Survival of Newborn Infants With Severe Respiratory Failure Before and After Establishing an Extracorporeal Membrane Oxygenation Program

Javier Kattan, MD; Alvaro González, MD; Pedro Becker, MD; Miriam Faunes, RN; Alberto Estay, MD; Paulina Toso, MD; Soledad Urzúa, MD; Andrés Castillo, MD; Jorge Fabres, MD, MSPH


Pediatr Crit Care Med. 2013;14(9):876-883. 

In This Article


Patient Characteristics

Two hundred fifty-nine infants with severe HRF were admitted to our NICU during the study periods; 100 in the pre-ECMO period and 159 after ECMO was instituted. Patient data, including BW, gender, Apgar scores, and GA and postnatal age, were similar between the two periods, except that during the ECMO period, there was a significantly greater proportion of outborn infants, a higher pretreatment OI (OI before iNO/HFOV), and a higher proportion of infants admitted with CDH (Table 1). The "other" diagnosis included other causes of HRF associated to omphaloceles, gastroschisis, hydrops, congenital cystic adenomatoid malformation, lymphangiectasia, alveolar capillary dysplasia, and idiopathic pulmonary fibrosis (Table 1).

In the ECMO period, 29.5% of the newborns (38 of 129) were referred after 48 hours of age to our NICU. The same percentage of CDH cases were transferred before and after 48 hours of age, 30% and 32%, respectively. From the infants transferred after 48 hours of age, a higher proportion had pneumothoraces (42% vs 29%), and a higher percentage died or required ECMO (63.2% vs 43.9%), with no difference in pretreatment OI mean (OI > 48 hr, 46 vs OI < 48 hr, 52).

Most infants in both periods received iNO, and a slightly higher proportion received HFOV or the combination of both in the ECMO period. ECMO was used by 33% of patients in the second period (Table 1).

Patient Survival and ECMO

During the pre-ECMO period, 72% of all infants (72 of 100) were rescued and survived with iNO, HFOV, or a combination of both; 28 patients expired, 12 of whom had CDH (Fig. 1).

Figure 1.

Comparison of survival per diagnosis between pre-extracorporeal membrane oxygenation (ECMO) and ECMO periods. Numbers below the bars = n/total. MAS = meconium aspiration syndrome, RDS = respiratory distress syndrome, Pneum = pneumonia, PPHN = persistent pulmonary hypertension in newborns, CDH-LH = congenital diaphragmatic hernia and other causes of lung hypoplasia, Other = other causes of hypoxic respiratory failure, Non-CDH HRF = includes infants with MAS, RDS-Pneumonia, and PPHN and Other, without other causes of lung hypoplasia.

In contrast, during the ECMO period, 62% of patients (98 of 159) with HRF were rescued with iNO, HFOV, or the combination of HFOV + iNO, whereas 38% (61) did not improve; 52 of these nonresponders were placed on ECMO; the other 9 expired, all of whom were not eligible for ECMO. Of these nine, four had signs of severe hypoxic-ischemic encephalopathy and/or a severe intracranial hemorrhage, three had a cardiac arrest, not responding to resuscitation prior to cannulation, and two had other associated anomalies that precluded placing them on ECMO (one had renal dysplasia, and the other Fryns syndrome phenotype).

Overall survival of infants with severe HRF significantly increased from 72% in the pre-ECMO period to 89% during the ECMO period (p < 0.01) as shown in Figure 1. This was especially important among infants with CDH and other causes of lung hypoplasia (CDH-LH), among which survival increased from 32% to 81% (p < 0.01). Survival of infants with non-CDH HRF also had a significant improvement from 83% to 94% (p < 0.04). Figure 1 shows the survival rate depicted by diagnoses. Table 2 shows that survival to discharge or transfer among infants placed on ECMO reached 85% (44 of 52).

A survival comparison between inborn and outborn patients in the pre-ECMO and ECMO period demonstrated the same increasing survival tendency between both periods, but without a significant difference in survival for inborn infants (Table 3).

During the ECMO period, 52 patients were treated with ECMO; 30 (58%) of them were male; BW ranged from 2,460 to 4,140 g. The most common diagnoses for ECMO were CDH (30), RDS-Pneumonia (13), MAS (7), and PPHN (2). Of the 52 patients who were placed on ECMO, seven were treated with venovenous ECMO, 45 with venoarterial ECMO, and one patient was switched from venovenous to venoarterial ECMO. Forty-four out of the 52 patients (85%) survived with ECMO and were discharged home or transferred to their referring hospital (Table 2). Among the eight ECMO-treated patients who died, seven were transferred from other institutions. Two of the eight had severe infections and multiple organ failure, one was associated with Bordetella pertussis pneumonia, and the other with Listeria monocytogenes sepsis. Both had evidence of severe cerebral injury probably due to pre-ECMO hypoxic-ischemic insult. One patient had severe RDS/PPHN due to a surfactant protein B deficiency. The other five deaths were patients with severe CDH, four of them whom had multiple complications prior to ECMO, including recurrent pneumothoraces.

A univariate comparison for the ECMO period between groups exploring survival without ECMO versus death or ECMO as an outcome served to identify factors associated with adverse outcomes (Table 4). No differences were found in terms of GA, BW, gender, or Apgar scores. An OI greater than 40 at admission (before iNO/HFOV), the presence of a pneumothrorax, and the diagnosis of a CDH were significantly associated with death or ECMO. The percentage of outborn infants and the age at admission to the referral center were different, but not statistically significant in this primary analysis.

After adjusting for confounders using multiple logistic regression analyses, the severity of the pretreatment OI (odds ratio [OR] = 13.4; 95% CI, 5.2–38.6), a late admission/referral to the tertiary NICU after 48 hours of life (OR = 3.8; 95% CI, 1.3–7.9), the presence of a previous pneumothorax (OR = 3.2; 95% CI, 1.3–7.9), and the diagnosis of a CDH (OR = 6.4; 95% CI, 2.3–19.4) were all significantly associated with the need for ECMO or death as an outcome (Fig. 2).

Figure 2.

Multivariate analysis of factors associated with extracorporeal membrane oxygenation (ECMO) or death in the ECMO period. BW = birthweight, CDH = congenital diaphragmatic hernia, OI = oxygenation index.

Cost-effectiveness Analysis

An initial cost-effectiveness analysis showed that, in the ECMO period, the cost per survivor was USD 31,104 in infants with non-CDH HRF and USD 73,515 in infants with CDH. Interestingly, the cost per survivor was USD 65,060 for ECMO-treated infant and USD 32,867 for conventional management. This analysis accounted for time to hospital discharge or transfer only. The mean charge per infant for ECMO run was estimated as USD 27,225.