Conclusions can change following reanalysis of randomized clinical trial data

1. 35% of reanalyzed clinical trials demonstrated conclusions that differed from the conclusions reached by the original article.

2. No statistically significant association was noted between changes in article conclusions following reanalysis and the use of independent versus original study authors.

Evidence Rating Level: 2 (Good)           

Study Rundown: Randomized controlled trials (RCT) are considered the gold standard of medical research. This study aimed to identify those RCTs that underwent reanalysis and to characterize variations in analysis techniques, study authorship, and the effect of this reanalyzed data on trial conclusions. Strengths of this study include the thorough methodology used to identify reanalysis studies. While only 37 reanalyses were ultimately included in this paper, the authors evaluated 2948 studies for appropriateness for inclusion. The results of this study are limited by factors inherent to the journal publication process. As journals may have demonstrated higher interest in the publication of reanalysis studies that disputed original study findings rather than those that confirmed them, the true percentage of clinical trials demonstrating conclusions that differed from the conclusions reached by the original article may in fact be smaller than the percentage identified by this study. Overall, however, this study highlights the paucity of reanalysis studies in the medical literature and reignites the debate surrounding the availability of open access to raw study data.

Click to read the study, published today in JAMA

Click to read the accompanying editorial, published today in JAMA

Relevant Reading: Data: to share or not to share?

In-Depth [systematic review]: This study utilized MEDLINE search terminology to identify 2948 articles reporting reanalysis data. 37 reanalysis were ultimately included and were examined for differences in study characteristics, authorship, statistical analysis methods, and study conclusions. 13 of these studies (35%) demonstrated conclusions different from those reached by the original article. While only 5 reanalyses utilized a panel of authors that was completely different from those individuals involved in the original study, no statistically significant association was noted between changes in article conclusions following reanalysis and the use of independent versus original study authors (Odds Ratio 0.3, Confidence Interval, 0.04-2.11, P=0.32).

More from this author: No mortality benefit with bilateral mastectomy vs breast-conserving surgery, Intermittent vagal nerve block ineffective for weight loss in obesity, Erythropoietin linked with reduced risk of brain injury in preterm infants, Guided patient self-management of blood pressure medications effective, Collaborative model effective in treating teen depression

Image: PD

©2012-2014 2minutemedicine.com. All rights reserved. No works may be reproduced without expressed written consent from 2minutemedicine.com. Disclaimer: We present factual information directly from peer reviewed medical journals. No post should be construed as medical advice and is not intended as such by the authors, editors, staff or by 2minutemedicine.com. PLEASE SEE A HEALTHCARE PROVIDER IN YOUR AREA IF YOU SEEK MEDICAL ADVICE OF ANY SORT.