Until more is known about the potential for unblinding, clinicians need to keep in mind that patients and physicians could often guess accurately who was getting placebo or active drug, first author Cody Bruggemeyer, MD, a resident at the Medical College of Wisconsin, Milwaukee, said in an interview.
Dr. Bruggemeyer
Dr. Cody Bruggemeyer
“It’s important that rheumatologists be aware of this potential issue and use their clinical reasoning and their ability to critically assess papers to evaluate the study design” of research on treatments, he said in an interview.
Dr. Bruggemeyer and coauthors at the Medical College of Wisconsin presented their assessment of the potential for unblinding in a Viewpoint article in The Lancet Rheumatology.
A sample of pivotal clinical trials
The authors selected a sample of pivotal studies of 14 commonly prescribed drugs for rheumatic conditions for which double-blind randomized controlled trials (RCTs) that compared the active ingredient with a placebo were available.
The 14 trials involved treatments classified as disease-modifying antirheumatic drugs (DMARDs), some of which were likely to produce side effects that placebos would not mimic, such as injection site and infusion reactions and difference in readings in lab reports, the authors wrote.
In their analysis, Dr. Bruggemeyer and colleagues evaluated discrepancies in the rates of adverse events reported between active drugs and placebos and classified the 14 studies as follows:
- High unblinding risk: Nine studies had a high estimated risk of unblinding, including trials of adalimumab with citrate (Humira), anakinra (Kineret), anifrolumab (Saphnelo), apremilast (Otezla), ixekizumab (Taltz), leflunomide (Arava), methotrexate, risankizumab (Skyrizi) and tofacitinib (Xeljanz).
- Moderate unblinding risk: Three studies had a moderate estimated risk of unblinding, including trials of azathioprine (Imuran), mycophenolate mofetil and tocilizumab (Actemra).
- Low unblinding risk: Two studies had a low estimated risk of unblinding. These involved tests of belimumab (Benlysta) and rituximab (Rituxan).
Many of the effectiveness measurements of treatments used in rheumatology depend on patients’ reports of relief of pain and other disease symptoms. For example, the widely used American College of Rheumatology 20% response for rheumatoid arthritis includes components that rely on patient and physician assessment of disease activity.
Unblinding risk to clinical trial validity
CTs are the highest level of evidence to establish efficacy, because the study design aims to mask whether the experimental treatment is a drug or placebo. In cases where patients and physicians are more likely to correctly detect use of an active drug, there can be biases that skew results toward reports of symptom improvement. Other patients’ views of their treatment may be distorted by accurate guesses that they have been given placebo, Dr. Bruggemeyer and coauthors wrote.
“The degree of these effects cannot be predicted, but they tend to erroneously inflate the perceived benefit of novel interventions,” they wrote.
The consequences of this unblinding may be minimal in cases where there’s a clear difference between the placebo and active drug, they said. As an example, they cited trials of interleukin-23 inhibitors for psoriasis, where skin clearance as measured by the Psoriatic Area and Severity Index 75 differed by more than 50% in absolute terms between the treatment and placebo groups.
But in other cases, there needs to be more attention paid to the potential role of unblinding, they wrote.
“Studies where effect sizes were small, contradictory, or dependent on subgroup analyses might be especially problematic, but commentary rarely reflects this issue or acknowledges the potential influence of unblinding,” they wrote.
In the paper, they call for more analysis of previous trials to look for unreported assessments of unblinding, while also asking that researchers consider surveying participants in future trials to evaluate the degree to which unblinding occurs.
“Advocacy from professional societies and the U.S. Food and Drug Administration itself might be necessary, but in the interim, rheumatologists should assume unblinding has occurred to some degree in most trials,” they wrote.