Electronic Thesis and Dissertation Repository


Master of Science


Epidemiology and Biostatistics


Dr. Janet Martin


BACKGROUND: Evidence Reversal (ER) is the phenomenon whereby new and stronger evidence contradicts previously established evidence.

OBJECTIVES: To quantify evidence reversals and to determine characteristics associated with reversibility.

METHODS: Original articles from the New England Journal of Medicine (2000 to 2016) were screened for three inclusion criteria: tested a clinical practice; Randomized Controlled Trial design; and tested an established clinical practice. The proportion of RCTs that represented ER was determined. Association of trial characteristics with reversal was explored using logistic regression in order to inform a potential framework of reversibility.

RESULTS: In total, 611 RCTs met the inclusion criteria, of which 54% were evidence reversals. Based on variables associated with ER, a reversibility framework was proposed, comprised of eight trial characteristics.

CONCLUSION: More than 50% of RCTs published in the NEJM that test established practices are evidence reversals. The characteristics of RCTs that are associated with reversal will inform future research to further understand reversibility.