Format

Send to

Choose Destination
J Clin Epidemiol. 2015 Sep;68(9):1036-45. doi: 10.1016/j.jclinepi.2014.06.001. Epub 2015 Jul 27.

Risk of bias of randomized trials over time.

Author information

1
Knowledge Management, Bioethics and Research Department, Pan American Health Organization 525 23rd St, NW, Washington DC 20037-2895, USA. Electronic address: reveizl@paho.org.
2
Knowledge Management, Bioethics and Research Department, Pan American Health Organization 525 23rd St, NW, Washington DC 20037-2895, USA.
3
Departamento de Salud Pública, Facultad de Medicina, Universidad de La Frontera, Claro Solar 115, Temuco, Chile.
4
Iberoamerican Cochrane Center-CIBER Epidemiología y Salud Pública (CIBERESP), Biomedical Research Institute Sant Pau (IIB Sant Pau). Sant Antoni M(a) Claret 167, 08025, Barcelona, España.
5
Service of Clinical Epidemiology, Hospital de la Santa Creu i Sant Pau, Biomedical Research Institute Sant Pau (IIB Sant Pau), CIBER Epidemiología y Salud Pública (CIBERESP), Universitat Autônoma de Barcelona Sant Antoni M(a) Claret 167, 08025, Barcelona, España.

Abstract

OBJECTIVES:

To determine the variation in the risk of bias (RoB) of randomized controlled trials (RCTs) in time.

STUDY DESIGN AND SETTING:

We reviewed all included RCTs from systematic reviews (SRs) published in the issue 12 (2012) of the Cochrane Databases of Systematic Reviews. We extracted the RoB author's evaluation per domain and other RCT characteristics. Multivariate logistic regression was used to evaluate association between the presence of a low RoB according to RoB domains and other characteristics.

RESULTS:

We included 1,732 RCTs from 97 SRs. The rates of RCTs judged as having low and high RoB significantly increased over time, whereas the rates of unclear RoB decreased for several domains. Increased rates of low RoB were consistent when considering the type of intervention (drugs vs. others), sample size, and country income level. Multivariate logistic regression shows that RCTs published between 2006 and 2012, compared with those published before 1990, were more likely to be considered at low RoB for sequence generation (odds ratio [OR] = 3.96; 95% confidence interval [CI]: 2.29, 6.87), allocation concealment (OR = 3.56; 95% CI: 1.96, 6.46), incomplete outcome data (objective outcomes; OR = 1.89; 95% CI: 1.13, 3.15), and selective reporting (OR = 4.14; 95% CI: 2.35, 7.29) domains.

CONCLUSION:

RCTs have improved reporting during the last decades decreasing the uncertainty for the RoB assessment.

KEYWORDS:

Methods; Quality improvement; Randomized controlled trials; Research design; Risk of bias; Systematic reviews

PMID:
26227423
DOI:
10.1016/j.jclinepi.2014.06.001
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Elsevier Science
Loading ...
Support Center