Patz also looked back into Gallinger's nursery records. Even though they usually didn't record the exact oxygen flow rate, it was pretty clear that nearly all the babies who went on to develop RLF had had prolonged exposure to high concentrations of the gas. In fact, over three years it was 18 out of 21.
"I said, 'Roy, the only thing we can do is a carefully controlled prospective study to test the oxygen,' " Patz recalled.
Marc Maurer, President of the National Federation of the Blind, with his wife Patricia. Both were born prematurely and both became blind from oxygen given to them in their first weeks of life.
(Michael Williamson - The Washington Post)
(The fact that Patz understood the concept of a controlled trial -- and Hoeck apparently did, too -- is amazing in itself. The first such study, an English trial that proved the antibiotic streptomycin could cure tuberculosis, had only been published in October 1948.)
Patz got a $4,000 grant from the newly created National Institutes of Health (NIH) after his initial application was rejected as unscientific and unethical. He reassured the reviewers that every premature baby who needed extra oxygen to stay pink and healthy would get it. But all the premature babies wouldn't get it automatically.
On Jan. 1, 1951, they started their experiment. Babies weighing under 3.5 pounds were alternately assigned, based on time of birth, to get either 80 percent oxygen for at least 28 days or 40 percent oxygen "only for specific clinical indications." Parents were told about the study after their child was assigned. But they weren't asked whether they wanted their child included, and they signed nothing. This lack of "informed consent" in the modern sense was standard for the time.
"We weren't doing those babies any harm," Hoeck recalled thinking. "In fact, we were doing what we thought was beneficial in every way."
There was a group worried about harm, though. It was the nurses.
Part of the mythology of this trial is that nurses would go around at night turning up the oxygen taps to the low-oxygen babies in a guerrilla operation to save the infants' lives. That story heightens the drama, but Patz today says it's mostly wrong.
One or two night nurses did crank up the oxygen to all the incubators when they came on shift, but that occurred during a dry run for the experiment. Patz says he explained the protocol and asked them to stop, which they did.
In May 1953, he and Hoeck stopped enrolling babies. This was what they found: 12 out of 60 premature babies assigned to standard care were blind from RLF. In the curtailed-oxygen group, one out of 60 was. There were also a lot of near-misses. In the high-oxygen group, 21 babies developed early RLF, which eventually regressed to normal. In the low-oxygen group, nine babies showed those changes.
The young doctors published their first-year results in September 1952 (a year after Kate Campbell's report from Australia), and the final results two years later. By that time, Patz had successfully reproduced the disease in newborn mice, rats, kittens and puppies. However, even before the study was finished, the American Academy of Ophthalmology and Otolaryngology (AAOO) -- the professional organization of eye, ear, nose and throat doctors -- was making plans to do its own trial of oxygen use.
As is usually the case in medicine, a single well-conducted trial wasn't enough to convince doctors to change their practice. The Gallinger study had the additional drawback of being relatively small (and therefore of greater uncertainty) and the product of two unknown researchers.
The AAOO proposed a trial in 18 hospitals east of the Mississippi River. It would enroll enough babies to answer the question beyond any doubt -- provided people could agree how to run it. Some doctors felt that because the chief problem of premature babies was respiratory distress, limiting oxygen to them was likely to cause brain damage or death. Others felt that giving supplemental oxygen indiscriminately would cause avoidable cases of blindness. Few people were in the middle -- and certainly not Arnall Patz and Leroy Hoeck. Neither Johns Hopkins Hospital, where by then Patz had an affiliation, nor Gallinger participated in the study.
"I couldn't take part in it," Hoeck said. "I didn't have to be convinced any more."
At a meeting in the mansion house of NIH's new campus in Bethesda, about 45 scientists met to draw up plans. They wrangled through an entire night, at one point placing a transatlantic telephone call to Austin Bradford Hill, the British biostatistician who had designed the streptomycin-for-TB study. He suggested a Solomonic compromise.
The trial would basically be two studies run sequentially. For three months, premature infants would be randomly assigned to routine high oxygen, or oxygen only if they needed it, in a 1:2 ratio. That would answer the question whether oxygen caused RLF. Then -- assuming the study wasn't stopped because of higher mortality in the curtailed-oxygen group -- all babies born in the following nine months would get oxygen only for clinical need. That would provide an estimate of the incidence of RLF under oxygen-sparing conditions -- a statistic nobody actually knew.
Cobbled together with a speed inconceivable today, the 18-hospital study commenced on July 1, 1953. It was run out of a central office at the Kresge Eye Institute in Detroit by a biochemist named V. Everett Kinsey, who'd been interested in RLF since working with Theodore Terry in Boston a decade earlier. The study used random numbers, not every-other-baby assignment as in Washington, to determine which infants would get what treatments. Western Union provided a teletype machine for free, and seven days a week telegrams came and went from the 18 hospitals.
"This was at the height of the McCarthy era," recalled William A. Silverman, who was in charge of enrolling preemies into the trial at Columbia's Babies Hospital in New York. "Some people from the FBI came up to see me to find out what kind of subversive activity was going on with all these suspicious telegrams." (Silverman died last December.)
The trial closed on June 30, 1954. Before the summer was over, Kinsey and his collaborators had answers, which they presented to a meeting of eye doctors in New York that fall.
RLF severe enough to cause blindness occurred in 17 percent of the babies getting routine high oxygen, but in only 5 percent of the curtailed-oxygen group. The death rate in the two groups was similar -- 22 versus 25 percent. Oxygen -- and nothing else -- was responsible for the epidemic of blindness in premature infants. Doctors could safely turn it down.
In fact, not to do so would soon be unforgivable.