Page 49 - PCC13

Heterogeneity in meta‐analysis
1. Was it a good idea to combine the trials?
2. Too much clinical heterogeneity to make sense?
3. Do the forest plots look consistent?
4. Do statistical tests suggest that heterogeneity is a problem?
Publication bias in meta‐analysis
• positive results:
– more likely to be published
– published rapidly
– in English
– more than one source (duplication)
– cited
• literature may be selective and biased subset of studies and outcomes
Page 43 of 84
PRECONGRESS COURSE 13 I BARCELONA, SPAIN – 1 JULY 2018 49