A study of babies in intensive care suggests that doctors are getting better at recognizing situations where infants are sure to die or have severe brain damage -- and are often holding back on life support when that's the case.

The findings "reflect increasing awareness by the medical community of the need to limit interventions of minimal or very questionable benefit, and particularly if those interventions potentially include significant pain or suffering to the patient," said Dr. Renee Boss, a neonatologist at Johns Hopkins Hospital in Baltimore, who wasn't part of the new research.

Over the past 30 years, Boss said, doctors have gotten better at keeping very premature babies and babies with severe birth defects alive.

But more recently, those survival rates have flattened out -- possibly because "the treatments that we have now simply have reached their maximum potential for increasing survival," Boss told Reuters Health, forcing doctors and families to address cases where survival, or survival with a positive outlook, doesn't seem likely.

About six of every 1,000 infants die in the U.S. before their first birthday -- with more than half of those deaths coming in the first 28 days.

For the new study, Julie Weiner and her colleagues from Children's Mercy Hospital in Kansas City, Missouri, looked back on the medical records of all infant deaths in their neonatal intensive care unit over a 10-year period.

Those included just over 400 deaths, of more than 7,000 intensive care admissions. The majority of deaths occurred in babies with major birth defects, including heart problems, and those born very premature -- at 32 weeks or earlier. In general, any baby born before the 37th week of pregnancy is considered premature.

But the researchers were more interested in how those babies died -- whether it was while doctors were trying to save or prolong their lives, or whether babies had been taken off ventilators, feeding tubes, and other life support when doctors realized there was nothing more they could do.

Over the study period, which ran from 1999 through 2008, less than one in five of the deaths happened while doctors were trying to perform CPR to keep infants alive.

About 60 percent of infants died when doctors took them off ventilators or otherwise stopped giving life support, and the remaining 20 percent occurred when medical staff withheld life-prolonging treatment altogether.

Deaths that happened when doctors decided not to start giving treatment became more common over the course of the study, rising by about one percent each year. That was especially true in the very premature group, according to findings published in the Archives of Pediatrics & Adolescent Medicine.

That result "makes us think that maybe we're better recognizing futility of care and providing a better and better situation for these babies that are dying anyway," Weiner told Reuters Health.

Dr. William Meadow, a neonatologist at the University of Chicago Medical Center, pointed out that these patterns might look different at another intensive care unit.

For example, at his hospital, most very sick babies who are in "stable" condition -- meaning they aren't obviously dying while on life support, but might have extensive brain damage -- don't have that care taken away.

He said that's because poor, religious parents at his hospital seem to be more okay with the idea that their child might survive, though remain very impaired.

In those types of cases, "it's a reflection mostly of parent preferences," Meadow told Reuters Health.

"Decisions to withdraw or limit care are easier for doctors and families alike when the baby is actively, critically ill despite interventions," Boss agreed.

In sick but stable babies, "it can be a little harder to find the point at which you say, 'Nothing more can or should be done,'" she said.

The study also suggested that parents have become more involved in discussing end-of-life options.

"We've seen an increasing recognition that parents (and) family members should have a very central role in these decisions," Boss said. In these cases, she added, "most parents do want to participate to some degree in decision-making."

"For parents it's overwhelming anyway to be in the (neonatal intensive care unit), and to have an infant that is dying -- nobody expects to have a baby that's going to die," Weiner said. "Our hope when we do provide end-of-life care is trying to provide compassionate loving, care to ... our families and our dying neonates."