Evidence-based medicine considers randomized clinical trials (RCTs) to be the strongest form of evidence for clinical decision making. To test the hypothesis that RCTs have fewer methodological flaws than non-RCTs, limitations of 17,591 RCTs and 39,029 non-RCTs were characterized. Panels of experts assembled to write meta-analyses evaluated this literature to determine which articles should be included in 316 meta-analytic reviews. Overall, 38.7% of RCTs evaluated were excluded from review for an identified flaw. Commonly identified flaws in RCTs were as follows: insufficient data provided to evaluate the study (9.6% of 17,591 RCTs); inadequate randomization (9.0%); inadequate blinding (4.9%); and duplicative publication (4.4%). Overall, 20.2% of all published medical research has an identified methodological flaw, with RCTs having as many limitations as non-RCTs.
Keywords: meta-analysis; meta-analytic review; prospective study; retrospective study; study design.