Bias in meta-epidemiological studies: an exchange of ideas

Workshop category: 

  • Investigating bias
Date and Location

Date: 

Sunday 16 September 2018 - 14:00 to 15:30

Location: 

Harris 1
Contact persons and facilitators

Contact person:

Bada Yang

Facilitators:

Yang B1, Ghannad M1, Olsen M1, Langendam M1, Leeflang M2
1 Amsterdam UMC, University of Amsterdam, Netherlands
2 Amsterdam UMC - University of Amsterdam, Netherlands

Acknowledgements:

Bossuyt P1
1 University of Amsterdam, Netherlands
Target audience

Target audience: 

Resarchers with an interest in research on research

Level of difficulty: 

Intermediate
Type of workshop

Type of workshop : 

Discussion
Abstract

Abstract:

Background:
Studies have been described in the literature that examine associations between study characteristics and predefined outcomes, e.g. the association between methodological flaws and the magnitude of treatment effect. These studies are essentially analogous to typical observational-epidemiological designs with patients, and are therefore referred to as meta-epidemiological (henceforth: meta-epi) studies. In a meta-epi study, a study is the unit of analysis. Meta-epidemiology provides empirical evidence for bias and informs the development of new research methods. Just like the studies that are subjects of meta-epidemiology, meta-epi studies are probably also susceptible to bias. However, guidance on how to critically appraise such studies is scarce. In this workshop, we aim to discuss with the participants what constitutes bias in meta-epi studies, in order to gain a clearer idea of which types of bias may exist, and to apply the newly obtained knowledge to critically appraise and to better conduct meta-epi studies.

Objectives:
1) To share knowledge and perspectives on what constitutes bias in meta-epidemiological studies and to discuss strategies to avoid any biases.
2) To identify practical challenges in conducting a meta-epidemiological study.

Description:
The primary target audience is anyone interested in research on research, regardless of level of experience with meta-epi studies. The workshop will start with an introduction to the concept of meta-epidemiology. We will also present preliminary data from a literature survey of meta-epi studies to illustrate the range of meta-epi study characteristics. This will be followed by a more interactive session, in which we will engage participants with questions using one or multiple formats, e.g. a round table discussion, a survey, an electronic voting session or a brainstorming session in smaller groups. Some text examples will be provided from existing meta-epi studies to facilitate discussion. A summary of the most important findings during the workshop will be shared with all participants by email after the workshop.

Relevance to patients and consumers: 

Bias in research can misinform clinical practice, potentially leading to adverse patient outcomes. Despite increasing awareness of bias, shortcomings in research methods are still common. Identification of these shortcomings is necessary to improve value of research to healthcare consumers. However the process of empirically identifying bias – often achieved by meta-epidemiological research (also called ‘research on research’) – is rarely scrutinized. Identifying something as bias when it isn’t, or vice versa, can lead to underutilization of good quality research or the incorporation of low-quality evidence in clinical practice.