Methodology of administration Edit

Placebos are things like sugar pills, that look like real treatments but in fact have no physical effect. They are used to create "blind" trials in which the participants do not know whether they are getting the active treatment or not, so that physical effects can be measured independently of the participants' expectations. There are various effects of expectations, and blind trials control all of these together by making whatever expectations there are equal for all cases. Placebos are not the only possible technique for creating "blindness" (= unawareness of the treatment): to test the effectiveness of prayer by others, you just don't tell the participants who has and has not had prayers said for them. To test the effect of changing the frequency of fluorescent lights on headaches. you just change the light fittings at night in the absence of the office workers (this is a real case).

Related to this is the widespread opinion that placebo effects exist, where belief in the presence of a promising treatment (even though it is in fact an inert placebo) creates a real result e. g. recovery from disease. Placebos as a technique for "blinding" will remain important even if there is no placebo effect, but obviously it is in itself interesting to discover whether placebo effects exist, how common they are, and how large they are. After all, if they cure people then we probably want to employ them for that.

Claims that placebo effects are large and widespread go back to at least Beecher (1955). However Kienle and Kiene (1997) did a reanalysis of his reported work, and concluded his claims had no basis in his evidence. Beecher misinterpreted his data. Also, Beecher's methodology was very questionable. Then Hrobjartsson & Gotzsche (2001) did a meta-analysis or review of the evidence, and concluded that most of these claims have no basis in the clinical trials published to date. This opinion is widely spread in the placebo literature. The chief points of their skeptical argument are:

Only trials that compare a group that gets no treatment with another group that gets a placebo can test the effect. Most claims are based on looking at the size of the improvement measured in placebo groups in trials comparing only placebo and experimental (active) treatments. This is misleading since (for instance) most diseases have a substantial clearup rate with no treatment: seeing improvements does not mean the placebo had an effect. (Put more technically, comparing with the baseline (pretest measure) is vulnerable to regression to the mean.)

Nevertheless, even they conclude that there is a real placebo effect for pain (not surprising since this is partly understood theoretically: Wall, 1999)); and for some other continuously-valued subjectively-assessed effects. A recent experimental demonstration was reported: Zubieta et al. (2005) "Endogenous Opiates and the Placebo Effect" The journal of neuroscience vol.25 no.34 p.7754-7762

This seems to show that the psychological cause (belief that the placebo treatment might be effective in reducing pain) causes opioid release in the brain, which then presumably operates in an analogous way to externally administered morphine.

A recent and more extensive review of the overall dispute is: M. Nimmo (2005) Placebo: Real, Imagined or Expected? A Critical Experimental Exploration Final year undergraduate Critical Review, Dept. of Psychology, University of Glasgow. PDF copy.

It has been observed often enough, yet still frequently overlooked, that the notion of placebo as it occurs in modern clinical discourse is not a rigorous concept, and despite many attempts, it has not been possible to provide a coherent definition. The most cited one is probably that of Arthur Shapiro [4]. which together with others in the same vein, makes use of the idea of "non-specific" therapeutic effects, and thus merely shifts the explanatory burden to this equally undefined concept. In addition, as Moerman has pointed out [5]. such definitions assert at the same time that placebos are inert and that they produce the placebo effect, which is self-contradictory. In a detailed critique of Shapiro's approach, Grunbaum attempted to formulate a logically consistent definition, [6] but this in turn has been criticized on the grounds that it is uninformative in respect of the phenomenon itself – that is, the behaviour in a clinical setting of patients, bodies, physicians, and therapeutic maneuvers. In the view of Moerman and others, clinical science has burdened itself with a pseudo-concept, more confusing than useful, the correction of which will entail a revision of some foundational thinking in clinical medicine.

Recent studies published in Proceedings of the National Academy of Sciences using advances in neuroscience (PET scans) have shown that placebos can demonstrably reduce pain in humans. Researchers at Columbia and Michigan University have shown that the brains of volunteers who believed that what they were taking was pain medication were shown to be spontaneously releasing opioids, or natural pain relief. [7] According to that ABC report the Food and Drug Administration contends that as many as 75 percent of patients have had responses to sugar pills. It pointed out that all major clinical trials use placebo groups because the effect is significant and to be expected.

This effect has been known for years. Generally, one third of a control group taking a placebo shows improvement and Harvard’s Herbert Benson says that the placebo effect yields beneficial clinical results in 60–90% of diseases, including angina pectoris, bronchial asthma, herpes simplex, and duodenal ulcers. [8]

The following are some of the issues pointing to a fundamental problem:

Ever since Beecher's 1955 study appeared [9]. it has been claimed that about one third of the therapeutic effect observed in a typical trial is attributable to the placebo effect. But this is not what Beecher showed at all. In the "meta-analytic" section of his paper he gave the proportion of subjects across 15 trials deemed to have "been satisfactorily relieved by placebo" as 35.2% +/- 2.2%. This, if anything, is an estimate of the frequency of 'placebo-responders' in the aggregate trial group, but says nothing about the magnitude of the effect. Beecher, intentionally or otherwise, gave currency to the idea that the placebo effects were roughly constant at around 35%, and that the term could be usefully applied to all those variables otherwise called "non-specific" contributors to therapeutic outcomes - the natural (and unknowable) course of diseases, regression to the mean, expectation effects, changes in effect and other unquantifiable psycho-somatic features of illness, beliefs and therapeutic communication, etc. If anything is clear from subsequent studies, it is that the placebo effect is not constant, but strikingly variable. Placebo response rates all the way from zero to 100% have been reported in virtually every clinical condition studied (the variation in Beecher's own series was 15-58%). The so-called effect appears to be both universal and utterly unpredictable. Beecher, who was concerned to promote the use of Randomised controlled trials (RCTs) in clinical research, made an unjustified assumption which is almost certainly false - that placebo effects in the intervention and control arms of a trial will be identical, or nearly so, and independent of the therapeutic effects. In the rationalization of RCTs which followed, this claim has never been rigorously defended, and in specific instances, can be easily refuted. [10] The original 1955 article of Beecher "The Powerful Placebo" claimed a 35% placebo effect in 15 studies. The original article was in 1997 re-analysed and "no evidence was found of any placebo effect in any of the studies" used by Beecher [11]. The claimed "effects" were produced by spontaneous improvement, fluctuation of symptoms, regression to the mean, additional treatment, conditional switching of placebo treatment, scaling bias, irrelevant response variables, answers of politeness, experimental subordination, conditioned answers, neurotic or psychotic misjudgment, psychosomatic phenomena, misquotation, etc. Kaptchuk has shown [12] that both the name and the concept of placebo were transferred from at least 200 years of use in clinical practice, in the decade following the second world war, to a new role required by the methodology of what was then the new discipline of 'clinical research'. Earlier usage corresponded to its Latin etymology - a harmless pill or potion given knowingly to patients who were either hard to please or hard to cure. The first clear example cited in the OED is from 1811. But during the post-war therapeutic revolution, it became the trashcan into which all the confounding factors that disturb therapeutic assessments were tipped. In Beecher's terms, it became a powerful if enigmatic distraction to researchers, whose results would be contaminated without rigorous procedures for its exclusion. Its modern use is therefore quite recent, and closely related to the adoption of the RCT as the methodological gold standard for trials of therapy. A considerable body of work has attempted to elucidate the 'mechanism' of the placebo effect - but without much success. Proposals ranging from 'suggestibility' and various other psychological hypotheses, to neuro-endocrine studies, and attribution of the effect to statistical artefacts, have turned out to be flawed in various ways, so that clinical researchers have no more idea of what is really going on in the control arms of their trials than did Hippocrates. It seems unlikely that this deeply unsatisfactory situation will be resolved by a new attempt to answer the old question; instead, as has been suggested by some of the most thoughtful students, we should expect to find that some part of the conceptual landscape in which this problematic entity resides must be reconstructed before it will come into focus. This view commends itself specially to those scholars who bring to the problem a perspective from outside the clinic - from medical anthropology, history of medicine, philosophy, and statistics. [13]