[ Neo Home | New | Jobs | Technology | Sociology & Ethics | History | Gallery | Careers | About ]

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

Chapter 6

The National Cooperative Study

By the year 1953, approximately 10,000 children had been blinded by RLF, 7000 of whom were born in the United States. Government agencies and private organizations were gearing up for the greatly increased need for special facilities. At the same time, they (notably the National Society for the Prevention of Blindness) clamored for action to halt the epidemic. The most promising clue to the solution of the problem lay with the evidence against the use of supplemental oxygen.

At the 1952 meeting of the American Academy of Ophthalmology and Otolaryngology the idea was conceived of organizing a national cooperative study. Spurred by V. Everett Kinsey, a research biochemist then at the Kresge Eye Institute in Detroit, and Franklin M. Foote, then Director of the National Society for the Prevention of Blindness, a large meeting was convened in Bethesda, Maryland, early in 1953 under the auspices of the National Institute of Neurological Diseases and Blindness of the U.S. Public Health Service. Also in attendance were most of the American pediatricians and ophthalmologists who had been studying the RLF epidemic.

It soon became clear that there were two camps. The majority felt that a formal controlled trial must be carried out immediately to answer the outstanding questions concerning oxygen treatment. A minority felt that there was sufficient evidence to indict oxygen therapy as the cause of RLF, and argued that the proposal for a new nursery study was not justified, and even immoral. The dissidents cited the results of the only properly controlled trial (p 27). However, We majority were impressed that the persons who conducted the Gallinger Hospital trial and knew its defects were solidly on the side of the cooperative study. One splinter group felt that limiting oxygen would require, infants to undertake an unjustified risk of death and brain damage. After much debate, most of it heated, it was agreed that a joint effort should be undertaken to determine whether the frequency of occurrence of RLF is dependent upon the amount of oxygen used in the management of premature infants. An eight-man subcommittee (Appendix B) was appointed to draw up specific plans for the cooperative study.

The planners met in Atlantic City in spring 1953 and debated the dilemma, i.e., whether the administration of substantially lesser amounts of oxygen might increase mortality, or whether continued use of the customary amounts might result in an unnecessarily high frequency of RLF. In the end, a strategy was chosen with the intention of minimizing both risks:

[Only] one-twelfth of the total number of infants expected to be in the study was to be assigned to the ... routine [unrestricted] oxygen group. In this way, should oxygen indeed prove to be positively associated with retrolental fibroplasia, the number of infants in the participating hospitals who would receive routine [unrestricted] oxygen during the . . . [study period] would be substantially less than normal. To obtain an answer to the questionable role of oxygen as soon as possible whilst not materially sacrificing [concurrent] temporal controls, one infant was to be assigned to the routine, [unrestricted] oxygen group for every two babies assigned to curtailed oxygen group during the first three months of the study. If at the end of the [first] three-month period there was no difference in mortality [italics added], all babies were to be assigned to the curtailed group until such time as the incidence of retrolental fibroplasia in the two groups could be evaluated.

(For further details of the study design see chapter notes.)

Eighteen hospitals (Appendix C) agreed to abide by a uniform protocol which was drawn up by the Coordinating Committee for the treatment of infants. The study began on July 1, 1953, and continued until June 30, 1954. Infants who weighed 1.5 kg or less (less than or equal to 3 lb 5 oz) and who survived 48 hours were enrolled in the study. The Coordination Center at Kresge Eye Institute, Detroit, was notified by telegram of the entry of each infant and assignment to a treatment category was made by return telegram. Until 68 infants were assigned to the "routine oxygen" (unrestricted) group, assignment was made in separate sets of 3 infants for each hospital -- 1 infant was allotted to the routine oxygen group for every 2 infants assigned to the "curtailed-oxygen" treatment category, the order of assignment randomized within each set of 3.

Mortality rates in the two oxygen-treatment groups were monitored week by week. Three months after the trial began, there appeared to be no systematic difference in survival between infants in the curtailed-oxygen group and the 68 infants who had been enrolled in routine oxygen; according to the prearranged plan all subsequent infants received curtailedoxygen and enrollment continued for nine more motnhs. A total of 786 premature infants were admitted to the Cooperative Study; 166 died prior to 40 days, and the eyes of 34 were not examined adequately (i.e., for a minimum follow-up of 21/z months after birth); this left 586 infants in whom the relationship between oxygen treatment and RLF could be analyzed.

As the Cooperative Study proceeded there was understandable impatience for information among all concerned in the trial and in the medical community at large. But, in view of the many dashed hopes in the checkered history of the RLF epidemic, all participants in the investigation agreed that no results should be released until the carefully planned trial was completed and all the accumulated data were analyzed; September 1954 was considered to be the earliest date for a responsible report. As a result there was considerable consternation among all concerned when a memorandum was circulated to hospitals with maternity and newborn services in New York City, dated April, 1954; it read (in part)

Recent studies suggest a possible interrelationship between high concentrations of oxygen and the development of retrolental fibroplasia in premature infants. It has therefore been recommended by the Pediatric Advisory Committee that premature infants receive oxygen only as needed and then in concentrations not to exceed 40 per cent.

The bulletin did not identify the "recent studies," but clearly the Cooperative Study was not being cited. The physicians who were participating in the national study were baffled by this turn of events. The minutes of the meeting of the Pediatric Advisory Committee of the New York City Health Department, dated March 18, 1954, shed no light on the source of the information which led to the action of this group.

In May 1954, another document added to the apprehensions of those who were awaiting the results of the national study. A randomized controlled clinical trial was reported which compared the effects of high versus low oxygen therapy among 64 surviving premature infants at Bellevue Hospital. The results clearly indicted high oxygen. However, puzzled skeptics raised questions which were not answered in the publication. For example, the report did not mention the Cooperative Study which was underway, despite the fact that Bellevue Hospital was one of the 18 cooperating hospitals in the national effort. Since the calendar dates of the trial period were not reported, it was unclear whether (1) the Bellevue report included some infants who were also enrolled in the Cooperative Study, in which case the interpretation would be clouded by a number of irreconcilable problems, real and/or theoretical, or (2) the trial reported in May 1954 had been completed before the national study, in which case it was difficult to understand why the Bellevue group joined the 18-hospital national effort. Although it would seem that an explanation would be simple to find, to this day the exact circumstances are unclear.

The safety of oxygen restriction was not established convincingly by the Bellevue controlled trial. Mortality in the low-oxygen group was greater than in the group that received oxygen liberally, but the difference was thought to be explained by chance. The authors concluded from this evidence and from a review of their uncontrolled experience during the previous 10 years that a policy of oxygen restriction did not appear to be harmful. Many observers, noting the small number of infants in the Bellevue study, chose to await the results of the National Cooperative Study before altering the methods of care for premature infants which might increase mortality or brain damage.

In the Bellevue trial, infants in the low oxygen group who required oxygen received it from prepared tanks containing a mixture of 50-percent oxygen and 50-percent nitrogen to prevent accidental administration of higher concentrations of oxygen. The concentration of oxygen measured in incubators when this gas mix was administered averaged 38 percent (standard deviation ± 7.7 percent). Among 28 infants assigned to the low oxygen group, there was no cicatricial (scarring) RLF. These results led the authors to propose giving oxygen only to definitely cyanotic infants, and then at measured concentrations below 40 percent. This ceiling was the same as that advised in the April 1954 recommendation of the New York City Health Department.

Two other reports appeared in 1954 while the national trial was underway, and raised the debates concerning oxygen therapy to new heights of intensity. In the first report, from Philadelphia, the effects of rapid versus slow withdrawal from oxygen were evaluated by a controlled trial. However, each infant who developed progressive RLF was, in the authors' words, ". . . placed back in oxygen at the same concentration from which he had been removed. Because our results with oxygen therapy had been uniformly successful, we did not feel it was fair to withhold treatment in order to obtain control cases . . ." Although the results of this experience condemned excessive use of oxygen (and rapid withdrawal from oxygen in particular), many observers found it very difficult to interpret the results because of the confounding factor introduced by oxygen "treatment" of early RLF.

The second report reviewed the experience at the Colorado General Hospital from 1947 through 1953. Coincident rise and fall of RLF frequency with increase and decrease in use of oxygen was similar to experiences reported by some, but not all, others. The experience was widely quoted even before it was published, principally because the retrospective analysis of mortality led the highly respected authors to the conclusion that oxygen restriction was safe.

The suspense which preceded the first announcement of the Cooperative Study was almost unbearable to those who had waited so long for a definitive answer to their questions. They were not disappointed. On September 19, 1954, at the Fifty-Ninth Annual Session of the American Academy of Ophthalmology and Otolaryngology in New York City, Kinsey presented the preliminary results (Table 6-1), and all doubt concerning the causal role of oxygen in RLF seemed to be dispelled. Mortality among infants allotted to the two contrasting oxygen-therapy groups was declared "not significantly different" (see chapter notes).

When the final report of the Cooperative Study was released several unsuspected associations were disclosed; for example,

(See chapter notes for a comparison of these results with those found in animal studies.)

The epidemic of RLF came to a dramatic halt following widespread publicity given to the report of the Cooperative Study. The Great Oxygen Debate all but ceased.


Table 6-1
Frequency of RLF in the Cooperative Study, By Multiplicity
of Birth and Severity of RLF (July 1, 1953-June 30, 1954)



Blood Vessel Changes of RLF
Scarring RLF

Oxygen Group

Number of Infants

Routine Oxygen


Curtailed Oxygen


Multiple Birth

Routine Oxygen


Curtailed Oxygen


*These little-remembered details of the one-year trial deserve emphasis: Not all of the infants who received "curtailed" oxygen in 1953-54 escaped without RLF. There were 35 instances of scarring (cicatricial) RLF among infants in the "curtailed" group, compared with 12 infants who incurred residual lesions in the "routine oxygen" group.


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