Meta-analysis

1. Meta-analysis Dr Adam Tooley GPVTS ST2

2. Trials, systematic reviews and meta-analysis • “Does the new treatment confer significant benefits compared with conventional treatment?” • Often several trials asking the same question. • Often individual trials don’t show statistical clinical significance. • Combination of these results may reveal significant benefits of treatment. • E.g. Thrombolytic therapy for prevention of MI. • Meta-analysis now a hallmark of EBM. • Systematic reviews: • The heart of meta-analysis. • Find all published and unpublished research. • Present a balanced and impartial summary of existing research.

3. Benefits of meta-analysis • Overcoming bias • Analysis carried out on a rigorous systematic review offers an unbiased review of data. • Precision • Analysis of combined results from many trials has more poweR to detect small but clinically significant effects • Transparency • Demonstrate all decisions made throughout the process.

4. Conducting a meta-analysis • Location of studies: • Comprehensive search strategy which interrogates several electronic databases (MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials). • Hand searching key journals and checking reference lists of papers obtained. • Search strategy developed with care as a sequence of requirements for inclusion. • Quality assessment: • Decisions which studies to include. • Use explicit and objective criteria for inclusion or rejection based on quality e.g. only including RCT.

5. Conducting a meta-analysis • Calculating effect size: • Trials often present results as a frequency of some outcome in intervention and control groups. • Meta-analysis usually summarised as a ratio of one versus other (e.g. odds ratio or relative risk). • Findings combined using appropriate statistical method. • Different methods depending on outcome (OR, RR, HR or risk difference). • Checking for publication bias: • Trials with negative findings less likely to be published. • Examine a funnel plot.

6. Funnel plot

7. Conducting a meta-analysis • Sensitivity analysis • Inclusion and aggregation of data may affect main findings. • Affect on outcome by altering approach to aggregation. • With no sensitivity analyses reader has to guess on likely impact of these important factors.

8. Heterogeneity • Concern regarding meta-analysis is the extent to which they mix studies that are different in kind. • Difficulty is deciding which studies are combinable. • Studies can differ by types of patient (severity, comorbidity), intervention given and primary end point (death, disease, disability). • This can be tested for using statistical methods. • Depending on presence/absence of heterogeneity will influence the method of analysis of the meta-analysis data.

9. Presenting findings

10. limitations • Was the search strategy comprehensive and likely to avoid bias in the studies identified for inclusion • Was the publication bias assessed • Was the quality of the individual studies assessed using appropriate checklist of criteria • Was the combined effect size calculated using appropriate statistical methods • Was heterogeneity considered and tested for.

11. Summary • Statistical technique for combining the findings from independent studies. • Most often used to assess the clinical effectiveness of healthcare interventions • Combining data from two or more RCTs • Precise estimate of treatment effect giving due weight to the size of the trial • The validity of the analysis depends on the quality of the systematic review on which it is based • Good analyses: • Complete coverage of all relevant studies • Look for presence of heterogeniety • Assess robustness of findings using sensitivity analysis