"The suggested framework can be summarised as follows: before planning a trial, the existing network of all competing treatments is considered, and its conclusiveness is statistically evaluated. If further trials are needed these are designed with an aim to render the existing evidence conclusive after considering patient’s preferences. The results from the new trial are used to update the existing systematic review.
This is described in the attached graph.
Methods underpinning the suggested framework differ from those in conventional network meta-analysis in two major aspects; a) inference about treatment effects is drawn considering the frequent updating of the evidence b) the need for further research and updating are expressed quantitatively via sample-size calculations for future trials. More details about the framework, can be found in the project’s webpage; an online document (
http://www.ispm.unibe.ch/research/research_groups/evidence_synthesis_methods/evidence_based_planning_of_clinical_research/index_eng.html#pane680818(opens in new window)) a video (
https://www.youtube.com/watch?v=0vOAK_YMx_Q#action=share(opens in new window)) several talks and lectures, ten published scientific articles and one submitted article. Below we outline the methods and results as presented in the most important developments and findings.
• In a technical article we explained the statistical models needed to continuously update the evidence (in the form of a network meta-analysis). In an empirical follow-up article we showed that continuously updated network meta-analysis (also called “living” network meta-analysis) is the most efficient way to answer clinical questions of policy importance. More specifically, network meta-analysis was 20% more likely to provide strong evidence against the hypothesis of treatment equality than pairwise meta-analysis; also it provided that evidence four years earlier than pairwise meta-analysis. Consequently, prospectively planned living network meta-analysis can facilitate timely recommendations and contribute to reduce research waste by providing strong evidence against the null hypothesis earlier than living pairwise meta-analysis.
• We developed and explained methods of conditional trial design that has the potential to reduce the resources needed to answer a clinical question of relevance to health-policy. We virtually designed a new study to resolve uncertainties about the efficacy of biologics in rheumatoid arthritis using data from an existing NMA as “historical data”. Reduced sample size and flexibility in the randomized arms included are the main advantages of the method. Using data from NMA resulted to reducing more than one third the required sample size compared with using data from pairwise meta-analysis.
• We conducted a survey of trial methodologists about their perceptions for NMA and their opinions about using NMA to design a new clinical trial. This survey found that the level of acceptance of network meta-analysis and it use on designing future studies is moderate. Three out of four participants of the survey were willing to definitely or possibly consider using NMA to design a new clinical trial. However, the major constraint in adopting our framework to plan future studies is the current paradigm of funders of clinical research and stakeholders as it represents a methodological frontier.
"
