Format

Send to

Choose Destination
J Clin Epidemiol. 2015 Aug;68(8):944-9. doi: 10.1016/j.jclinepi.2015.03.007. Epub 2015 Mar 21.

Unbalanced rather than balanced randomized controlled trials are more often positive in favor of the new treatment: an exposed and nonexposed study.

Author information

1
INSERM U 1153, Hôtel-Dieu, 1 place du parvis Notre-Dame, 75004 Paris, France; INSERM CIC 1415, Hôpital Bretonneau, CHRU de Tours, 2 boulevard Tonnellé, 37044 Tours cedex 9, France; Université François-Rabelais de Tours, PRES Centre-Val de Loire Université, 60 rue du Plat d'Etain, 37020 Tours cedex 1, France; Département Universitaire de Médecine Générale, Université François-Rabelais de Tours, PRES Centre-Val de Loire Université, 10 boulevard Tonnellé, BP 3223, 37044 Tours cedex 1, France. Electronic address: clarisse.dibao-dina@univ-tours.fr.
2
INSERM U 1153, Hôtel-Dieu, 1 place du parvis Notre-Dame, 75004 Paris, France; INSERM CIC 1415, Hôpital Bretonneau, CHRU de Tours, 2 boulevard Tonnellé, 37044 Tours cedex 9, France; Université François-Rabelais de Tours, PRES Centre-Val de Loire Université, 60 rue du Plat d'Etain, 37020 Tours cedex 1, France; CHRU de Tours, 2 boulevard Tonnellé, 37044 Tours cedex 9, France.

Abstract

OBJECTIVES:

We aimed to assess whether the clinical equipoise principle is satisfied in unbalanced randomized controlled trials (RCTs) (i.e., with an unequal probability of subjects being allocated to one group than another).

STUDY DESIGN AND SETTING:

Observational and comparative study between unbalanced and balanced RCTs. We searched the "core clinical journals" of MEDLINE to identify reports of two-parallel group superiority unbalanced RCTs published between January 2009 and December 2010. For each unbalanced RCT, we identified a maximum of four reports (to maximize power) of matched balanced RCTs dealing with the same population. Our primary outcome was the proportion of positive RCTs [i.e., with statistically significant results for the primary outcome (P < 0.05), showing greater efficacy with the new treatment than the control treatment].

RESULTS:

We selected 46 reports of unbalanced RCTs and 164 of balanced RCTs; 65.2% unbalanced RCTs and 43.9% of balanced RCTs were positive (odds ratio, 2.38; 95% confidence interval: 1.23, 4.63). As compared with balanced RCTs, unbalanced RCTs were more often industry funded and their control treatments were more often inactive. Adjusting for these latter variables did not modify the results.

CONCLUSION:

This result questions the respect of clinical equipoise in unbalanced RCTs.

KEYWORDS:

Clinical equipoise; Ethics; Meta-study; P-value; Randomized controlled trials; Unbalanced randomization

PMID:
25892193
DOI:
10.1016/j.jclinepi.2015.03.007
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Elsevier Science
Loading ...
Support Center