• English
  • Español

Meta-analysis of continuous outcomes: helping review authors negotiate the change score versus final value debate

Topic category Statistical methods and meta-analysis
Date and Location
Saturday 22 October 2011 - 14:00 - 15:30
Methods Group
Methods Group: 
Statistical Methods Group
Contact person
Contact person: 
Joanne McKenzie (Contact this person)
First nameLast nameAffiliation and Country
First name: 
Last name: 
Affiliation and Country: 
School of Public Health and Preventive Medicine, Monash University, Australia
First name: 
Last name: 
Affiliation and Country: 
Centre for Statistics in Medicine, University of Oxford, United Kingdom
Target audience
Target audience: 
Review authors
Is your workshop restricted to a specific audience or open to all Colloquium participants?: 
Level of knowledge required: 
Type of workshop
Type of workshop: 
Abstract text
In this workshop we will illustrate, discuss, and provide guidance on issues associated with meta-analysing intervention effects estimated from various analyses including: (i) analysis of change from baseline scores, (ii) analysis of final values, and (iii) analysis of final values adjusted for baseline values using regression.

Continuous data are commonly measured in randomised trials pre and post intervention. When this occurs, several choices are available to analyse the data. Commonly trialists will choose to analyse the final values, change scores, or the final values adjusted for baseline values using regression. Generally only one type of analysis will be reported in the trial publications. Less frequently, an alternative analysis will be provided, or enough information to allow an alternative analysis to be performed. Across the set of trials, the analytical approaches are likely to differ (e.g. a mix of analyses of final values and change scores), which provides many challenges for review authors.

In this workshop we present worked examples to highlight the issues, provide a hierarchy of potential solutions (discussing the assumptions and limitations), and discuss the importance of pre-specification of an approach in the protocol. Opportunities will be available to discuss specific examples and for general discussion.