This national retrospective cohort study is the first to provide contemporary population-based US demographic and survival statistics for preterm infants with CCHD by week of gestation at birth, including extremely preterm infants. This study suggests that the pattern of mortality for preterm infants with CCHD is distinct from that of preterm infants without CCHD. Understanding these patterns in this understudied group of vulnerable infants with national, population-based data are extremely valuable despite inherent limitations in birth certificate data and is critical to advancing care.

CCHD was over represented in preterm infants, with the highest prevalence occurring in infants born between 25 and 28 weeks GA. Preterm infants with CCHD born ≥ 23 weeks GA were less likely to survive to 1 year compared to GA-matched infants without CCHD. The largest survival gap occurred between 28 and 31 weeks GA. Between 23- and 31-week GA, early mortality accounted for more than half of the deaths in infants born with CCHD, substantially more than those without CCHD. Finally, 1-year survival of preterm infants born with CCHD in the US did not improve over the study period and in fact worsened in 35–36-week infants. In contrast, survival improved in almost all GA categories among preterm infants without CCHD.

Infants with CCHD in this study were more likely to be born premature, IUGR, and to birthing persons with pre-existing or antepartum health conditions. It is known that infants with CHD are more likely to be preterm [3, 4] and growth restricted [12]. Moreover, these factors are predictive of poorer outcomes [3, 4, 13]. Additionally, health complications including hypertension and diabetes are known to be more common in pregnancies with CHD [14, 15], but their impact on outcomes in infants with CHD is less clear [16].

It is well established that preterm infants with CHD are less likely to survive than preterm infants without CHD [17]. However, the majority of studies use GA as a dichotomous variable (term/preterm), and few studies have evaluated survival by GA as a continuous variable and with the inclusion of extremely preterm infants. With this unique lens, this study was able to demonstrate that differences in survival between preterm infants with and without CCHD were not consistent across gestational ages. At the extreme end of viability (< 23 weeks), the difference between the two groups was the smallest. Given the known high mortality rates in this periviable population [5], it is likely that mortality in these infants was primarily driven by extreme prematurity. The survival gap increased from 24 to 28 weeks, was widest from 28 to 31 weeks, and decreased modestly, but remained present thereafter. The widening gap in 1-year survival from 24 to 31 weeks GA is driven by an asymptotic improvement in survival in infants not affected by CCHD with a relative stagnation of the probability of survival in infants with CCHD.

An international cohort study of infants 24–31 weeks GA with severe CHD also found no significant difference in mortality between infants with and without CHD at the youngest GAs studied, 24–26 weeks, but an increasing survival gap thereafter [7]. Similarly, a California cohort study published by our group demonstrated an increasing survival gap between infants with and without serious CHD in infants ≥ 26 weeks GA up until 31 weeks GA and a slight decrease thereafter [4]. This California cohort study, however, did not include infants < 26 weeks. Notably, point estimates for mortality are lower in these two studies [4, 7] compared to our study likely due to inclusion of non-cyanotic CHD. Other recent studies combine GAs < 32 weeks [9, 18] making it difficult to ascertain these GA-specific trends.

There is much to gain to mirror the improved outcomes of infants born at preterm GAs without CCHD. The field of neonatology has made progress in survival of preterm infants over the last few decades [5, 6, 19]. Even over the 5-year period from 2014 to 2019, 1-year survival improved in almost all GA categories among preterm infants without CCHD in this study. The same was not true for infants with CCHD. 1-year survival of preterm neonates with CCHD in the US did not improve over the study period and in fact, it worsened in 35- to 36-week infants. This lack of improvement on a national level is concerning. It may be due to the limits of surgical techniques but it may also be due to limitations of medical intensive care in the pre- or post-operative period and beyond.

Neonates who die shortly after birth in the neonatal intensive care unit without ever being transferred to a cardiac intensive care unit or undergoing a surgical intervention are missed in existing CHD registry data. This represents a significant percentage of infants. In fact, early mortality, before three days of age, accounted for more than half of the deaths among infants with CCHD born at ≤ 31 weeks GA in this study. Similarly, the EPICARD population-based cohort from France looked at early neonatal mortality, defined as death less than 1 week of age, and also found that it accounted for more than half of neonatal deaths [9]. Therefore, one of the major benefits of using this population-based dataset is that it includes all liveborn preterm infants with CCHD.

The burden of early mortality among preterm infants with CCHD is striking, particularly among very preterm neonates. It may in part reflect decisions around the provision of active care. However, it also further underscores the need for improved early, pre-operative neonatal intensive care for this vulnerable population, many of whom can be stabilized on prostaglandin E — especially, as surgical techniques improve and more preterm infants are able to undergo technically successful cardiac surgery [8, 20].

One approach might be specialized units designed to care for this dually affected population where the complicated and interlacing physiology of congenital heart disease and comorbidities of prematurity, including fetal lung immaturity and neurologic vulnerability, can be considered and balanced by a skilled neonatal cardiac intensive care team. Small baby units have been successful in delivering evidence-based care and improving outcomes for extremely preterm infants [21, 22] and as such, represent an interesting model to consider for the preterm CHD population. A recent publication demonstrated success with this type of care model in reducing in-hospital mortality for very preterm infants with CHD at a single center [23].

The main strengths of this study come from the nationally representative birth certificate data inclusive of all liveborn preterm infants with and without CCHD down to 21 weeks GA and available timing of death in days. This allows for a granular understanding of outcome metrics for this understudied group of vulnerable infants at a population level that is not well captured in cardiac registry data. Nevertheless, population-based data also comes with limitations. One of the main limitations of this study is the exclusion of all non-cyanotic CHD. Thus, the results of this study, particularly the survival predictions, are not generalizable to all CHD, and especially not to the milder diagnoses. Additionally, specific type of CCHD, major neonatal, or cardiac morbidities and actual timing of surgery are not available from the database. It is therefore impossible to actually distinguish between pre-operative mortality, operative mortality, and post-operative mortality. Furthermore, information around decisions related to providing or with-holding interventions including neonatal resuscitation is not available. A portion of infants in the early mortality group likely did not receive active care but the exact percentage cannot be known. Additionally, prenatal data are not available and thus fetuses with CHD who are not liveborn are excluded. Although rare, infants not identified to have cyanotic congenital heart disease during their initial admission, when birth certificates are completed, are also missed by this database. Finally, morbidity and longer-term neurologic outcomes are outside the scope of this work. They are, however, critically important considerations and have been previously reported to be poorer in infants with CHD [4, 24].