Boxes of Xanax in a medication drawer.Share on Pinterest
Xanax is a medication that is often prescribed to treat anxiety. BSIP/Getty Images
  • Panic is an acute form of anxiety, and frequent panic attacks may signal a panic disorder, which is a type of anxiety disorder.
  • Xanax is a medication commonly prescribed for panic disorder in the US.
  • Findings from a recent study have suggested that Xanax for panic disorder may be less effective than previously thought.

About 4% of the world’s population currently experience an anxiety disorder.

Panic describes acute symptoms of anxiety, and panic attacks describe intense panic experienced for a short period of time. If they have recurring panic attacks, they may be diagnosed with a panic disorder.

The most common treatment for panic disorder is a combination of psychotherapy and medications, such as Xanax. Xanax (alprazolam) is the most commonly prescribed psychotropic medication in the U.S, as of 2023.

Now, a team of researchers from Oregon Health & Science University School of Medicine and Harvard Medical School have found that Xanax XR — the extended-release version of Xanax — may be less effective than previously thought for panic disorder.

They report that publication bias in medical journals has inflated the effectiveness of Xanax for panic disorder by more than 40%.

Their findings were recently published in the journal Psychological Medicine.

Dr. Erick Turner, professor of psychiatry at the Oregon Health & Science University School of Medicine, former U.S. FDA reviewer, and co-author of this study, told Medical News Today that over the years, he and his colleagues have looked at several other classes of psychotropic drugs — comparing study results in U.S. FDA reviews with those in journal articles — and found evidence for publication bias.

“More recently, I was approached by an OHSU medical student who was interested in working on a project with me,” he said.

“Knowing that benzodiazepines had not been looked at in this way, I suggested that we do so. However, because most of the benzodiazepines had been approved many decades ago (e.g. Valium in 1963), it was not possible to get access to FDA reviews for most of the drugs,” he continued.

“The review for the original Xanax, being approved in 1981, was also unavailable. However, the extended-release (XR) formulation was approved relatively recently (2003), so that FDA review was available to us,” Dr. Turner added.

For the study, researchers examined publicly available U.S. FDA data from phase 2 and 3 clinical trials conducted for Xanax XR for the treatment of panic disorder.

The researchers reported only five trials had been conducted, and only three had been published in medical journals. When the FDA reviewed the drug company’s clinical trial results on how well Xanax XR performed compared to a placebo, they determined that only one in five trials had a clearly positive outcome.

After conducting a statistical analysis of the combined results of all five studies, scientists reported that Xanax XR was still better than the placebo, but not as much as the published data conveyed.

They stated that publication bias inflated Xanax XR’s efficacy by more than 40%.

“Having seen this phenomenon with other drug classes, we half-expected to find some evidence for publication bias, but we did not know how much to expect,” Dr. Turner said. “I must say that there were more negative studies than I expected.”

“Having said that, the difference between the two ‘looks’ at the trial data — FDA vs journal articles — was less dramatic using meta-analysis compared to the ‘vote-counting’ approach of looking at the number of positive vs negative studies,” he added.

Medical journals publish the findings of scientific research that has been peer-reviewed, or reviewed by other experts in the same field that were not part of the study.

Papers published in medical journals include clinical trials, systematic reviews, meta-analyses, and narrative reviews.

Publication bias occurs when a study is not approved for publication based on the direction or strength of its findings. For example, studies that show positive results are published over those that show negative or statistically insignificant outcomes.

“With publication bias, aka selective publication, there is a process of picking and choosing which study results do and do not get published, with the result of making the drug look better,” Dr. Turner explained.

“If the results are flattering, in terms of how well the drug works or how safe it is, then of course that gets published. If, on the other hand, the results are unflattering, then those results might go unpublished, or they might be published using alternate statistical analyses that produce a more flattering result,” he continued.

After reviewing this study, Dr. Nathan A. Carroll, associate chief resident psychiatrist at Hackensack Meridian Health, said he was not surprised by its findings.

“There is always a concern that decision-making is influenced by the data available to us — a sort of availability bias. All too often, there can be an invisible hand on the scales regarding research and outcomes. This study’s results pull back the curtain on that invisible hand,” he explained to MNT.

Dr. Carroll said it’s essential for the general public to know about publication bias.

“Readers should be aware that there is a strong tendency to publish only positive results, and available information often influences decision-making. It could skew their decision-making if clinicians don’t have the full picture. However, identifying publication bias is difficult because we don’t know what we don’t know,” he said.

Dr. David Merrill, a geriatric psychiatrist and director of the Pacific Neuroscience Institute’s Pacific Brain Health Center in Santa Monica, CA, agreed, as these medical publications are used to persuade both doctors and patients.

“Readers should be concerned that if their doctors and themselves are only being shown [p]ositive data supporting the use of the drug, [b]ut then negative data says, well, wait a minute, maybe this is not as effective as we thought,” he said.

“That could be misleading and could affect the informed consent process where you want to have a genuine risk-benefit analysis,” he added.

When asked how researchers and medical journals can help remove publication bias, Dr. Turner said researchers who do systematic reviews of drugs should not rely so much on the published literature to give them the truth, the whole truth, and nothing but the truth.

“Wherever available, they should search for unpublished clinical trial data from regulatory agencies, such as the FDA,” he continued.

“Another big thing they can do, before they start their study, is to see if they can use the new ‘Registered Reports’ peer review model. Unfortunately, though there are over 300 journals that participate in this model, few of them, if any, publish in the area of clinical drug trials,” he said.

Dr. Merrill suggested clinical trials committing to having the data shared publicly in a published way right from the start.

“So a priori or ahead of time agreeing we’re going to do this trial and we’re going to release the outcomes and data no matter how it turns out,” he said.

“There has been publication bias not just in drug trials, but in science in general, where it is much easier to get ‘positive results’ published. If our hypothesis doesn’t pan out and the results are ‘negative’ — meaning that there’s no difference from placebo or there’s no benefit — quite often we see those works get rejected for publication because the scientific system and the medical system have been set up to reinforce positive results.”
— Dr. David Merrill

Three changes to limit bias

Dr. Carroll said three changes can help eliminate publication bias that will require the assistance of both researchers and publishers.

“First, publishers can commit researchers willing to publish negative results. Second, a database should be created so researchers can register their hypotheses. This will make it harder to change or cherry-pick results later. This would also be consistent with the study’s suggestion regarding ‘the value of regulatory data to the public,’” he said.

“Finally, there is the question of transparency and whether the industry has an ethical obligation to publish negative results,” Dr. Carroll added.