Video-on-Demand Library

Play

This content is restricted to members of PSI. If you are already a member please login. To join PSI or to see all the benefits of membership click here.

PSI Conference 2018: Evidence synthesis with limited studies: incorporating genuine prior information about between-study heterogeneity

05 June 2018

Background:

Meta-analyses using fixed effect and random effects models are commonly applied to synthesise evidence from randomised controlled trials in health technology assessment. The models differ in their assumptions and the interpretation of the results. Fixed effect models are often used because there are too few studies with which to estimate the between-study standard deviation from the data alone, but not that heterogeneity is unlikely to be expected.

Objectives:

The aim is to propose a framework for eliciting an informative prior distribution for the between-study standard deviation in a Bayesian random effects meta-analysis model to genuinely represent heterogeneity when data are sparse.

Methods:

We developed an elicitation method using external information such as empirical evidence and experts’ beliefs on the ‘range’ of treatment effects in order to infer the prior distribution for the between-study standard deviation. We also developed the method to be implemented in R.

Results:

The three-stage elicitation approach allows uncertainty to be represented by a genuine prior distribution to avoid making misleading inferences. It is flexible to what judgments an expert can provide, and is applicable to all common types of outcome measure.

Conclusions:

The choice between using a fixed effect or random effects meta-analysis model depends on the inferences required and not on the number of available studies. Our elicitation framework captures external evidence about heterogeneity and overcomes the often implausible assumption that studies are estimating the same treatment effect, thereby improving the quality of inferences in decision making.

Related Documents

Part of Collection(s)

Topic(s)

Upcoming Events

PSI Introduction to Industry Training (ITIT) Course - 2025/2026

An introductory course giving an overview of the pharmaceutical industry and the drug development process as a whole, aimed at those with 1-3 years' experience. It comprises of six 2-day sessions covering a range of topics including Research and Development, Toxicology, Data Management and the Role of a CRO, Clinical Trials, Reimbursement, and Marketing.

Joint PSI/EFSPI Visualisation SIG 'Wonderful Wednesday' Webinars

Our monthly webinar explores examples of innovative data visualisations relevant to our day to day work. Each month a new dataset is provided from a clinical trial or other relevant example, and participants are invited to submit a graphic that communicates interesting and relevant characteristics of the data.

One-day Event: Change Management for Moving to R/Open-Source

This one-day event focuses on the comprehensive management of transitioning to R/Open-Source, addressing the challenges and providing actionable insights. Attendees will participate in sessions covering essential topics such as training best practices, creating strategic plans, making the case to senior management, and managing both statistical and programming aspects of the transition.

PSI Vaccines SIG Webinar: Challenges in Estimating Vaccine Effectiveness Against Progression to Severe Disease

This is a free webinar on Challenges in Estimating Vaccine Effectiveness Against Progression to Severe Disease.

PSI Book Club - The Art of Explanation: How to Communicate with Clarity and Confidence

Develop your non-technical skills by reading The Art of Explanation by Ros Atkins and joining the Sept-Dec 2025 book club. You will be invited to join facilitated discussions of the concepts and ideas and apply skills from the book in-between sessions.