The placebo effect
The treatment that works without working
Description
Italy, 1944. A field hospital near Anzio is running out of morphine. The American anesthesiologist Henry Beecher is treating wounded soldiers under bombardment, and the supply has not arrived. A nurse, in what may have been desperation or improvisation, injects a saline solution into a screaming soldier and tells him it is morphine. The soldier calms down. He becomes operable. Beecher watched it happen, then saw it happen again, with other soldiers, with other nurses. Salt water in a syringe was producing what looked like the same effect as a powerful opioid. He spent the rest of his career trying to figure out what he had seen.
Beecher returned to Harvard, and in 1955 published a paper called The Powerful Placebo in the Journal of the American Medical Association. He pulled together fifteen clinical studies and argued that around 35% of patients responded to inactive treatments — sugar pills, saline injections, sham procedures. The number became famous. It also turned out to be sloppy: later researchers showed Beecher had conflated several different things, including spontaneous recovery and regression to the mean. But the underlying observation was real. People given a treatment they believe is real often improve. The improvement is not imagined.
The placebo effect is now the foundation of how we test drugs. Every randomized controlled trial built since the 1960s has had to demonstrate that the active compound outperforms an inactive one. The placebo arm is the bar the drug must clear. In some conditions — depression, pain, irritable bowel — the bar has been moving upward, with placebo response rates climbing in trial after trial. The phenomenon Beecher named is closer to the center of medicine than most patients realize.
The question we're asking: what is the placebo effect, and what does its existence say about how medicine measures whether something works?
What we'll see: the wartime origin, the trial architecture built on it, the surgery studies that nobody expected, and what the effect tells us about the framing of treatment itself.
Table of contents
01A nurse at Anzio
Beecher's wartime memory is the founding scene of placebo research, but the phenomenon was not new. Physicians for centuries had noticed that patients sometimes improved on treatments now known to be inert — bread pills, colored water, theatrical hand-waving over the bed. The word placebo comes from medieval Catholic vespers for the dead, where the chant began with placebo Domino. By the eighteenth century, English physicians used the term semi-pejoratively for prescriptions given to please the patient rather than to heal. What changed in the twentieth century was that someone began measuring it.
Beecher was the right person to ask the question. He had watched soldiers describe pain in conditions where pain should have been overwhelming, and noticed that battlefield injury produced less reported suffering than the same injury would in a civilian. The context, he concluded, was doing something the tissue damage could not explain. Back at Harvard, he ran experiments giving patients saline and telling them it was a painkiller, then giving them morphine and telling them it was saline. The results varied wildly depending on what the patient was told. The variable doing the work was the expectation.
02Building trials around it
The randomized controlled trial is the workhorse of modern medicine, and the placebo group is what makes it work. Patients are randomly assigned to receive either the experimental treatment or an indistinguishable inactive substitute. Neither patients nor clinicians know who received what — the double-blind design — until the data is unblinded. The 1948 streptomycin trial for tuberculosis is often cited as the first modern RCT. By the 1970s, double-blind placebo-controlled trials were the standard.
The architecture works because the placebo group does most of the same things the treatment group does. Both populations are anxious about being sick, hopeful about being treated, and given pills on a schedule. The only difference is the active compound. If the treatment group improves more than the placebo group by a margin that can't be explained by chance, the drug is doing something. The framing forces an honest accounting of what medicine is contributing beyond context.
03The sham surgery problem
If placebo effects were limited to mild subjective complaints, the phenomenon would be a footnote. The studies that made it impossible to dismiss involved operating rooms. In the late 1950s, the cardiologist Leonard Cobb compared internal mammary artery ligation, then a popular treatment for angina, against a sham operation in which the surgeon made the skin incision but did nothing inside. Both groups improved. The sham group improved as much as the real-surgery group. The procedure was abandoned.
Forty years later, the orthopedic surgeon Bruce Moseley ran a similar trial at the Houston VA Medical Center on arthroscopic knee surgery for osteoarthritis. Patients were randomized to one of three arms: real arthroscopic debridement, real arthroscopic lavage, or a sham surgery in which incisions were made and the surgeon mimed the procedure. The 2002 paper in the New England Journal of Medicine reported no difference between the three groups at two years. Hundreds of thousands of arthroscopic knee procedures done annually in the United States were producing no benefit beyond the sham. The finding has since been replicated for vertebroplasty and several arthroscopic shoulder procedures.
04What the effect tells us
The placebo effect is sometimes presented as evidence that belief heals — a soft, almost mystical claim. The serious version is narrower. Expectation, context, attention, and ritual modulate measurable neurochemistry in specific systems, particularly those regulating pain, mood, and visceral sensation. They do this whether the patient knows the treatment is inert or not. Open-label placebo trials, where patients are explicitly told the pill is inactive and still report benefit, have replicated in conditions including irritable bowel syndrome and chronic low back pain. The mechanism is not deception.
This has implications for the question, does this drug work? The framing assumes a binary: either the molecule does something or it does not. The reality is layered. Many treatments work partly because of the molecule, partly because of the ritual surrounding it, and partly because of the natural course of the disease. Disentangling those three contributions is what trials are for. The placebo effect is what makes the disentangling possible — and also what makes some weak drugs look better than they are when the placebo arm happens to underperform.
05Conclusion
Henry Beecher's saline injection at Anzio was not a discovery in the strict sense physicians had been administering inert substances and watching patients improve for centuries. What Beecher started was the willingness to treat the placebo response as something that needed to be measured rather than dismissed. The result is the modern clinical trial, an apparatus that has caught a great many ineffective treatments and approved a great many real ones. It has also revealed that the line between the two is sometimes harder to find than the original framing suggested.