Summary: New analyses of both published and unpublished clinical trials show notable discrepancies in the apparent effectiveness of alprazolam extended-release (Xanax XR), a commonly prescribed benzodiazepine sedative. Researchers report that publication bias may have substantially overstated the drug’s benefit for panic disorder.
A review of the five randomized controlled trials submitted to the U.S. Food and Drug Administration (FDA) found that only one trial produced a clearly positive outcome. When the full set of FDA-reviewed data was combined using meta-analysis, the measured effect of alprazolam XR versus placebo was smaller than what published journal articles alone suggested. The investigators estimate that publication bias inflated reported efficacy by roughly 40%.
Erick Turner, M.D., professor of psychiatry at Oregon Health & Science University and a former FDA reviewer, who led the study, urges caution for prescribing alprazolam, particularly in patients who have not previously used benzodiazepines.
Key facts:
- Alprazolam extended-release (Xanax XR) is subject to re-evaluation: recent analyses suggest its benefit for panic disorder may be smaller than previously reported.
- Of five phase 2/3 trials reviewed by the FDA, only one produced a clearly positive result according to FDA assessments.
- Comparing published reports to FDA data indicates publication bias increased the apparent effect size by about 42% (an increase of 0.14 in Hedges’s g, from 0.33 to 0.47).
Background
Alprazolam, marketed in extended-release form as Xanax XR, belongs to the benzodiazepine class of medications, widely prescribed since the 1970s for anxiety, panic disorder and sleep disturbances. In addition to therapeutic effects, benzodiazepines are associated with well-established risks including dependence, withdrawal syndromes, impaired balance and cognition, and increased risk of falls—risks clinicians commonly weigh when deciding to prescribe these drugs.
While safety concerns have received substantial attention, Turner and colleagues argue that the evidence base on effectiveness has not been scrutinized as closely. Their paper, published in Psychological Medicine, examined public FDA review documents for all phase 2 and phase 3 efficacy trials of alprazolam XR for panic disorder and compared those trial results with what has appeared in the medical literature.
The FDA records show five trials were conducted. Only one of these trials met the FDA standard for a clearly positive outcome. Of the four trials that were not positive by FDA criteria, two were nonetheless published in a manner that conveyed a positive result, while the other two were not published at all. As a result, the published literature presented three trials—all portrayed as positive—creating a misleading picture of consistent efficacy.
By pooling the trial data using meta-analytic techniques, the research team calculated an effect size (Hedges’s g) of 0.33 (95% CI 0.07–0.60) using the complete FDA dataset, compared with an effect size of 0.47 (95% CI 0.30–0.65) when using the published literature alone. That difference corresponds to an approximate 42% inflation in apparent efficacy attributable to publication bias.
Turner notes the implications are particularly important for patients who have never previously used benzodiazepines: clinicians should carefully weigh whether to start a prescription, given the combination of smaller-than-expected benefit and well-known safety risks such as dependence and withdrawal.
About this psychopharmacology research
Author: Erik Robinson
Source: Oregon Health and Science University
Contact: Erik Robinson – Oregon Health and Science University
Image credit: Neuroscience News
Original research (open access): “Unpublished trials of alprazolam XR and their influence on its apparent efficacy for panic disorder” by Erick Turner et al., Psychological Medicine. The study compares FDA-reviewed phase 2/3 trial data for alprazolam extended-release against corresponding published reports to evaluate publication bias and its effect on estimated efficacy.
Abstract
Title: Unpublished trials of alprazolam XR and their influence on its apparent efficacy for panic disorder
Objective: To assess publication bias for alprazolam, the most widely prescribed benzodiazepine, by comparing efficacy estimates for panic disorder derived from published studies versus FDA review data.
Methods: The authors extracted results from all phase 2 and phase 3 efficacy trials of alprazolam extended-release for panic disorder as documented in FDA reviews. They searched bibliographic databases to identify matching journal publications. Publication bias was evaluated by contrasting (1) overall trial outcomes (positive vs. not positive) as recorded by the FDA with corresponding published reports, and (2) pooled effect sizes (Hedges’s g) calculated from FDA data versus published data.
Results: FDA documentation identified five trials, only one of which was positive by FDA criteria (20%). Of the four not-positive trials, two were published in ways that conveyed positive outcomes and two were not published. Consequently, the published literature suggested three trials had been conducted and that all were positive. The pooled effect size using FDA data was 0.33 (95% CI 0.07–0.60) versus 0.47 (95% CI 0.30–0.65) using published data, an absolute increase of 0.14—or approximately 42%—reflecting inflation from publication bias.
Conclusions: Publication bias substantially inflates the apparent efficacy of alprazolam XR for panic disorder. These findings support cautious consideration before initiating benzodiazepine therapy, especially in benzodiazepine-naive patients.