Korva is a scroll of the internet’s best ideas.
Saline can make joints feel better
A shoulder injury
A physician with a complete labral tear in his shoulder asked orthopedic surgeons whether infusing a million stem cells could spare him surgery, and the answer was unambiguously no. A thousand people with that injury, injected with stem cells, would very likely see nothing change, given what the injury is.
The trial design problem
Some injuries might respond. The difficulty is trial design: a heterogeneous patient population buries a small signal. If only 10% of the patients are the right ones, the trial is underpowered to detect the effect and returns a null result.
The incentive problem
The incentive to run such a trial is hard to see, because providers make money from the procedures as they stand. A pharma angle is what it would take.
How strong is the placebo effect for procedures?
A 2006 randomized trial in the BMJ led by Ted Kaptchuk gave 270 people with arm pain either a sham acupuncture device or an inert pill. Both reduced pain, and the sham device did slightly better. Procedures that look more elaborate appear to signal a stronger treatment.
What did sham knee surgery show?
In a 2002 trial in the New England Journal of Medicine led by J. Bruce Moseley, 180 patients with knee osteoarthritis were randomized to arthroscopic debridement, lavage, or a placebo procedure in which surgeons made incisions but did not operate. Pain and function scores were no better in the surgical groups than in the placebo group over two years. The study contributed to a decline in arthroscopic surgery for knee osteoarthritis.
Why does a mixed patient group hide results?
A trial's statistical power depends on the size of the average effect. If only 10 percent of patients can respond, the average effect shrinks to a tenth, and the sample needed grows roughly with the inverse square of the effect size, which makes it about a hundred times larger. Oncology trials increasingly enroll only patients whose tumors carry a specific mutation for this reason.
More from Andrew Huberman
