Accelerating evidence synthesis for safety assessment through the platform: A large-scale feasibility analysis

Session Type
Evidence synthesis innovations and technology
Yang X1, Clark J2, Yu T3, Furuya-Kanamori L4, Lin L5, Xu C1
1School of Public Health, Anhui Medical University, China
2Institute for Evidence-Based Healthcare (IEBH), Faculty of Health Sciences and Medicine, Bond University, Australia
3Université Paris Cité, Research Center of Epidemiology and Statistics (CRESS-U1153), INSERM, Paris, France, France
4School of Public Health, Faculty of Medicine, The University of Queensland, Australia
5Department of Epidemiology and Biostatistics, Mel and Enid Zuckerman College of Public Health, University of Arizona, USA

Background: Systematic reviews, long considered the best way to answer questions in healthcare, are labor-intensive and time-consuming, making them impractical as a method to provide timely evidence for an urgent public pandemic. The platform provides a promising way to accelerate systematic review evidence production.
Objectives: In this study, we aim to investigate the feasibility of accelerating evidence synthesis practice for safety assessment through the platform.
Methods: We searched PubMed (from 1st Jan, 2015 to 31st Dec, 2020) for systematic reviews of safety using randomized controlled trials (RCT) of pharmacological interventions. This established a real-world dataset consisting of meta-analyses with five or more studies. The registration information of each trial from eligible reviews was further collected and verified to emulate rapid synthesis process through The meta-analytic data were then re-analyzed by using 1) the full meta-analytic data with all trials and 2) emulated rapid data with trials that had been registered and posted results on, under the same synthesis methods. The effect estimates of the full meta-analysis and rapid meta-analysis were compared in terms of the magnitude and direction of the effects, as well as the significance of P-value.
Results:The real world dataset consisted of 151 eligible systematic reviews, contaning 558 meta-analyses. Amongst these meta-analyses, 56 (10.0%) had none of their RCTs, therefore, they were not feasible for a rapid synthesis. For the remaining 502, a median proportion of 70.1% (interquartile range: 33.3% to 88.9%) of the RCTs were registered on and posted results per meta-analysis. Under a 20% tolerable bias, rapid synthesis through can achieve an accurate effect estimate in 77.4% to 83.1% of the situation and an accurate direction prediction in 91.0% to 95.3% of the situation.
Conclusions: Synthesising evidence through the platform may be a feasible rapid approach to producing evidence to support urgent decision-making, which could be considered in the future.
Patient, public and/or healthcare consumer involvement: None

Figure 1.pdf
Figure 2.pdf