“Do placebo responders exist?” This is a question raised by many and answered by few (1). Are those who respond with symptom improvement after placebo applications, e.g. in a randomized, placebo-controlled drug trial (RCT), people that always respond the same way and exhibit personality traits that allow for the prediction of their future behavior in drug or placebo application based on past behavior? Drug companies testing new compounds would love to know this in advance, as would doctors, pharmacists, and scientists. The latter, however, for more non-economical reasons.
The above-cited paper (1) reviewed the literature until about 2008 and concluded that the evidence for a stable response pattern (a “placebo personality”) is poor, but solid experimental proof is missing; few studies have investigated what would be required to underscore such an assumption: stability of the behavior across time within the same condition and stability across different conditions.
We are aware of only one study (paper) that explicitly has investigated this in an experimental set-up (2), and found that consistency within a condition, e.g. pain and placebo analgesia, may be possible, but not necessarily across different conditions, e.g. from pain to nausea. This throws the question back to a conclusion reached long before, namely that at some point we all are placebo responders, it just depends on the existence of a “weak” point in each of our biographies, where we long for relief.
Too little data from drug trials: Who’s afraid of what?
The paucity of data in support of or in disagreement with the assumption of the existence of a placebo personality has, once again, a reason — or better, there are several reasons, depending on the angle from which you look at it. Let’s review the situation from the point of a drug company doing an RCT of a novel compound to be tested for the treatment of chronic pain (but any other clinical condition would work as well as an example).
Currently, more than 50,000 drug trials are conducted worldwide every year, including between a few hundred to a few thousand patients each. Even if we consider only the last 10 years (because of changed trials designs over time) and assume that about half of these patients would have received placebos, one wonders that no firm answer can be generated from the archives of the drug industry.
When conducting a drug trial and recruiting patients for it, a few prerequisites are conventional, one of which is that patients likely responding to placebo would best be excluded before the study starts, either during recruitment or at least during a placebo run-in phase of the study, during which patients receive placebo. Responders here would be excluded from further participation. Sometimes, run-in phased are done without placebo provision, as some patients already respond with symptom improvement knowing that they will receive treatment in the future, a kind of “waiting list effect” (3). While both strategies may allow identifying and excluding some placebo responders prior to the main phase of the RCT, they cannot and have not prevented the occurrence of strong placebo response rates during the study. Alternative study designs, including randomized run-in and withdrawal phases, have been developed to minimize this (4), but this will be the topic of a future posting.
Two other options to achieve the same result (identifying and excluding placebo responders) are instead not realized in RCT from the drug industry, for different reasons:
For one, if patients recruited for a new study would allow to be linked to past RCT they were involved in, whether or not they had received drug or placebo and whether they had responded positively to placebo or drug, this would answer one of the above-listed questions of consistency (across time). In times of country-wide organization of patient recruitment and the risk of guinea-pigging (semi-professional patients in drug studies), the need for a centralized registry of patients and the protection of individualized patient data may overcome ethical concerns and may protect drug companies’ interests at the same time. Until then, the archives of the drug industry that may allow answers for many placebo questions only occasionally provide a glimpse into it, e.g. (5); there is a need for change in the interest of better patient management.
The other alternative would be to include psychometric personality characterization of patients in all RCT beyond those in psychiatry, where assessment of personality profiles may be a part of the condition to be treated. This is not done for an entirely different reason: If the outcome of such an assessment would indicate that only a subgroup of patients would profit from the drug, this would put the drug at risk for a selective indication. Therefore, sensitivity analyses of approval trials are usually only performed for the absolute minimum of data reported, e.g. age, sex, and some disease specific conditions (severity, disease, sub-classification, duration, etc.). No drug company would currently take such a risk, unless approval authorities would either require it or it would except the outcome of such explorative analyses from the question of indication, e.g. by analyzing only the subset of data of those patients that had received placebo in the RCT. Also, the patients need protection from these sensitive (psychological, psychiatric) personality data when they are not part of the disease for which they requested treatment.
Conflicting data from experiments
In sharp contrast to RCT, researchers interested in identifying placebo responders and their profiles have installed quite a number of different psychometric tests to predict an outcome, with poor response, however, so far. Here, we will present a few of the personality concepts that have been linked to higher placebo responses in experimental settings before discussing reasons for their inconsistency and poor replicability. They have been identified in a number of systematic reviews and meta-analyses over the past decade (6, 7). These concepts refer to a few tests that have shown some potency of predictability, especially in relations to dispositional optimism, behavioral inhibition/activation (BIS/BAS), extraversion, and self-efficacy and locus of control. Most have been tested in healthy student volunteers but not in patients.
Dispositional optimism refers to the higher likelihood of a placebo response in optimistic (as a trait) people, but only when they are provided with positive expectations (dispositional) and not with only a chance for a positive outcome; pessimists are more likely to follow negative expectations (8). Offering more choices increases the positive outcome of a treatment (9), but there may be a limit: too many choices may be counter-productive (the Goldilocks principle of “just right.” Optimism not only affected subjective well-being but also pain perception (10), e.g. during the cold pressor test and other experimental conditions.
Extraversion: Among the personality traits tested in IBS patients (11), high extraversion was the only one yielding significance and predicting placebo response in a sham-acupuncture trial. Extraversion is one of the five dimensions tested with the Big Five Test, the others are neuroticism, agreeableness, conscientiousness, and openness to experience. The extraversion-introversion dualistic concept is, similar to BIS/BAS, a very traditional personality model that is seen on a continuous scale and can be found in many theories, e.g. in Eysenck’s personality concept.
Locus of control: Rotter’s concept of a “locus of control” subdivides individuals based on their predominant self-perception of being in control of one’s own life, or that life is controlled by outside factors, e.g. by fate, which cannot be influenced. The model can also relate to the self-perception of one’s health (health-related locus of control). In this view, an internal locus of control resembles similarities with the concept of “self-efficacy” of Bandura: high self-efficacy and an internal locus of control were therefore expected to be associated with low placebo responses (12) in an experimentally-induced nausea provocation test via body rotation (13).
These traits, as was noted (7), link (high) placebo response to outward orientation (optimism, extraversion, external locus of control), as for altruism . However, a number of other personality traits, e.g. suggestibility (14), hypnotizability and absorption (15) are inward orientations, as is high self-efficacy (6), but they are also associated with high placebo responsiveness. These apparent contradictions cannot be resolved by conceptual or semantic discussions, they need another explanation.
One possibility: Underreporting of psychometric results
One could be that the reported significant associations of single traits (or subscales of traits) are purely random and have occurred as a consequence of a beta-error, but this remains undisclosed because all the scale that were used in the studies but did not yield significance are simply not reported, and reporting is restricted to one or two test results. Beta-errors are named findings that appear as significant but have occurred by chance: Including 10 psychometric tests with 5 subscales each in a comparison of two groups will result in 50 single but not independent comparisons (t-tests, for matters of simplicity), and would require statistical correction (e.g. according to Bonferroni) to account for such errors. At the same time, the number of subjects entering the testing should be at least double the number of tests performed — in our case, 100. These requirements are often neglected or violated in research.
If a publication does not report the tests that have been used but only those with significant results (or, to avoid embarrassment, one that was significant and one that was not) it would be possible to check for the total number of tests implemented by checking the trial registration, if there is any; however, registration is often not needed for psychological experiments. As psychologists, we can confirm that using only one or two tests if you are interested in psychometric properties of the placebo response is extremely unlikely. In studies we were involved, the number of psychological constructs (= test scales used) tested were 13 in one case (12) (Table 1), and 8 tests (with more than 15 subscales) in another one (16). At least for all the studies involving (psychology) students, the restriction to one or two tests is simply unbelievable. Under-reporting seems to be the rule rather than the exception, checking the above discussed papers.
While there may be an indication of the existence of a “placebo personality,” current research provides no appropriate answer, because of self-limited restrictions for RCT as well as experimental approached, but for entirely different reasons. Until these limitations are given up, no final answer can or will be given.
This is part 10 of a series covering “placebo” provided by Paul Enck and Sibylle Klosterhalfen from the Tübingen University Hospital. Continuous updates on placebo research can be found at www.jips.online.
- Kaptchuk TJ, Kelley JM, Deykin A, Wayne PM, Lasagna LC, Epstein IO, et al. Do “placebo responders” exist? Contemporary clinical trials. 2008;29(4):587-95.
- Whalley B, Hyland ME, Kirsch I. Consistency of the placebo effect. Journal of psychosomatic research. 2008;64(5):537-41.
- Hesser H, Weise C, Rief W, Andersson G. The effect of waiting: A meta-analysis of wait-list control groups in trials for tinnitus distress. Journal of psychosomatic research. 2011;70(4):378-84.
- Weimer K, Enck P. Traditional and innovative experimental and clinical trial designs and their advantages and pitfalls. Handbook of experimental pharmacology. 2014;225:237-72.
- Enck P, Vinson B, Malfertheiner P, Zipfel S, Klosterhalfen S. The placebo response in functional dyspepsia–reanalysis of trial data. Neurogastroenterology and motility. 2009;21(4):370-7.
- Horing B, Weimer K, Muth ER, Enck P. Prediction of placebo responses: a systematic review of the literature. Frontiers in psychology. 2014;5:1079.
- Darragh M, Booth RJ, Consedine NS. Who responds to placebos? Considering the “placebo personality” via a transactional model. Psychology, health & medicine. 2015;20(3):287-95.
- Geers AL, Helfer SG, Kosbab K, Weiland PE, Landry SJ. Reconsidering the role of personality in placebo effects: dispositional optimism, situational expectations, and the placebo response. Journal of psychosomatic research. 2005;58(2):121-7.
- Geers AL, Rose JP, Fowler SL, Rasinski HM, Brown JA, Helfer SG. Why does choice enhance treatment effectiveness? Using placebo treatments to demonstrate the role of personal control. Journal of personality and social psychology. 2013;105(4):549-66.
- Geers AL, Wellman JA, Fowler SL, Helfer SG, France CR. Dispositional optimism predicts placebo analgesia. The journal of pain : official journal of the American Pain Society. 2010;11(11):1165-71.
- Kelley JM, Lembo AJ, Ablon JS, Villanueva JJ, Conboy LA, Levy R, et al. Patient and practitioner influences on the placebo effect in irritable bowel syndrome. Psychosomatic medicine. 2009;71(7):789-97.
- Horing B, Weimer K, Muth ER, Enck P. Prediction of Symptom Change in Placebo Versus No-Treatment Group in Experimentally Induced Motion Sickness. Applied psychophysiology and biofeedback. 2015;40(3):163-72.
- Weimer K, Horing B, Muth ER, Enck P. How to study placebo responses in motion sickness with a rotation chair paradigm in healthy participants. Journal of visualized experiments : JoVE. 2014(94).
- De Pascalis V, Chiaradia C, Carotenuto E. The contribution of suggestibility and expectation to placebo analgesia phenomenon in an experimental setting. Pain. 2002;96(3):393-402.
- De Pascalis V, Scacchia P. Hypnotizability and Placebo Analgesia in Waking and Hypnosis as Modulators of Auditory Startle Responses in Healthy Women: An ERP Study. PloS one. 2016;11(8):e0159135.
- Roderigo T, Benson S, Schols M, Hetkamp M, Schedlowski M, Enck P, et al. Effects of acute psychological stress on placebo and nocebo responses in a clinically relevant model of visceroception. Pain. 2017;158(8):1489-98.