Automated Adjustment of Oxygen in Ventilated Preterm Infants: Turn on, Tune in, ROP out?

The death of newborns from respiratory failure has been recognized and noted to occur more frequently in preterm infants for >2000 years.1 In the mid 20th century, the term respiratory distress syndrome (RDS) was coined to describe the often lethal respiratory distress and episodic apnoea, cyanosis, and bradycardia seen in preterm newborns, which began at or shortly after birth and progressed over 72 hours. Hyaline membrane disease (HMD) described the striking pathological features seen on post-mortem examination of lung tissue of infants who died.

Illustrating doctors' enduring enthusiasm for introducing new and unproven therapies, oxygen was first given to newborns in 1780, within 6 years of its discovery.1 Oxygen supplementation for newborns began in earnest in the 1940s. Although oxygen given in high concentrations likely improved survival from RDS, it resulted in an epidemic of blindness caused by retinopathy of prematurity, thereby starting modern-day neonatologists' ambivalent affair with oxygen.2 In an attempt to prevent blindness, oxygen use was curtailed, which led to an increase in deaths from HMD and was associated with a rise in the frequency of spastic diplegia.2 Therapy for RDS has progressed markedly in the last 60 years. The introduction of mechanical ventilation was met initially with limited success initially and, with the emergence of bronchopulmonary dysplasia, which was attributed partly to excessive oxygen exposure.3 Continuous positive airways pressure showed an impressive reduction in mortality in a small case series of infants with established severe RDS. 4 Antenatal steroids were demonstrated to reduce the frequency of RDS and death in preterm infants.5 Techniques of mechanical ventilation, which included the use of positive end-expiratory pressure, were refined, and exogenous surfactant therapy followed, resulting in immature infants surviving in greater numbers.6 These survivors, however, had substantial rates of chronic lung disease (CLD) and neurological dysfunction. Vitamin A supplementation7 and caffeine therapy8 were subsequently demonstrated (modestly and substantially, respectively) to reduce the rate of chronic lung disease. Despite showing promise in cohort studies, use of nasal continuous positive airways pressure in preference to endotracheal ventilation from birth did not reduce the rate of CLD in a recently reported randomized trial.9

Although the fraction of inspired oxygen (FiO2) given to infants with respiratory distress was once determined with clinical assessment of color, in time this was aided and usurped by intermittent measurement of oxygen tension in arterial blood, and subsequently by continuous oxygen saturation (SpO2) monitoring. We now monitor the oxygen levels of our smallest patients more closely than before, and the rate of visual impairment in survivors has fallen with time. Despite this, many survivors still have chronic pulmonary dysfunction and, more concerning, neurodevelopmental impairment. Cohort studies have shown associations between higher SpO2 targets and poorer visual and pulmonary outcomes in premature infants. 10 A randomized trial showed that targeting higher SpO2 in preterm infants who remained receiving oxygen at 32 weeks resulted in worse pulmonary outcomes (prolonged oxygen therapy, higher rates of CLD and home oxygen therapy).11 The “optimum level of oxygenation (to balance 4 competing risks: mortality, retinopathy of prematurity blindness, chronic lung disease, and brain damage)…remains unknown,”2 and several large trials comparing relatively lower and higher SpO2 target ranges for preterm infants are ongoing. Although it is not yet known which is preferable, 3 things are clear: the optimal range is likely lower than many that were previously targeted, there is widespread interest in maintaining infants SpO2 within predefined ranges, and maintaining infants within these ranges is often difficult.

In this issue of The Journal, Claure et al report the results of their randomized crossover trial comparing the ability of a computer system that automatically adjusts FiO2 to that of a bedside nurse manually adjusting the FiO2 in keeping the SpO2 of ventilated infants within a target range.12 The 16 extremely preterm infants they studied had been ventilated for a mean of 4 weeks and were receiving a mean of 32% oxygen at the time of enrollment. Infants were studied for 8 hours, 4 hours of automated FiO2 control and 4 hours during which the bedside nurse adjusted the FiO2. They found that infants had shorter periods outside the target SpO2 range during automated FiO2 control. In particular, they had shorter periods of SpO2 > 95% and ≥98%. Infants spent more time with SpO2 < 88%, but not <85%. Also, the nurses made fewer manual FiO2 adjustments during the automated period. The authors are to be commended on their well-designed and executed study. Although the number of infants enrolled is small, the crossover design of the study strengthens its findings. Although some readers might be disappointed that automated FiO2 control did not maintain SpO2 exclusively within the target range, it would be unreasonable to expect this of such a strategy, because chronically ventilated infants have falls in SpO2 for reasons that are not amenable to correction by adjusting FiO2 alone (eg, forceful expiration against the ventilator with loss of residual lung volume).

Should we all rush out to buy this technology to use in the nursery today? In my opinion, no. Although this novel technology might spare 2 treasured resources in any nursery—the time and sanity of the bedside nurses—we don't yet know whether it helps the babies. Whether it results in tangible benefits for preterm infants can only be adequately addressed in the context of a well-designed randomized trial of sufficient size and power to detect meaningful differences in clinically important outcomes. The focus on oxygen and its potential for causing harm in infants has now retreated to the first minutes of life. Randomized trials have demonstrated that air is as effective as 100% oxygen for resuscitation of asphyxiated term infants and is associated with lower mortality.13 Although the experience of using <100% oxygen in preterm infants at birth is limited, lower concentrations have been advocated14 because preterm infants have limited anti-oxidant defenses and a large oxygen load early in life could trigger inflammatory changes that induce end-organ damage. It would be intriguing to see whether automated FiO2 control would benefit not only the chronically ventilated and oxygen-dependent infants Claure et al describe (who, although infrequent, seem ever-present), but also whether using this approach earlier rather than later might reduce the number of extremely preterm infants reaching this state.