276 / 1983
Limits
• No control over previous work
•
selection bias
•
obsolete staging / treatment
•
inadequate evaluation of toxicity
• Trial exclusion = loss of information
• Individual data not available or complete
• Publication bias fully avoided ?