We need meta-analysis
• Non significant differences
•
low effect or low power ?
• Estimates of treatment effects
inconsistent / contradictory
• “Suboptimal” quality of trials
small sample size
analysis, reporting