Neonatal Outcomes of Extremely Preterm Infants From the NICHD Neonatal Research Network

DISCUSSION

Although VLBW infant mortality rates in the United States decreased substantially in the 1980s and early 1990s,^{3,–,6,15,–,18} most reports, including findings for this cohort, failed to demonstrate further progress in reducing neonatal morbidity and mortality rates.^6,16,–,19 In contrast, a population cohort of all preterm infants born at GAs of <27 weeks in Sweden in 2004–2007 demonstrated survival rates higher than rates reported for other countries or reported previously for Sweden.²⁰ Our study reviewed neonatal morbidity and mortality rates for a large cohort of extremely preterm infants, to evaluate changes in clinical practice and contemporary outcomes at US academic centers. Although previous reports from the NRN used BW as the reference for morbidity and survival rates, the current study assessed outcomes according to GA. Appreciation of GA-based outcomes is particularly valuable for prenatal counseling and physician/family decision-making.

The decisions to provide active obstetric care and to initiate neonatal intensive care for the most-immature infants remain controversial. Center differences in obstetric/early neonatal interventions were identified, but we did not collect sufficiently detailed information on decision-making processes to help explain differences. In our cohort, rates of active obstetric intervention, as indicated by prenatal steroid administration and cesarean section delivery, increased markedly after 23 weeks of gestation. Prenatal steroid use was almost twice as frequent for infants born at GAs of 24 to 28 weeks, compared with infants born earlier. Similarly, rates of neonatal interventions and intensive care, measured as active resuscitation with ventilation in the delivery room, increased substantially between 22 and 23 weeks (19% vs 68%). Rates of death at ≤12 hours, which in part reflect willingness to provide intensive care to the most-immature infants, decreased with increasing GA, from 85% of infants at 22 weeks to 2% of infants at 28 weeks.

In-hospital morbidity rates remain high among extremely preterm infants, and morbidities contribute to adverse neurodevelopmental outcomes. The majority of infants studied experienced a major complication during the initial hospitalization, with the risk of morbidity being inversely related to GA at birth. Center differences in the proportions of infants with specific morbidities were noted. At the lowest GAs (22–24 weeks), small numbers of infants at some centers contributed to the variability. The registry does not collect data on the reasons behind the choice of interventions for individual infants and has limited data on the severity of illness at birth, information that might permit more-detailed evaluation and understanding of center differences. Reducing the high rates of in-hospital morbidity among extremely low GA infants who are provided ongoing intensive care remains a challenge for clinicians and investigators.

To reduce rates of BPD, attention is being paid to avoidance of intubation, less prophylactic use of surfactant, and alternative modes of respiratory support. Rates of endotracheal intubation in the delivery room decreased in recent years among infants of >24 weeks, with a corresponding increase in CPAP therapy use at 24 hours of life. At GA of 28 weeks, use of surfactant decreased in the most-recent years. Furthermore, the proportion of infants who survived >12 hours without ever undergoing intubation and ventilation increased with increasing GA and more-recent year of birth. With substantially increased use of CPAP therapy, it was surprising that overall rates of BPD were unchanged, although the adjusted RR for BPD decreased over the study period.

This is the first study to report ophthalmologic status as favorable, unfavorable, or undetermined at the time of the last in-hospital examination. Although 7% of all infants had severe ROP, the rate was 30% for infants with GAs of 22/23 weeks. Of concern, 53% of infants had undetermined ophthalmologic status at the last examination before discharge. This finding has implications for discharge planning and underscores the importance of a medical home, to ensure careful ophthalmologic follow-up monitoring of these vulnerable infants after discharge home or transport to a community hospital.

Although ours is not a population-based study, we included all extremely low gestation births at 20 academic centers across the United States that together represent >110 000 live births per year, an annual birth cohort equal in size to the Swedish national cohort described recently.²⁰ The rate of extremely low gestation birth was fivefold higher in our NRN cohort (∼10 births at <27 weeks per 1000 infants) than in the Swedish cohort (2.3 births at <27 weeks per 1000 infants). This remarkable difference may be explained in part by Sweden's universal health insurance, with free prenatal care and associated social services, as well as an ethnically more homogeneous and somewhat older pregnant population. The high rates of prematurity in our cohort underscore the importance of the current health care debate in the United States. Survival rates for extremely low gestation infants born at NRN centers are lower than those reported from Sweden. For nearly all infants in the Swedish cohort, GA was estimated on the basis of ultrasound findings. The authors of the Swedish study noted that a limitation of the use of ultrasonography to determine GA is that erroneously low GAs might be estimated for infants with growth restriction. Given the decrease in mortality rates with increasing GA, underestimation of GA by as little as 1 week might explain in part the difference in mortality rates between the 2 cohorts. Greater use of prenatal steroid treatment at all GAs and of surfactant therapy at 22 to 23 weeks also might have contributed to differences between the 2 cohorts.

During the 5-year study period, there was no substantial improvement in rates of survival to discharge for extremely low gestation infants born at NRN centers. However, each additional week of GA at birth had substantial survival advantage; the most marked changes were between GAs of 22 and 25 weeks, with survival rates increasing from 6% to 72%. Furthermore, rates of survival to discharge without major morbidities increased dramatically between 22 and 25 weeks, with continued steady improvement for each additional week of gestation. PMA at discharge for VLBW infants, a proxy measure of length of stay and a reflection of the cost of care, was inversely related to GA at birth. Each additional week of GA at birth reduced PMA at discharge by almost 1 week and total length of hospital stay by ∼2 weeks, a reflection of both severity of illness and complications of prematurity among these very immature infants. Although adjusted RRs for survival without morbidity increased over time, the burden of in-hospital complications remained high. Retrospective analyses of center differences and benchmarking studies to identify best performance have been unable to identify modifiable practices that consistently improve outcomes, which underscores the need for hypothesis-driven clinical trials to assess the efficacy of current neonatal interventions.^21,–,24 Clinicians and investigators are challenged to identify and to test currently available interventions and resources that yield consistently lower morbidity and mortality rates at some centers, so that we can improve rates of survival without major morbidities and reduce long-term neurodevelopmental impairments for all infants.