For decades, scientists have tried to understand why some people experience the placebo effect, a positive outcome from an inert drug or other inactive treatment. This is often attributed to an individual’s expectations that the treatment will work, among other factors. But, while it’s now well established that a patient’s expectations can affect treatment outcomes, how the social interaction between a healthcare provider and a patient might influence those outcomes, and in particular what role the provider’s own expectations about the treatment might play, is less well understood.
Now, new research led by Luke Chang, Dartmouth College, Hanover, US, shows that a provider’s expectations of treatment success can directly alter a patient’s experience of pain. This effect was at least partially transmitted to the patient socially, through the provider’s facial expressions. The researchers replicated these findings in two additional independent groups of study participants.
Marta Pecina, University of Pittsburgh, US, who studies the response to treatments and to placebo but was not involved with the new research, said the study is innovative “on many levels.”
“From the topic itself to the creative way the researchers used classical conditioning [a type of learning] to manipulate the doctor’s expectations, the study is very innovative,” said Pecina. “In addition, the study was well replicated. It offers new insights on how the ways doctors deliver treatments to patients could influence their effectiveness.”
“A lot remains to be done,” wrote Harald Walach, Poznan Medical University, Poland, and University Witten-Herdecke, Germany, in an accompanying News and Views, “but this study will surely be a landmark because it started to disentangle the question of how placebo effects might be transmitted. The answer is: socially, very likely through subtle facial cues.”
The study and related News and Views were published in the December 2019 issue of the journal Nature Human Behavior.
From doctor to patient
While training as a clinical psychologist, Chang said he was always curious why some patients showed dramatic improvement after only a few sessions of cognitive behavioral therapy while others would not respond at all.
“It’s clear that what doctors call ‘non-specific’ factors, which include things like aspects of the doctor-patient relationship and therapeutic expectancies, both on the part of the doctor and patient, can alter outcomes,” Chang explained. “But it’s hard to study these sorts of factors in a realistic way.”
Nearly 50 years ago, psychologist Robert Rosenthal, University of California, Riverside, US, demonstrated that social interactions played an unanticipated role in human studies—that an experimenter’s own expectations of what should happen in response to an experimental condition, whether it involved a response to medical treatment or performance on a cognitive task, shaped outcomes. Several studies over the past few decades also suggested that a healthcare provider’s expectations can also influence patient outcomes after treatment, said Pin-Hao Chen, first author of the new study.
“These studies found that patients’ pain experience is influenced by a healthcare provider’s expectancy of a medication’s benefit,” he explained. “But they didn’t answer the question of how providers reveal that expectancy and influence patients.”
To understand how placebo effects might be transmitted from doctor to patient, the researchers developed a unique experimental paradigm using a placebo-pain model. For the first phase of the study, they recruited 48 undergraduate participants and randomly assigned them to the role of “doctor” or “patient.”
The doctors then went through a classical conditioning phase where their expectancies regarding the supposed effectiveness of a placebo cream named thermedol were manipulated. Here, the investigators placed thermal electrodes on the doctors’ forearms to cause a painful burning sensation at 47° Celsius. They then applied the thermedol cream while simultaneously lowering the temperature of the electrodes to 43° Celsius, thus creating an association between the reduction in sensation and the cream.
However, while the doctors were told that thermedol provided relief of thermal pain, in reality this cream had no analgesic [pain-relieving] effects. Nonetheless, the doctors reported less thermal pain, in comparison to an identical control cream that also had no analgesic properties.
The camera doesn’t lie
In the second phase of the experiment, the doctor-patient interaction phase, both patients and doctors were fitted with special headsets that could record the facial expressions of the person with whom they were interacting.
Prior to meeting, the doctors were instructed to first apply the control cream to patients and then the thermedol, but not to reveal the differences between the two creams, while the patients received thermal arm stimulation at 47° Celsius across both treatments. The patients then reported their beliefs about the effectiveness of the creams as well as their pain levels. The patients reported lower pain levels and also found the doctors more empathic after receiving thermedol, compared to the control cream.
Further, the researchers used a neural network trained on the doctors’ facial expressions, such as brow lowering and nose wrinkling, which had been captured by the head-mounted cameras; this is an approach where computers are put to work to make sense of visual data. This model was able to predict how much pain the patients reported, with lower pain reported with the thermedol treatment, compared to the control treatment. In addition, both the patients as well as the doctors themselves showed fewer pain-related facial expressions during thermedol administration, compared to control treatment.
The researchers were able to replicate the results in a second sample of 86 participants, and then a third group of 60 participants.
Chen said he was surprised the results were so robust.
“When we started analyzing the data in the first study, we worried there may be other factors causing the effects we found,” he said. “But after running the second and third study, the findings strongly suggested that social cues like facial expressions might play a role in transmitting placebo effects.”
Still, it remains unclear what the facial cues were actually communicating to patients, wrote Walach in the News and Views.
“Whether this [subtle facial cues picked up by patients] just increases the patients’ own confidence, conveys information about what to expect, makes the doctors more empathic in the supposedly real treatment or has some other effect, we do not know. But we do know now that subtle cues are being transmitted by doctors and read by patients,” according to Walach.
Another key question, wrote Walach, is whether the results would apply to an actual clinical setting with real doctors and real patients.
Both Chen and Chang said it’s likely that facial expressions are only one of a variety of social cues that can transmit provider expectations—and, by extension, a placebo effect—to a patient. Chang added that he would like to do follow-on studies to look at other potential social cues, as well as look at brain activity in both doctors and patients as they interact, to further understand the social transmission of placebo effects.
“This study is the first demonstration that you can manipulate a provider’s expectations and manipulate the outcome of a treatment,” Chang said. “This provides a framework where we can start studying these social cues more carefully and better understand the underlying mechanisms, and do so in a setting that is closer to what you might actually find in a hospital or clinic.”
Pecina added that the study provides evidence that clinicians should be mindful of their expectations because these may play a significant role in overall treatment response in the patients they treat. And, the research has wider implications.
“This study is very relevant not only to medicine, but also to areas like education, where the beliefs of a teacher could influence outcomes,” she said. “There is a lot more that we need to understand and study. But this research is the first step to understanding how expectations and social cues can bring about more positive outcomes—and it is a strong first step.”
This story first appeared on the IASP Pain Research Forum and has been lightly edited for RELIEF.
Kayt Sukel is a freelance writer based outside Houston, Texas.