In the 1970s, Pittsburgh paramedics began inserting plastic breathing tubes down the throats of trauma patients -- a procedure previously performed only in hospitals -- to help keep them alive until they could reach a hospital.

It made so much sense that today virtually every U.S. city has paramedics who intubate critically injured patients. The problem is that it doesn't seem to be working as intended.

Researchers at the University of Pittsburgh School of Medicine who reviewed thousands of cases from across Pennsylvania found that patients with traumatic brain injuries who are intubated in the field are four times more likely to die than those who are intubated in the hospital.

What's more, those patients who survive after being intubated at the accident scene are more likely to suffer neurologic damage than those intubated in the hospital, the Pitt researchers report in this month's issue of the Annals of Emergency Medicine.

It's too soon to say that paramedics should stop intubations, emergency medical specialists say, but it's becoming clear that researchers need to figure out what's going wrong.

"Up until five years ago, people thought this was a good idea," said Dr. Donald Yealy, Pitt's vice chairman of emergency medicine and one of the study's authors. But then several studies began to show that the intubated patients fared poorly.

The Pitt study, led by Dr. Henry Wang, is by far the largest of these studies, so its conclusions are difficult to dismiss, according to Drs. Brian Zink and Ronald Maio, emergency medicine physicians at the University of Michigan who wrote an accompanying editorial.

"Patients with severe traumatic brain injury already have the deck stacked against them," they wrote. "If current out-of-hospital airway standards are dealing these patients a bum hand, we need to reshuffle the cards and take a fresh look."

Still, the findings are puzzling. After all, placing an endotracheal tube down a patient's throat has long been considered the definitive method for keeping a patient's airway open. If a patient has stopped breathing or is having difficulty breathing, the tube provides a means of pumping air into the lungs. And if the patient is still breathing, the tube keeps him from choking on his own vomit should he regurgitate.

The use of breathing tubes by paramedics was one of the innovations pioneered in Pittsburgh by the Freedom House Ambulance Service, one of the first modern emergency medical systems in the country, in the late '60s and '70s.

Just why intubation should increase the death rate for patients is not known. It's possible, of course, that intubation by paramedics keeps alive patients who otherwise would have died before reaching a hospital; they obviously would fare worse than less seriously injured patients intubated in a hospital. But Yealy said the researchers tried to compare only patients with similar levels of injury.

It could be that conditions in the field are simply less conducive to intubation than the hospital, Yealy said. If that's the case, he said, paramedics should use other means to help patients breathe. Or, intubation might only be appropriate for certain subsets of patients.

It could be that paramedics may not all be equally skilled. Yealy said an unpublished study found that most paramedics do intubations infrequently. Someone who doesn't intubate regularly may do so only slowly, or may place the tube incorrectly.

Though Wang and his colleagues reviewed the cases of 4,098 trauma patients from across Pennsylvania who were intubated between 2000 and 2002, one shortcoming of the study is that the researchers were forced to glean details from medical records.

A better test would be a randomized, prospective study -- resolving to intubate some patients in the field and to delay intubation until reaching the hospital for others. But Yealy contended that would be expensive and controversial. Failing to intubate patients who are unconscious and unable to consent to such a study would require much community debate, he noted.