When testing drugs for pain relief, the information doctors give matters almost as much as the medicine subjects receive, according to a study published January 8 in Science Translational Medicine. Led by Rami Burstein and Ted Kaptchuk, both at Harvard Medical School in Boston, US, the study explored how labeling of pills affected pain relief in people experiencing migraine headaches. They found that mislabeling a placebo pill as Maxalt, a powerful migraine drug, produced pain relief similar to that obtained by mislabeling a true Maxalt pill as placebo.
“What we told them was almost as powerful as the drug itself,” Burstein told PRF. Burstein is a migraine researcher who teamed up with Kaptchuk, who studies placebo, to see if there was anything they could do or say to patients to increase the benefits of migraine medication.
Previous studies have picked up on an influence of messaging that manipulates subjects’ expectations, with information given by a clinician producing self-fulfilling results. When patients receive positive information about the benefits of a treatment, a greater therapeutic effect results, whereas negative information dilutes real therapeutic effects (Benedetti, 2013). This not only works for placebos (Pollo et al., 2001), but also for active drugs (e.g., Schenk et al., 2013, and PRF related news story).
The new study had a design that allowed, for the first time, a systematic examination of different information regarding a pill’s action on both placebo and drug responses within the same patients. The labeling of pill packaging was designed to vary patient expectations from negative to neutral to positive.
“It’s a brilliant study because they were able to compare the relative importance of the message versus the pharmacology of the drug,” said Nathaniel Katz, president of Analgesic Solutions, a company in Natick, Massachusetts, that conducts clinical trials of pain drugs. Katz was not involved in the study.
The size of expectation's effect complicates comparisons of results from different clinical trials, which do not typically control for expectation.
But that is beginning to change, Katz said. For example, his company is starting to track expectation in their trials, or to standardize it across participants with training (see PRF related news story). The Food and Drug Administration is also starting to pay attention to expectation as an important variable in clinical trials, he said.
The outpatient setting of the new study also suggests that the effect of expectation extends to real-world clinical trial settings, rather than just being a laboratory-based curiosity, he added.
“This particular study is one example of a whole new emerging science of clinical trials that will help us design smarter, more efficient, and more informative trials,” Katz said.
Great expectations
First author Slavenka Kam-Hansen and colleagues studied 66 people who suffered from repeated migraine headaches. To get a baseline for pain intensity, each subject left a first headache untreated for two and a half hours. For the next six headaches, subjects rated their pain 30 minutes after migraine onset, then took a pill—either placebo or rizatriptan (Maxalt)—from an envelope labeled with negative information (“placebo”), neutral information (“placebo or Maxalt”), or positive information (“Maxalt”) about the pill’s active ingredient. Two and a half hours later, participants rated their pain again. If it had not subsided, they could take a rescue dose of medicine. Each participant experienced all six conditions, taking placebo and Maxalt each under three different conditions of information.
For both placebo and Maxalt, information about the pill modulated pain relief. For example, when placebo was mislabeled as “Maxalt” or “placebo or Maxalt,” it gave greater pain relief than when it was labeled as “placebo.” A similar effect transpired for Maxalt, in which labeling it as “Maxalt” or “placebo or Maxalt” relieved pain more than when Maxalt was mislabeled as “placebo.”
Positive information also boosted placebo responses into the realm of Maxalt: Placebo labeled as “Maxalt” relieved pain nearly as well as Maxalt labeled as “placebo.” This was also true for the percentage of patients reporting they were pain free after 2.5 hours in these conditions.
But positive information had its limits, as it could not increase pain relief more than the neutral condition, in which pills were labeled as “placebo or Maxalt.”
“To our surprise, we could not increase the effectiveness of the drug by giving a lot of positive information over the uncertain condition,” Burstein said.
The placebo gave potent pain relief—about half that as Maxalt—under each expectation condition. Even when placebo was accurately labeled as “placebo,” pain scores decreased by 14.5 percent relative to untreated headaches. This adds to Kaptchuk’s previous findings of benefit from open-label placebo for irritable bowel syndrome (Kaptchuk et al., 2010) and depression (Kelley et al., 2012). Burstein speculates this has to do with some kind of learned physiological responses to the act of pill taking that develop over a person’s life.
The results also suggest that finding the optimal messaging to extract a drug’s true effect could be a challenge for clinical trials of other drugs. Being too upbeat could create a potent placebo effect that narrows the difference between placebo and drug; aiming too low could neutralize the effects of powerful drugs.
“We probably want the patients’ expectation levels to be in a middle zone,” Katz said. “If that’s true, then how do we get there?”
Manipulating expectations could invite ethical problems, too. For example, clinicians may choose to withhold information about a drug from patients in order to increase its effectiveness. “Do patients have the right to know everything about a drug?” Burstein asked. “This is a question for medical ethicists, not me.”
Michele Solis is a science writer and former neuroscientist who lives in Seattle, Washington, US.
Image: © Roman Rodionov | Dreamstime.com
Comments on Related Content
Fabrizio Benedetti, University of Turin Medical School; National Institute of Neuroscience
This is an interesting and
This is an interesting and well-conducted study. The methodology is sound and the interpretation of the data appropriate. It is a confirmation of previous studies, with some interesting methodological improvements. In their Discussion the authors say “Our study improved on this previous study [Pollo et al., 2011] by using a within-subjects design…” Indeed, we performed a similar study in 2011 by using a between-subjects design, in which three different verbal suggestions were given to three different groups of patients. In addition, our study was on acute pain (postoperative pain), whereas the present study is on chronic pain (migraine).