Content area
Full Text
Introduction
Cluster Randomised Trials (CRTs) are an important research design in health services, quality improvement (QI) and knowledge translation research. Unlike individually randomised trials, in which individual subjects are randomly assigned to receive either an experimental or control intervention, CRTs randomly assign groups of individuals, such as medical practices, hospitals or entire communities. Some CRTs evaluate the effect of a therapeutic intervention directed at individual subjects, but assign subjects to trial arms in groups for a variety of logistical and methodological reasons or to avoid experimental contamination within groups. 1 Alternatively, some CRTs randomly assign healthcare professionals or organisations to trial arms to evaluate the effect of an intervention on the professional or healthcare organisation. These trials may evaluate patient outcomes as well as professional or organisation-level outcomes.
CRTs pose several challenges with respect to obtaining individual informed consent from study subjects. Several commentators, 1-3 as well as the UK Medical Research Council, 4 have suggested that the need to seek informed consent from cluster members depends on the type of experimental interventions being evaluated in a particular CRT. In studies with cluster-level experimental interventions, the interventions may be difficult for cluster members to avoid, effectively making refusal of consent meaningless. In such studies, it has been suggested that consent need not be sought. 1-4 Conversely, in studies with patient-level experimental interventions in which the CRT design is used for logistical reasons or to avoid treatment contamination, consent should be sought as it would be in individually randomised trials.
Other factors may also influence consent practices in CRTs. In trials randomising large clusters, such as entire communities, it may be logistically difficult to seek consent from all cluster members. 1 2 Some methodologists have expressed concern that the information disclosed during the consent process may bias the findings of CRTs of behavioural interventions. 5 6 Consent requirements for CRTs have been noted to vary between jurisdictions and ethics committees. 7 Uncertainty over whether some CRTs constitute research or QI 8 9 may have also led to variation in consent practices in CRTs that evaluate healthcare QI interventions.
In this paper, we document the frequency of reporting whether individual informed consent was sought from subjects in healthcare CRTs. Using multivariable logistic regression models, we examine whether...