Clinical Research: Phase 1 - Phase 4

The Placebo Problem in Pain Research: Keeping Up With the ‘Mrs. Joneses’

We were conducting a trial for a painful diabetic neuropathy (PDN) drug and were investigating why one site had an especially high placebo response rate. Then one of our representatives, waiting in the site’s lobby to meet with the principal investigator, pinpointed the likely reason when a patient walked in the door.

“Mrs. Jones,” the receptionist beamed, “you look fantastic today.” The compliment, however benign and well-intentioned, was exactly the kind of interaction sites need to avoid. Positive reinforcement from the staff can easily color a patient’s perception of how well the treatment is working and must be avoided. That’s especially true in pain trials, which are often undermined by high placebo response.

Numerous pain studies have failed not because the drugs were ineffective but because researchers couldn’t statistically separate placebo response from real therapeutic benefit. It’s a subject we addressed in our webcast Neuropathic Pain Studies: Patient Selection, Study Design, and Clinical Trial Challenges.

Pain trials are uniquely challenging because they rely so heavily on patients’ own assessments. While oncology trials measure tumor size and changes in blood chemistry, analgesic trials rely mostly on patient perceptions of how they feel, and that creates a wide range of obstacles. Most notably:

  • Patients like our “Mrs. Jones” are easily influenced by their environment and by the people surrounding them
  • Patients interpret pain scales and rating criteria differently
  • Encouraged by their acceptance in a trial, hopeful patients may convince themselves to perceive improvements that aren’t real
  • Patients suffering from chronic pain may enjoy the attention they’re receiving and subconsciously want to please the physicians and site staff by showing an encouraging response

How do we counteract this?

Site staff should maintain a clinical distance from patients and provide only neutral reinforcement when working with them. Patients themselves need to be trained on how to objectively evaluate their pain, and that’s something that should begin immediately after the consenting process.

Some patients may not even be familiar with the concept of placebo response, so defining it is a vital first step. Varying degrees of training, online and in person, can provide a deeper understanding of the issue and how to minimize its impact.

From the clinical side, maintaining the blind is essential. That may sound obvious, but it can be difficult to achieve. For one thing, it’s important to match the method of drug delivery: If you are administering pills (or capsules or patches or creams or injections) to patients receiving the study drug, you want to do the same for the placebo population. When that’s not possible, you should have an unblinded, independent administrator on site. That requires another physician-level person to administer the medication, especially if it’s an injectable.

It’s also important to avoid excipients that have distinctive scents, tastes, and physical sensations. We ran a trial of a topical treatment that had a menthol component, and the placebo had the same component, but we found that that the cooling sensation gave placebo patients the impression that the treatment was effective.

We’ve also worked with study medication that had a metallic aftertaste. The study was easily unblinded since 100 percent of the patients receiving study medication had this dysgusia while placebo patients did not. If the patients reported this adverse event to the staff or investigator, then it was clear that they received study medication and not placebo and they were unblinded.

Using a study design that includes a treatment washout during baseline as well as a flare score is important in many studies. These measures help ensure that we’re measuring patients’ true pain off medication and minimize the likelihood that they will falsely respond to placebo treatment. It’s also often advisable to limit the number of sites because placebo response can vary from one to another. By applying basic analytics after the study is unblinded, you will likely learn which sites are least susceptible to exaggerated placebo response.

Researchers will always need to account for the placebo effect when studying indications as subjective as neuropathic pain, but there are ways to increase the likelihood of achieving objective trial outcomes. For more on this subject, check out our neuropathic pain webcast.