2.4. Poor Science in the Oxygen Trial

It is thus no surprise to find a heavy thumb on the risk-weighing balance of that trial. To support their prejudiced accusation against oxygen, the trial designers knowingly killed many of those who needed this life-saving gas.

They knew that the preemies most at risk for ROP were the most immature ones with the lowest birth weights and the most immature eyes, and that these had also the most immature lungs^{39, 40, 41}. They further knew that babies with immature lungs who need help with breathing need this help right away, particularly during the first 12 to 24 hours^{42, 43}, or else they will die from their respiratory problems.
 
The medical experts until that time had written about preemie breathing problems and oxygen in terms like these:

"This is a condition of incomplete or delayed expansion of the lungs due to a persistence of the fetal state in one or several pulmonary areas. It is usually a sequel to asphyxia and the causes are those of asphyxia neonatorum. Premature and debilitated infants are predisposed. (...) Oxygen inhalations are indicated for cyanosis. (...) The prognosis is good if the condition is of short duration. Persistence of symptoms (...) offer an unfavorable prognosis; and the development of pneumonia ."  The same author added in his chapter on "Feeding the Premature: "During the first 12 to 24 hours, premature babies should be given plenty of oxygen on the slightest provocation and otherwise left alone."⁴²

Or, in the words of another pre-stampede authority:

"It is advisable to place every premature infant, even if there are no signs of cyanosis, in an incubator containing a 40 to 50 per cent oxygen concentration, for at least the first 24 hours. Cyanosis may occur in apparently healthy premature children from causes which would not bring about such a condition in infants born at term. Any infection or minor obstruction of the respiratory tract may prove fatal, and the tissues become edematous if under-oxygenation is permitted for any length of time. The oxygen concentration must be increased for premature infants with any indication of asphyxia. The lungs of asphyxiated infants, being atelectatic, congested and edematous, are particularly susceptible to infection. Bronchopneumonia should be suspected in any newborn premature infant with signs of asphyxia. It is now conceded that many lesions formerly ascribed to trauma of delivery are in reality owing to asphyxiation."⁴³

Knowing all this, the trial designers delayed the enrollment and most oxygen administration until the babies were 48 hours old.  By that time, 634 babies had died; 786 survived to be enrolled. Only 52 of these survivors had received any supplementary oxygen during their first two days⁴⁴.

Those who had died included, of course, the great majority of those who would otherwise have survived to grow up blind. This protocol of killing the patients most at risk for blindness before their deaths would be counted concealed the rise in mortality which the perpetrators of this grotesque experiment knew their cure worse than the disease would produce. It thus allowed them to announce their marsupial verdict that the oxygen withholding had not affected the mortality rate but had virtually ended the blinding.

There were many additional flaws in that trial, including statistical slight-of-hand such as alleging to separate the babies' need for oxygen from their immaturity although one is clearly a function of the other. A correlation between the pulmonary immaturity which makes the babies need oxygen and the ocular immaturity which predisposes them to ROP must be expected but is meaningless, as "The Oxford Textbook of Medicine" explains in more general terms:

"If factor A [the baby's immaturity] influences both factor B [the need for oxygen] and disease C [ROP], factor B and disease C will be associated statistically. The association between B and C will, however, be non-causal and there would be no hope of producing a change in disease C by manipulating factor B."⁴⁵

Despite the contrived statistics and other examples of bias, the reported results still did not condemn oxygen. They specifically exonerated variations in the concentration of the gas; these had exerted no influence on either the incidence or the severity of the disease⁴⁶. The results also showed that the severity of the damage was unrelated to the length of exposure to supplemental oxygen⁴⁷.

The entire blame against oxygen was based on a correlation between that length of exposure and the incidence of the disease, as asserted in the "Comment" and "Summary" sections of the paper. However, buried in the middle of the 48-page text you find the qualification that even this unsurprising correlation was statistically significant only for the first one and a half to three days after the enrollment in the study at the age of two days⁴⁸.

All the study had achieved was thus to confirm weakly what many other observers had previously seen and what no one had ever doubted: that the more immature preemies are more likely to suffer from ROP than the more mature ones, and that they also need oxygen for a longer time.

The mountains had labored mightily to bring forth a minuscule mouse, but the mouse was made to look like a lion. This unscientific, lethal, and result-less study received much praise as a medical breakthrough, a splendid success for the then relatively new conceptual tool of double-blind controlled clinical trials. 

This study is also said to have been one of the major reasons for the large expansion of Congressional appropriations to medical research which began in the late 1950s⁴⁹. Its biased verdict against innocent oxygen is now a cornerstone of the intensive care nursery industry's intellectual foundation, and its recommendations for oxygen rationing have affected the daily life in intensive care nurseries around the World probably more than any other single study ever has.

Even before the final report on that study's dressed-up non-findings was printed, many neonatologists began to institute policies of limiting preemies to oxygen concentrations of not more than 40%.  This was an arbitrary number not based on any results from the trial but on the personal belief of a few physicians at the Bellevue Hospital in New York^{50, 51}.

The author of the Cooperative Study report warned in a letter to the editor of Pediatrics against this unwarranted conclusion⁵². But his warning did not stop a physicians' stampede towards this arbitrary and unjustifiable 40% limit as the easy final solution for seeing less babies blinded by ROP leave the nurseries.

The adversarial legal system reinforced the bias against oxygen. The authoritative and much publicized condemnation of this gas as the alleged cause of blindness led quickly to many malpractice suits alleging its improper use⁵³. Parents were much more likely to sue over a baby's blindness than over her or his death, and behind every nurse who wanted to open an oxygen faucet to help a struggling preemie breathe for survival stood the specter of a lawyer who would accuse her if that preemie happened to wind up blind.

Continue reading