A new paper in the Deutsche Ärzteblatt argues that there should be alternatives to RCTs. The reasons are manifold

• selection bias towards more severely ill patients
• selection towards too homogeneous samples
• patients may decline participation
• physician may decline participation
• bias towards larger cities and universities
• usually “hard” endpoints that ignore quality of life, compliance, side effects
• usually only short time studies
• protocol may deviate from daily practice in medical routine

Community-based studies may therefore not be as bad, yea, yea.

Addendum

An extended and reworked version of this blog can be found in issue 45 of the “Deutsche Ärzteblatt“, page A3019, 10th November 2006.