Ben Colagiuri, PhD, is an associate professor in the Faculty of Science and leader of the Placebo Research Network at the Charles Perkins Centre at the University of Sydney, Australia. His research focuses on placebo and nocebo effects, associative learning, and psycho-oncology. Colagiuri sat down with Lincoln Tracy, a research fellow from Monash University, Melbourne, Australia, at the 2019 Australian Pain Society Annual Scientific Meeting, which took place April 7-10, 2019, on the Gold Coast, Australia. He discussed his path to placebo and nocebo research, the development of a novel model for experimentally-induced nausea, and how the placebo research field has progressed over time. Below is an edited transcript of their conversation.
What first got you interested in science?
I have always been interested in psychology—in how people’s minds work and what affects their behavior. Being interested in people and wanting to understand why, from a scientific perspective, they do the things they do has been the main driver.
What got you interested in the placebo effect?
I was actually interested in anorexia and wanted to be a clinical psychologist. But at the start of my honors year I was allocated to a supervisor, Professor Robert Boakes, who had the placebo effect listed as one of his research areas. I thought that sounded quite interesting, and he gave me a book about it to read. That’s what really got me into it—I found it fascinating that placebos could cause side effects. After reading that book, I wanted to study the placebo effect.
Your lab has developed the world’s first model of placebo-induced nausea. What was your line of thinking in developing this kind of model?
By coincidence I was involved in psycho-oncology work through one of my PhD supervisors and was interested in expectancies causing nausea following chemotherapy. We were interested in doing some conditioning studies with nausea, but there weren’t any good models available, unlike the models available for experimental pain.
Most pre-existing models of nausea were essentially rotation based—such as getting people to stare at an optokinetic drum [a rotating instrument used to assess vision], or to rotate in a chair. However, the main limitation with these types of models is that they don’t have a placebo condition—people know when they aren’t rotating.
Veronica Quinn, my PhD student at the time, was really interested in the idea of an experimental model of nausea. Over a couple of years, we tried various things to find a model we were happy with. For example, we tried using nicotine gum to make people nauseous, but that didn’t really work. We briefly considered injecting people with drugs that made them nauseous, but that was going to be ethically challenging. We also thought about using virtual reality to put people on virtual roller coasters.
Eventually we came up with using galvanic vestibular stimulation [GVS]. It was a technique that people were using to study balance, visual perception, and stability that we knew had nausea as a side effect. We thought that could work really well as an actual model of nausea itself. But the critical thing about the GVS model is that we have a real nauseating stimulation and a placebo stimulation. I don’t think anyone is doing something like this.
We have also started using a model of virtual reality [VR] to cause nausea. Essentially, you can alter settings such as the lag between a person’s head movements and what they see on the screen. The early VR models had this problem, and people got nauseous. Although they have improved these issues with advances in VR technology, we are looking at using VR as a more fixed model of nausea compared to the older, rotation-based models.
In a previous study with Veronica you investigated why nocebo hyperalgesia persists longer than placebo analgesia. What were the findings from that study, and why are they important?
There is lots of evidence suggesting that nocebo hyperalgesia doesn’t extinguish. The evidence for placebo analgesia is a bit more mixed; I always find that it extinguishes, whereas other people have found a lack of extinction. But it seems to me that if you take all the evidence, placebo analgesia is more likely to extinguish than nocebo hyperalgesia.
We were interested in whether the persistent nocebo hyperalgesia had something to do with anxiety or autonomic arousal. We found that people with higher autonomic arousal failed to extinguish their nocebo hyperalgesia, suggesting that anticipatory anxiety might be prolonging nocebo effects.
This is very important, because if nocebo effects do genuinely fail to extinguish, that means an early negative experience in a clinic could set you up for a long-term nocebo effect. Understanding what is preventing the nocebo effects from extinguishing means that we can try to target those mechanisms. Hopefully this means that people don’t end up with recurring nocebo effects, for example, every time they go to the dentist or get an injection.
You also led a recent meta-analysis that assessed blinding in pharmacological trials for chronic pain. Why is this such an important area of research?
Double-blind trials are the cornerstone of evidence-based medicine; we rely on them for most of our decisions about whether a treatment is effective or not. We know from other areas, including general medicine and schizophrenia, that the rate of testing for blinding is low in clinical trials. This means that people are attempting to blind their patients as part of the trial, but at the end don’t actually test whether the patients were blinded or not. The general medicine and schizophrenia work showed that people weren’t testing for blinding very often at all, and when it was tested, that it often failed. We wanted to see what was happening in terms of blinding in chronic pain trials.
Our meta-analysis found that only about 5 to 8 percent of more than 400 chronic pain trials over the last 10 years actually tested for blinding. More surprisingly, on average the blinding failed in the trials that did assess blinding. Put simply, people can guess which treatment they’ve been allocated to. This obviously has implications for the validity of the trial. If people know what treatment they are taking, then have we really controlled for expectancies, placebo effects, and other types of patient bias?
However, it is important to realize that failed blinding doesn’t inherently mean a trial is invalid. If you compare a really effective treatment to a placebo, people in the active treatment group may notice that their pain has gone down or that they are feeling better, so of course they are going to believe that they are taking the active treatment. But we also know that side effects predict whether or not people believe they are on the active treatment. Therefore, blinding can fail due to reasons other than the drug itself being effective.
I would argue that we just need to assess blinding and see whether it has been maintained. If it has been maintained, we don’t have concerns about expectancy in the trial. If blinding hasn’t been maintained, then we should investigate why it failed. Did it fail because the treatment was effective and everyone in that group got better? If so, then that probably isn’t a threat to the trial’s validity. But did it fail because people had side effects? If so, then maybe there is a threat to the trial’s validity.
What other kinds of studies are you working on?
One thing we’re really interested in at the moment is how we can reduce nocebo side effects. We have run a few studies looking at latent inhibition, which is when you pre-expose a treatment or the treatment cues that are going to later be paired with something aversive. In the context of nocebo hyperalgesia, you have a cue paired with pain, and later that cue elicits increased pain.
We find that if you present the cue by itself prior to painful events, you get less nocebo hyperalgesia later. That is, the pre-exposure attenuates the nocebo conditioning. We are looking at doing these types of studies in clinical settings, both for pain and nausea.
We think that latent inhibition is one thing that could be used clinically because it is ethical. You aren’t deceiving anyone; you are just pre-exposing them. Imagine pre-exposing people to a dentist’s room before you then take them in there and perform some painful procedure. That could basically protect them from the nocebo conditioning and prevent later nocebo hyperalgesia.
How has the approach toward research on placebo and nocebo effects changed over the course of your career?
There has been a big shift recently to try to do more work in patient populations; there has been a lot of experimental research, but now there is more and more interest in seeing whether what we understand of placebo and nocebo effects in the lab actually translates to the clinic. This is probably making placebo research more “mainstream.”
Where would you like the field to be in five to 10 years?
One of the big remaining challenges is the lack of replication going on in placebo research. There are a number of prominent studies that we cite all the time that haven't been replicated. They may well hold up; there is no a priori reason to believe any particular study would fail to replicate. But science as a whole is having lots of problems with replication. This is something that we have as a field been slow to do anything about, so that’s an important goal. It just becomes challenging to do replications when these studies cost a lot of money, and the field keeps moving forward.
What are some of the biggest misconceptions people have about placebo and nocebo effects?
One of the biggest misconceptions is about the difference between what we would call a placebo response [any change in the placebo arm of a clinical trial] versus a placebo effect [the difference between the placebo response and the change in the no treatment group]. There are a lot of data from randomized controlled trials where people see a huge improvement in the placebo group—and therefore it is often assumed that this reflects a large placebo effect. However, when there isn’t a comparison group to provide a baseline against the placebo group, then the improvement could be due to natural history and disease progression, the Hawthorne effect [when people change their behavior because they know someone is watching them], or demand characteristics [when experimental results are biased because the experimenters’ expectancies create an implicit demand for participants to perform as expected]. But I think people are slowly becoming aware of this.
If you had new students coming to work with you on placebo and nocebo effects, is there a paper or book that you would recommend they read?
I have always really liked the Stewart-Williams and Podd review of the placebo effect because it does a really good job describing conditioning and its relationship to the placebo effect. I think it is very grounded in the evidence but also is a nice theoretical piece about how we should conceive of placebo effects with respect to expectancy versus conditioning.
Lincoln Tracy is a postdoctoral research fellow in the School of Public Health and Preventive Medicine at Monash University and a freelance writer from Melbourne, Australia. You can follow him on Twitter @lincolntracy.
The Blind Leading the Not-So-Blind: A Meta-Analysis of Blinding in Pharmacological Trials for Chronic Pain.
Colagiuri B, Sharpe L, Scott A
J Pain. 2019 May; 20(5):489-500.
Autonomic Arousal as a Mechanism of the Persistence of Nocebo Hyperalgesia.
Colagiuri B, Quinn VF
J Pain. 2018 May; 19(5):476-486.
Galvanic Vestibular Stimulation: A new model of placebo-induced nausea.
Quinn VF, MacDougall HG, Colagiuri B
J Psychosom Res. 2015 May; 78(5):484-8.
The placebo effect: dissolving the expectancy versus conditioning debate.
Stewart-Williams S, Podd J.
Psychol Bull. 2004 Mar; 130(2):324-40.