Placebo Trends across the Border: US versus Canada

Background Physicians around the world report to using placebos in a variety of situations and with varying degrees of frequency. Inconsistent methodologies, however, complicate interpretation and prevent direct comparisons across studies. While US- and Canada-based physicians share similar professional standards, Canada harbours a less-litigious universal healthcare model with no formal placebo-related policy—factors that may impact how physicians view and use placebos. Methods To compare American and Canadian data, we circulated an online survey to academic physicians practicing in Canada, collected anonymous responses, and extracted those of internists and rheumatologists for comparison to US data obtained through parallel methodologies. Results Whereas our data show overall concordance across the border—from definitions to ethical limitations and therapeutic potential—differences between American- and Canadian-based placebo practices merit acknowledgement. For example, compared to 45%-80% among US-based respondents, only 23±7% of Canada-based respondents reported using placebos in clinical practice. However, 79±7% of Canada-respondents—a figure comparable to US data—professed to prescribing at least one form of treatment without proven or expected efficacy. Placebo interventions including unwarranted vitamins and herbal supplements (impure placebos) as well as sugar pills and saline injections (pure placebos) appear more common in Canada, where more doctors described placebos as “placebos” (rather than “medications”) and used them as a “diagnostic” tool (rather than a means of placating patient demands for treatment). Interpretation Cross-border variation in the use of clinical placebos appears minor despite substantial differences in health care delivery system, malpractice climate, and placebo-related policy. The prevalence of impure placebos in both Canadian and US clinics raises ethical and practical questions currently unaddressed by policy and warranting investigation.

Congruent with the working definition assumed in the high-powered world of pharmacology, most physicians construe placebos as the non-specific effects of medical treatment that, in clinical trials, must be controlled for to assess the specific effects of new (drug) interventions. 1 Placebo-like treatments, accordingly, refer to any short-term or illusory impression of improved health that some patients experience when they receive what appears to be effective treatment but actually isn't for the condition being treated. As such, the placebo effect is a powerful mindbody phenomenon with a specific underlying biology that health professionals should investigate and exploit. 2 Exemplifying the link between psychosocial factors and physiological processes, placebos are central to medicine 3 but even more pivotal to psychiatry. 4 Furthermore, placebos bind behavioural science to the techniques of neuroscience. [5][6][7][8] Several scholars grant placebos a prominent place in clinical psychiatry 4,9,10 and mounting evidence suggests a large placebo component even in drugs forming the backbone of biological psychiatry. [11][12][13][14][15][16][17] Together with the majority of physicians, however, most modern psychiatrists find the science of placebos difficult to swallow. 18 Shrouded in a checkered history, placebo use in a therapeutic context remains controversial.
Indeed, in 2006 the American Medical Association (AMA) cautioned that "[p]hysicians may use [a] placebo for diagnosis or treatment only if the patient is informed of and agrees to its use". 19 The AMA admonition followed a controversial meta-analysis of clinical trials suggesting that placebo effects are either minimal or non-existent and that "outside the setting of clinical trials, there is no justification for the use of placebos". 20 Multiple researchers have critiqued many aspects of this controversial meta-analysis, [21][22][23][24] and reanalysis of the data yielded findings of a "robust" placebo effect 25 resulting in a flurry of rebuttals and debates. [26][27][28] The charged AMA statement, however, still colours the views of many clinicians. 29 Despite subsequent discussions of this issue in bioethical circles, 30,31 the AMA tenor still guides many of the assumptions that the medical community maintains about placebos. 18 The Canadian Medical Association is yet to draft a formal policy regarding the use of placebos in clinical practice.
The placebo flame has been recently rekindled with reports of placebos being dispensed as part of routine care. 32 Publications concerning placebos now span research studies, 12 reviews, 33,34 books, 8,35 and popular media coverage, 36,37 including legal scholarship 29 and social science. [38][39][40][41] The widespread use of placebos in clinical practice has been demonstrated in a recent survey of internists and rheumatologists in the United States (US) 42 revealing that of the 679 physicians who replied, more than half said they prescribed "placebo treatments" every now and then, and that they deemed the practice ethical. About 40% of respondents reported they used painkillers or vitamins as placebos and 13% acknowledged using antibiotics and sedatives for this purpose; barely 3% said that they used sugar pills. Over two-thirds, however, reported that rather than calling them placebos they described the pills to patients as "a potentially beneficial medicine or treatment not typically used for their condition." Five percent of physicians reported telling their patients that they were receiving a placebo and 62% believed that prescribing placebos was an ethically acceptable practice.
A number of similar studies have been conducted in select geographic locations outside of Canada. [43][44][45][46][47][48][49] For example, a Danish study reported that 86% of general practitioners have used placebos at least once, with 48% using placebos more than 10 times in the previous year. 47 A separate study from Israel found that 60% of respondents prescribed placebos. 48 Among those, 62% reported that they prescribed placebos as often as once a month. Another US study targeting academic physicians in the Chicago area reported that placebos were being used in everyday clinical practice. 49 Forty five percent of physicians reported that they had used placebos and 96% of physicians believed that placebos had a therapeutic effect. The sparse data from physicians practicing in Canada motivated us to probe the role of placebos in clinical care.
Here we show results from an online survey comparing academic psychiatrists to other academic physicians across Canada. Because placebo responses and effects often occur more readily when the endpoint of treatment is a change in behaviour, 50 we expected psychiatrists to differ from other physicians. Compared to non-psychiatrists, therefore, we hypothesized that psychiatrists would display better placebo knowledge, different beliefs, more tolerant attitudes, and heightened patterns of use. In addition, we expected sex-based differences between male and female physicians. We envisaged that female psychiatrists would have a tendency to me more compassionate toward and more innovative about treating their patients than would male psychiatrists. Accordingly, we hypothesized that male psychiatrists, relative to female psychiatrists, would be less likely to integrate placebos into their medical practice.

Method
Using the open source LimeSurvey® web-based application tool, we designed our survey to collect self-report information concerning placebos in clinical practice. Our 5-minute survey implemented a number of computerized checks to preclude invalid data, and ensured expediency as well as data anonymity. Following seven demographic questions, fourteen placebo questions Medical schools and psychiatry departments that did not reply to our initial e-mail received follow-up phone calls. E-mails to the deans and chairs provided English-French information regarding the nature and relevance of the study, as well as the institutional ethics approval. We requested that the deans encourage all physicians to complete the web-based survey. For the chairs of psychiatry, the e-mail also outlined the importance of placebos in psychiatry and the value of receiving feedback from practicing academic psychiatrists.
A brief e-mail, crafted for the physicians, described the research study and provided live links to the survey in both French and English. We informed participants that the survey was completely anonymous. In accordance with certain provincial constraints (e.g., section 30. Services provided support and maintenance of the online survey and ascertained data confidentiality through the Educational Technologies team.

Statistical Analysis
We analyzed the data using descriptive statistics and frequency distributions using SAS statistical software, version 9.2 (Statistical Analysis Systems, SAS Institute Inc, Cary, NC), including Chi-square and Fisher exact tests.
Male and female respondents represented 65% and 35% of the sample, respectively. Age ranged from 24 to 88 years (median=52; mean=51.1). Table 1 shows statistically significant differences between psychiatrists and other physicians concerning characterizations of placebo.

Administration in Clinical Practice
About 20% of physiciansbe they psychiatrists or non-psychiatristsreported that they had either prescribed or administered a placebo in the course of routine clinical practice. Only 2% of psychiatrists reported that placebos had no clinical benefit compared to 6% of other physicians (χ 2 (1) = 4.72, P = 0.03). Forty-three percent of psychiatrists indicated that the use of placebos might be permitted after notifying patients that they are receiving a placebo whereas 28% of nonpsychiatrists concurred (χ 2 (1) = 14.36, P< 0.001). treatments. For example, 38% of psychiatrists reported giving subtherapeutic doses of medication to their patients compared to 6% of non-psychiatrists (χ 2 (1) = 97.36, P< 0.001). In addition, 16% of psychiatrists, relative to 9% of the remaining sample, have prescribed prefabricated placebo tablets (χ 2 (1) = 6.39, P = 0.01). Table 2 outlines what physicians contemplating a hypothetical situation involving placebo administration would say to their patients.
Different levels of agreement arose between psychiatrists and non-psychiatrists when responding to the statement "the placebo effect is real" and "placebos have therapeutic effects".
Approximately 90% of physicians reported that psychological factors played a role in explaining how patients may benefit from a placebo. In contrast, more psychiatrists (47%) compared to the rest of the sample (25%) accounted for biological factors (χ 2 (1) = 33.14, p< 0.001). About 70% of females reported that the mind-body connection was at play; however, a difference exists between the 67% of male psychiatrists and 56% of male non-psychiatrists that agreed (χ 2 (1) = 4.77, P= 0.03).

Health Benefits of Placebos and Other Alternative Methods
A difference emerged between psychiatrists and other physicians when asked about the benefits placebos may have in a variety of health problems, as displayed in Table 3

Discussion
Compared to other medical specialties, psychiatrists appear more complaisant in their attitudes and beliefs towards placebos. Although about 20% of psychiatristscomparable to other physicians -responded affirmatively to the question "Have you ever prescribed a placebo in the course of routine clinical practice?" psychiatrists reported using subtherapeutic doses of medication significantly more than non-psychiatrists (see Figure 1). This spike in the administration of subtherapeutic drugs was prevalent irrespective of sex and age of psychiatrist.
Fewer psychiatrists (2%), compared to non-psychiatrists (6%), reported that placebos had no clinical benefit. This finding suggests that psychiatrists may better appreciate the clinical merits of using placebos in routine care. Our findings, moreover, suggest that physicians may only partially heed the AMA admonition. 20 About 90% of respondents agreed that psychological factors play a role in explaining placebo benefits. More psychiatrists (47%) than nonpsychiatrists (25%), however, reported that biological factors explain how placebos may benefit patients. This attitude extends to other categories (see Table 3). For example, over 95% of psychiatrists report believing that relaxation techniques have both psychological and physiological benefits. Furthermore, findings indicate that psychiatrists appreciate the effects placebos can engender in a range of disorders (see Table 2). Because of their continuous exposure to the effects of a disrupted mind on health, psychiatrists may better appreciate the therapeutic effects of placebos. Figure 1 shows that two variations on the placebo theme seem palatable as treatment options in situations without expected clinical efficacy. One variation refers to pseudoplacebosplacebolike interventions that may be active in principle but unlikely effective for the condition being treatedwhich comprise such treatments as vitamins for chronic insomnia. 51 In the present study we show that the use of pseudoplacebos is rampant in clinical practice. This trend is increasingly prevalent probably because using pseudoplacebos reduces some of the logistical and ethical problems associated with inert placebo administration. In other words, ethical concerns appear less tenuous when a physician prescribes an active substance, albeit speciously. 52,53 Figure   1outlines how psychiatrists as well as non-psychiatrists prescribe a variety of pseudoplacebos, including vitamins, herbal supplements, and other treatments. This figure also demonstrates that non-psychiatrists prescribe significantly more antibiotics, ibuprofen, and saline infusions than psychiatrists. In line with the disorders that they see and treat, psychiatrists should seldom prescribe patients with antibiotics and ibuprofen; however, they do appear to prescribe more prepared placebo pills (e.g., commercially available lactose pills) relative to the other responding physicians.
A second variation has to do with the notion of a superplacebo-a treatment that is an actual placebo wherein neither the prescribing practitioner nor the receiving patient is aware of the absence of evidence to recommend it therapeutically. 54 Having gleaned the insights of multiple clinical psychiatrists, our findings suggest that at least some psychiatrists view prescribing subtherapeutic doses of psychiatric medication as clinically therapeutic. 55 As a case in point, in the 1980s, haloperidol dosing of up to 100 mg/day was not unusual and a dose of 2-4 mg/day would have been considered "homeopathic" if not a downright placebo. Subsequent studies, however, have suggested even such low doses as potentially therapeutic. Thus, when administering subtherapeutic doses of medication, at least some psychiatrists may be under the impression that they are instigating an effect that may have therapeutic value. 51 Placebo confusion appears deeply-entrenched because although nearly half of physicians reported that they "would never give a placebo outside of a clinical research trial" (on Item #11), many more indicated that they have prescribed placebo-like treatments (on Item #9). It is likely that fewer physicians explicitly report to prescribing placebos in clinical practice because such admission implies bad professional form: congruent with the AMA policy, the dominant view among medical researchers and clinicians deems placebo administration ethically problematic and most doctors feel effectively prohibited from using placebos in clinical practice.
Clinicians who purposefully prescribe unwarranted treatments run the risk of both legal and ethical transgressions. Prescribing treatments without demonstrated clinical efficacy is tenuous; however, at least some psychiatrists appear to believe that subtherapeutic doses have therapeutic effects. Two common scenarios leading to the prescription of subtherapeutic doses include: 1.
the practice of "start-low-and-go-slow"psychiatrists often start patients on an ineffective dose of medication that they intend to gradually increase, but some patients display improvement at doses that remain far below a standard pharmacological threshold (e.g., prescribing 25mg of chlorpromazine while the recommended dose is 600-1000mg). 56 2. Receiving new patients that are already taking subtherapeutic doses of medication, the receiving psychiatrist continues to prescribe the same low dose because the patient appears to benefit.
Either deliberately or unwittingly, psychiatrists appear to be savvy placebo users. A recent metaanalysis, for example, reported that antidepressantsflagship drugs of modern psychiatrywere largely comparable to placebos for most individuals suffering from depression; antidepressants were clinically superior to placebos in people with extreme depression only. 57 Although this controversial account has been the focus of heated debates, additional data have supported the notion that antidepressants are certainly less effective than we have been led to believe, and in many cases possibly as effective as inert placebos. 11,58,59 Other examples span polypharmacyusing more than one drug for the same underlying condition without evidence-based research to support itand off-label medicationsusing a drug for a purpose different from its intended indication or using an atypical dosage-related interpretation. These increasingly rampant methods of treatment, especially in the elderly, 60 appear to gel with recent studies reporting a dramatic surge in placebo response since the 1980s. 61 That psychiatrists prescribe more subtherapeutic doses than other physicians is contrary to accounts suggesting that general practitioners are more likely to prescribe such doses; [62][63][64] however, it supports the notion that psychiatrists prescribe a broad range of doses. The term subtherapeutic has many interpretations and in our survey we left those to the discretion of the physician. Whereas some physicians may interpret subtherapeutic through the lens of drug blood-levels and as such, a function of the patient's metabolism, others construe subtherapeutic dose as any prescription that is below the recommended therapeutic level. Most physicians surely appreciate that homeopathy is incongruent with some basic principles of modern science and likely distinguish subtherapeutic dosage from homeopathic quantities. With over 35% of responding psychiatrists prescribing what they believe to be subtherapeutic doses, however, further investigation should elucidate this lacuna.This issue becomes all the more complicated when even "therapeutic" doses of antidepressants seem to resonate, at least in large part, with the appellation of placebos. 11,12,58

Limitations and Caveats
In addressing the relative merits and drawbacks of Internet surveys we refer the reader to a recent special issue in Public Opinion Quarterly (Vol. 72, No. 5, 2008). On the one hand, a few of these shortcomings include the challenge of drawing representative samples of the general population; dealing with the issue of people without Internet access; and minimizing the potential for non-response bias. These potential caveats weaken the generalizability of Internet surveys, especially those focused on broad and diffuse populations. On the other hand, Internet surveys entail advantages, such as reduced social desirability, turning them into valuable research tools under certain conditions. 65 In this section, we provide a detailed account showing that these latter conditions apply to the present study.
Unlike typical surveys of specifically named persons requiring a response rate of at least 60%, the present Internet survey targeted academic physicians without referring to specific individuals.
As such, adhering to the definitions and metrics proposed by the American Association for Public Opinion Research is unsuitable. [66][67][68] In addition, response rates of online surveys using email invitations outperform other electronic media, such as mobile short messaging service, without compromising the sample composition of respondents. 69 On the other hand, response rates for Internet surveys such as the present study differ from mail surveys, [70][71][72][73] with characteristic values falling below 10%. [74][75][76][77] Our calculations show that the response rate in the present study was about 10%. Response representativeness, however, denotes more than response rate. 78,79 The present demographic data are congruent with data drawing on more than 62,000 physicians practicing in Canada. 80 Our findings, therefore, likely represent a valuable contribution to preliminary investigations of placebo use among physicians and their beliefs about placebo mechanisms and effectiveness.
The advantages of web-based surveys are multiple: they are expedient, allowing for efficient data collection and timely results; they permit casting a wide net while reducing the cost relative to the sample size; 71 they eliminate the need for a full mailing address, and thus provide respondents with a guarantee of anonymity. 81 Consequently, respondents benefit from social advantages such as an increased willingness to answer charged (e.g., socially threatening) questions 82 as well as a reduction, or elimination, of social desirability effects. 83 This feature is of special importance when addressing the ethically tenuous topic of placebos in clinical care.
The disadvantages of web-based surveys include several aspects. For example, they exclude responses from individuals without internet access thereby introducing coverage error. 65 The majority of university professors, physicians and government officials, however, generally have internet access, thus minimizing the coverage error. 71,84 Furthermore, in any survey, including a web-based survey, respondents differ from the non-respondents in terms of demographics and attitude resulting in non-response error. 85 Nonetheless, research reports comparing Internetsuch as the one we report in the present surveyand mail survey methodology suggest that differences between responders and non-responders are likely small. 86 In addition, this literature contains no account of response bias based on demographic characteristics. Finally, web-based surveys are susceptible to multiple survey completions by the same person (i.e., "ballot stuffing"). We have implemented certain technological measures, such as the use of cookies and IP addresses, to avoid duplicate responses. 65 The present survey attempts to estimate the prevalence of attitudes and behaviours in a population of physicians using a self-selected sample. Although it is theoretically possible that the physicians who chose to complete our survey were already those most likely to use placebos, in light of the abovementioned explanations this possibility is unlikely. It is likely however, that our results represent specific trends and capture clinical undercurrents that may be of general interest.

Conclusions
With Internet access becoming ubiquitous, online surveys loom as potentially powerful tools to probe such populations as academic physicians. Web-based technologysuch as the one we used in the present studyis hardly a remedy to all survey research problems. Similar to other tools in a researcher's toolkit, however, Internet surveys fit some tasks better than others.
Although such methods require further refinement, using this tool appropriately paves the road to a more scientific way of practicing eHealth. 65 Although no difference existed between the 20% of physicians reporting that they had either prescribed or administered a placebo, psychiatrists were less likely to report that placebos had no clinical benefit. Overall, psychiatrists' attitudes appear less stringent towards applying placebos in clinical practice.
Defining placebos continues to be a source of debate and confusion within the medical community. 1,3,4,18,87 This tenuous gray zone probably plays at least some role in obtaining results from over 35% of psychiatrists, who report prescribing subtherapeutic doses without expected clinical efficacy, while considerably fewer of them report prescribing placebos.  Percentages may not add up to 100% because each physician could select multiple options.  Percent rating of agreement from psychiatrists (inner circle) and other physicians (outer circle) to Item #13: "I believe placebos have therapeutic effects." Answers ranged from 1 (strongly disagree) to 5 (strongly agree). χ 2 (4) = 22.74, P < 0.001 between the two groups.