The words popped up this week in two separate journals that I read regularly. In the New England Journal of Medicine, there was an opinion piece on Placebo Effects in Medicine. In the Journal of Clinical Psychiatry there was an article on Nocebo Effects in the Treatment of Major Depression. Most people are familiar with the definition of placebo, or an apparent therapeutic effect from an otherwise inert medication or therapeutic intervention. Even though the nocebo effect has been known for some time, it is less familiar. A nocebo effect is an apparent adverse reaction to an inert substance. I first became aware of it about 30 years ago as a clinical investigator working on a double blind placebo controlled study of an experimental anxiolytic medication. In that study, the blind could be broken and the research subject informed of whether they were receiving active drug or placebo. I had to inform several distressed subjects that they were receiving placebo after they insisted on stopping the study due to medication side effects. In clinical practice, nocebo effects are also apparent typically as adverse reactions to low doses of medication or very atypical responses to medication. In clinical practice, the determination is always probabilistic because placebos can't be given.
In the nocebo paper (1), the authors analyze treatment-emergent adverse events (TEAEs) in 1,565 or 2,457 placebo treated participants in 20 industry sponsored, randomized, placebo controlled trials of duloxetine. There were 16 different study designs typically in terms of length. The Hamilton Depression Rating Scale (HDRS) was the primary outcome measure in 17 of the trials. The authors looked at worsening ratings of depression score, TEAEs, and discontinuation rates. The authors hypothesized that prior conditioning by previous treatments (especially within the same class of drug) and negative expectations regarding treatment might predict nocebo responses. They could find no results to support either of these theories.
The nocebo response has clear implications for interpreting the results of clinical trials in psychiatry and clinical practice. Nocebo has really not been a term in the discussion, even by some authors who have basically declared that there is no antidepressant response (3). Practically all of the naysayers doing meta-analyses to prove that antidepressants don't work don't include any discussion of it in their work. All that you hear is that there is really no difference or not much difference between active drug and placebo. What if 10% of the worsening depressive symptoms and 5% of the dropouts were due to a nocebo response? That is a significant proportion of the trial subjects carried forward in an intent-to-treat design.
Should nocebo response rates be calculated for all clinical trials? I think that they should and that data should be collected in a standardized manner as a part of clinical trials redesign in psychiatry.
What happens in clinical practice? If a patient tells me that he or she is cutting up the lowest dose of a medication into sixteenths and can only tolerate 1/16 at a time due to side effects, I am not going to tell them to gradually titrate the dose up by sixteenths. I know that this is probably a nocebo response, and it will likely occur with other medications. I tell them to stop whatever they are doing immediately and we will try something else. That could be a treatment focus on insomnia, supportive psychotherapy, exercise, meditation, relaxation techniques, mindfulness approaches - anything but that medication. If a patient tells me that they are basically "allergic to everything" my approach is the same. I have no interest in prescribing a medication that makes a patient feel worse, irrespective of the purported clinical phenomenon. Often, the patient's response is surprise. Many people with these reactions are accustomed to physicians anguishing with them over the fact that they "cannot take any medication" and going through all of the excruciating misadventures associated with that nocebo response. I certainly don't. There are many other approaches and many other doctors who they can see. One of the behaviors that I have observed in this population is a tendency to seek out complementary medicine providers where there is a risk that nocebo responders will find other treatments that may be more expensive with no proven efficacy in the context of improved tolerability. There is also a tendency for non-psychiatrists to "kick the can down the road" and tell the patient experiencing a nocebo response that they have a psychiatric problem and need to see a psychiatrist. I think it is useful to discuss the placebo and nocebo effects with patients and provide them with as much detail as possible. I tend to focus on what is known rather than speculative neurobiology, especially in any conversation about endorphins. Endorphins have already been excessively hyped as being associated with the "high" associated with exercise but the evidence is weak (4).
The opinion piece on placebo effects (2) is an interesting contrast. One of the authors of this piece is associated with the Program in Placebo Studies & Therapeutic Encounter (PiPS). A white paper on their web site describes their observations about the placebo effect and an action plan to conduct further research and a possible introduction of it as an action plan in medicine. In the opening paragraphs they allude to the theoretical neurobiology of placebo effects including the early genetics of placebo responders. They summarize three major findings of current research on placebo effects. The first is that they are not curative. The best example of this was the placebo effects with asthmatics. Their subjective symptoms are relieved but their forced expiratory volume in one second (FEV1.0) - stays the same. The same is true of placebo in cancer treatment where the side effects of treatment improve but there are no changes in tumor size. The second is the expectation effect, best illustrated by an example the authors give having to do with rizatriptan - a standard migraine medication. If the active drug is labeled "placebo" the results are no better than placebo. If the active drug is correctly labeled the antimigraine effect increases by 50%. They list a number of medications with similar expectation effects. Lastly, they touch on the nocebo effect as "the psychosocial factors that promote therapeutic placebo effects also have the potential to cause adverse consequences." In a clinical trial of finasteride for benign prostatic hypertrophy, patients informed of sexual side effects report them at three times the rate of men who have not been informed. They quote the statistic that 4-26% of placebo treated patients in clinical trials discontinue the study due to perceived side effects. The philosophical aspects of this commentary are probably the most interesting. The authors correctly point out that the placebo effect has pejorative connotations. There is perhaps no better example than in psychiatry. They suggest a further understanding and application of the various facets of this response to create a better therapeutic alliance with patients and alleviate their suffering.
This is a fascinating area of psychiatry. I am generally compulsive about informed consent in general and more so in high risk situations. For the highest risk warnings that I give patients - serotonin syndrome, tardive dyskinesia, agranulocytosis, cardiovascular complications, seizures, renal failure, and liver failure - I have never seen a nocebo effect. That may have to do with the clear objective markers of these problems or the fact that I describe them as rare complications. On the issue of sexual side effects, I clearly explain what they are and give people the exact numbers from clinical trials. When it comes to explanations about medication side effects, the one that leads to the most problems is increased appetite and weight gain. Even though that side effect is common with medications that psychiatrists prescribe, people tend to flee from it independent of the statistics and how much weight they have recently lost or gained due to either the primary psychiatric diagnosis or substance use. It seems that most people who are likely to be nocebo responders, are well known before it gets to the informed consent stage. In the initial evaluation stages they have clear histories of not be able to tolerate much of anything and the side effects described are very atypical.
Another area where placebo-nocebo comes into play is when the primary disorder has been treated and a patient presents with the idea that the "medication has lost its effect". There are papers written on this effect and some give statistics about how often it occurs. In my experience, these outcomes are most often not due to a medication, but prevailing psychosocial factors and/or substance use. Clarifying and addressing those issues frequently leads to better outcomes than changing medications or adding another one. In many way it seems that some elements of a placebo response are an antidote to psychosocial stressors that affect medication responses.
Translating life into a medication mediated process needs to be averted at all costs.
George Dawson, MD, DFAPA
1: Dodd S, Schacht A, Kelin K, Dueñas H, Reed VA, Williams LJ, Quirk FH, MalhiGS, Berk M. Nocebo effects in the treatment of major depression: results from an individual study participant-level meta-analysis of the placebo arm of duloxetine clinical trials. J Clin Psychiatry. 2015 Jun;76(6):702-11. doi: 10.4088/JCP.13r08858. PubMed PMID: 26132671.
3: Ioannidis JP. Effectiveness of antidepressants: an evidence myth constructed from a thousand randomized trials? Philos Ethics Humanit Med. 2008 May 27;3:14. doi: 10.1186/1747-5341-3-14. PubMed PMID: 18505564.
The graphic in this case is from 2,000 Plus Royalty Free Images from the Apple App store.