the joint distribution of the estimated basic parameters of the model and the precision in the treatment ranking. We quantify the precision in the estimated effects by considering their variance-covariance matrix and estimate the precision in ranking by quantifying the dissimilarity of the density functions of summary effect estimates. Then, based on a desirable improvement in precision we calculate the required sample size for each possible study design and number of study arms and we present graphical tools that can help trialists select the optimal study design.
Results: We used a published network of three interventions for the treatment of hepatocellular carcinoma to illustrate the suggested methodology. Although the three-arm design is the most efficient in terms of required sample size, choosing a two-arm design can also decrease the uncertainty about the relative effects substantially, depending on the chosen comparison.
Conclusions: The methodology presented can be used to inform the future research agenda by indicating which parts of existing networks need further investigation. Through this process, unnecessary waste of research that leads to human and monetary cost may be considerably reduced.
Objectives: Our objective is to provide a general framework for using NMA evidence in planning future studies.
Methods: The targeted parameter is the precision of the results obtained from NMA: the precision of the joint distribution of the estimated basic parameters of the model and the precision in the treatment ranking. We quantify the precision in the estimated effects by considering their variance-covariance matrix and estimate the precision in ranking by quantifying the dissimilarity of the density functions of summary effect estimates. Then, based on a desirable improvement in precision we calculate the required sample size for each possible study design and number of study arms and we present graphical tools that can help trialists select the optimal study design.
Results: We used a published network of three interventions for the treatment of hepatocellular carcinoma to illustrate the suggested methodology. Although the three-arm design is the most efficient in terms of required sample size, choosing a two-arm design can also decrease the uncertainty about the relative effects substantially, depending on the chosen comparison.
Conclusions: The methodology presented can be used to inform the future research agenda by indicating which parts of existing networks need further investigation. Through this process, unnecessary waste of research that leads to human and monetary cost may be considerably reduced.