– Newly unearthed documents show that the Food and Drug Administration (FDA) failed to use its policing powers to make sure a program to curb improper prescribing of opioids was effective, researchers say.

The lax oversight, they point out, occurred as the epidemic was growing and tens of thousands of people were dying from overdoses each year.

In 2011, the FDA began asking the makers of OxyContin and other addictive long-acting opioids to pay for safety training for more than half the physicians prescribing the drugs, and to track the effectiveness of the training and other measures in reducing addiction, overdoses and deaths.

But the FDA was never able to determine whether the program worked, researchers at Johns Hopkins Bloomberg School of Public Health found in a new review, because the manufacturers did not gather the right kind of data. Although the agency's approval of OxyContin in 1995 has long come under fire, its efforts to ensure the safe use of opioids since then have not been scrutinized nearly as much.

The documents show that even when deficiencies in these efforts became obvious through the FDA's own review process, the agency never insisted on improvements to the program, called a risk evaluation and mitigation strategy, or REMS. The new research was published Monday in JAMA Internal Medicine.

"What's surprising here is the design of the program was deficient from the start," said Caleb Alexander, the senior author of the study, who serves as a paid expert witness in litigation against opioid manufacturers and distributors. "It's unclear why the FDA didn't insist upon a more scientifically rigorous evaluation of this safety program."

Dr. Andrew Kolodny, the co-director of opioid policy research at the Heller School for Social Policy and Management at Brandeis, said the safety program was a missed opportunity. He is a leader of a group of physicians who had encouraged the FDA to adopt stronger controls.

He called the program "a really good example of the way FDA has failed to regulate opioid manufacturers. If FDA had really been doing its job properly, I don't believe we'd have an opioid crisis today."

In 2007, Congress gave the agency the authority to require drug manufacturers to train physicians to safely prescribe certain dangerous drugs and to monitor the companies' performance. The bill, lawmakers said, was influenced by agency lapses in oversight of drugs like Vioxx, a popular painkiller withdrawn from the market in 2004 after it was found to pose a substantial heart risk.

The main goals of the 2011 program for long-lasting opioids were to train more than half of an estimated 320,000 doctors who prescribed the drugs on how to do so safely, and to inform patients about the significant risks of taking the drugs.

But Hopkins researchers, who relied on thousands of pages of internal FDA documents obtained through the Freedom of Information Act, found the agency repeatedly could not determine whether the companies' safety strategies were working because of poor study designs, which the agency itself had approved.

Asked for a response, the FDA had no immediate comment.

There were early warning signs that the safety program for long-acting opioids had weaknesses. In 2010, an FDA advisory committee of experts in the treatment of pain voted 25-10 against the proposed program design. Committee members suggested that the agency require training for prescribers and that it redesign curriculums to lessen industry influence. The FDA moved forward anyway.

A 2013 report by the inspector general of the Department of Health and Human Services found that only 14% of the safety programs reviewed by the FDA met their goals. Another inspector general report last year noted that the agency did not have the authority to take enforcement actions against companies that don't provide it with enough information to assess their safety programs — a common problem, it found.

"If FDA does not have comprehensive data to monitor the performance" of the program, the report said, "it cannot ensure that the public is provided maximum protection from a drug's known or potential risks."

In the case of long-acting opioids, the FDA wanted 60% of prescribers to take the classes developed as part of the program. But only about 27% did so within the specified time frame, 2012 to 2016.

Perhaps most important, researchers found that manufacturers didn't do a good job of assessing whether the safety program led to fewer overdoses and deaths. Instead they relied on broad national data that made no distinction between patients of doctors who had taken the safety classes and those who had not — a problem the FDA pointed out repeatedly, in the third, fourth and fifth years of the program.