Elsevier

Sleep Medicine

Volume 4, Issue 1, January 2003, Pages 57-62
Sleep Medicine

Original article
A meta-analysis of sleep changes associated with placebo in hypnotic clinical trials

https://doi.org/10.1016/s1389-9457(02)00242-3Get rights and content

Abstract

Objectives: The effects associated with placebo (EAP) have been incompletely described in clinical trials of insomnia treatment. We conducted a meta-analysis of insomnia medication trials for the purpose of estimating the magnitude of sleep EAP.

Method: We reviewed Medline for 1966 through 2000 for the meta-analysis. The subject heading of insomnia restricted to the subheading of drug therapy was crossed against the results of a search on the subjects heading placebo and text word placebo. We selected only papers that examined primary insomnia, incorporating both placebo and active medication therapies in a randomized, double-blind, parallel-group design. We required that results be reported for 1, 2, 3, or 4 weeks of treatment, and that outcomes be reported in hours/minutes.

Results: Five papers satisfied our requirements for eligibility, comprising 213 patients receiving placebo for a 2-week interval. Subjective sleep latency demonstrated a significant reduction (mean±S.E.) of 13.1±2.0 min (95% confidence interval (CI) 9.2, 17.0) for the placebo group after combining the data across studies. Subjective total sleep time demonstrated a significant increase of 13.5±5.4 min (95% CI 2.9, 24.0). Polysomnographic (PSG) sleep latency demonstrated a non-significant reduction of 2.5±4.3 min (95% CI −5.9, 10.9).

Conclusions: The confirmation of EAP in insomnia clinical trials argues for the retention of a placebo control in future insomnia clinical trials.

Introduction

The symptoms of many different psychological disorders improve during administration of placebo in the context of controlled clinical trials of psychotropic medication. This phenomenon is commonly termed ‘the placebo effect’, but it is unclear to what extent the change in symptom severity is due to the expectant hopefulness in consumption of the placebo, or whether the change is attributable to the passage of time, or non-specific effects of participation in a research project. In the interest of linguistic precision, we will use the term ‘placebo effect’ only to refer to the patient's expectation associated with placebo ingestion, and will use the more general term ‘effects associated with placebo’ (EAP) to describe the sum of all factors leading to symptom change during placebo administration. Also, although the term placebo has been applied to non-pharmaceutical, psychological interventions such as quasi-desensitization, this paper will define placebo as an inert substance designed for ingestion unless otherwise specified.

Although EAP have been described for most psychiatric disorders, including even more severe disorders such as bipolar disorders and schizophrenia [1], [2], there is surprisingly little information about EAP during insomnia treatment. If EAP exist at all, then it seems likely that EAP would occur during clinical trials for primary insomnia, given that the overall level of psychopathology is by definition less than that seen in psychiatric disorders [3], and less severe disorders might be more susceptible to EAP [4].

Conclusions are mixed regarding whether there are EAP associated with insomnia, with some authors deducing that EAP do exist, [5], [6], [7], [8] and others take the opposing view [9], [10], [11], [12]. Resolution to this issue is not trivial, as administration of placebo in clinical trials has both economic and ethical implications. We sought to address this question through combining the information provided in several clinical studies using meta-analytic techniques. Our goal is not to establish the clinical significance of placebo, but rather to establish whether there is any change associated with placebo that might serve to guide the design of future clinical trials.

Section snippets

Strategy

Inspection of recently published hypnotic clinical trials reveals that many study designs include up to 4 weeks of treatment, and that changes in key outcome variables are typically reported for the end of each week of treatment. The possible number of outcome variables is large including subjective sleep latency (SL), subjective wake after sleep onset (WASO), subjective number of awakenings, subjective total sleep time (TST), polysomnographic (PSG) SL, PSG WASO, PSG TST, PSG sleep stages, PSG

Results

Our combined search strategy produced 402 candidate papers, and the abstract of each paper was read to ascertain qualification for the meta-analysis. The full paper was inspected if the abstract did not clarify the qualifications for analysis. Many studies prior to 1990 were eliminated on the basis of reporting only qualitative outcomes (i.e. changes in SL reported as ‘better’, ‘worse’, or ‘no change’) or failure to use a parallel group design. Papers from 1990 through 2000 were more likely to

Discussion

Our selection strategy identified only a small number of eligible papers. Many candidate papers with otherwise excellent designs were excluded because the data of interest were not reported, or were reported only in figures. The small number of eligible papers had a profound effect on our original ambition of examining EAP for four variables at four follow-up time points. Instead, we presented findings for only three variables at one time point.

Our results found statistically significant EAP

References (34)

  • M.N. Musa

    Paroxetine for primary insomnia: possible placebo effect?

    J Clin Psychiatry

    (1999)
  • G. Hajak et al.

    Doxepin in the treatment of primary insomnia: a placebo-controlled, double-blind, polysomnographic study

    J Clin Psychiatry

    (2001)
  • E.O. Bixler et al.

    Methodological issues in pharmacological studies of sleep

  • K. Adam et al.

    Do placebos alter sleep?

    Br Med J

    (1976)
  • P.D. Nowell et al.

    Benzodiazepines and zolpidem for chronic insomnia: a meta-analysis of treatment efficacy

    J Am Med Assoc

    (1997)
  • A.M. Holbrook et al.

    Meta-analysis of benzodiazepine use in the treatment of insomnia

    Can Med Assoc J

    (2000)
  • J.C. Gillin et al.

    A comparison of nefazodone and fluoxetine on mood and on objective, subjective, and clinician-rated measures of sleep in depressed patients: a double-blind, 8-week clinical trial

    J Clin Psychiatry

    (1997)
  • Cited by (60)

    • Efficacy and safety of Zolpidem in the treatment of insomnia disorder for one month: a meta-analysis of a randomized controlled trial

      2021, Sleep Medicine
      Citation Excerpt :

      This was consistent with the findings of this study. However, McCall et al. found that the increase in subjective total sleep time and the decrease in sleep latency in the placebo group were clinically significant [32]. In addition, a meta-analysis found that after non-benzodiazepine hypnotic drugs and placebo treated adult insomnia, the polysomnography and subjective sleep latency of the two groups were significantly reduced, and the drug group changed more than the placebo group [33].

    • The influence of placebo administration on the first- night effect in patients with insomnia disorder

      2020, Sleep Medicine
      Citation Excerpt :

      A number of meta-analyses and studies have described the size of placebo-related effects in clinical trials of insomnia treatment. For example, two weeks of pharmacological placebo improved subjective sleep onset latency (SOL) and TST in insomnia patients with moderate to large effect sizes but had no significant effects on objective SOL [10–12]. Placebo administration for one week significantly reduced sleep difficulty compared with the control group [13].

    View all citing articles on Scopus
    View full text