0472 Do the Subjective Effects of Hypnotic Medications Result in Unblinding of Treatment Assignment in Hypnotic Randomized Clinical Trials?

Abstract

Introduction

Some hypnotic medications have obvious subjective effects, including therapeutic effects (i.e., anti-insomnia effects), and side effects (i.e., feelings of impairment/intoxication). Information is lacking regarding whether the subjective effects of hypnotics result in unblinding of treatment assignment (active drug versus placebo) in hypnotic randomized clinic trials (RCTs), thus undermining internal validity of study results. In response, we now report the ‘best guesses’ of clinical trial participants, versus study coordinators, versus study physicians in the study Reducing Suicidal Ideation Through Insomnia Treatment (REST-IT).

Methods

Eighty-nine of the 103 participants in the REST-IT study completed their ‘best guess’ regarding which randomized treatment they had been assigned. REST-IT was a blinded RCT, comparing zolpidem controlled release (CR) versus placebo at bedtime given over 8 weeks in adults with major depressive disorder who also had insomnia and suicidal ideation, and who also received open label fluoxetine. At the conclusion of study participation, the study participants, the study coordinators and the study physician each independently recorded their ‘best guess’ regarding which treatment arm the patient had been assigned. The study physicians and the study coordinators had access to the participants’ Insomnia Severity Index scores when the ‘best guess’ was made.

Results

Patients guessed correctly 58.4% of the time, coordinators 53.9% of the time, and physicians 49.4% of the time. The percentage guessed correctly did not differ significantly between groups. Physicians most often guessed placebo (56.2%) while study coordinators most often guessed zolpidem-CR (55.1%). Agreement between pairs of study participants with the study coordinators and the study physician was moderately high, with all kappa values 0.49- 0.57, and all kappa differences between zolpidem and placebo with p-values >0.8.

Conclusion

The blind was maintained in this RCT of zolpidem-CR versus placebo, especially for the physicians. Our results may not apply to other hypnotics or other RCT designs.

Support

NIMH MH095776, MH095780, MH95778