Evidence Synthesis Across Study Designs and Time Horizons

Wednesday, 16 December 2026, 15:00-16:15

Room: Zoom

Presenter: Seitidis Giorgos, University of Ioannina

Randomized controlled trials (RCTs) are considered the most reliable source of evidence for evaluating the relative efficacy of interventions. When several trials address the same question, a systematic review provides a transparent and rigorous way to identify and assess the relevant studies, and meta-analysis can combine their results when the studies are comparable. Well-conducted systematic reviews with metaanalysis of RCTs are typically placed at the top of the evidence pyramid and can inform recommendations by organizations such as the World Health Organization. However, the evidence from RCTs may be limited by narrow eligibility criteria or short followup, and studies at high risk of bias may inflate the pooled effect estimate. Nonrandomized evidence (NRE), though prone to confounding, can enhance RCT findings by providing evidence from routine practice, broader patient populations, or longer periods of follow-up. A robust meta-analysis framework is proposed that utilizes Bayesian variable selection techniques to account for potential bias arising from both RCTs and NRE. Studies included in a systematic review can also differ in how long they follow participants. For time-to-event outcomes, this creates a trade-off when choosing a single time horizon for meta-analysis. An early horizon can leave longerterm information unused. At a later horizon, some studies may have insufficient followup to contribute. A meta-analytic framework is proposed to jointly synthesize evidence from single-arm studies across multiple time horizons, accounting for between-study heterogeneity, correlation among time-points and study-level characteristics.

Zoom link: https://uoc-gr.zoom.us/j/88177475079?pwd=ik9vJRdyknP8xneS5NumlqgQdnplaz.1

Department Of Economics Website

myEcon Newsletter

Join the notification list of the Department of Economics.