Medical Scans Not Accurate

Cheryl Weinstein’s left knee bothered her for years, but when it started clicking and hurting when she straightened it, she told her internist that something was definitely wrong.

It was the start of her medical odyssey, a journey that led her to specialists, physical therapy, Internet searches and, finally, an MRI scan that showed a torn cartilage and convinced her that her only hope for relief was to have surgery to repair it. But in fact, fixing the torn cartilage that was picked up on the scan was not going to solve her problem, which, eventually, she found was caused by arthritis.

Scans — more sensitive and easily available than ever — are increasingly finding abnormalities that may not be the cause of the problem for which they are blamed.

The scans are expensive — Medicare and its beneficiaries pay about $750 to $950 for an MRI scan of a knee or back, for example. Many doctors own their own scanners, which can provide an incentive to offer scans to their patients.

It is common for doctors and patients to assume that any abnormalities found are the reason for the pain.

But in many cases it is just not known whether what is seen on a scan is the cause of the pain. The problem is that all too often, no one knows what is normal.

“Every time we get a new technology that provides insights into structures we didn’t encounter before, we end up saying, ‘Oh, my God, look at all those abnormalities.’ They might be dangerous,” said Dr. David Felson, a professor of medicine and epidemiology at Boston University Medical School. “Some are, some aren’t, but it ends up leading to a lot of care that’s unnecessary.”

That was what almost happened with Weinstein, an active, athletic 64-year-old who lives in London, N.H. And it was her great fortune to finally visit a surgeon who told her so. He told her bluntly that her pain was caused by arthritis, not the torn cartilage.

No one had told her that before, Weinstein said, and looking back on her quest to get a scan and get the ligament fixed, she shook her head in dismay. There’s no surgical procedure short of a knee replacement that will help, and she’s not ready for a knee replacement.

Weinstein was lucky her problem was with her knee. It’s one of only two body parts — the other is the back — where there are good data on abnormalities that turn up in people who feel just fine, indicating that the abnormalities may not be so abnormal after all.

But even the data on knees come from just one study, and researchers say the problem is far from fixed. It is difficult to conduct scans on people who feel fine — most do not want to spend time in an MRI machine, and CT scans require that people be exposed to radiation.

But that leaves patients and doctors in an untenable situation.

“It’s a concern, isn’t it?” said Dr. Jeffrey Jarvik, a professor of radiology and neurosurgery at the University of Washington. “We are trying to fix things that shouldn’t be fixed.”

As a rheumatologist, Felson saw patient after patient with knee pain, many of whom had already had scans. And he was becoming concerned about their findings.

Often, a scan would show that a person with arthritis had a torn meniscus, a ligament that stabilizes the knee. And often the result was surgery — orthopedic surgeons do more meniscus surgery than any other operation. But, Felson wondered, was the torn ligament an injury causing pain or was the arthritis causing pain and the tear a consequence of arthritis?

That led Felson and his colleagues to do the first and so far the only large study of knees, asking what is normal. It involved MRI scans on 991 people ages 50 to 90. Some had knee pain, others did not.

On Sept. 11, Felson and his colleagues published their results in The New England Journal of Medicine: Meniscal tears were just as common in people with knee arthritis who did not complain of pain as they were in people with knee arthritis who did have pain. They tended to occur along with arthritis and were a part of the disease process itself.

“The rule is, as you get older, you will get a meniscal tear,” Felson said. “It’s a function of aging and disease. If you are a 60-year-old guy, the chance that you have a meniscal tear is 40 percent.”

It is a result that paralleled what spine researchers found over the past decade in what is perhaps the best evidence of what shows up on scans of healthy people. “If you’re going to look at a spine, you need to know what that spine might look like in a normal patient,” said Dr. Michael Modic, chairman of the Neurological Institute at the Cleveland Clinic.

After Modic and others scanned hundreds of asymptomatic people, they learned abnormalities were common.

“Somewhere between 20 and 25 percent of people who climb into a scanner will have a herniated disk,” Modic said. As many as 60 percent of healthy adults with no back pain, he said, have degenerative changes in their spines.

Those findings made Modic ask: Why do a scan in the first place? Some may benefit from surgery, but does it make sense to routinely do scans for nearly everyone with back pain? After all, one-third of herniated disks disappear on their own in six weeks, and two-thirds in six months.

So Modic did another study, this time with 250 patients. All had MRI scans when they first arrived complaining of back pain or shooting pains down their leg, which can be caused by a herniated disc pressing on a nerve in the spine. And all had scans again six weeks later. Sixty percent had herniated disks, the scans showed.

Modic gave the results to only half of the patients and their doctors — the others had no idea what the MRIs revealed.

In 13 percent of the patients, the second scan showed that the herniated disk had become bigger or a new herniated disk had appeared. In 15 percent, the herniated disk had disappeared. But there was no relationship between the scan findings at six weeks and patients’ symptoms. Some continued to complain of pain even though their herniated disk had disappeared; others said they felt better even though their herniation had grown bigger.

The question, though, was whether it helped the patients and their doctors to know what the MRIs had found. And the answer, Modic reported, is that it did not. The patients who knew recovered no faster than those who did not know.

Scans, he said, are presurgical tools, not screening tools. A scan can help a surgeon before he or she operates, but it does not help with a diagnosis.

“If a patient has back or leg pain, they should be treated conservatively for at least eight weeks,” Modic said, meaning that they take pain relievers and go about their normal lives. “Then you should do imaging only if you are going to do surgery.”

When Weinstein started looking up her symptoms on the Internet, she decided she probably had a meniscus tear. “I was very forceful in asking for an MRI,” she said.

And when the scan showed that her meniscus was torn, she went to a surgeon expecting an operation.

He x-rayed her knee and told her she had arthritis. Then, Weinstein said, the surgeon looked at her and said, “Let me get this straight. Are you here for a knee replacement?”

She said no, of course not. She skis, she does aerobics, she was nowhere near ready for something so drastic.

Then the surgeon told her that there was no point in repairing her meniscus because that was not her problem. And if he repaired the cartilage, her arthritic bones would just grind it down again.

For now, Weinstein says she is finished with her medical odyssey.