Does the Word "Placebo" Evoke a Placebo Response?

Gary J. Bennett


Pain. 2018;159(10):1928-1931. 

Over the past decade or two, there has been a significant increase in the magnitude of what is generally called the placebo response (ie, the response of the control group) in clinical trials of potential analgesics for the treatment of neuropathic pain,[11,21] and there is evidence that this change has been especially prominent in trials conducted in the United States.[21] The discussion here explores the hypothesis that the increase in the placebo response is due to a change in the understanding of what "placebo" means among the general public. This hypothesis may be germane to the increase in the placebo response seen in trials of new medicines for other neurological conditions.[1,15,20]

History of the Meaning of "Placebo"

Placebo has two meanings. Since at least the 18th century, "placebo" has been a term of the medical professional. It refers to a method of treatment: a well-intentioned deceit in which the physician attempts to persuade his patient that an inactive substance will aide what is in actuality an untreatable condition.[7,9]

A second meaning appeared around 1950 when it became accepted that a trial of a clinical intervention required that both physician and patient be blinded as to the presence of the intervention. This required that the patients in the control and treatment groups be treated in an apparently identical way. If an oral medicine was being tested, both groups had to be given pills that looked identical. Of course, the control group was to receive a pill that contained no active medicine, and in this way, it was the same as the pill that was used to deceive the patient that he was being treated. Thus, the pill that was to be given to the control group was similarly referred to as a placebo. This was an error. The two situations are fundamentally different.

Placebo treatment involved intentional deceit with the aim of helping the patient. But there is no deceit in a clinical trial, and the aim is not to treat but rather to mask the identity of the pill's contents. As early as 1954, Gaddum[10] noted this error of terminology and advocated that the inactive pill for the control group be called a "dummy" rather than a "placebo." In his seminal article, "The powerful placebo," Beecher[2] acknowledged Gaddum's concern but nevertheless endorsed the term "placebo," arguing that "If it falls a bit short of precision, perhaps the language will have to grow a little to include the new use." The language has indeed grown in the more than 60 years since Beecher's pronouncement, but not in ways that have promoted clarity of expression or thought in this regard. Thus, we speak of the "placebo group" when we should say the "control group" and the "placebo response" when we should say the "response of the control group." Even the standard name for a proper clinical trial endorses misuse of the word—"a randomized, double-blind, placebo-controlled trial." Note that "placebo-controlled" is (1) redundant because the "double-blind" is achieved by using dummy pills and (2) is simply inaccurate because a placebo would not control for anything.

Placebo Analgesia: Imaginary or Real?

In Beecher's day, physicians believed that placebo analgesia was imaginary, in the sense that there was no actual reduction in the neural processes that generate the sensation of pain, but rather a decrease in the patient's report of pain because of personality and psychological factors.[2,3,13] It is reasonable to assume that if in those days the general public considered placebo analgesia at all, then they had a similar opinion. But modern research, especially the results of neuroimaging studies, has shown unequivocally that placebo-induced analgesia is real in the sense that the neural correlates of the sensation of pain are reduced, in the same way that they would be reduced by a decrease in the intensity of a pain-evoking stimulus or by the administration of an analgesic drug.[8,22,23] Today, pain researchers are united in the belief that placebo analgesia is real, in the sense that the neural processes that represent pain are reduced.

Patient Expectations Today

Patient expectation is a major factor in the generation of placebo analgesia.[4,17–19] Consider what the expectations are when a patient is told that he will be randomly assigned to a placebo group or to a treatment group when he believes that placebo analgesia is imaginary vs when he believes that placebo analgesia is real (Table 1).

Different beliefs will certainly lead to different expectations. For the patient who believes in the reality of placebo analgesia, being told that he may receive a placebo creates the expectation that he may obtain authentic pain relief. In effect, the use of the word "placebo" evokes a placebo response. It seems highly probable that the condition, "Maybe pain reduction" vs "Maybe pain reduction," will yield a large reduction in pain in the control group, and perhaps make it more difficult to detect a treatment effect.

Evidence That the Placebo Concept has Entered the Realm of General Knowledge

If placebo analgesia is accepted as authentic among medical and scientific professionals, then it is at least possible that the idea has also been accepted to some degree by the general public. Evidence for this might come from culturomics, ie, the quantitative analysis of digitized texts. To this end, analyses of the frequency of usage of the word "placebo" were conducted based on data from the Google Books Ngram Viewer[14] and The Corpus of Contemporary American English.[6]

The Google Books Ngram Viewer ( consists of the digitized texts of over 5 million books, containing over 500 billion words published in the following languages: English (361 billion, with separate entries for American and British publications), French (45 billion), German (37 billion), Italian (not stated), Spanish (45 billion), Russian (35 billion), Chinese (13 billion), and Hebrew (2 billion). The first 5 of these language categories were examined because each uses the Latin alphabet and each has the same spelling for "placebo."

Figure 1 shows the frequency of appearance of the word "placebo" for the years 1950 to 2008 in American English, British English, French, German, Italian, and Spanish. For all these languages, usage of "placebo" was very low from 1950 to 1955. From approximately 1965 onwards, the frequency of usage of "placebo" increased in a dramatic way in both American and British English publications, but not in French, German, Italian, or Spanish. Usage in French, Italian, and Spanish books seems to have accelerated modestly starting between 2000 and 2005, but the word is still unaccountably rare in books published in Germany.

Figure 1.

Google Books Ngram Viewer data. Frequency of usage of the word "placebo" in books published in the United States, Great Britain (GB), France (FRA), Germany (GER), Italy (ITA), and Spain (SPA) for the years 1950 to 2008. The ordinate plots the frequency of use of "placebo" as a percentage of the use of all words in the language category, normalized for the number of books in each year. Data are shown as 3-year moving averages.

The Google Books corpus of texts has been criticized in several respects.[16] Particularly germane to the point here is that books intended for an academic audience, which would include medical and scientific professionals, have become an increasing portion of the total. It is thus uncertain to what extent the increased usage of "placebo" shown in Figure 1 might reflect the knowledge base of the general public.

The Corpus of Contemporary American English (COCA; is composed of more than 560 million words from more than 220,250 texts, including 20 million words each year from 1990 to 2017. For each year, the corpus is evenly divided between texts from the 5 genres shown in Table 2. It seems likely that genre-specific analysis of the usage of "placebo" might be especially informative for the question of the knowledge base of the general public.

As shown in Table 2, for the period 1990 to 2017, the usage frequency for "placebo" was highest in the genre Academic Journals; this is not surprising for a word with a technical meaning. But it is perhaps surprising that relatively high usage frequencies are seen in transcripts of television and radio programs (Spoken), in Popular Magazines, and in Newspapers.

All the genres except Academic Journals are likely to represent texts widely available to the general public. The total for Spoken, Fiction, Popular Magazines, and Newspapers is 11.49, which is higher than the total (9.72) for Academic Journals alone. The frequencies of usage of "placebo" in both nonacademic and academic texts for the period covered by the COCA corpus (1990-2017) do not show the dramatic yearly increases that are seen in the corresponding period of the Google Books corpus (data not shown). However, it should be noted that the COCA data begin in 1990, whereas the Google Books data show increasingly high usage in English speakers beginning about 1965.

Conclusions and Recommendations

It is indisputable that a change has occurred in the conception of placebo analgesia among pain physicians and researchers during the last few decades. Unequivocal evidence dates from at least 2004.[22] It is not unreasonable to suspect that this change has been gradually absorbed by the general public from which clinical trial patients are drawn.

Analysis of digitized text in the Google Books corpus for 1950 to 2012 shows a clear increase in the frequency of use of the word "placebo." This increase began in approximately 1955 to 1960, which is just after the start of a period of intense discussion among clinical scientists about proper clinical trial design and the necessity of double-blind control. It is of interest to note that Henry Beecher was a leader in the movement demanding double-blind controls in all kinds of clinical trials, but opiate analgesia was his personal research focus. Thus, the history of placebo research became intertwined with the history of research on analgesia.

The increase in usage of "placebo" shown by the Google Books data is prominent in books in English (both American and British), but not for those in French, German, Italian, or Spanish. One wonders whether this is related to the finding that the increase seen in the magnitude of the placebo response (or rather, the magnitude of change in the control group) over the past 15 years has occurred in trials conducted in the United States, but not those performed in Europe.[21] Tuttle et al.[21] did not categorize European trials by country, and it is thus not known to what extent British sites contributed to the European trials considered in their report.

The Google Books data show that English language books containing information about the placebo concept were increasingly common beginning about 1965. However, it is not possible to say to what extent the Google Books data refer to books that were likely to influence the general public. By contrast, the COCA data are categorized by genres and the texts of academic journals are tabulated separately. Texts from the other 4 genres (Fiction, Newspapers, Popular Magazines, and Spoken) are clearly those that would be accessed by the general public. With the exception of Fiction, the usage frequencies for the nonacademic genres are relatively high and have been so since at least 1990 (COCA's inception date). The results of the COCA data analysis support the idea that the general public has become increasing familiar with the idea of placebo. Whether this includes an increased familiarity with the authenticity of placebo analgesia is of course not proven but seems at least possible.

Familiarity with the authenticity of placebo analgesia may have contributed to the outcome of recent studies in pain patients with the "open-label placebo" protocol. Statistically robust analgesia has been found with an open-label placebo protocol in patients with irritable bowel syndrome[12] and also in patients with chronic low-back pain.[5] In these studies, the control group received no treatment, whereas the other group was told that they would receive "…placebo pills made of an inert substance, such as sugar pills, that have been shown in clinical studies to produce significant improvement in irritable bowel syndrome symptoms through mind-body self-healing processes,"[12] or they were told that they would receive "…placebo pill … an inactive substance, such as a flour pill, that contained no active medication in it" followed by exposure to a 15-minute script containing positive information concerning placebo effects.[5] It may be that their general knowledge predisposed the patients in these studies to accept the information endorsing placebo efficacy.


The foregoing discussion leads to an obvious recommendation for clinical trial procedure. The word "placebo" should be avoided in all information and instructions given to the patients. Patient instructions should have the goal of forcing the patient's expectations to the form: "I may receive pain relief" vs "I will not obtain pain relief." Thus, eg, patients should be told that they will be randomly assigned to receive pills (tablets, capsules, etc.) that contain "a drug that might relieve their pain" (the treatment group) or "pills that do not contain any medicine and will not affect their pain" (the control group). If the patient were to ask whether he might receive a placebo, he should be told "No, the pills will contain a medicine or no medicine."

Recommendation for terminology: Gaddum's[10] suggestion of the term "dummy" instead of "placebo" does not seem right to the modern ear. We say "sham surgery" for the control procedure in a clinical trial of a surgical intervention; thus, "sham medicine" or "sham treatment" could replace "placebo."