This lecture was given by Tomas Furmark, Professor of Psychology at Uppsala University. My rough notes are below. Note that all of this, including the question/answer sections, is paraphrased - not exact quotes.
Furmark has worked with comparison between psychological and pharmacological treatment of anxiety disorders, studying similarities and differences between antidepressant drugs and psychotherapies. At some point, his team started comparing placebo responders and non-responders, and how those brain responses are different.
Expectancy Effect and placebo
Today, “placebo” refers to a substance or procedure that lacks specific activity for the condition being treated.
Placebos can be active or inert.
An active placebo is a placebo that produces noticeable side effects that may convince the person being treated that they are receiving a legitimate treatment. It is very seldom used - but should be used more, as not using it can ruin blinding.
An inert placebo has no effect.
Why do drugs work?
How much improvement after taking a drug is due to the active ingredient (pharmacodynamic effect) vs placebo effect?
Irving Kirsch has conducted a meta-analysis of SSRIs. According to him, inactive placebos explain 75% of antidepressant effects in clinical trials. Active placebos explain the remaining 25%. Suggesting SSRIs are active placebos.
Kirsch argues that based on side effects participants can understand if they get an active drug or a placebo, and that will result in different expectancies.
[L: I clarified whether Kirsch bases his 75%/25% conclusions on anything other than 'the participants can guess if they're taking a placebo', and apparently there are other factors taken into accounnt - not just this one.]Later in the lecture it was mentioned that difference between placebo and medication could only be distinguished by those who have taken antidepressants before - not those who are new to them.
Psychotherapy
There’s a technique vs relation debate in psychotherapy. Common factors (relation, expectancies, etc.) explain up to 70% of the variation in therapy outcome. According to some, specific technique (eg CBT, IFS, etc) explains only about 1% of the variation.
Question: Are there any studies that show therapy works less on people that don’t believe in therapy?
Answer: Yes. There is data suggesting rather convincingly that how much you like and believe in a specific type of treatment is associated with outcome. if you are a fan of psychodynamic therapy and not CBT that could impact the outcome.
Question [Liza]: Could a preference to type of therapy entail a certain psychological makeup for which that therapy is actually more effective rather than just a purely arbitrary bias? I.e., maybe if Anne prefers CBT it is because CBT is going to be more effective for her system, not because she’s just arbitrarily biased.
Answer: Yes, that could be the case. And it is not only the technique but the relationship between patient and therapist. If you don’t have a good relation it’s unlikely that the outcome will be that good.
[L: It is hard for me to relate placebo effect with psychotherapy as both entail primarily mental processes. Is it placebo if it is _supposed_ to be a mental process to be effective?]Types of placebo
Placebo is not limited to pills - placebo surgery, placebo creams, placebo injections, etc exist. Effects can be very long-lasting. In Parkinson’s disease, after 1.5 years sham vs real surgery could not be discriminated. Only after 2 years did differences appear.
The effects are not limited to subjective measures or self-reports. For example, improved motor performance and increased dopamine release was observed in Parkinson disorder after placebo administration.
Placebo response in different disorders
- Major depression tends to have a strong placebo response compared to psychosis or OCD
- Placebo surgery response: obesity tends to have strong placebo response. Pain surgery tends to have low placebo response.
Nocebo
When you expect something bad to happen, you can be affected by it emotionally and physiologically. Expectancy effect can be very strong.
- A person expects something bad or harmful will occur, and then experiences some or all of those effects.
- Are side effects of drugs caused by drug or by expectations?
- 26% of patients allocated to placebo group drop out due to side effects
History of placebo
In some ways, the history of medicine is the history of placebo. Ancient therapies were built upon belief systems and social influence.
- Huan-Ti (emperor in China about 2700 BC): more than 2000 drugs were known at the time.
- Sumerian/Babylonian sources: >600 drugs were used in ancient India
- Most, if not all, were likely placebos with a few exceptions
- Michel de Montaigne (1572) “There are men on whom the mere sight of medicine is operative”
The first known demonstration of a plcebo
John Haygarth, British physician: one of the first known demonstrations of a placebo in 1799
- Fake wooden pointers were compared to expensive metal “Perkins tractors” (used to ‘draw out’ diseases like rheumatism).
- Haygarth showed effects could be achieved with fakes vs expensive metal versions
- Book: Of the imagination, as a cause and as a cure of disorders of the body
More history
- 1700–1800: placebo referred to substances used for calming or pleasing (rather than healing) the patient
- William Cullen (1772): talked about placebo in clinical lectures
- T. C. Graves (The Lancet 1920): referred to “placebo effect” to explain certain treatment effects “in those cases where a real psychotherapeutic effect appears to have been produced”
- The notion that psychotherapy is placebo cannot be entirely ruled out - but maybe placebo is psychotherapy.
- Henry K. Beecher (1904–1976) - American anesthesiologist.
- Observed WWII soldiers who underwent surgery without morphine
- Just telling them they’d receive morphine was sufficient to get some pain relief
- Placebo can provide objective physiological effect
Open-hidden paradigm
You can administer painkillers with open (which is standard) or hidden applications.
Open applications: physician administers painkiller at given time points and patient knows for sure now they’re getting the medication.
Hidden application: pain medication is administered via a computer-controlled infusion pump at random time points during the day. The patient does not know when they get the painkiller - just that they get it.
Open administration adds an active component (the relation). Subjective pain relief with open administration is substantially greater compared to hidden administration.
In individuals with postoperative pain Levin & Gordon (1984) showed that open intravenous injections of saline as a placebo gained the same effect as 6–8 mg of morphine administered as a hidden intravenous injection. In other words, just telling the patient that a pain-relieving drug is injected was as potent as giving 6–8 mg of morphine.
Inducing expectancies
- Verbal instructions - information given to patient
- Therapeutic rituals
- Social contact
Example - same treatment, different information
Lecturer described a study in which they treated patients with an SSRI for 9 weeks. They randomly allocated to 1 of 2 groups: receiving correct or incorrect information.
The group receiving correct info were told they’d be treated with an effective SSRI for 9 weeks.
The group receiving incorrect info were told they’d be treated with an active placebo for 9 weeks, which has the same side effects as the real thing but is ineffective for the target issue. So the patients were asked to take a placebo for 9 weeks and then have their brains scanned with fMRI and PET.
After 9 weeks all participants were offered CBT treatment for the condition of social anxiety.
2 groups treated with same drug. Expectancies were the only difference.
Result:
- Percentage “responders”:
- Overt SSRI: 50% (12 of 24)
- Covert SSRI: 14% (3 of 22)
- Reduced social anxiety
- Better effect for overt SSRI
- The difference between the two groups is an estimate of the placebo effect
Brain functioning was observed using PET before and after treatment (scanned dopamine and serotonin transporters)
- Occupancy of serotonin transporter was identical. Same effect on serotonin system in both groups.
- But dopamine transporters were clearly different between two groups.
- So in overt SSRI group, participants had an increased release of dopamine that occupied dopamine transporters.
Placebo effect accounted for at least 50% of SSRI outcome. The mechanism underlying improvement seems to have a lot to do with dopamine and not so much with serotonin.
Labelling of medication
A group of patients with migraine headaches were treated with Rizatriptan. The efficacy of Rizatriptan labelled as placebo, and placebo labelled as Rizatriptan, was the same: not very effective. Correctly labelled Rizatriptan improved the beneficial effect by about 50%.
The Honest Placebo
The honest placebo refers to using placebo without deception. A study of IBS patients compared placebo + quality interactions vs no treatment + quality interactions, for a 3-week duration.
Patients knew they were receiving placebo, and all received a 15-minute lecture on the powerful placebo and how beneficial it can be. They were exposed to messages like ‘You just need to take it regularly’ and ‘It will work over time’
- 59% of patients on the placebo reported substantial improvement
- This amount to an even better effect than previously shown for the IBS medication alosetron.
Open-label placebos have seen effect in IBS, back pain, cancer-related fatigue, ADHD, allergic rhinitis, major depression, menopausal hot flashes…
Question [Liza]: How far can we push the border of the placebo effect? If we can see physical effects of placebo, what is the line between placebo effects for cancer-related fatigue and placebo effects for cancer?
Answer: Even though placebos can’t be used to treat cancer, it can still have an impact on rehab after surgery and potentially have an impact on to what extent and for how long a patient survives. We have been surprised many times that placebos work when they shouldn’t. Maybe we can push the limits even further than we thought was possible. More research is needed.
Commment: An article in The Lancet doesn’t show a specific boundary but takes Alzheimer’s as example. It can clearly affect your quality of life positively but sooner or later the degenerative process of the brain will still be the same. So it seems more of a quality-of-life effect rather than affecting disease itself.
Answer: Yes. And while most placebo research concerns subjective measures - wellbeing, pain, anxiety, etc. But it is not limited to that only. You can see an effect on some objective measures as well.
Inducing expectancies
Therapeutic rituals
- Taking a pill
- Injection
- Surgery
- Using a medical device
Different rituals, different effects
- Knee osteoarthritis: meta-analysis 149 trials
- Analgesic effect size differed significantly across different placebo formats
- Placebo injection and placebo cream > oral placebo
- Injected placebo was more effective than paracetamol (!)
Changing the features of the placecbo
- Colored > white placebo pills
- Blue = calming/relaxing
- Red = activating
- Brown = laxative
- Big pills > small pills
- Taking several pills each time > taking just one pill
- Expensive > cheap pills
Inducing expectancies in doctor-patient relation
Medical treatment is usually done in a psychosocial context. Good clinical care - attention, support, understanding, friendliness, respect, listening, etc induces expectancies. That supportive care affects placebo response.
- The number of visits can have an impact. In a study on late-life depression (Rutherford et al. 2014), increasing the number of visits to the clinic led to significantly increased placebo (but not antidepressant) response. Is the difference between drug and placebo larger in trials with fewer visits?
How may expectancies change us?
- Cognitive-emotional changes: increased optimism, hope, remoralization, positive thoughts regarding treatment (calming/comforting)
- Behavior changes: increased moderation, activity level and approach behavior, decreased avoidance.
- Increased adherence to taking medication/increased treatment adherence
- Increased therapeutic alliance
- Biology: brain top-down processes; “reward-expectancy” - dopamine release, opioids
- Placebo: increased release of dopamine, e.g. in the nucleus accumbens - the brain’s reward system.
Drugs have a pharmacodynamic response. Placebo, on the other hand, has a “meaning response” ascribed by the brain. Which leads to expectations, hope and desire to be helped, conditioning processes.. which then leads to a physiological placebo effect.
Other unspecific effects than placebo
There are other reasons people improve that do not have to do with placebos.
Internal validity
To what extent can alternative causes of the results be excluded? e.g. other factors underlying improvement rather than the evaluated treatment, and rather than placebo.
Threats to internal validity
- Maturation - naturally occurring change processes or development within individual over time. This can lead to spontaneous remission or the patient state changing over the natural course of the disease.
- History: Uncontrolled life events that occur during the study may influence results (e.g., a high stress world event)
- Regression to the mean: Extreme values tend to change towards the mean with repeated testing.
- Instrumentation: Measured values change due to variable precision or calibration changes in the instrument over time
Randomization helps this, but does not take care of all potential problems. e.g. subject and experimenter effects that can occur regardless of randomization.
We also have to keep in mind the Hawthorne effect: participants behave differently when observed.
Unwanted effects on subjects
Response biases, such as conformity, “yea-saying”, demand characteristics (adjust answer according to perceived aim/hypothesis).
Expectancy effects
- Expect improvement - report improvement
- Expect side effects - report side effects
- Placebo and nocebo - special cases of S&E effects?
In researchers
Expectancy effects
- Confirmation bias
- “See what you want to see”
- “Find what you want to find”
- Allegiance effect in psychotherapy research. I a therapist or a group develops a new treatment and themselves evaluate the treatment in a trial they may not be totally objective and take pride in the therapy they developed, which can sometimes affect objectivity and scientific conduct.
How do we deal with this?
- Blinding!
- Blinding protects against subject and experimenter effects.
- Double-blinding is ideal - neither administrator nor patient knows what treatment they get
Summary of challenges
If you have a fake medicine or a placebo, induced expectations that treatment will help, conditioning, and good clinical care can lead to a placebo response.
If you also add other nonspecific factors: maturation, spontaneous improvement, life events, Hawthorne effect, S&E effects, etc., all in combination can result in a good treatment effect.
So the question is: Does active treatment add anything to the above effect? How much of the treatment effect is due to placebo, and how much is due to other non-specific causes?
To answer that you need another control group for placebo effect. You have real drug, placebo, and control for non-specific factors.
Placebo response: within-group effect
Placebo effect: between-group effect
A control group is needed to estimate placebo effect.
Improving clinical trials
Clinical trials can be improved by comparing drugs with ‘active placebo’, comparing active treatments with each other, adding a no treatment control group, measuring/evaluating expectancies, and using open-hidden paradigms and balanced placebo designs.
But what is a balanced placebo design?
Balanced placebo design
4 groups:
- Baseline: Subjects expect, and get, placebo
- Placebo: Subjects expect treatment, but get placebo (ethically hard to do)
- Treatment effect: Subjects expect placebo, get treatment
- Treatment + placebo: Subjects expect, and get, treatment
There is room for improvement with regards to clinical training and education. All students who will eventually be involved in treatment work need better education about placebos and nocebos. How can we use placebos as a companion in treatment and improve therapy outcome? And how can we avoid nocebo effects?
Questions and answers
Question [Liza]: What kinds of people are more susceptible to placebo effect?
We’ve tried to find out, but nothing convincing has resulted from this type of research. There are no clear predictors with regards to personality traits that could really explain why some people respond better to placebo than others.
About 15 years ago there was a line of research seeking candidate genes for various traits. For example, looking for genes that could be linked to things like depression, or a serotonin transporter gene - if you carried it, you were more prone to develop depression.
We thought we may have found a “placebo gene” that could explain why people responded better to a placebo. We looked at a gene that coded for serotonin synthesis in the brain, T-allele or G-allele carriers. T-allele was the risk variant. If people had the T-allele, people with social phobia had increased reactivity of the amygdala. Researchers saw in one trial that carriers of the T-allele gene did not attenuate their amygdala ith placebos to the extent that G-carriers did. So T-carriers did not get as high placebo response as G-carriers. They did a psychotherapy trial based on that, looking for outcomes in G vs T carriers, and noticed the same effect in the same direction as the plaecbo trial.
They then got a phone call from Karolinska, who it turned out had done the same thing. They also noticed an effect of this TPH2 gene coding for serotonin synthesis in the brain. They suggested pooling all this data and publishing it…
It turned out Karolinska’s effect was in opposite direction. And when looking at long-term outcome there was no genetic effect at all.
So, we now don’t thnk there is a placebo gene.
Comment/question from another participant: If we look at people who are more into alternative medicine, there is a distrust of regular doctors. I spontaneously think that people who are more inclined to be part of clinical trials are also more susceptible to placebo than someone who would not be inclined to take part in a clinical trial as they are skeptical towards medicine…
Probably right. When you try to recruit participants to clinical trials you get some selection bias as you only get volunteers. You don’t reach those in need of treatment but reluctant to participate. This is problematic. If you have a selection bias you can still say the drug works better than placebo if you do it in a double-blind randomized fashion, but those factors can affect both the drug response and placebo response.
Followup comment: Thinking about deception - would be interesting to give actual real medicine to people who believe in alternative medicine but not real medicine.
There was an attempted study on electricity allergy involving a double-blind procedure turning magnetic fields on or off. People who claimed they hhad el-allergy responded very strongly when the electromagnetic field was off. When they knew about this and were made aware of the false response, they were hostile and looked for various procedural faults. They refused to believe the procedure was true.
Question: How would you set up a study to determine effectiveness of a random new therapy, like CBT?
In psychotherapy, you can’t do placebo research as you do in medical trials, where you can just remove the active component from the pill or capsule. Some claim you shouldn’t talk about placebos in psychotherapy… maybe about pseudoplacebo, but it’s not the same thing.
So it’s very difficult, and lecturer is not aware of any study managing to do a double-blind placebo-controlled trial in psychotherapy reserach.
Comment: IPT (Interpersonal Psychotherapy), the therapy developed as a placebo therapy, was found to be effective instead and is now in guidelines for treatment.
A: Yes, many therapeutic procedures like behavioural activation for depression were originally thought to be a placebo procedure. Turned out it was highly effective on its own.
Question [Liza]: Has there been any medical use of placebo effect as a legitimate treatment?
Although it would be good to delve deeper into the power of placebo, as we know it can legitimately affect physiology, there are vast ethical concerns with this kind of research [in terms of deception].