A major study led by Cooper University Hospital suggests a new way to improve dismal survival rates after cardiac arrest: Turn down the patient's oxygen.

Stinting on oxygen may seem counterintuitive, since the brain begins to die when deprived of oxygen-rich blood for more than five minutes. But studies in dogs and in premature infants have long shown that too much oxygen can be harmful.

The new study, published in Wednesday's Journal of the American Medical Association, is the first solid evidence that this vital gas is also a double-edged therapy after the heart suddenly stops - as it does in 220,000 Americans every year.

In fact, the researchers found that patients on ventilators who got too much oxygen fared worse than those who got too little; 63 percent of patients with excessive blood oxygen died, compared with 57 percent with insufficient blood oxygen.

"We think of oxygen as fairly innocuous," said Stephen Trzeciak, a Cooper critical-care and emergency-medicine physician who was senior author of the study. "But to the injured brain, it may not be."

When the heart stops - whether because of an abnormal rhythm, a birth defect, or a clogged artery that triggers a heart attack - the brain is extremely vulnerable to injury.

Yet when normal blood flow resumes, it unleashes what doctors call "reperfusion injury," a cascade of inflammation and oxidation - the same oxygen reaction that causes metal to rust.

The bottom line: Only 6 percent of cardiac-arrest patients survive, and only half of those come through without heart or brain damage, or both.

The one practice that has been shown to reduce this brain injury is cold therapy, or "therapeutic hypothermia," in which the patient is kept in a coma while the body is slowly cooled to about 91 degrees Fahrenheit. Cooper, in Camden, and other leading hospitals provide this complex therapy, but it has not yet become standard.

In terms of ventilation, standard practice is to turn the breathing machines on full blast - 100 percent oxygen - when cardiac-arrest patients arrive in the intensive-care unit.

International resuscitation guidelines recommend aiming to keep the blood 94 percent to 96 percent saturated with oxygen.

But Trzeciak said there was no guidance for how much or how quickly to turn down the ventilator, which can force oxygen into the lungs as it controls the rate of respiration.

Nor have experts figured out the threshold at which a high blood oxygen level should immediately trigger adjustments in care, resuscitation researchers wrote in an editorial published with the study.

The new study assessed whether patients had too much or too little blood oxygen based on their first blood-gas measurement taken within 24 hours of their arrival in intensive care.

About 1,150 patients, 18 percent, had excessive oxygen, or "hyperoxia," defined as blood that is 100 percent saturated with oxygen. Of these, 63 percent died in the hospital.

In comparision, about 4,000, or nearly two-thirds, had insufficient oxygen, or "hypoxia," defined as less than 90 percent saturation. Fifty-seven percent of them died in the hospital.

Those in the middle range, with "normoxia," had a dramatically better rate. Still, 45 percent died in the hospital, an indicator of the dire complexity of cardiac arrest.

"This raises a lot of questions, but in my practice, I will immediately see if I should be more vigilant" in adjusting down the ventilator, said Clifton Callaway, vice chair of emergency medicine at University of Pittsburgh Medical Center.

"There's a right oxygen dose. We haven't been able to find that dose," said Michael Sayre, an Ohio State University emergency-medicine physician and American Heart Association spokesman. "This study focuses attention on an area that hasn't gotten much attention and shows there are opportunities for improvement."

Cooper researchers conducted the study with experts from the Carolinas Medical Center in Charlotte, N.C., Beth Israel Deaconess Medical Center in Boston, and Ohio State University in Columbus.

Cooper is planning clinical studies designed to answer some of the many questions about how much oxygen to give during and after resuscitaion.

"As we speak, I'm writing my next NIH [National Institutes of Health] grant application," Trzeciak quipped.

Contact staff writer Marie McCullough at 215-854-2720 or mmccullough@phillynews.com.