The results, published in Thursday's New England Journal of Medicine, could lead to big changes in the way health care providers handle a surgery that is done on more than 300,000 Americans each year.
The arthroscopic surgery is performed to ease the pain of osteoarthritis, the steadily worsening, wear and tear on joints that affects 12 percent of senior citizens. The operation involves clearing out debris or repairing damaged cartilage after making tiny incisions that allows patients to heal quickly.
In a type of study only rarely conducted, some patients got a real knee operation, while others underwent sham surgery.
At every point over the next two years, those who had the fake surgery could climb stairs and walk slightly faster on average than those who had gotten real operations.
Arthroscopic knee surgery for osteoarthritis costs roughly $5,000, or a total of $1.5 billion a year in the United States, said Dr. Nelda Wray, a researcher at the Houston VA Medical Center who led the study.
"We think that money could be better spent," she said.
The findings could prompt insurance companies to refuse to pay for the procedure. Other treatments for arthritic knees include pain relievers, cortisone shots, hot packs, exercise and, as a last resort, knee-replacement surgery.
Dr. William W. Tipton Jr., executive vice president of the American Academy of Orthopedic Surgeons, said other researchers should duplicate the results before doctors, patients and insurance companies react. But he said the study confirms some doctors' growing suspicions.
He said many orthopedic surgeons already feel that arthroscopic surgery should not be used except in certain circumstances, such as acute sports injuries to ligaments. Tipton added that he has arthritis but is opting for knee replacement surgery.
Surgical procedures are often not subjected to the kind of rigorous testing done on medications. Some critics consider sham surgery unethical because of the risk of harm to study volunteers.
But Sam Horng and Franklin G. Miller of the National Institutes of Health said in an accompanying editorial that the VA study "exemplifies the ethically justified use of placebo surgery."
In the study, 180 veterans 75 or younger were divided into three groups for arthroscopic surgery at the medical center, all done by the same expert surgeon between 1995 and 1998.
One-third had arthroscopic debridement, in which debris and rough and torn pieces of cartilage are removed or sanded down. Another third had arthroscopic lavage, in which debris and calcium phosphate crystals are flushed out. The remaining third likewise were anesthetized and had three incisions made in the knee, but the surgeon only manipulated the knee as if working on it.
No one knew which procedure they had for two years, during which they received regular evaluations.
During that period, all three groups said they had slightly less pain and better knee movement, with the sham-surgery group often reporting the best results. That apparently reflects the well-known "placebo effect," in which patients feel better simply because they believe they have been treated.
However, tests in which the patients were timed as they walked and climbed up and down stairs found that patients in the two treatment groups were slower than they were before surgery.
Dr. Todd P. Stitik, associate professor of rehabilitation at University of Medicine and Dentistry of New Jersey in Newark, said patients he has sent for arthroscopic surgery were not helped much.
"I don't think this is now necessarily a reasonable option," he said.
He said his patients have benefited more from a relatively new treatment — injections of a synthetic version of fluid inside the knee that acts as a shock absorber and lubricant.