Are There Scenarios When the Use of Non–Placebo-Control Groups in Experimental Trial Designs Increase Expected Value to Society?. (January 2016)

• Record Type: Journal Article
• Authors: Uyei, Jennifer; Braithwaite, R. Scott
• Is Part Of: Medical decision making. Volume 36:Number 1(2016)
• Page Start: 20
• Page End: 30
• Publication Date: 2016-01
• Subjects: placebo-controlled trial -- clinical trial -- RCT; Medical policy -- Periodicals; Clinical medicine -- Decision making -- Periodicals; Medicine -- Periodicals; Médecine clinique -- Prise de décision -- Périodiques; 362.1
• DOI: 10.1177/0272989X15584770
• Languages: English
• ISSNs: 0272-989X

Abstract

Background . Despite the benefits of the placebo-controlled trial design, it is limited by its inability to quantify total benefits and harms. Such trials, for example, are not designed to detect an intervention's placebo or nocebo effects, which if detected could alter the benefit-to-harm balance and change a decision to adopt or reject an intervention.Objective . In this article, we explore scenarios in which alternative experimental trial designs, which differ in the type of control used, influence expected value across a range of pretest assumptions and study sample sizes.Method . We developed a decision model to compare 3 trial designs and their implications for decision making: 2-arm placebo-controlled trial ("placebo-control"), 2-arm intervention v. do nothing trial ("null-control"), and an innovative 3-arm trial design: intervention v. do nothing v. placebo trial ("novel design"). Four scenarios were explored regarding particular attributes of a hypothetical intervention: 1) all benefits and no harm, 2) no biological effect, 3) only biological effects, and 4) surreptitious harm (no biological benefit or nocebo effect).Results . Scenario 1: When sample sizes were very small, the null-control was preferred, but as sample sizes increased, expected value of all 3 designs converged. Scenario 2: The null-control was preferred regardless of sample size when the ratio of placebo to nocebo effect was >1; otherwise, the placebo-control was preferred. Scenario 3: When sample size was very small, the placebo-control was preferred when benefits outweighed harms, but the novel design was preferred when harms outweighed benefits. Scenario 4: The placebo-control was preferred when harms outweighed placebo benefits; otherwise, preference went to the null-control.Limitations . Scenarios are hypothetical, study designs have not been tested in a real-world setting, blinding is not possible in all designs, and some may argue the novel design poses ethical concerns.Conclusions . We identified scenarios in which alternative experimental study designs would confer greater expected value than the placebo-controlled trial design. The likelihood and prevalence of such situations warrant further study.