The study was published online Tuesday in the journal the Lancet. It was paid for by the National Joint Registry.
Ashley Blom, head of orthopedic research at the University of Bristol and one of the study authors, emphasized most people with a metal hip haven't needed a replacement. But with so many alternatives, he said, there was no reason to take that risk.
"If I were a patient, I would not choose a metal-on-metal hip," he said.
He said the rates of failing metal hips were likely an underestimate since not all patients report their symptoms or have surgery to fix the problem.
In 2010, DePuy, part of Johnson & Johnson, recalled a metal hip implant after it was linked to high failure rates. Blom said the new analysis suggested the problem applied to all metal-on-metal hips, not just one brand. Doctors began using metal-on-metal implants after laboratory tests suggested the devices would be more resistant to wear and reduced the chances of dislocation. They aren't sure why that isn't the case once they are used in patients.
Last year, the U.S. Food and Drug Administration asked all manufacturers of metal-on-metal hips to conduct safety studies. Use of the devices has dropped dramatically in recent years worldwide. In the United Kingdom, only about 5 percent of patients are believed to be getting the metal hips.
In the United States, estimates are 500,000 people have them.
Some experts called for tighter regulation, warning there might be similar problems with other joint replacements, such as those for knees and shoulders.
"I wouldn't be surprised if this was just the beginning of the storm," said Art Sedrakyan, an associate professor of public health at Weill Cornell Medical College in New York, who wrote an accompanying commentary in Lancet. "A lot of products have been allowed onto the market without clinical evidence they work."