Correction: An earlier version of this article incorrectly described the role of the Office for Human Research Protections. It does not investigate allegations of research fraud. Its investigations are limited to issues involving the protection of human subjects in research. This version has been corrected.

Kathleen Sebelius, U.S. Secretary of Health and Human Services. (Matt Rourke/AP)

A clinical experiment involving 1,300 premature infants at two dozen hospitals “failed to adequately inform parents” of the risks of the treatment, which included blindness, brain damage and death, according to a watchdog agency at the Department of Health and Human Services.

The description of the experiment that parents read before deciding whether to enroll their babies downplayed the risks and emphasized possible benefits of getting a less-than-average amount of supplementary oxygen in the critical weeks after birth.

“The consent form was written in a slanted way,” said Jerry A. Menikoff, director of the Office for Human Research Protections (OHRP), which found that the study was “in violation of the regulatory requirements for informed consent” required by federal law.

“They went out of their way to tell you that your kid might benefit,” he said in an interview. “But they didn’t give the flip side, which is that there is a chance your kid might end up worse off. You can’t have it both ways.”

Research done more than 50 years ago showed that supplemental oxygen can cause blindness in preemies but that withholding it can increase the risk of death. The study in question was trying to discover whether a sweet spot exists between too much and too little oxygen where premature babies avoid both blindness and a higher risk of death.

It found — contrary to expectations — that babies who got supplementary oxygen in the higher end of the acceptable range had no more blindness, but greater survival, than babies that got slightly less oxygen. As a result, practice is starting to shift.

“People are thinking twice about keeping babies in the lower part of the range,” said Yvonne E. Vaucher, a neonatologist at the University of California at San Diego, who published a paper in December describing the health of the infants at age 2.

The study, called SUPPORT, was funded by the National Institutes of Health and ran from 2005 to 2009 at 23 research hospitals. Some of the babies died and some became blind, but most avoided both of those tragic outcomes. As a group they fared no worse — and possibly better — than preemies not enrolled in the experiment.

People associated with SUPPORT defended the consent form and noted that it was approved by 23 “institutional review boards’” (IRBs), which are the committees each research hospital must have to oversee the design and ethics of medical studies involving human beings.

“I don’t have any regrets. Everybody went into this with their best intention. Nobody was trying to deceive anybody,” said Neil N. Finer, a neonatologist at the University of California at San Diego who ran the trial.

“Because all the infants were being treated within the standard of care, investigators agreed that the extent to which risks and benefits were delineated in the consent form was appropriate,” said Richard B. Marchase, vice president for research at the University of Alabama at Birmingham, which was one of the study’s sites.

OHRP, which investigates issues involving the protection of human subjects in research for the Department of Health and Human Services, sent a chastising letter to Marchase on March 7. Its contents became widely known this week when advocacy organization Public Citizen’s Health Research Group called upon Kathleen Sebelius, the HHS secretary, to publicly apologize to the parents of the infants for what it called “indefensible, highly unethical” research.

The lungs of premature infants are often underdeveloped. They can’t extract enough oxygen from atmospheric air to support the body’s needs, in particular the brain’s. To make up for that, physicians since the 1940s have given premature infants air with extra oxygen in it.

However, research in the 1950s revealed that supplementary oxygen causes an overgrowth of blood vessels in the back of the eye. In its severe form, that condition, called retinopathy of prematurity, can lead to blindness.

At the time the study began, the practice was to give premature infants enough extra oxygen to keep the hemoglobin in their blood 85 to 95 percent “saturated” with the gas. A small device, called a pulse oximeter, that clips painlessly on the hand takes the measurement.

That range — 85 to 95 percent saturated — is quite broad. Many pediatricians favored the lower end of the range. There were hints that keeping hemoglobin saturation in the low 80s, or even high 70s, produced greater survival and less blindness than keeping it in the 90s. Most experts believed there was an optimal (and narrower) range somewhere between 85 percent and 95 percent that maximized brain health and minimized the risk of blindness.

The SUPPORT study was designed to find it.

It enrolled tiny, premature babies at the highest risk of dying from lack of oxygen or going blind from too much of it. Their average weight was just under two pounds, and their average gestational age was 26 weeks, or more than three months premature.

Half were randomly assigned to get enough oxygen to keep their hemoglobin saturation from 85 to 89 percent. The other half were kept at 91 to 95 percent.

The consent form made a big point that all babies would get oxygen within the “acceptable” range of 85 to 95 percent. It noted that babies in the lower range might have less eye damage. But it didn’t mention that those in the higher range might have more eye damage. And there was no mention that in the days before oximetry — when oxygen was measured more crudely — that premature infants that got higher concentrations of oxygen tended to have greater survival, although often at the cost of their eyesight.

Instead, the only risk the consent form mentioned was skin irritation from the oximeter device, a trivial problem.

What wasn’t made clear to parents was that the study created two groups of babies with different treatments — although both fell within “standard of care” treatment — for which the researchers expected there would be different outcomes. That was the point of the study.

“You’re intentionally shoving [the babies] into one end or the other of the range. And they were studying very real consequences in the kids,” Menikoff said.

In fact, the two groups did have different outcomes.

In the lower oxygen group, 20 percent of infants died compared to 16 percent of those in the higher oxygen group. Severe eye damage, however, occurred more frequently in the high oxygen group — 18 percent — than in the low-oxygen group, at 9 percent. The eye damage, however, is partly treatable. By age 2, there were equal rates of total blindness in the surviving children, at about 1 percent of each group.

For that reason, the 91-95 percent range is now favored. Not a perfect sweet spot, but a better and narrower target, the study investigators say.