Introduction

Late-preterm infants, defined by birth at 340/7 through 366/7 weeks of gestation (239th–259th day), are at higher risk of neonatal morbidity than their full-term counterparts [7, 16]. These infants are more likely to present with respiratory distress at birth and to be admitted in the Neonatal Intensive Care Unit (NICU) with respiratory complications, including transient tachypnea of the newborn (TTN) and respiratory distress syndrome (RDS) [4, 8, 11, 22, 29]. Late-preterm infants presenting with respiratory distress are also at increased risk of respiratory failure and need for mechanical ventilation [6, 11, 15]. Their increased respiratory morbidity is related to the functional immaturity of the lung structure at 34 through 37 weeks of gestation, which predisposes to delayed intrapulmonary fluid absorption, surfactant insufficiency, and, ultimately, inefficient gas exchange [6].

Several indices, such as the alveolar-arterial oxygen tension difference (A-aDO2), arterial to alveolar oxygen tension ratio (a/A ratio), and partial arterial oxygen tension to inspired oxygen fraction ratio (PaO2/FiO2), have been used to quantify the severity of respiratory failure, compare treatment modalities, and assess short- and long-term outcomes in ventilated neonates [5, 13, 14, 23, 24, 26]. It has been suggested that these indices may be more objective measures of respiratory compromise than commonly used parameters such as inspired oxygen fraction (FiO2), partial arterial oxygen tension (PaO2), and partial arterial carbon dioxide tension (PaCO2) [3, 10, 19, 24]. However, their value in late-preterm infants with respiratory morbidity has not been systematically evaluated.

The aim of this study was to investigate the ability of commonly used respiratory failure indices to predict respiratory impairment and need for mechanical ventilation in late-preterm neonates with respiratory distress at birth. We hypothesized that composite parameters such as the A-aDO2, a/A ratio, and PaO2/FiO2 could be affected earlier in the course of acute lung disease and thus, could predict imminent respiratory failure and need for mechanical ventilation in these infants, allowing for early medical intervention or transfer to a level III NICU.

Materials and methods

Patients

Between January 2006 and June 2008, all neonates admitted for respiratory distress to the NICU of the University Hospital of Patras (Patras, Greece) were screened for eligibility by one of the investigators. Neonates with gestational age 340/7–366/7 weeks only were included in this prospective observational study. Late-preterm infants with respiratory distress which (1) presented signs of severe respiratory impairment and required intubation and mechanical ventilation during the first hour after birth, and (2) had major congenital abnormalities, were excluded.

Protocol

The study was strictly adhered to the standard of care protocols applied for late-preterm neonates with signs of respiratory distress in our institution. Late-preterm infants with signs of moderate respiratory distress (respiratory rate of more than 60 per minute, retractions, and expiratory grunt), and those with signs of mild respiratory distress (tachypnea and/or intermittent grunt), but unable to maintain arterial oxygen saturation levels above 94% without supplemental oxygen administration at 1 h after birth, were admitted to the NICU. Routine management consisted of close clinical observation and supplemental oxygen administration to maintain arterial oxygen saturation levels between 94% and 98%. Arterial blood gases were checked routinely every 4–6 h, depending on the severity of the respiratory disease. In addition, infants with oxygen requirement higher than 30% had arterial access. The decision for mechanical ventilation was based on clinical signs of respiratory impairment and arterial blood gas parameters, if one of the following was present: (1) signs of severe respiratory distress (tachypnea, retractions, and grunting) with an inspired oxygen fraction of greater than 0.5; (2) either a major apnea requiring medical intervention or recurrent minor apneas; (3) an inspired oxygen fraction of greater than 0.5 with PaO2 < 6.7 kPa (50 mmHg); and (4) respiratory acidosis (pH < 7.25) with PaCO2 > 8.0 kPa (60 mmHg). The diagnosis of respiratory distress and the assignment of any related medical intervention (admission in the NICU, frequency of arterial gas determination, and decision for mechanical ventilation) were made by the attending neonatologists, according to the protocol applied in our NICU, and were not at the discretion of the investigators. Attending neonatologists were blinded to the calculated indices.

Blood samples were collected from an indwelling arterial umbilical catheter or via peripheral arterial puncture. Arterial blood gas analyses were routinely performed using a Stat Profile pHOx Series analyzer (Nova Biomedical, Waltham, MA, USA). For each arterial blood gas obtained within the first 12 postnatal hours or until intubation, PaO2, PaCO2, and the respective FiO2 were recorded, and the A-aDO2 ([FiO2 × 713 − PaCO2] − PaO2; expressed in mm Hg), a/A ratio (PaO2/[FiO2 × 713 − PaCO2]), and PaO2/FiO2 (expressed in mm Hg) were calculated [14].

The gestational age of the infants was determined by the attending neonatologists by the date of the last menstrual period and/or by early second-trimester ultrasound, and confirmed by the Expanded New Ballard Score [1]. Demographic, antenatal and perinatal data, such as neonatal gender, gestational age, birth weight, mode of delivery, antenatal steroids, Apgar score at 5 min, hypertensive disorders of pregnancy, and gestational diabetes, were also recorded. Diagnoses were based on clinical presentation and chest radiography (TTN, RDS, pneumothorax, congenital pneumonia, meconium aspiration), blood cultures (infection), and echocardiography (persistent pulmonary hypertension of the newborn – PPHN).

The study was approved by the Ethics Committee of the University Hospital of Patras, and an informed consent was obtained from the parents on admission.

Statistical analysis

Differences between groups were assessed for statistical significance using either the Mann–Whitney U test or chi-squared test, as appropriate. The overall performance of individual parameters in predicting the need for mechanical ventilation was assessed by receiver operating characteristic (ROC) curve analyses and area under the curve (AUC) pairwise comparisons [12]. The parameters evaluated were maximum FiO2, maximum PaCO2, maximum A-aDO2, minimum a/A ratio, and minimum PaO2/FiO2 within the first 12 postnatal hours or until intubation.

An earlier pilot study indicated that, in our NICU, late-preterm neonates with respiratory distress, who ultimately developed respiratory failure, were intubated at a median of 14 h. Hence, we used the time period of first 12 postnatal hours in order to evaluate the predictive ability of the studied indices. In addition, this time period permitted a sufficient number of arterial blood gases (at least three) to be included in the analysis. In the pilot study (n = 30 late-preterm neonates), the intubation rate was 30%, the mean difference in the A-aDO2 between the two groups was 55 mmHg, and the SD in both groups was approximately 60 mmHg. Therefore, 144 patients would be needed to detect a 55-mmHg or greater difference in A-aDO2 between late-preterm infants who developed respiratory failure and those that did not, with a power of 99% and a significance level of 0.01. Since 2002, a mean of 73 late-preterm infants with respiratory distress were admitted in our NICU per year; 10% of these would not meet the inclusion criteria of the present study. Assuming an enrolment failure of 10%, a time period of 2.5 years would be necessary to achieve a goal of 150 neonates (60 eligible neonates per year).

Further analysis was performed by calculating the sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios for values with the highest accuracy (defined as values with minimal false positive and false negative results) of the studied parameters. Statistical analysis was made using the MedCalc 8.1 statistical software (MedCalc, Mariakerke, Belgium).

Results

Of a total of 346 late-preterm neonates born in our institution during the study period, 191 were admitted to the NICU for respiratory distress; 36 of these were excluded from the study (33 with severe respiratory impairment who required intubation during the first hour after birth). Of the 155 neonates finally included, 55 (35.5%) developed severe respiratory impairment and required mechanical ventilatory support. The median age at intubation was 13 h (interquartile range 6–18 h). Respiratory distress in the infants that did not develop respiratory failure was attributed to TTN (n = 94; 94%), pneumothorax (n = 3; 3%), and infection (n = 3; 3%). In the neonates who required mechanical ventilation, the following diagnoses were established: RDS (n = 40; 70.9%), infection (n = 7; 12.7%), congenital pneumonia (n = 4; 7.3%), meconium aspiration (n = 3; 5.4%), and pneumothorax (n = 2; 3.7%). Of them, 5 (9.1%) developed PPHN and received inspired nitric oxide. A total of 612 arterial blood gas determinations during the first 12 postnatal hours or until intubation were recorded and analyzed.

Demographic data and maternal risk factors were similar between neonates who required and those who did not require mechanical ventilation (Table 1). However, a significantly lower percentage of neonates who developed respiratory failure were exposed to antenatal steroids (Table 1). Maximum FiO2, maximum PaCO2, and respiratory failure indices were significantly higher during the first 12 h in infants that ultimately required respiratory support (Table 1).

Table 1 Patient characteristics according to mechanical ventilation requirement
Full size table

The ROC curves for the maximum A-aDO2, minimum a/A ratio, minimum PaO2/FiO2, and maximum FiO2 within the first 12 postnatal hours or until intubation are presented in Fig. 1. No significant differences were found between the AUCs for maximum A-aDO2, minimum a/A ratio, and minimum PaO2/FiO2 (Table 2). All three AUCs were significantly higher than the AUC for maximum PaCO2 and tended to be higher than that for maximum FiO2, but only for the A-aDO2, the difference reached statistical significance. A maximum A-aDO2 greater than 200 mmHg proved to be a highly effective index for predicting respiratory failure (Table 3). This threshold was achieved by 54 of the 55 late-preterm infants (98.25%) who ultimately developed respiratory failure at a median age of 10 h (interquartile range 3–15 h) and 3 h (interquartile range 2.5–6 h) before the definite decision for ventilatory support.

Fig. 1
figure 1

Receiver operating characteristics curves for prediction of respiratory failure in late-preterm neonates with respiratory distress at birth. A-aDO 2 alveolar-arterial oxygen tension difference, a/A ratio arterial to alveolar oxygen tension ratio, PaO 2 /FiO 2 partial arterial oxygen tension to inspired oxygen fraction ratio, FiO 2 inspired oxygen fraction

Full size image
Table 2 Predictive performance of indices of respiratory failure for the requirement of mechanical ventilation in late-preterm neonates with respiratory distress
Full size table
Table 3 Predictive characteristics of the values with the highest accuracy (minimal false positive and false negative results) for the maximum FiO2, maximum PaCO2, maximum A-aDO2, maximum PaO2/FiO2, and minimum PaO2/FiO2
Full size table

Discussion

Births between 34 and 37 weeks’ gestation account for a significant proportion (>70%) of preterm births in the United States and other industrialized countries [8, 20, 29]. Incorrectly, these neonates are often treated like being developmentally mature term infants, although they manifest signs of physiologic immaturity or delayed transition in the neonatal period. Among a number of frequent neonatal complications in this population, respiratory morbidity is of primary concern [4]. Late-preterm infants with respiratory distress represent a significant workload in neonatal nurseries. Nearly 50% of infants born at 34 weeks gestation require intensive care; this number drops to 15% at 35 weeks and 8% at 36 weeks gestation [6]. The majority of these will only require oxygen treatment and close clinical observation, but a proportion the disease process will continue to the point where they require respiratory support and mechanical ventilation [6, 15]. It would, therefore, be very helpful to determine who of the late-preterm infants with respiratory distress requiring oxygen supplementation are at risk for developing respiratory failure, and if those at risk can be identified early in their course.

Respiratory failure indices have been widely used in critical care settings to identify and monitor acute lung injury and respiratory failure both in adult and pediatric patients [2, 18, 21, 25]. In preterm neonates with RDS, the A-aDO2 and a/A ratio have been successfully used to quantify disease severity, compare different ventilation strategies, and predict extubation failure [5, 13, 14, 23, 26]. It has been also suggested that these parameters may be better predictors of the short- and long-term outcomes than other variables, such as birth weight, gestational age, Apgar score, maximal ventilator settings, maximum FiO2, or maximum PaCO2 [24]. However, data on their value in assessing the risk of developing respiratory failure are sparse, and to our knowledge, this is the first study in which these indices were used to predict short-term respiratory outcomes in late-preterm neonates with respiratory distress. In a recent study, it has been suggested that the prediction of the need for ventilation in near-term neonates can be based on the oxygen requirement within the first hours of age, with a reasonable sensitivity and specificity [17]. The FiO2 represents a simple and readily available index; however, it is operator-dependent [3] and has been shown to be a less effective predictor in comparison to other indices of respiratory failure in neonates [5, 13, 23, 24, 26].

In the present study, the predictive performances of the A-aDO2, a/A ratio, and PaO2/FiO2 were quite similar. All three indices outperformed PaCO2 in terms of their ability to predict respiratory impairment, but only the A-aDO2 proved to have a statistically superior performance compared to FiO2. An A-aDO2 of >200 mmHg had better performance characteristics than cut-offs based on a/A ratio and PaO2/FiO2 in predicting imminent respiratory failure and the need for mechanical ventilation. Notably, A-aDO2 values of >200 mmHg were achieved by the majority of neonates early in the course of the disease at a median of 3 h before the ventilatory support to be definitely decided.

In order to assess the predictive ability of the respiratory failure indices, the “worst” single value for each parameter was used. We believe that most neonatologists are familiar with such thresholds, which are more appropriate for assessing the risk of respiratory impairment during the natural course of a respiratory disease. Arterial blood gas analyses are available in level III NICUs and in the majority of level I and II nurseries. Parameters such as the PaO2 and PaCO2, in combination with the FiO2 and other specific clinical variables, are routinely used to monitor neonates with respiratory distress and to guide management strategies. The results of the present study suggest that, in face of a late-preterm infant with respiratory distress (not severe enough to impose ventilatory support soon after birth), neonatologists could also consider the value of composite respiratory failure indices in the monitoring process, in conjunction with the aforementioned conventional parameters. Such an approach could be useful, in terms of implementing appropriate management strategies for these neonates. In addition, in a population where the majority of the infants will not require ventilatory support, any index that could predict respiratory failure early enough would allow the transfer of the high-risk infants to a level III neonatal unit on time.

To our knowledge, this is the first study examining the ability of composite respiratory failure indices to predict respiratory failure in late-preterm neonates with respiratory distress at birth. In a previous study, a simple score was developed to predict short-term respiratory outcome in infants greater than 34 weeks’ gestation [9]. However, the target outcome of that study was the need for assisted ventilation for more than 3 days or death. Although it is important to know whether a neonate is likely to require prolonged assisted ventilation, from a logistical standpoint, it is more important to predict whether the neonate will require mechanical ventilation in the first place.

Demographic data and maternal risk factors were similar between the neonates who required and those who did not require mechanical ventilation. Also, there was no difference in the cesarean section rates among these two groups, although a higher occurrence of respiratory morbidity in late-preterm infants delivered by elective cesarean section has been observed by many investigators [6, 27]. Our data also indicated that the exposure to antenatal steroids led to an improvement in short-term respiratory outcomes among infants with respiratory distress at birth, a finding that corroborates the results of other published reports [28]. It should be noted, however, that in our study, only neonates admitted in the NICU were included, excluding asymptomatic late-preterm infants, as well as those who required mechanical ventilation shortly after birth. Therefore, firm conclusions regarding the association of cesarean section and antenatal steroid administration with respiratory morbidity in late-preterm infants, cannot be drawn.

In conclusion, our data suggest that composite indices of respiratory failure, such as the A-aDO2, a/A ratio, and PaO2/FiO2, in conjunction with clinical signs and conventional blood gas parameters, could predict respiratory impairment and need for mechanical ventilation in late-preterm neonates with respiratory distress at birth. Among these indices, the A-aDO2 at levels of >200 mmHg proved to be more effective compared to the other parameters, and had superior performance characteristics for predicting respiratory failure. Further studies are needed to investigate the clinical implications of these findings in terms of improving short- and long-term respiratory outcomes in late-preterm neonates with respiratory distress at birth.