The "reader" genre is becoming more difficult today than in any previous decade: in the pre-digital repositories age, compilations granted easy access in print to papers, otherwise difficult and expensive to obtain. Nowadays, we will only buy such compilations if they combine an excellence of editorial taste and readability that beats the temptation of downloading our own selection directly from a journal archive. In this regard, I must admit that the editors of this placebo reader have succeeded in producing a volume worth buying.
The book compiles 52 papers on the placebo effect divided in four sections. Eight articles feature in the first one, on the concept and significance of the placebo effect. 30 more are classified under four headings in the second section, experimental studies of the placebo effect: the headings are pioneering efforts, psychological mechanisms neurobiological mechanisms and contextual factors. The third and final section, on the ethics of using placebos, presents the last 14 papers about research and clinical practice. The papers are chosen and presented in such a way that the compilation reads as a historically motivated introduction to our current understanding of the placebo effects.
The uninitiated reader will probably appraise the placebo as a treatment without active ingredient that nonetheless makes (some) patients improve (e.g., sugar pills). As Ted Kaptchuk observes in his introduction, this is a relatively modern concept. Using placebos implies that we are somehow able to differentiate between effective and ineffective treatments and such a distinction could only be properly quantified with the emergence of clinical trials in the 1940s. There was an initial enthusiasm about the healing power of placebos that gradually vanished with more sophisticated analyses (documented in the anthology with papers from the 1980s and 1990s): the statistical data of most clinically relevant variables tend to regress from extreme values (disease) to the mean (health) of their distribution, and we can only take properly into account this spontaneous improvement if we compare in a trial a group of patients treated with a placebo with another one with no treatment at all. Taking all this into account, the mainstream view today is that the only placebo effect statistically documented appears when measuring patient reported outcomes, such as pain. The papers complied in the first section document how this view emerges.
It is worth recalling here that clinical trials were introduced in medicine as a yardstick for assessing the effective of treatments without a real causal understanding of why they were effective. E.g., the underlying mechanism of the antidepressant action of benzodiazepines was understood in the 1970s almost two decades after the first trials, when Valium was already one of the best selling drugs of the past century. Trials often guided our investigation of such mechanisms, showing how a drug operated under a range of different circumstances. In this respect, placebos would be like any other treatment: we lack a "robust and comprehensive" (Kaptchuk) theory, but there is a growing body of experiments documenting how placebos work. First, there are psychological mechanisms, among which the most prominent are behavioral conditioning (often unconscious) and expectations (usually conscious) induced verbally or, e.g., through social observation. Through a number of experimental designs we learn how these mechanisms operate and the physiological responses they trigger (e.g., the release of opioids). Then there are neurobiological mechanisms underlying placebo, illustrated in a number of experiments supported by various forms of brain imaging. Finally, there are contextual factors, often related to the interaction between physicians and patients in a given setting.
Half of the third section, on the ethics of placebo, hinges on clinical trials as well. Here the approach is mostly methodological: to what extent do we actually need placebo in order to ground solid experimental designs? More precisely, do placebos provide a real benchmark to gauge the efficacy of a new treatment? The obvious ethical implication is that, were the response negative, there would be no reason to use a placebo instead of an active treatment (if there was one) as a comparison. Moreover, since placebos partly operate through the patients' expectations, it is an open question how much shall we deceive them about the (lack of) treatment they are receiving. There is some evidence about placebos working when they are almost openly presented as such. And there is also evidence, presented in the second half of this section, about physicians prescribing placebos outside trials and a discussion about the ethics of such a practice.
The book closes with a paper by Miller and Colloca, two of the editors of the volume -there are 10 papers co-authored by, at least, one of the members of the editorial team, so there is no pretence of neutrality in the compilation. Their conclusion summarizes somehow the agenda this volume seems to promote: for certain conditions, placebo might actually be an effective treatment; if there is evidence gathered in well-designed trials and the nature of the treatment is properly disclose to the patient, it is acceptable to prescribe it. Hence, acupuncture for pain relief in various conditions could be legitimately prescribed on evidence-based grounds. But, for the time being, no other placebo meets such a scientific standard.
On a more general note, I'd say that a major thread in this book is that it contributes evidence to ground quite a paradoxical intuition: the mere act of diagnosing a patient and administering a treatment (at least for a few conditions) has effects that break the equivalence between "placebo" and "lack of treatment". My favorite illustration in the book is an trial with Kaptchuk as first author (pp. 226-32) testing different "treatments" for the irritable bowel syndrome: being in a waiting list (measuring patients' response to observation and assessment); sham and real acupuncture (placebo) and an augmented placebo: the needling was accompanied by a scripted positive interaction with the physician. The trial showed that these three components add up progressively, reaching a maximum effect with the augmented placebo, reaching a clinically significant effect in the treatment of the condition.
In the traditional approach to trials, the preferences of patients were considered a source of bias, since they could make a difference on the outcome. The evidence gathered in placebo research shows that, except for pain, there is no trace, so far, of a significant effect of the patients’ expectations on the treatment outcome. Perhaps this volume should prompt us to reconsider the status of blinding as a debiasing method: its usual justification is precisely that it controls for placebo effects derived from the patients preferences about treatments. Making them entirely alike breaks any systematic correlation between such preferences and the treatment outcome. But if there is no placebo effect for most conditions, perhaps we should reappraise blinding as a method to enforce the treatment protocol: since some patients would drop off the trial if they knew they were receiving an unwanted treatment, blinding secures their compliance, independently of the placebo effect. Clinical trials are indeed strategic interactions between agents with different, sometimes conflicting, interests that we should take into account in designing the trial. Controlling for some of these interactions (more than the placebo effect) might be the real justification of masking devices. In one of the compiled papers, R. Temple and S. Ellenberg contribute another argument in this same vein against active-control equivalence trials (p. 258): if you compare a drug against placebo, you have every incentive to enforce compliance with the trial protocol, since every deviation will usually reduce the differences between treatment groups, making your drug equivalent to a placebo. If you make a comparison with a standard treatment in order to prove lack of difference, there are weaker incentives to enforce protocol compliance. In other words, placebos may play a role in making both patients and researchers alike play by the rules of the experiment.
Two final positive comments. One surprising feature of this collection is how well it reads: each section is preceded by a short (but incisive) introduction intended as a road map of the papers to come. These are short and clear, and very accessible for the lay reader. The design of the experiments, their findings and the issues they raise are often so puzzling that the collection becomes engaging: I found myself eager to know whether an experiment had passed the test of replication, whether counter-arguments existed, if there was a final word on a topic. The papers are selected and ordered in such a way as to elicit this sort of engagement. Another surprising trait is the size of the volume (10.9 x 8.4 x 0.9 inches): it may initially look difficult to handle (in this age of palm-sized readers), but I found it very pleasant to work with.
© 2014 David Teira
David Teira, Ph.D., UNED (Madrid)