This is a story about placebos.
Placebos offer a simple conceit – a sugar pill, consumed by a suitably optimistic patient, provides the same effect as the drug it mimics. Headaches are relieved. Heart rates are lowered. Confidence is built.
Except the whole idea – the “placebo effect” – is in fact a carefully constructed fraud.
To understand this story – how the entire medical establishment came to believe a complete fabrication – we have to first understand a bit about the history of medicine, a bit about the ethics of treatment, and a bit about the structure of science.
So strap in for a ride.
It was late August of 1793, and panic gripped the residents of Philadelphia.
An epidemic of Yellow Fever terrorized the city; one third of the infected succumbed to the illness. Half of Philadelphia’s residents fled to the countryside or nearby cities; local committees closed bridges and roads to stop the contagion; children threw parents into the streets at the first sign of illness. Citizens who remained buried bodies in mass graves.
Into the abyss stepped a small coterie of brave doctors, led by Dr. Benjamin Rush. A signer of the Declaration of Independence, a professor at the University of Pennsylvania, and a prolific writer, Dr. Rush was the leading American physician of the 18th century. He authored America’s first chemistry textbook, as well as over 80 other significant publications. A brilliant student, Rush’s medical school expenses were reportedly borne by Benjamin Franklin himself, in return for a promise to practice medicine and teach in Pennsylvania.
Dr. Rush had developed a hypothesis of the mechanism of action of not just Yellow Fever, but all viruses. He theorized that viral infections operate via a uniform process, an “irregular convulsive action of the blood vessels”. And he proposed a simple, radical solution:
In 1794, he wrote:
I preferred frequent and small, to large bleedings in the beginning of September, but toward the height and close of the epidemic, I saw no inconvenience from the loss of a pint and even 20 ounces of blood at a time. I drew from many persons 70 and 80 ounces in five days, and from a few a much larger quantity.
Eighty ounces represents roughly 40% of the total blood volume of an adult. In his clinic, Dr. Rush treated over 100 patients a day, disposing of so much blood that the front yard of his clinic buzzed with flies.
How could a man later described as “the father of American medicine” adopt a practice that, by today’s standards, seems little short of an atrocity?
By Dr. Rush’s time, rational testing of a hypothesis had been practiced for centuries in physics. Yet the scientific method was slow to come to medicine, where the subject of study was not lead balls or glass prisms, but living, breathing humans.
A proper medical experiment requires a “control” group: remedies will be withheld from some patients, while the doctor worked to heal others. With humans involved, how could such a construct be considered? Would not a doctor be condemned for the suffering and deaths of those who he could have helped, but left untreated for the betterment of Science?
By the end of the 18th century, there were no reports of controlled trials in the medical literature. Experiments on humans were rightly considered an ethical abomination. Instead, doctors practiced as they always had through history, following the learning cobbled from their teachers and peers, and modifying treatments based on their personal insights.
Insights that led Dr. Benjamin Rush to, quite literally, let blood run freely onto the streets of Philadelphia.
By 1800, medicine had hardly progressed from the time of Aristotle. Doctors used bloodletting, leeches, and salts of mercury to “heal” their patients. No long term records were kept, and no comparisons made between treatments. Another 150 years would pass before science came to dominate medicine.
And in one of history’s greatest ironies, the trigger that pushed medicine into science’s embrace was itself a lie.
A statue of Dr. Benjamin Rush outside the Bureau of Medicine and Surgery in Washington, D.C. The statue was erected with funding provided by the American Medical Association. Photo from https://www.nlm.nih.gov/hmd/medtour/rush.html
Can an ethical doctor possibly experiment on his patients? If so, how?
A well-constructed experiment is, first an foremost, an admission of humility. The doctor – or, to use a more modern term, the medical researcher – applies his unique knowledge and experience to improve a treatment. Yet despite his obvious learning, the researcher cannot assume he is correct without proof.
More humbling still, the researcher must admit the possibility that his own beliefs and biases may sway the outcome. Some subjects will receive the experimental treatment, and some will not; if the researcher knows which group is which, he is prone to care for the two groups differently, and possibly influence the result.
And so, over the first half of the twentieth century, the medical profession established guidelines for what would become known as the randomized controlled trial. Prospective patients would be randomly placed into one of two groups: the first group would receive the experimental treatment, while the second would received an ineffective treatment, a “placebo”. Critically, neither the doctor nor patients would know who was assigned to each group – the experiment was “doubly blind”, minimizing bias. The patients, informed of the nature of the experiment, would participate only if they consented to the risks therein.
Was this ethical?
Scientifically, the logic behind the randomized controlled trial was sound. The approach clearly distinguished an effective from an ineffective treatment. In the long run, we would all benefit from the improved knowledge that controlled trials create. Medicine would slowly but surely improve.
Yet in the short run, such trials seemed to violate the original Hippocratic oath, “Into whatsoever houses I enter, I will enter to help the sick.” How could a doctor enter into the houses of the infirm with promises of healing, only to step back and let a roll of the dice control their destinies?
The resistance of the medical community to this new, scientific medicine was fierce. A randomized controlled trial was a complex, ethically dubious procedure, one that doctors resisted to protect their patients.
And this resistance held through World War II, until the publication of a highly influential paper in 1955, “The Powerful Placebo”, by Harvard anesthesiologist Dr. Henry Knowles Beecher.
“The Powerful Placebo” was a landmark in modern medicine, introducing the term “placebo effect” to the global lexicon.
Its thesis was simple: In a randomized controlled study, a placebo itself was more than just a technique to eliminate bias – the sugar pill could actually heal a patient. A patient’s belief that he was treated was enough to cause remission.
This idea was entirely credible to physicians, who had prescribed sham treatments to difficult patients for centuries. To wit, in 1775 the English Bishop John Douglas anticipated this finding, stating “That the eager Confidence of the Patient in the Skill of his Physician, and the firm Expectation of Relief by his Means, have sometimes a wonderful Efficacy in restoring Health, is a Point not to be doubted of.’’
Thomas Jefferson more concisely (and disapprovingly) summarized the condition of pre-scientific medicine, stating in 1807, “One of the most successful physicians I have ever known has assured me that he used more bread pills, drops of colored water, and powders of hickory ashes, than of all other medicines put together.”
Dr. Beecher’s evidence in favor of the “placebo effect” – a physiological impact of a sugar pill – was indeed powerful. 35% of patients improved with placebos alone. This improvement held for pain, nausea, and mood changes, and with a shocking consistency (± 2%), that “suggests that a fundamental mechanism in common is operating in these several cases, one that surely deserves further study.”
This placebo effect was revolutionary not merely for its scientific findings. Its biggest impact was to spark the scientific revolution by giving doctors permission to approve of randomized controlled trials.
Think about it: If patients improved on placebos alone, then a trial posed little ethical dilemma. Participation in the trial itself was a powerful act of healing.
Better still, Beecher had shown that placebo treatments that doctors had individually prescribed to quiet intractable patients were actually effective. A spoonful of sugar conferred respectability to both old medicine and new. The incentives to believe in this paper were profound for all doctors, both skeptics and advocates of the new, scientific medicine.
In a single paper, the ethical barriers that held back support for randomized controlled trials were swept away. The placebo effect provided the perfect story to entice doctors to embrace the modern era. Dr. Beecher became, rightly, a hero of medicine.
Except that the paper was, in fact, a sham.
A Google n-gram, measuring the prevalence of the word “placebo”.
In 1997, researchers in Germany revisited “The Powerful Placebo” and were stunned by what they found. They reviewed the details of the original 15 trials cited by Beecher in evidence of the placebo effect.
None of them supported his thesis.
Yes, patients in the placebo group often got better. But some also got worse (a fact that Beecher neglected to include in his statistics). A natural fluctuation of disease is common without any treatment, so improvement in a few patients provides no evidence of a special “placebo effect”. Other causes of improvement, such as observer bias, switching of treatments in the middle of the experiment, or a desire of the patient to please the experimenters, were lumped together into a single category, the placebo.
Worse, Beecher was found guilty of repeatedly “misquoting” research results.
As an extreme example of misquotation, the authors describe “Beecher’s claim that in a study of antitussive [anti-coughing] agents there was a placebo effect in 36% of 22 patients and in 43% of another 22 patients. However… under none of the placebo administrations could any significant change be demonstrated… Thus, Beecher’s quotation was wrong (which is amazing, as Beecher himself had been one of the authors of the original publication).” [emphasis added.]
Surely Beecher knew of the concept of spontaneous remission of disease. Even more surely he had been familiar with the details of his own papers. He was as pre-eminent a scientist in the 1950s and 1960s as Dr. Rush had been in the 1790s. Yet the 1997 review found that he had misquoted results in 10 out of the 15 papers studied, and clearly stretched his interpretations to make data fit the concept of a “powerful placebo” in the others.
Could Beecher have really made such basic mistakes? Or did he purposefully bias his own paper to demonstrate a “powerful placebo”, knowing that this would attract doctors to support his true goal: the establishment of controlled trials as central to medicine?
Whatever the cause, the paper was one of the most influential in modern medicine. Google Scholar tracks almost 2000 citations. Even grade school children have heard of the “placebo effect”.
And we hold to this belief to this day, even though subsequent studies have confirmed that the placebo effect does not actually exist.
In 2001, researchers published a paper comparing trials which had both a placebo and a “no-treatment” control, and found there were no differences between the two. Taking a placebo did not change patient outcomes.
At best, the reviewers detected a placebo effect only for self-reported results such as pain, and only when there is a continuous, graded outcome (such as a scale of 1-10) rather than a binary outcome (pain or “no pain”). And even in these cases, there is a great chance that patients are claiming mild improvements in order to meet the perceived wishes of the experimenter, not because they actually felt better.
The placebo effect can only be seen when patients have an opportunity to fudge the results.
Philadelphia experienced two more Yellow Fever epidemics in 1794 and 1797, and Dr. Benjamin Rush continued his practices of massive bloodletting. During the last outbreak, the journalist William Cobbett became a relentless critic, demonstrating that mortality rates in 1793 increased substantially after implementation of Rush’s methods. One article, entitled “Medical Puffing”, included the inflammatory doublet: “The times are ominous indeed, when quack to quack cries purge and bleed.”
Dr. Rush’s response was to sue Cobbett for libel. On December 14, 1799, a jury found Cobbett guilty and fined $5000, then the largest verdict in Pennsylvania history. For the moment, Rush was vindicated.
On the same day of the verdict, the first President of the United States, George Washington, lay dying in his bed. He was being treated by three physicians, the most senior of which had served with Rush in the Revolutionary War. Another had been a colleague of Rush in medical school. A third had been a student of Rush in Pennsylvania. All had clearly been influenced by the work of this great doctor.
As President Washington declined, his doctors let 80 ounces of his blood in less than 12 hours. Yet this treatment failed to revitalize him. Washington died that night, and scholars to this day question whether his physicians hastened his death.
Beecher and his placebo-controlled studies are rightly lauded for steering medicine away from such mythology and barbarism, to a world of treatments based on rigorous scientific inquiry. A world of science now protects us from Washington’s fate.
A doctor prescribed a placebo for the practice of medicine. And, against all odds, medicine got better.