Simon Hausberger / Getty

After giving birth to a baby, a young woman told her nurses at Boston Medical Center that she was having pain in her hip. That happens sometimes after births, says Ali Guermazi, one of the doctors involved. As he recounts the case from a few years ago, he looked at X-rays and saw a small amount of extra fluid in the joint. Otherwise things looked normal. “We injected her hip with steroids, hoping to help with the pain,” Guermazi says. They seemed to help, and the women went home with her baby.

Guermazi didn’t think more of it until the woman returned to the hospital six months later, unable to walk. “The head of her femur was gone,” says Guermazi, who is now the chief of radiology at VA Boston Healthcare System. The bone appeared to have simply vanished. The new mother needed a total hip replacement. “We didn’t know what happened, and still can’t know for certain,” Guermazi says. “But I feared it was related to the injection.”

This is not a typical suspicion. Doctors have long considered a single injection of steroids—the type that come from the adrenal glands and modulate the body’s stress response—to be a pretty harmless way to temporarily relieve pain in a joint. The worst-case scenario was that the shot didn’t help the pain. Some people get temporary relief, and some do not. Such injections are done by podiatrists, rheumatologists, orthopedists, spine neurosurgeons, anesthesiologists, and others at major hospitals around the world.

As a specialist in joint pain, Guermazi has done thousands of steroid injections over decades of work. He has trained other doctors as he was trained: to believe that the injections are safe as long as they aren’t overused. But now he has come to believe that the procedure is more dangerous than he knew. And he and a group of his Boston University colleagues are raising a warning flag for doctors and patients alike.

Millions of times every year, people with joint pain allow doctors to run a needle through their skin, then their muscle, then their tendons, and into the fluid-filled space of a painful joint to calm inflammation. Such inflammation can be the result of many types of injury or disease, but most commonly it is the result of gradual wear and tear known as osteoarthritis, in which the cartilage diminishes, the space between the bones narrows, and eventually bones start to rub on one another. At that stage, a person may need a surgical joint replacement. The progression of the disease itself can’t be reversed with drugs, so medical treatment is aimed at easing pain and maximizing mobility. Steroid injections are one of the chief ways this is attempted.

In the journal Radiology this week, Guermazi and his colleagues at Boston University published a study of 459 patients at their hospital who got injections, in the hips or knees, in 2018. Of those patients, 8 percent had complications that worsened the state of their joints. In some cases, the arthritis actually sped up. Others developed small fractures under the cartilage or had complications that compromised the blood supply to bone. In the worst cases, patients had what Guermazi and his colleagues described as “rapid joint destruction.”

Patterns of harm can be slow to emerge in medicine, and causal relationships are difficult to prove. But these findings build on a gradual accretion of evidence challenging the widespread use of steroid injections. In 2015, Cochrane Musculoskeletal did a meta-analysis to see if the intervention was even helpful. After collating data from 27 knee-arthritis trials carried out around the world, the authors concluded that the quality of evidence was low and overall inconclusive. Some of the studies they analyzed found small to moderate improvements in pain and physical function, but the results were not statistically reliable. Whether there is truly any positive effect, the authors concluded, is “unclear.”

Since then, the role of the placebo effect in steroid injections has gotten attention. In 2017, rheumatologists at Tufts University and Boston University did a randomized controlled trial in people with knee pain. A control group got a “sham” injection that contained no steroids. In what became a bombshell paper in the journal JAMA, people with knee arthritis reported that their pain was no different if they received injections of steroids or saline. What’s more, the people who got the steroid injections saw more erosion in the cartilage in their knees.

These less-than-promising findings tend to be overshadowed by anecdotes from many people who receive the injections and say they feel like they’ve magically received a new knee. Doctors and patients hoping to keep a person ambulatory, and to stave off a major surgery such as a joint replacement, might have a bias toward hoping that the injections are indeed a wise choice. Short on other options, steroid injections are still recommended in certain cases by the American College of Rheumatology and the Osteoarthritis Research Society International, with caution. The latest guidelines from the American Academy of Orthopaedic Surgeons equivocate on the injections, saying the evidence is not strong enough to recommend for or against them.

“The unfortunate thing is that there is no pharmaceutical treatment for osteoarthritis,” Guermazi says. The injections were only ever thought to be a temporary measure, but they were one of the few things in a doctor’s tool kit to help people with an often debilitating condition. “All the guidelines tell you to lose weight, exercise, and improve lifestyle. Those are the treatments,” Guermazi says.

He and his colleagues emphasized that two groups in particular should be cautious: young patients and anyone with pain that seems dramatically worse than might be expected (based on the history, imaging, and physical exam). Such disproportionate pain suggests a subtle problem that, perhaps, is being overlooked. Adding steroids to the mix could only make things worse, or delay an important finding. This may well have been the case for the young mother Guermazi treated. A tiny stress fracture could have been invisible in the X-ray. It would have required treatment by keeping weight off the leg. Instead, with steroids or a placebo creating some sense of relief, the woman felt able to walk on the hip, precipitating the collapse of the bone.

The procedure still likely has a role in helping people with arthritis in some cases, Guermazi believes. But he says that more research is “urgently needed” to help figure out what makes some people develop seemingly related complications, and how they might be prevented. Performing fewer injections could have massive financial ramifications for hospitals and doctors, and medicine is notoriously slow to change its ways in the face of new evidence. Fundamentally, though, Guermazi sees this as an ethical issue—as a matter of consent. Patients at least deserve to know about these possible complications. “As a doctor, I want to protect patients,” he says. “We are just saying we need to be careful.”

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.