|
In this module we look at issues related to the unit of analysis, including the incorporation of crossover trials and cluster randomized trials into a Cochrane review, or meta-analysis. The main aim of the module is to help you to recognize when unit of analysis errors can occur, rather than to learn how to deal with every eventuality. Many of the methods for dealing with unit of analysis errors are tricky, and if you do have to deal with these issues, you should ensure that you have access to appropriate methodological expertise.
Learning objectives
- Identify a variety of sources of unit-of-analysis errors
- Identify a crossover trial and understand when it is an appropriate design to use
- Understand that the correct analysis of a crossover trial is a paired analysis, in which within-patient differences are the focus
- Identify a cluster randomized trial
- Understand that the appropriate analysis of a cluster randomized trial should recognize that individuals are members of clusters
- Identify repeated outcome measurements, and be aware of strategies for incorporating them into meta-analyses
- Be aware that the inverse variance method for meta-analysis can be used for crossover, cluster randomized trials and trials with other designs
|
