[ Neo Home | New | Clinical | Computers | Jobs | Diversions | Links ]

Retrolental Fibroplasia: A Modern Parable
by William A. Silverman, M.D.

Chapter 4

The Oxygen Hypothesis

On July 14, 1951, the Medical Journal of Australia printed an article by Kate Campbell of Melbourne, in which she noted

... I heard from colleagues returning from overseas, the suggestion that oxygen might be responsible for producing retrolental fibroplasia. The idea arose apparently from a comparison of the treatment of premature infants in America, where retrolental fibroplasia is a problem and where oxygen was used freely, with the treatment in England, where retrolental fibroplasia is seen rarely, and whole oxygen was used sparingly.

Although Campbell did not identify the source of the suggestion "from overseas," it developed that her informants had visited Birmingham, England, where the recent appearance of RLF was being pondered by V. Mary Crosse and Philip Jameson Evans. In private, Doctor Crosse noted that liberal use of oxygen in England started after the institution of the National Health Service, because government funds now made it possible for hospitals to afford modern incubators and the use of the expensive gas for medical purposes.

As Kinsey and Zacharias had done earlier in Boston (Chapter 3), Crosse and Evans traced the history and course of RLF in England beginning in 1946 and concluded that the occurrence of the disease followed a period of free use of oxygen "on the State," and subsided again with its restriction (Table 4-1). This experience suggested a general political evil to Doctor Evans:

. . .as in domestic and national policies, a prolonged subsidy paralyzes the ability to struggle which would otherwise have had the opportunity to develop, and that the disease [RLF] has in fact been artificially induced by a well intentioned, but misguided, change in the management of such cases. That a return to a less indulgent care of the premature infant prevents the disease should be appreciated as soon as possible, and cannot be too quickly undertaken.


Table 4-1
RLF in Birmingham


Number of Infants

Number with RLF

High oxygena (1944-50)



Limited oxygenb (1951)



a "Free" administration of oxygen (concentration not estimated).

b Shortest possible exposure to oxygen, just enough oxygen to prevent cyanosis; graduated return to normal air as soon as possible.

The rise in frequency of RLF coincident with change in oxygen practice in certain hospitals in Melbourne, Australia was reported by Campbell (Table 4-2) and by Ryan, who noted

. . . no case of retrolental fibroplasia had occurred at the Women's Hospital [Melbourne] prior to the introduction of a most efficient oxygen cot. With this apparatus it was the practice of the nursing staff to give oxygen liberally to all babies, even when apparently not requiring it.

On the other hand, Houlton reported that although a rise in RLF did correlate with the use of higher concentrations of oxygen in Oxford, new cases continued to occur even after a reduction in the level of oxygen use.

Thaddeus Szewczyk, an ophthalmologist in East St. Louis, Illinois, postulated that RLF was due to unrecognized oxygen-lack at a time in development when oxygen requirement of the retina was particularly high. He criticized the clinical practice of exposing premature infants to high concentrations of oxygen and subsequent rapid withdrawal in the change from incubator to room air, before acclimation could take place. Szewczyk advised that oxygen be administered to premature infants in minimal amounts and a slow withdrawal from enriched environments. He reported that early changes of RLF appeared when 7 premature infants were moved abruptly from incubators (50-percent oxygen) to bassinettes in room air (21-percent oxygen). Moreover, the changes in the eyes regressed rapidly when the infants were returned to oxygen-rich (60-percent) incubators. In a later trial, blood vessel dilation, new vessel formation and hemorrhages in the retina appeared in 19 premature infants removed rapidly from incubators; when they were returned to high oxygen all improved within 24 hours and their eyes became normal after four days.


Table 4-2
RLF in Melbourne

High Oxygena
(Institution I)
Moderate Oxygenb
(Institutions II and III)


Number of Infants
Number with RLF
Number of Infants
Number with RLF











a In "Institution P' oxygen was piped into the ward and was given in an oxygen cot, in which the percentage of oxygen was 40-60 percent. Oxygen was given prophylactically as well as during periods of cyanosis.

b In "Institutions II and III" most, but not all, premature infants were nursed in an electrically heated cot; oxygen was administered by intranasal catheter or funnel, sometimes by tent or cot.

However, not all the evidence pointed toward excessive use of oxygen as a cause of RLF. In New Orleans, at a large premature infant center in Charity Hospital, Exline and Harrington examined or traced the outcome in 96 of 131 ex-premature children with very low birthweights (under 1.5 kg = under 3 lb 5 oz) born during a two-year period ending December, 1948. All these children received oxygen at an estimated concentration of 50 percent for four or more weeks after birth; none had RLF.

At the Clinique de Puériculture in Paris, two groups of premature infants were compared; in those who received continuous high concentrations of oxygen, the final results (scarring RLF) were no worse than among others raised with a minimum of oxygen (Table 4-3).

While confusion reigned because of the conflicting indications of the role of oxygen from these and other observational studies, a pioneering experimental* clinical trial was undertaken at the Gallinger Municipal Hospital in Washington, D.C. One of the first infants to be placed in a closed incubator in this nursery developed RLF. Leroy Hoeck, the director of the nursery, recalled that the oxygen was delivered through a funnel strapped to the baby's face. He remarked to Arnall Patz, then a young ophthalmology resident, "All that oxygen might be doing something." Patz recalls that he ". . . explained to Hoeck how unlikely extra oxygen was a factor because the disease seemed to be a response of the type that is associated with anoxia [that is] proliferation of vessels; if indeed, oxygen was related at all." Patz dropped the matter for a while until he came across an article by Stadie describing toxic effects of oxygen on cells, where mention was made of the questionable change in adult retinal vessels upon inhalation of oxygen. In late 1949, toward the end of his hospital residency training, Patz designed a controlled clinical trial and some animal tests, both to explore the idea that excess oxygen might be a factor in causing RLF. With Hoeck, he applied for a research grant of $4000 from the then very small National Institutes of Health for a nursery trial to compare the outcome in infants weighing less than 1.6 kg (under 3 1/2 lb) at birth assigned in alternate order to incubators with either routine oxygen or low oxygen. Oxygen concentration in the incubators was to be monitored by measurement with an electronic analyzer. The grant application was criticized by reviewers as being weak in scientific merit and with extreme concern by the pediatric referees that ". . . these guys are going to kill a lot of babies by anoxia to test a wild idea." Patz and Hoeck satisfied the objections of the referees by stating, "to avoid having deaths from lack of oxygen, every baby in the low oxygen group would be maintained at a healthy pink color." The $4000 was granted.


Table 4-3
Experience with Oxygen in Paris (1948-1952)

Vascular Changes
Cicatricial RLF

Oxygen Group

Number of Infants

Highly oxygenated groupa


Sparely oxygenated groupb


a Continuous oxygen (60 percent) never less than 21 days, occasionally 6-8 weeks.

b Minimum oxygen (50 percent), discontinuous, given only if indicated for respiratory difficulty or cyanosis; stopped by eighth day at the latest.

The nursery trial proved to be difficult for the investigators. Patz found that "The nurses were convinced that we were going to kill the babies in the low oxygen group, and indeed, at night some of the older nurses would turn the oxygen on for a baby who was not receiving oxygen, then turn it off when they would go off duty in the morning." Seventy-six infants were enrolled during the first year of the proposed three-year controlled study; but 11 were dropped from consideration. because of technical complications or failure to return for follow-up. The results of the first year's experience were published in 1952; the outcome damned continuous high oxygen (Table 4-4).


Table 4-4
Gallinger Municipal Hospital Controlled Trial of Oxygen (1952)

Oxygen Administration

Number of Infants
Minor RLFa
Severe RLFb





a Blood vessel dilation, tortuosity of retinal vessels, retinal edema, retinal hemorrhages, vitreous clouding.

b Detachment of the retina, new vessel formation, retinal hemorrhages, and complete retinal membrane behind the lens in some infants.

c 65-70 percent oxygen for 4-7 weeks.

d Under 40 percent oxygen for 24 hours up to 2 weeks, depending on infant clinical condition, especially cyanosis.

The debates about the role of oxygen were mounting at the end of 1952, and the results of the first controlled trial did not do away with uncertainties in the minds of many observers who were wary after ten years of false leads. Because of the number of infants in the Gallinger Hospital study, concern about the overriding matter of safety of oxygen restriction had not been answered to the satisfaction of many skeptics. Indeed, the survival statistics of the trial were not provided in the report. The bias introduced by alternate case assignment could not be evaluated, and the investigators themselves had a number of unanswered questions about their first experience with a clinical trial. They concluded their report by noting the following:

The data cited here [and the anecdotal reports of others] suggest strongly that high oxygen is a factor in the pathogenesis of retrolental fibroplasia. However, in view of the bizarre manner in which the incidence of the disease fluctuates, additional rigidly controlled observations are necessary to establish this concept.

* "Experiment is fundamentally induced observation ... to reason experimentally we must have an idea and afterwards induce or produce facts, i.e., observations to control our preconceived idea." -- Claude Bernard

Return to the Table of Contents Page
Return to the Classics Page

Created 7/27/2002 / Last modified 7/27/2002
Copyright © 1980 by Grune & Stratton, Inc.
Neonatology on the Web / webmaster@neonatology.org