Doctors have long known that patients’ expectations have a significant impact on the effectiveness of treatments prescribed for them. But much less is known about the effect of providers’ expectations for those treatments.
Researchers at Dartmouth College recently devised an intriguing experiment to begin investigating these questions. They simulated clinical interactions in which “providers” (actually undergraduate students) administered what they’d been told were two different kinds of creams—one supposedly with analgesic properties, and one without—to “patients” (also students) who’d been subjected to mild thermal pain.
In reality, and unknown to the providers or patients, the creams were identical. The goal was to see how the doctors’ expectations of the effectiveness of both kinds of creams affected the level of pain relief patients experienced, and how those expectations were transmitted between doctor and patient.
Results of the study were published in 2019 in Nature Human Behaviour. Senior author Luke J. Chang, Ph.D., assistant professor of psychological and brain sciences at Dartmouth, spoke recently with Medical Economics about the study and what its results mean for primary care doctors.
Medical Economics: What was the purpose of your study?
Luke Chang: We’ve known for a long time that expectations matter in treatment effects. And to-date, most of the studies have been on the patient side, about how their beliefs influence their own outcomes. But we’ve known that the doctors’ expectations also really matter. And so basically we were interested in trying to figure out how strong these effects were. And there hadn’t really been any studies that have tried to causally manipulate doctors’ expectations.
ME: Describe the methods you used.
LC: We randomly assigned participants to play the role of “doctor” or “patient.” Some of the participants were students at the college who were getting research credit and some were just people around town, and they got paid.
Then we explained to the doctors that we were interested in studying simulated clinical interactions, and that they’d be administering a treatment to the patients. We showed them what looked like a real sort of medical-grade cream that we called Thermedol and it came out of tubes. And then we showed them another cream that came out of like a Vaseline jar, and we mixed it with food colorings so they could tell which was which.
We administered both of the creams to the doctors, and then we applied thermal heat pain to the doctor’s arm on top of each of the creams, and when we applied the heat to the Thermedol, we lowered the temperature, which it made it seem like the treatment was working from their perspective.
Then, after we’d manipulated their beliefs, we had them administer the creams to the patient. So they had to introduce themselves and then explain what the study was about with the treatment, and how the treatment works. Then they applied the creams and administered the thermal stimulation to the patient. But the key part was they couldn’t tell the patient which treatment was which.
The temperature was, on average, 47 degrees Celsius, which is about 112 degrees Fahrenheit. It feels basically like if you pour hot coffee on yourself. So it would turn your skin red afterwards but it wouldn’t blister or actually leave any permanent damage.
And we found that when the doctors believed that they were giving the real treatment, the patient reported experiencing less pain. And we did a couple of different versions, but it seemed in all of them the patients reported somewhere between a 15 to 30 percent decrease in the amount of pain they were experiencing.
Now, one kind of problem in the pain world is that we don’t really have good ways to measure how much pain someone’s experiencing outside of just them self-reporting it. So we also tried to use a couple other objective measures of pain. One is just overall physiological arousal, and for that we measured skin conductance while they were receiving the pain and found that the patient had lower arousal levels when the doctors believed the treatment was real.
The other way is to try to see if there’s any evidence of their behavior changes based on the manipulation. And what we did for that is, while they were undergoing this procedure, both the doctor and the patient wore video cameras.
They were basically just GoPros that were placed on this thing we developed that mounts on their heads so that we can record their expressions but not disrupt their interactions. Then we used computer vision techniques that turned the video feeds into predictions for which facial muscles were being expressed.
And then we tried to figure out, is there any combination of facial expressions that predicted how much pain people felt? We did this by training the model on the doctors when they were receiving the pain. And basically it was if you showed more eyes widening and clenching of your jaw, things like that, then that meant you were experiencing more pain.
And then we applied the model to the patients’ data while they were receiving the pain and found that the model thought that they were experiencing less pain when the doctors believed the treatment to be valid.
ME: How did you explain the use of the cameras to the doctors and the patients?
LC: They were told that we were interested in studying how the dynamics of clinical interactions between people affected one another.
ME: What was your reaction to the findings?
LC: It seemed like too good to be true in the beginning, how robust the findings were. But we did a bunch of different manipulations to rule out other hypotheses like habituation, or maybe it’s because of the cameras, or maybe because there was also an experimenter in the room to make sure that the pain was being delivered safely. All of it basically to make sure it was the doctors’ beliefs that were being transmitted.
Importantly, the patients perceived the doctors to be more empathetic when the doctors believed the treatment to be true. We don’t know exactly what they were doing differently. They could have been paying more attention to the patients, they could have been more caring, more eye contact, those are things we’re exploring more.
ME: What is the main takeaway of your study for primary care doctors? How can they use what you found to improve outcomes in their patients and their connections with patients?
LC: There’s not a ton of research on this, at least that I’m aware of, and the research findings are intriguing, but not fully compelling. There was some work by David Rakel showing that for the common cold when patients perceive their primary care providers to be more empathic, they had a slightly lower duration in their cold and slight changes in their white blood cell counts and other biological indicators.
So I think what that tells me, along with our results, is that finding ways in which you can show compassion and being empathetic are probably going to help, although we haven’t actually done those manipulations yet to see if that’s the mechanism for how these beliefs are being transmitted.
And then I think in general—and this is a bit speculative—but trying to instill hope and things like that, which would be trying to maximize these expectancy effects, could also be ways to augment treatments.
ME: What you’re talking about sounds a lot like bedside manner, and the importance of that has been known in medicine for a long time, correct?
LC: Yeah, but interestingly, there hasn’t been as much training on it. So we know that you’ll be sued less, you’ll probably get more treatment compliance. Those things have been documented, but it’s not like most students are trained in how to have better bedside manner. And I think it’s also really hard with how many patients physicians have to see to deliver that to everyone equally and effectively. But I think this provides some support that it’s worth spending more time thinking about how to do that.
The other thing is, and it sounds like it’s kind of deceptive, but [showing empathy] doesn’t always have to be genuine, as long as the patient believes it. People in sales know this. If you’re trying to build rapport, if you’re doing a business deal, there’s a whole bunch of training that they try to do to increase perceptions of trust and connection with people.
And so I do think that those things could be trained, and we’re not doing those manipulations yet. But we are trying to study just basic things about social interactions. So when people are having a conversation, what makes the conversation good, or what leads to social connection between people? And I hope eventually that we’ll be able to start studying this a little bit more and trying to do some manipulation to say like, if you do this thing, it will increase the trust that one person has in the other.