Year: 1998
Type: review
Abstract: Clinicians and patients make many decisions in situations where optimal treatment is uncertain. Despite well-published advantages of clinical trials for reducing such uncertainties, a trial may not be conducted because the sample size indicated by classical, hypothesis-testing methods is so large that no one institution could afford to sponsor the trial. By explicitly taking into consideration the costs and benefits of a trial, Bayesian statistical methods permit estimation of the value to a health care organization conducting a randomized trial instead of continuing to treat patients in the absence of more information. This paper describes a method for calculating the cost-benefit to a health care organization conducting a clinical trial, and the expected sample size to adequately resolve the uncertainties about which treatment is better. The method is illustrated in the case of a proposed clinical trial of a drug to prevent multiorgan system failure and death in patients admitted to the Stanford University surgical intensive care unit. This method should permit health care organizations to evaluate a proposed trial's expected cost-benefit and the expected sample size that will resolve the question of interest, and thereby assist in the process of deciding whether to conduct the trial.
Source: Controlled Clinical Trials
Authors: John Hornberger, Pirooz Eghtesady
Institutions: Stanford Medicine, Stanford University
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W2117365390