The Use of Oxygen in Neonatal Medicine
2003; American Academy of Pediatrics; Volume: 4; Issue: 12 Linguagem: Inglês
10.1542/neo.4.12.e340
ISSN1526-9906
Autores Tópico(s)Neuroscience of respiration and sleep
ResumoAfter completing this article, readers should be able to: Despite supplemental oxygen being probably the most common treatment provided to infants in the newborn period over the past 60 years, (1) uncertainty continues as to the most appropriate ranges to maintain oxygen levels for term or preterm infants or a threshold value below which oxygen should be administered. (2)(3) This situation has contributed to a wide variation in practice and has fueled the current controversy regarding the most appropriate ranges to target the blood oxygen levels of babies. (4)(5)Although the use of oxygen as a medicinal product has a long history, “routine” use of supplemental oxygen in the care of small or preterm infants originated in the 1940s. One report that had a marked influence on this practice was by Wilson, Long, and Howard. (6) They noted that the irregular pattern of periodic breathing commonly seen in infants of short gestation largely was abolished when these babies were given 70% oxygen or more to breathe (Fig. 1). The authors noted that they had no proof that a regular “normal” respiratory pattern was better for the preterm infant than the periodic type of breathing or that increased oxygen content of arterial blood was beneficial or necessarily important. They only noted that healthy preterm babies in their study breathed in a more “normal” manner in an oxygen-enriched atmosphere. By 1949, Howard and Bauer (7) were less cautious, recommending that irregular breathing in early infancy be treated with a mixture of high oxygen to increase breathing volume and exert a steadying effect on the rhythm of respiration. In their opinion, oxygen was the most valuable single agent available for any newborn infant showing any evidence of respiratory difficulty, and they suggested that it be used early and generously.It was within this climate of opinion that the widespread practice of unrestricted oxygen supplementation for small or sick infants came into force in the early 1950s. The ensuing epidemic of severe eye disease and blindness is now well known (Fig. 2). The first person to suggest in print that oxygen could be responsible for the rising epidemic of retinopathy of prematurity (or retrolental fibroplasia, as it was then known) was Kate Campbell in Melbourne, Australia, (8) although she generously noted that news of the idea came to her “from colleagues returning from overseas.” The hypothesis arose from a comparison of data from the treatment of preterm infants in America, where there was an increasing incidence of retrolental fibroplasia and oxygen was freely used, and data from England, where retrolental fibroplasia was rare and oxygen was used sparingly. More convincing evidence came within a year from Crosse and Evans in Birmingham, England, (9) and from a study started in 1948 by Arnall Patz in Washington. (10) In the Washington trial, babies weighing less than 3.5 lb (1.6 kg) were assigned 24 hours after birth alternately to care in a high-oxygen atmosphere (65% to 70% for 4 to 7 wk) or less than 40% oxygen for as short a time as possible (1 to 2 wk). Seven of the 28 babies nursed in high oxygen developed stage 3 to 4 retinopathy compared with none of the 37 nursed in as little oxygen as possible. A Cochrane review (11) of this trial and two others (12)(13) clearly demonstrated that unrestricted, high levels of ambient oxygen could cause severe eye disease in preterm infants (Fig. 3).Results of the large cooperative trial (13) were widely interpreted at the time as suggesting that oxygen was safe as long as it was not administered at more than 40% concentration. The fact that babies in one arm of the trial not only had received more oxygen, but also received it for much longer was almost entirely overlooked. So, too, was the fact that that some babies in the restricted exposure arm still developed eye damage. Even more seriously, it took some time for clinicians to realize that a policy of restricting oxygen exposure rather than restricting arterial oxygen levels almost certainly was causing a increase in the number of early neonatal deaths. (14)(15)(16) Even in the cooperative trial, where restricted exposure was started only on entry into the study 48 hours after birth, mortality was 10% higher among babies allowed a limited amount of oxygen.Since the bombshell of the 1950s oxygen-induced blindness epidemic, many investigators have tried to define a safe level of arterial oxygenation. Unfortunately, there is still considerable uncertainty regarding what constitutes “normal” blood oxygen levels for infants.Fetuses grow well in utero with arterial blood that is only 70% to 80% saturated with oxygen (Fig. 4). Children who have cyanotic heart disease make the transition to extrauterine life without difficulty with saturation levels similar to those of the fetus. Nonetheless, the aim of clinicians during the past 3 decades generally has been to attempt to maintain oxygen levels in all newborns in line with those of term and noncompromised preterm infants. Several studies have documented what are often termed “normal” or reference values for oxygen saturation (Spo2) or partial pressure of arterial oxygen (Pao2) levels for both term and well preterm infants. (3)(17)(18)(19)(20)(21)(22) These studies demonstrated a relatively narrow range of normal baseline Spo2 values during regular breathing, that is, in quiet sleep. For newborn preterm and term infants, the range is 93% to 100% (0 to 28 d age); for older term infants, the range is 97% to 100% (2 to 6 mo). These data correspond with the few existing studies of Pao2, which have demonstrated a mean of 70 to 76 mm Hg in term infants on postnatal days 2 to 7. (23) Desaturation episodes are common in both term and preterm infants in the early neonatal period, but decrease markedly with age. Preterm infants, particularly those who have prolonged dependency on supplemental oxygen, have lower baseline saturation levels, (24) more frequent desaturations episodes, (25) disturbed sleep patterns, (26) greater risk of pulmonary hypertension, (27)(28) increased adverse pulmonary sequelae, (29) and poorer long-term growth and development (30) compared with preterm infants who do not have chronic oxygen dependency or infants born at term. However, it is not known whether these associations are causal. It remains unclear whether attempts to ameliorate the “non-normal” states previously described do more harm than good. In other words, it is unknown whether oxygen-related interventions designed to, for example, reduce desaturations, decrease Pao2 variability, or improve sleep actually make any material difference in long-term outcomes that are meaningful to children and their families, such improving growth and development in infancy or reducing serious adverse pulmonary sequelae or death.Perhaps because of the enigma of what is “normal” for infants making the transition to extrauterine life, particularly when they are born very prematurely, the opinions as to what constitutes a safe minimum or maximum level of oxygenation for a small baby in the first weeks after birth vary widely. (31)(32) Figure 5 illustrates how practice varies in the United Kingdom. (33) A survey of Australasian practice is currently underway, and policies in the United States also vary substantially. (34)One intervention designed to keep infants’ oxygen levels within the “normal” range is oxygen monitoring. The primary aim of this intervention is to reduce hypoxic and hyperoxic episodes and to decrease the variability in an infant’s oxygen levels by enabling close or continuous monitoring and rapid changes in therapy as required.Direct, but intermittent, measurement of Pao2 via indwelling intra-arterial catheters may not reflect an infant’s steady-state oxygen level, and this measurement method also has potentially serious complications. Hence, continuous, noninvasive methods of measuring blood oxygen levels in neonates have been developed in recent years. Transcutaneous measurement of the partial pressure of oxygen (TcPo2), a technology developed in the 1970s, appears to approximate actual arterial oxygen levels well in most circumstances. (2) However, the use of continuous, noninvasive oxygen monitoring methods has not necessarily resulted in significant improvements in the outcomes that they were designed to affect, such as retinopathy of prematurity (ROP). (35) Investigators in some nonrandomized studies (36) have claimed a near abolition of retinopathy using TcPo2 monitoring; others (37) have reported no difference in the incidence or severity of ROP attributable to TcPo2 monitoring. The only randomized trial (38) to date that has examined the effect of transcutaneous monitoring (continuous TcPo2 monitoring versus standard care) on the incidence of ROP suggested a modest improvement for infants whose birthweights were greater than 1,000 g, but no effect on smaller infants in whom ROP occurs more frequently and is more severe. Conversely, there was a trend toward higher mortality in the group receiving continuous transcutaneous monitoring, and the rates of the combined outcome (death or ROP) were nearly identical in the two groups. Further, these investigators did not detect any effect of transcutaneous monitoring on the incidence of chronic lung disease. It has been hypothesized that the variability of oxygen levels rather than a threshold upper level might be the primary contributing factor to the development of ROP in at-risk infants. (39)Oxygen saturation monitoring using pulse oximetry has gained widespread use in neonatal nurseries since the early 1980s (40) due to its ease of use and lack of heat-related adverse effects, particularly in extremely preterm infants who have sensitive skin. However, there is very little evidence of its effectiveness on clinically important outcomes. (2) No randomized trials have attempted to assess whether oxygen saturation monitoring in itself can reduce the risks associated with oxygen exposure for newborns. The evidence from nonrandomized studies suggests that pulse oximetry is a reliable measure of oxygenation in infants who have chronic lung disease and prolonged oxygen dependency, particularly at lower Pao2 levels. (41)(42) The only randomized trial of pulse oximetry monitoring in infants was performed in patients undergoing surgery, and results suggested the value of pulse oximetry in detecting major hypoxic events in anesthetized children. (43) However, the ability of pulse oximeters to detect hyperoxia reliably (44)(45)(46) remains controversial, with most authors suggesting that oxygen levels should be corroborated with intermittent arterial blood gas estimations (47) or pulse oximeters used in conjunction with, rather than as a replacement for, transcutaneous monitoring where possible. (48)(49)Following the lessons learned from the oxygen-induced blindness epidemic in the 1950s and the ensuing proliferation of oxygen monitoring devices, many investigators have attempted to define a safe level of oxygenation for newborns. Unfortunately, very few studies in this area of medicine have employed randomized, controlled trials methodology.Restriction of inspired oxygen concentration to less than 40% regardless of Pao2 was designed to reduce the early 1950s epidemic of ROP (Fig. 2). Compelling evidence suggests that curtailing unrestricted high levels of ambient oxygen in the first postnatal weeks can reduce the incidence of severe eye disease significantly. (11) The results of the few early randomized trials addressing this issue were widely interpreted as suggesting that oxygen was safe in concentrations less than 40%, (50) but proponents of this approach overlooked the fact that babies in the unrestricted arms of the trials had longer, not just higher, oxygen exposure. A possibly more important factor was that infants enrolled in the early trials often were not entered until after 48 hours of age. Failure to recognize these study limitations resulted in a widely implemented policy of strictly curtailing oxygen to concentrations of less than 40% and almost certainly resulted in an increased number of early neonatal deaths in the ensuing decade. (14)(15)(16)In the 1950s and 1960s, attempts also were made to restrict ambient inspired oxygen to less than that of room air (commonly 15%, with room air having 21% oxygen), then gradually increase the inspired oxygen content. The hypothesis was that the sudden and premature exposure of preterm infants to blood oxygen levels far in excess of fetal levels resulted in the morbidity seen. Studies in both animals and human infants suggested that 15% oxygen can reduce oxygen consumption in the newborn infant. (51)(52)(53)(54) However, only a small number of infants was studied, and there was little meaningful follow-up of these infants to indicate whether this practice resulted in adverse long-term neurodevelopmental sequelae or excess death rates.Since the advent of techniques to monitor infants’ oxygen levels noninvasively, it has been possible to institute interventions designed to expose infants to lower Pao2/Spo2 levels in an attempt to reduce rates of severe ROP and other potential adverse effects of prolonged oxygen exposure. (55)Unfortunately, no randomized, controlled trials have assessed directly the effects of targeting a lower Pao2 or Spo2 range from birth or soon after on important outcomes, such as severe ROP, death rates, or long-term growth and development measures. However, recent evidence from several observational studies has suggested that this hypothesis is worth exploring (Tables 1 and 2). (34)(56)(57)(58) According to these recent studies, neonatal units that adopt a policy of lower oxygen saturation targeting (either compared with similar contemporaneous units or within their own unit over time) report a dramatic reduction in rates of severe ROP. In the 2001 study by Tin and associates, the significant difference in ROP rates was achieved without higher rates of death or poor neurodevelopmental outcome. (56) However, the association between a lower target Spo2 range and improved outcomes reported in these studies cannot be deemed causal, and because there are no randomized, controlled trials, uncertainty remains.Two recently completed trials have assessed outcomes when a lower versus higher oxygen saturation range is targeted in chronically oxygen-dependent preterm infants. Information derived from these two trials suggests that lower target ranges may be associated with less adverse pulmonary sequelae (STOP-ROP, USA) (59) and less health service burdens in the form of reduced oxygen dependency at 36 weeks’ postmenstrual age and significantly lower home oxygen rates (BOOST, Australia) (60) (Fig. 6). Although these trials have contributed evidence regarding the effects of differing oxygen saturation target ranges for chronically oxygen-dependent, preterm infants, they do not address the question of the most appropriate range at which to target such infants in the first, critical weeks after birth.A randomized trial (provisionally known as POST-ROP) is being planned to help redress this evidence gap by assessing the effects of lower Spo2 levels from birth on ROP rates and other short-term outcomes while ensuring no concomitant (even small) increase in death or major disability rates (personal communication, Cole C, Tarnow-Mordi WO, 2002). The question is whether targeting lower saturation values, thereby potentially gaining several secondary short-term benefits (eg, shorter duration of mechanical ventilation, better early growth, less lung scarring, less infection, lower costs, and other outcomes), can be achieved with equivalent survival and developmental and cognitive outcomes at 2 to 4 years. This will be, in effect, a noninferiority trial that requires large numbers. If the trial also shows that fewer babies required retinal surgery or had less severe eye disease, that would be a bonus. The challenge is that the trial will require more than 10 times the 212 infants used in the first cooperative trial. (13)The planned POST-ROP trial seeks to enroll babies of less than 28 weeks’ gestation (because gestation rather than birthweight is the best predictor of serious morbidity, including ROP) and randomize them within a few hours of birth to a target Spo2 range of either 85% to 90% or 91% to 96%. This intervention will continue until 32 weeks’ postmenstrual age or until the infants are stable in room air. The recent BOOST trial successfully used masked oximeters to ensure double-blinding of treatment allocation. (60) The POST-ROP trial plans to use a similar study design that entails the of use Masimo Radical pulse oximeters whose display values are offset by either +3% or −3%. Caregivers will be asked to target an enrolled infant’s saturation in the blinded target range of 88% to 93%. Although this technique appeared to be well accepted by BOOST trial participants, the feasibility of using this method in more acutely ill infants requires a careful pilot study to ensure the cooperation of both parents and staff. Masimo oximeters employ signal extraction technology and appear to reduce false-positive alarm rates significantly, which should enhance staff acceptance of the study oximeters. (61) The STOP-ROP trial monitored treatment allocation compliance via continuous bedside monitoring and feedback. (59) This may be difficult and costly to implement universally in a large, international, multicenter trial, but it may be feasible to collect data on actual versus intended oxygen saturation ranges in a subset of trial infants. Even more important in the planned POST-ROP trial is for participating centers to ensure reliable follow-up and assessment of infants at 2 to 4 years of age because even limited loss to follow-up could distort the perceived outcome.The COT trial is a smaller but important investigation currently being planned by the United States NICHD network to assess aspects of lower oxygen saturation targeting from birth. It also will use blinded offset Masimo oximeters to assess the impact of two strategies for initial stabilization after birth in infants of 24 to 27 weeks’ gestation. The 2 × 2 factorial design will allocate infants to oxygen saturation ranges of either 85% to 89% or 92% to 96%, with co-randomization to either face mask resuscitation followed by nasal CPAP or to the more routine strategy of intubation and prophylactic surfactant administration. Both short- and longer-term outcomes, such as chronic lung disease rates and neurodevelopment, will be assessed.The targeting of higher Pao2/Spo2 levels has been explored extensively over the past 2 decades, mostly with animal models and human physiologic and observational studies. This intervention is hypothesized to improve a variety of outcomes associated with preterm or oxygen-dependent infants. These include improving sleep architecture (62) and decreasing apnea and desaturation episodes, (24) oxygen consumption, (63) pulmonary hypertension, (64) and subclinical hypoxia. (27) Improvement in these surrogate outcomes is believed to be associated with improved long-term growth and development, the real outcome of interest to families. Other outcomes for which higher oxygen targeting is believed to be beneficial include reducing ROP progression (65)(66) and death rates after discharge from hospital. (67)Again, little direct evidence from randomized trials can inform clinicians of the effects of this increasingly common intervention. (11) The recently completed BOOST randomized trial found no difference in long-term (1 y corrected age) growth and development measures when a higher Spo2 range was targeted in extremely preterm infants who were still oxygen-dependent after 32 weeks postmenstrual age (Fig. 7). (60) This finding contradicts the substantial body of observational evidence that suggests higher oxygen targeting can improve growth, (68)(69) ameliorate sleep pattern abnormalities, (62) and reduce desaturation episodes. (24)The effect of higher oxygen targeting on adverse outcomes such as death and pulmonary sequelae is also unclear. Both the BOOST(60) and STOP-ROP trials (59) reported increased adverse pulmonary sequelae in the higher oxygen target groups, although as secondary outcomes in both trials, they lack the statistical power to assess such results reliably. This reinforces the need for a trial with sufficient power to detect such adverse outcomes conclusively. Infants for whom a higher oxygen saturation range was targeted in the BOOST trial had greater use of postnatal steroids (58% versus 50%) and diuretics (52% versus 44%), more readmissions (54% versus 48%), and more pulmonary deaths (6 versus 1).The infants for whom a higher oxygen saturation range was targeted in the STOP-ROP trial (59) had more exacerbations of chronic lung disease (13.2% versus 8.5%). In addition, there was more diuretic usage (35.8% versus 24.4%), and more babies were still receiving oxygen (46.8% versus 37.3%) and in the hospital (12.7% versus 6.8%) in the high oxygen exposure group at 50 weeks’ postmenstrual age. Even more revealing, as might be predicted if oxygen is toxic in high but not low concentrations, was that children in the STOP-ROP trial who already had impaired lungs at recruitment (and, therefore, would have been in a higher oxygen environment) suffered most of the pulmonary exacerbations. Although none of these findings was statistically significant, the consistency of the direction of effect lends weight to the hypothesis that high concentrations of oxygen may result in pulmonary toxicity.The effect of higher oxygen targeting on ROP outcomes also is mixed. It is clear that unrestricted, high ambient oxygen resulted in an epidemic of severe ROP during the 1950s, and there is accumulating evidence (trials, animal models, observational studies) that higher oxygen targeting may improve eye outcomes if used as a short-term treatment in a subset of extremely preterm infants who have severe ROP during the recovery phase of the disease. Again, no randomized trial evidence is available to inform clinicians about the effect of higher oxygen targeting from birth on the rates of severe ROP (stages 3 or 4). This remains a significant problem because approximately 12% of infants born at fewer than 28 weeks’ gestation have ROP of this severity, and more than 50% of these infants receive invasive ablative retinal surgery in an attempt to treat the condition. (57)(70) Although surgery can be effective, it is by no means always successful. Ophthalmologic outcome at 10 years in the CRYO-ROP trial (71) highlighted that even with treatment, acuity outcomes for eyes that have confirmed threshold ROP are favorable in only slightly more than 50% of the eyes, and retinal detachment and blindness still occurred in some infants. (72) There also have been reports of iris atrophy, cataracts, and hypotony following cryotherapy. (73) There is a pressing need to find methods of preventing the development of severe ROP to avoid the current dilemma of determining whether to provide slightly more oxygen at 35 to 36 weeks’ postmenstrual age to reduce the risk (slightly) of ROP progression at the potential expense of further exacerbating serious bronchopulmonary dysplasia.For more than half a century, clinicians have been administering supplemental oxygen to newborns in relative ignorance of this medicinal product’s potentially harmful long-term effects. The ongoing uncertainty surrounding the most appropriate ranges to maintain oxygen levels for either term or preterm infants has contributed to wide variation in practice and has fueled the current controversy regarding this issue.Perhaps the time has come to admit ignorance of how to optimize the delivery of supplemental oxygen to the very preterm baby and to agree to use the tool best suited to address that uncertainty: the randomized, controlled trial. Despite calls as long ago as 1977 for large, collaborative trials to resolve this important issue, (74) very few have been forthcoming. It was collaboration across three continents that initially helped identify the link between oxygen and retinopathy in the preterm baby. It seems likely that similar collaboration will be required to optimize the use of a powerful product that we now know can do great harm as well as great good.We are grateful to Dr. Edmund Hey for his help in the preparation of this paper.
Referência(s)