David Mark07.01.09
According to Hooper's Medical Dictionary (1811), the word "placebo" was an epithet given to any medicine adapted more to please than benefit the patient. Use of placebos was common to medical practice (and still is), but it was not until the early 20th century that the concept of a placebo-controlled clinical trial entered the medical lexicon. Perhaps the need for better quality of evidence was driven in part by the needs of nascent pharmaceutical companies to differentiate their medical products from those of the unregulated "snake oil" era. Regardless, it wasn't until the 1960s that the clinical trial ideal became the randomized, double-blind, placebo-controlled trial.
So why do clinical trials need to be placebo-controlled? Because the placebo effect for subjectively reported symptoms such as pain or mental state averages roughly 30%. As shown in Table 1, the average percent of responders varies somewhat from condition to condition, but individual trials have reported positive response to the placebo group as low as 0% and as high as 76%.
These meta-analyses are useful in determining what variables impact the placebo effect. In general, a larger placebo effect is seen with:
• High subject expectation in benefiting from the treatment product(s);
• More severe disease at onset of study;
• Large trials compared to small trials;
• Long trials compared to short trials;
• More subject/investigator interaction, i.e., more visits per trial;
• More expensive or more invasive treatments; and
• Parallel group controlled trials compared to crossover trials.
During the past couple of decades, drug treatments for depression have become more effective due to the placebo effect. Walsh reports that in 1980 about 40% of patients were responding, increasing to 55% by 2000. During the same period the placebo response rate increased from 20% to 35%. Osteoarthritis trials in general report a placebo response around 40%. Yet GAIT (The Glucosamine Arthritis Intervention Trial), a large, long clinical trial with four treatment groups and one placebo group reported 60% responders in the placebo group. Placebo responses were also stronger in longer trials with more visits for Crohn's disease, and for larger studies for osteoarthritis. The osteoarthritis meta-analysis also confirmed that placebo treatment was significantly better compared to non-treatment.
Less consistent evidence supports injections having a greater placebo effect over capsules, and capsules over tablets-and for subjects who complete the trial and are in compliance with product use. For a time there were theories that women, or people with less education or lower socioeconomic status were more likely to be placebo responders. These theories have been discredited, but there are recent attempts to develop a screening test for "placebo responders."
There is some evidence that as dementia becomes progressively more severe the placebo effect diminishes. This makes sense, as diminished ability to manifest expectations would go hand in hand with diminished cognitive capacity.
And there is no placebo effect when you are asleep. The 40% response for Restless Leg Syndrome shown in Table 1 was for symptoms while awake or trying to go to sleep. In the same patients there was no placebo benefit for number of times being awoken in the night.
Interestingly, treatments may also elicit placebo adverse side effects. These are not the same as nocebo effects, which are the responses to a substance believed by the subject to be able to cause harm. In the "Discussion" section of clinical trial articles authors often conclude that as adverse effects were comparable between treatment and placebo, the test product is safe and has no side effects. However, if the test product is known (or thought) to have side effects then the presence of the same responses in the placebo group may be an example of placebo side effects. For example, when the Women's Health Initiative Study ended, 60% of the women who had been getting hormone replacement therapy and 40% of the women getting placebo suffered symptoms typical of hormone cessation, such as hot flashes.
There are objective measures and objective measures. Blood tests such as cholesterol, glucose, cell counts and liver function do not appear to be responsive to the placebo effect. But physical measures such as blood pressure, respiratory function, circulation and pulse can be responsive to placebo treatment, albeit in general not as strongly as perception of pain or change to mental state.
Research on the causes of placebo effect has made great advances in recent years. Proposed mechanisms include conditioning and expectation, with the majority of experts favoring the latter as the primary force. Benedetti proposes that pain and mental state fit better into the expectation model, while classical conditioning may be an explanation for responses in immunological function, hormonal and respiratory responses (such as asthma). Functional MRI conducted while a placebo treatment is administered confirms activity in brain regions involved in processing of perception, pain and emotion. Use of the opioid antagonist naloxone reverses placebo-induced pain relief, suggesting production of endogenous opioids as a mechanism.
The mechanisms for nocebo (Latin: "I shall harm") effects are also conditioning and expectation, in the form of anxiety. Anticipatory anxiety can increase production of cholecystokinin (CCK), resulting in an increased pain response. Much as the naloxone reduces placebo effects, CCK antagonists blunt nocebo effects. Schenk points out certain "word traps" can function as conditioning aural nocebos: "Just breathe normally" "Don't worry, but…." and "This will only hurt a bit."
The worst-case scenario is a charismatic doctor who invents or champions a novel treatment-and then writes a book. As this doctor considers it unethical to deprive patients of the clearly worthy treatment, the published evidence is usually in the form of testimonials or an open-label, uncontrolled trial. The certainty of the healer raises expectations in the patients, leading to more positive results. Galen had it right when he said: "He cures most in whom most are confident."
"It works for me" is a common refrain when athletes are asked about sports performance dietary supplements and functional foods. Caffeine does work. A review of 21 studies reported an average reduction in perceived exertion of almost 6% and an exercise performance increase of more than 11%. And the caffeine effect is well enough known among athletes to manifest a placebo effect. Six trained cyclists did repeat 10K time trials after being given drinks they were told contained either 0, 300 mg or 600 mg caffeine. In reality, all of the drinks contained zero caffeine. When the cyclists thought they had consumed 300 mg their performance was more than 1% better; with 600 mg, more than 3% better.
The practice of complementary and alternative medicine owes to its own desire to be credible the need to use the same tools mainstream medicine has honed-controlled clinical trials of adequate design and execution, leading to results supportive of evidence-based practice. For health conditions known to be responsive to placebo treatment this means designing placebo-controlled trials of adequate size to fairly test the hypothesis.
A "Catch 22" type problem to overcome in designing and conducting clinical trials is that the greater the belief the product works, the larger the placebo effect will likely be. Increasing size, duration and number of visits also tend to increase the placebo effect, as do other variables. In theory, trials can be designed to reduce the placebo effect, thus requiring smaller numbers or enrollees to be adequately powered for statistical analysis.
In a sadly ironic note, there is a Cochrane Collaboration 2009 report titled "Placebo Interventions for All Clinical Conditions." Cochrane reviews are notorious for setting such high standards for inclusion that most of the evidence is excluded, and then typically reach a "theory not supported" conclusion. In this report the authors concluded, "There was no evidence that placebo interventions in general have clinically important effects."
References furnished upon request.
So why do clinical trials need to be placebo-controlled? Because the placebo effect for subjectively reported symptoms such as pain or mental state averages roughly 30%. As shown in Table 1, the average percent of responders varies somewhat from condition to condition, but individual trials have reported positive response to the placebo group as low as 0% and as high as 76%.
 |
• High subject expectation in benefiting from the treatment product(s);
• More severe disease at onset of study;
• Large trials compared to small trials;
• Long trials compared to short trials;
• More subject/investigator interaction, i.e., more visits per trial;
• More expensive or more invasive treatments; and
• Parallel group controlled trials compared to crossover trials.
During the past couple of decades, drug treatments for depression have become more effective due to the placebo effect. Walsh reports that in 1980 about 40% of patients were responding, increasing to 55% by 2000. During the same period the placebo response rate increased from 20% to 35%. Osteoarthritis trials in general report a placebo response around 40%. Yet GAIT (The Glucosamine Arthritis Intervention Trial), a large, long clinical trial with four treatment groups and one placebo group reported 60% responders in the placebo group. Placebo responses were also stronger in longer trials with more visits for Crohn's disease, and for larger studies for osteoarthritis. The osteoarthritis meta-analysis also confirmed that placebo treatment was significantly better compared to non-treatment.
Less consistent evidence supports injections having a greater placebo effect over capsules, and capsules over tablets-and for subjects who complete the trial and are in compliance with product use. For a time there were theories that women, or people with less education or lower socioeconomic status were more likely to be placebo responders. These theories have been discredited, but there are recent attempts to develop a screening test for "placebo responders."
There is some evidence that as dementia becomes progressively more severe the placebo effect diminishes. This makes sense, as diminished ability to manifest expectations would go hand in hand with diminished cognitive capacity.
And there is no placebo effect when you are asleep. The 40% response for Restless Leg Syndrome shown in Table 1 was for symptoms while awake or trying to go to sleep. In the same patients there was no placebo benefit for number of times being awoken in the night.
Interestingly, treatments may also elicit placebo adverse side effects. These are not the same as nocebo effects, which are the responses to a substance believed by the subject to be able to cause harm. In the "Discussion" section of clinical trial articles authors often conclude that as adverse effects were comparable between treatment and placebo, the test product is safe and has no side effects. However, if the test product is known (or thought) to have side effects then the presence of the same responses in the placebo group may be an example of placebo side effects. For example, when the Women's Health Initiative Study ended, 60% of the women who had been getting hormone replacement therapy and 40% of the women getting placebo suffered symptoms typical of hormone cessation, such as hot flashes.
What About Objective Measures of Health?
There are objective measures and objective measures. Blood tests such as cholesterol, glucose, cell counts and liver function do not appear to be responsive to the placebo effect. But physical measures such as blood pressure, respiratory function, circulation and pulse can be responsive to placebo treatment, albeit in general not as strongly as perception of pain or change to mental state.
Placebo Theory
Research on the causes of placebo effect has made great advances in recent years. Proposed mechanisms include conditioning and expectation, with the majority of experts favoring the latter as the primary force. Benedetti proposes that pain and mental state fit better into the expectation model, while classical conditioning may be an explanation for responses in immunological function, hormonal and respiratory responses (such as asthma). Functional MRI conducted while a placebo treatment is administered confirms activity in brain regions involved in processing of perception, pain and emotion. Use of the opioid antagonist naloxone reverses placebo-induced pain relief, suggesting production of endogenous opioids as a mechanism.
The mechanisms for nocebo (Latin: "I shall harm") effects are also conditioning and expectation, in the form of anxiety. Anticipatory anxiety can increase production of cholecystokinin (CCK), resulting in an increased pain response. Much as the naloxone reduces placebo effects, CCK antagonists blunt nocebo effects. Schenk points out certain "word traps" can function as conditioning aural nocebos: "Just breathe normally" "Don't worry, but…." and "This will only hurt a bit."
Placebo Effects in Medical Practice
The worst-case scenario is a charismatic doctor who invents or champions a novel treatment-and then writes a book. As this doctor considers it unethical to deprive patients of the clearly worthy treatment, the published evidence is usually in the form of testimonials or an open-label, uncontrolled trial. The certainty of the healer raises expectations in the patients, leading to more positive results. Galen had it right when he said: "He cures most in whom most are confident."
Sports Performance & Placebo Effects
"It works for me" is a common refrain when athletes are asked about sports performance dietary supplements and functional foods. Caffeine does work. A review of 21 studies reported an average reduction in perceived exertion of almost 6% and an exercise performance increase of more than 11%. And the caffeine effect is well enough known among athletes to manifest a placebo effect. Six trained cyclists did repeat 10K time trials after being given drinks they were told contained either 0, 300 mg or 600 mg caffeine. In reality, all of the drinks contained zero caffeine. When the cyclists thought they had consumed 300 mg their performance was more than 1% better; with 600 mg, more than 3% better.
Conclusion
The practice of complementary and alternative medicine owes to its own desire to be credible the need to use the same tools mainstream medicine has honed-controlled clinical trials of adequate design and execution, leading to results supportive of evidence-based practice. For health conditions known to be responsive to placebo treatment this means designing placebo-controlled trials of adequate size to fairly test the hypothesis.
A "Catch 22" type problem to overcome in designing and conducting clinical trials is that the greater the belief the product works, the larger the placebo effect will likely be. Increasing size, duration and number of visits also tend to increase the placebo effect, as do other variables. In theory, trials can be designed to reduce the placebo effect, thus requiring smaller numbers or enrollees to be adequately powered for statistical analysis.
In a sadly ironic note, there is a Cochrane Collaboration 2009 report titled "Placebo Interventions for All Clinical Conditions." Cochrane reviews are notorious for setting such high standards for inclusion that most of the evidence is excluded, and then typically reach a "theory not supported" conclusion. In this report the authors concluded, "There was no evidence that placebo interventions in general have clinically important effects."
References furnished upon request.
