A comparison of two worlds: How does Bayes hold up to the status quo for the analysis of clinical trials?

Pressman AR, Avins AL, Hubbard A, Satariano WA., Contemp Clin Trials. 32(4):561-8. doi: 10.1016/j.cct.2011.03.010. Epub 2011 Mar 29., 2011 Jul 01

Abstract

BACKGROUND: Modern clinical-research practice favors placebo controls over usual-care controls whenever a credible placebo exists. An unrecognized consequence of this preference is that clinicians are more limited in their ability to provide the benefits of the non-specific healing effects of placebos in clinical practice.

METHODS: We examined the issues in choosing between placebo and usual-care controls. We considered why placebo controls place constraints on clinicians and the trade-offs involved in the choice of control groups.

RESULTS: We find that, for certain studies, investigators should consider usual-care controls, even if an adequate placebo is available. Employing usual-care controls would be of greatest value for pragmatic trials evaluating treatments to improve clinical care and for which threats to internal validity can be adequately managed without a placebo-control condition.

CONCLUSIONS: Intentionally choosing usual-care controls, even when a satisfactory placebo exists, would allow clinicians to capture the value of non-specific therapeutic benefits that are common to all interventions. The result could be more effective, patient-centered care that makes the best use of both specific and non-specific benefits of medical interventions.

Pubmed Abstract

Pubmed AbstractOpens New Window