Respiratory function monitoring during early resuscitation and prediction of outcomes in prematurely born infants

Introduction

Over the last decade, there has been increased use of end-tidal carbon dioxide (ETCO₂) and oxygen saturation (SpO₂) monitoring [1] during resuscitation of prematurely born infants in the delivery suite This has demonstrated that preterm infants who develop an intracerebral haemorrhage (ICH) had lower and more variable ETCO₂ levels than infants who do not develop an ICH during resuscitation [2, 3]. Increasing ETCO₂ levels indicates pulmonary vasodilation has occurred and thus low levels may indicate a more compromised infant with poorer adaptation. Holte et al. showed that higher and faster rising ETCO₂ levels were associated with improved survival. Achieving adequate ETCO₂, however, must be carefully balanced alongside the risks of high VT_E which is associated with higher expired CO₂ [4], as potentially high VT_Es could increase the risk of BPD [5]. Furthermore, in one study, preterm infants who developed an ICH received more inflations in the delivery suite with a higher expired tidal volume (VT_E) [3]. The 2015 International Consensus for Newborn Resuscitation stated that more work is needed to define the role of ETCO₂ monitoring in neonatal resuscitation [6].

The TO2RPIDO study, which compared two year outcomes of infants born at less than 32 weeks of GA who were resuscitated with FiO₂ of 0.21 or 1.0, was prematurely terminated due to a lack of equipoise for the use of 100% oxygen. A post hoc analysis demonstrated higher mortality in the infants born at less than 28 weeks of gestation who were ventilated in air [7]. Earlier studies reported prolonged cerebral vasoconstriction [8], more BPD at discharge and increased requirements of oxygen supplementation and mechanical ventilation [9] in premature infants resuscitated with 80–90% FiO₂. The increase in BPD, however, was not confirmed in a subsequent meta-analysis [8]. The 2021 Resuscitation Council (UK) guidelines recommend low inspired oxygen (FiO₂ 0.21–0.3) for infants born between 28 and 32 weeks of GA, increasing to an FiO₂ of 0.3 for infants born at less than 28 weeks of GA [9]. A post hoc analysis demonstrated that infants born at less than 32 weeks of GA who did not achieve a SpO₂ of 80% by 5 min after birth were more likely to die or develop neurodevelopmental impairment [10]. A meta-analysis of eight studies demonstrated an increase in death or severe ICH in such infants [11] and this was recently confirmed by Katheria et al. [12].

The aim of this study, therefore, was to test the hypotheses that low ETCO₂, low SpO₂ and high VT_E levels during early resuscitation in the delivery suite would be associated with increased mortality and ICH or BPD development in prematurely born infants.

Materials and methods

Resuscitation recordings in the delivery suite from infants born at less than 34 weeks of GA were retrospectively reviewed. Ethical approval was given by the Outer London Ethics Committee for the data collection. The committee required parental consent only for analysis of the data, which was obtained once the mother was transferred to the postnatal ward.

Results

During the study period 370 infants were born less than 34 weeks of gestational age. There were recordings available from 150 infants for analysis, 56 had recordings of tidal volume or ETCO₂ traces which were unsuitable for analysis, a further 24 were excluded due to a large leak and a further 10 excluded for a lack of interpretable data at 5 min. The outcomes of the ten infants were as follows: one had an ICH and died, three developed an ICH and BPD, two developed BPD and four developed an ICH. The remaining 60 infants’ traces were analysed. The infants had a median gestational age of 27 + 2 (interquartile range (IQR) 25 + 1–29 + 0) weeks and a birth weight of 0.92 (IQR 0.77–1.19) kg.

Eleven of the 60 infants (18%) did not survive to discharge. Those who died had a lower Apgar score at 5 min (p=0.048) and were more likely to be born to mothers with chorioamnionitis (p=0.034) (Table 1). The SpO₂ at 5 min was lower in the infants who died than those who survived (p<0.001), which remained statistically significant after adjusting for the 5-min Apgar score and chorioamnionitis (p=0.021). The AUROC was 0.859 (Figure 1), with an SpO₂ of less than 47.75% having a sensitivity of 89% and specificity of 75% in predicting mortality. There were no significant differences in PIP, MAP, ETCO₂ or VT_E levels at 5 min in the infants who did and did not survive to discharge (Table 1).

Figure 1:

ROC analysis of mortality and oxygen saturation levels.

Cranial ultrasound examinations demonstrated that 25 of the 59 infants (42.4%) had an ICH. One infant died during resuscitation and did not have an admission cranial ultrasound. ETCO₂ levels at 5 min were lower in the infants that developed an ICH (28.5 mmHg, IQR: 17.3–34.7 mmHg) than those who did not (43.1 mmHg, IQR: 29.9–53.4 mmHg) (Table 2). This remained significant after adjusting for gestational age, coagulopathy and chorioamnionitis (p=0.03). ETCO₂ levels were lower in infants that developed ICH or died (27.8 mmHg, IQR: 17.8–34.5 mmHg) compared to those that survived without ICH (43.5 mmHg, IQR: 31.6–54.4 mmHg). This remained significant after adjustment for gestational age, Apgar score at 10 min, chorioamnionitis and coagulopathy (p=0.004). The AUROC was 0.767 (Figure 2) with an EtCO₂ of less than 33 mmHg having a sensitivity of 73% and specificity of 64% in predicting ICH. The only significant difference between the infants who developed severe ICH (grade three or four) and those who developed less severe ICH was in the 5 min Apgar scores (Supplementary Table).

Figure 2:

ROC analysis of ICH development and ETCO2 levels.

BPD was diagnosed in 23 of the 49 infants (46.9%) alive at 36 weeks gestational age. There were no significant associations of any of the respiratory function results and BPD development (Table 3).

Discussion

We have demonstrated that at 5 min of resuscitation in premature infants, a lower SpO₂ was associated with increased mortality and a lower ETCO₂ with an increased incidence of ICH. At 5 min, the SpO₂ was 42.8% in those that died an 84% in those who survived to discharge. In a cohort of 160 preterm infants (median GA 33 weeks), the median SpO₂ at 5 min was 86% (IQR: 80–92%) [13]. Our cohort, however, were of lower GA (median GA 27 + 2 weeks). Indeed, there is a paucity of reference values of SpO₂ levels in extremely preterm infants. Our findings are in keeping with previous work that demonstrated an association of a low SpO₂ with the composite outcome of death or severe ICH [12]. A systematic review and meta-analysis of eight randomised controlled trials found no significant differences in death, ICH or BPD in infants who received a low (0.3) compared with high (0.6) fraction of inspired oxygen (FiO₂) at delivery [8], but a lower mortality was noted in the high oxygen arm of unmasked studies (8 vs. 15.7%, p=0.04). Gandhi et al. evaluated a visual oxygen saturation target monitor and found that preterm infants resuscitated with the target system had SpO₂ values maintained in the target range for longer than infants resuscitated without the monitor [14].

We also observed that low ETCO₂ levels at 5 min after birth were associated with increased ICH development. Immediately after birth, carbon dioxide elimination only occurs if there is effective ventilation of the lungs and associated vasodilation of the pulmonary vascular bed. In the absence of pulmonary vasodilation only 10% of the cardiac output is available to perfuse the lungs greatly restricting the delivery of carbon dioxide to the lungs. Thus, assessment of ETCO₂ levels could be used to indicate that pulmonary vasodilation had occurred during resuscitation. Low ETCO₂ levels then may indicate a more compromised infant with poorer adaptation [15]. Currently the Resus Council (UK) guidelines specify detection of exhaled CO₂ only to ensure correct tube placement when undertaking intubation [9].

We did not detect a significant association between ETCO₂ levels and the development of BPD. A study from the post antenatal corticosteroid and postnatal surfactant era, in which blood gases within the first hour of birth were recorded in infants less than 33 weeks of GA, also did not detect a significant association between hypo- or hyper-carbia with BPD [16]. We also did not detect any significant associations of higher VT_E levels with poorer outcomes. Mian et al. assessed the tidal volumes given in the delivery room during the resuscitation of infants born at less than 29 weeks of GA. The found a greater proportion of infants who developed an ICH (51 vs. 13%, p=0.008) had received high tidal volumes (greater than 6 ml/kg) [17]. We however, measured expired tidal volumes, rather than the delivered tidal volumes assessed by the Mian group. A randomised controlled trial that used a respiratory function monitor, either masked or visible, showed significantly less excessive tidal volumes in the group with a visible monitor during the resuscitation of preterm infants [18]. Of note, however, a larger unmasked multicentre trial found no significant difference in tidal volumes when a respiratory function monitor was used [19].

This study has strengths and some limitation. To our knowledge this is the first study to evaluate morbidity and mortality outcomes related to ETCO₂ and VT_E at 5 min of age in preterm infants. We did not find any significant differences in the results of the respiratory function monitoring between infants with different grades of ICH which may reflect the relatively small sample size, but we did see significant differences between those who did an did not develop an ICH. Although the data were analysed retrospectively, the respiratory function monitoring were analysed independently of knowledge of outcomes. Some traces were excluded from the analysis because 5 min of recording was not achieved. This could skew our results by removing infants with adverse outcomes, in whom recording might have not been completed because of challenging resuscitations. Our results, however, would then be underestimating the effect size. It should be noted that our study did not have access to the FiO₂ data at the time of resuscitation, but we did continuously record oxygen saturation levels.

In conclusion, at 5 min of resuscitation in premature infants, a low SpO₂ was significantly associated with increased mortality and a low ETCO₂ with increased ICH development. The results suggest that those more compromised and with poorer adaptation as indicated by low oxygen saturations and low ETCO₂ levels, likely reflecting delayed pulmonary vasodilation, during early resuscitation were more likely to die or develop an ICH.