Article Text

Download PDFPDF

Predictors of early nasal CPAP failure and effects of various intubation criteria on the rate of mechanical ventilation in preterm infants of <29 weeks gestational age
  1. H Fuchs,
  2. W Lindner,
  3. A Leiprecht,
  4. M R Mendler,
  5. H D Hummler
  1. Division of Neonatology and Pediatric Critical Care, University Children's Hospital, Ulm, Germany
  1. Correspondence to Dr H Fuchs, Division of Neonatology and Pediatric Critical Care, University Children's Hospital, Eythstrasse 24, 89075 Ulm, Germany; hans.fuchs{at}


Introduction Delivery room management using early nasal continuous positive airway pressure (nCPAP) may delay surfactant therapy.

Objective To identify factors associated with early nCPAP failure and effects of various intubation criteria on rate and time of intubation.

Design Retrospective analysis of the first 48 h in infants of 23–28 weeks gestational age (GA) treated with sustained inflations followed by early nCPAP.

Results Of 225 infants (GA 26.2±1.6 weeks) 140 (62%) could be stabilised with nCPAP in the delivery room, of whom 68 (49%; GA 26.9±1.5 weeks) succeeded on nCPAP with favourable outcome and 72 infants (51%; GA 26.3±1.4 weeks) failed nCPAP within 48 h at a median (IQR) age of 5.6 (3.3–19.3) h. History or initial blood gases were poor predictors of subsequent nCPAP failure. Intubation at fraction of inspired oxygen (FiO2)≥0.35 versus 0.4 versus 0.45 instead of ≥0.6 would have resulted in unnecessary intubations of 16% versus 9% versus 6% of infants with nCPAP success but decreased the age at intubation of infants with nCPAP failure to 3.1 (2.2–5.2) versus 3.8 (2.5–8.7) versus 4.4 (2.7–10.9) h.

Conclusions Medical history or initial blood gas values are poor predictors of subsequent nCPAP failure. A threshold FiO2 of ≥0.35–0.45 compared to ≥0.6 for intubation would shorten the time to surfactant delivery without a relevant increase in intubation rate. An individualised approach with a trial of early nCPAP and prompt intubation and surfactant treatment at low thresholds may be the best approach in very low birthweight infants.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Mechanical ventilation or early nasal continuous positive airway pressure (nCPAP) may be used for early respiratory support in preterm infants.1,,4 Similar rates of bronchopulmonary dysplasia (BPD) or death were demonstrated with both interventions.3 5 6 Although nCPAP may help to avoid mechanical ventilation, some infants will need subsequent intubation and surfactant therapy.3 5 7

Animal experiments suggest that surfactant applied before or soon after delivery provides optimal lung protection,8,,10 which can be realised clinically by prophylactic surfactant therapy.11,,14 However, this strategy may result in unnecessary treatment15 and does not translate into superior outcomes compared to nCPAP, as demonstrated in recent studies.5 6 16

What is already known on this topic

  • Approximately 50% of infants of <29 weeks gestation managed with early nasal continuous positive airway pressure (nCPAP) will need intubation and mechanical ventilation.

  • There is uncertainty which thresholds should be used when deciding to intubate preterm infants with respiratory distress syndrome and early nCPAP failure.

What this study adds

  • There are no clinically adequate predictors of early nCPAP failure at the time of admission to the neonatal intensive care unit.

  • Selective intubation at low thresholds (fraction of inspired oxygen: 0.35-0.45) shortens time to surfactant therapy without relevant increase in intubation rate compared to higher thresholds.

If nCPAP is used, early identification of infants with subsequent nCPAP failure might be crucial to avoid negative effects from delayed surfactant application. However, no markers correctly predicting subsequent nCPAP failure have yet been identified.7 Furthermore, little information is available regarding the effects of using different criteria for intubation on the rate of infants needing mechanical ventilation.

Until recently, we used a high threshold for intubation in order to avoid mechanical ventilation in our very low birthweight infants. We retrospectively reviewed the first 48 h of life of all inborn infants of <29 weeks gestational age (GA) to identify early predictors of nCPAP failure in this cohort of preterm infants with early nCPAP as a first line therapy. In addition, we calculated the potential effects of various intubation criteria on the rate and age at intubation.

Patients and methods

This retrospective study was approved by the local ethics committee. Inborn infants of 23+0/7 to 28+6/7 weeks GA born alive from January 2005 to June 2008 at the University Hospital of Ulm, Germany were included. Four infants were excluded due to unavailable charts (n=2) or congenital abnormalities requiring intubation (n=2; neck giant cystic hygroma and diaphragmatic hernia).

Neonatal resuscitation (figure 1) was performed underneath an overhead warmer. Infants were wrapped in a plastic bag to avoid evaporative heat loss. Oxygen saturation (SpO2) was measured by pulse oximetry (Radical; Masimo, Irvine, California, USA) at the right hand. After excess amniotic fluid was suctioned from the mouth, a nasopharyngeal tube (2.5 mm internal diameter) was inserted at 3–4 cm and nCPAP (5 cm H2O; fraction of inspired oxygen (FiO2) 1.0) was applied using a F120 respirator (Stephan, Gackenbach, Germany). If the heart rate remained at <100 beats/min (bpm) or SpO2 was <70% (>15 s) up to three sustained lung inflations were applied at pressures of 20, 25 and 30 cm H2O for 15 s each as described before.17 Thereafter, nCPAP or nasal intermittent mandatory ventilation (nIMV; rate 60/min, inflation pressure 14–25/5 cm H2O) was continued with a FiO2 adjusted to maintain SpO2 at 80–92%. Criteria for intubation during resuscitation were: FiO2≥0.6, persistent heart rate <100 bpm and GA of <25 weeks for prophylactic surfactant therapy. Infants of <25 weeks GA remained on nCPAP if they appeared to be very stable. Chest compressions were applied if bradycardia of <60 bpm persisted after intubation and mechanical ventilation. Umbilical catheters were placed in most infants in the delivery room. Serial blood gases were used to guide further interventions. Indices of gas exchange partial pressure of arterial oxygen (PaO2)/FiO2, alveolar–arterial oxygen partial pressure difference (AaDO2), arterial/alveolar oxygen partial pressure ratio (a/APO2) were calculated from the first arterial or capillary blood gas after admission to the neonatal intensive care unit (NICU).

Figure 1

Delivery room resuscitation algorithm at the Children's Hospital, University of Ulm during the study period. Up to three sustained inflations are used as first line respiratory support if CPAP is not sufficient to stabilise the infant. During the study period initial FiO2 was 1.0. FiO2, fraction of inspired oxygen; GA, gestational age; HR, heart rate; nCPAP, nasal continuous positive airway pressure; nIMV, nasal intermittent mandatory ventilation; SpO2, oxygen saturation.

During the first days of life, nCPAP at 3–6 cm H2O was delivered by a nasopharyngeal tube or binasal prongs. Intubation criteria in the NICU were: FiO2≥0.6 for 1 h, FiO2≥0.5 with severe dyspnoea, partial pressure of carbon dioxide (PCO2)≥65 mm Hg or >6 apnoeas/h. Target SpO2 was 80–92%. Intubation for apnoeas was performed only if methylxanthines or nIMV were unsuccessful. Intubation before 48 h of age was considered a nCPAP failure. BPD was defined as need for supplemental oxygen at 36 weeks corrected age to maintain SpO2≥90%. The impact on the intubation rate and age at intubation applying different intubation criteria (FiO2, PCO2) during the first 48 h of life were analysed.

Statistical analysis

Means (SD) or medians (IQR) are given. Data were analysed using χ2 tests, Mann–Whitney U tests or using analysis of variance (ANOVA) or ANOVA on ranks where appropriate. Significance was set at p<0.05.


Of 225 infants, 85 (38%) had to be intubated in the delivery room for FiO2≥0.6, persistent bradycardia or for prophylactic surfactant therapy in infants of <25 weeks GA. Infants intubated in the delivery room were more immature and had a lower birth weight.

Of the 140 infants maintained on nCPAP within the subsequent 48 h of life, 60 (43%) had to be intubated for FiO2≥0.6, three (2%) for PCO2≥65 mm Hg and nine (6%) for occurrence of apnoeas unresponsive to methylxanthines or nIMV. Sixty-six of 140 (47%) infants were treated with surfactant. nCPAP response rate during the first 48 h of life was 68/140 (49%), corresponding to 30% of all study infants. Infant characteristics and additional outcome data are given in table 1.

Table 1

Clinical data of study patients

The rate of successful stabilisation with nCPAP in the delivery room was lower in more immature infants (figure 2).

Figure 2

Response to early nasal continuous positive airway pressure (nCPAP). The rate of nCPAP success in relation to gestational age (GA) is shown. Open symbols indicate the response rate in all infants including infants intubated in the delivery room and closed symbols indicate the response rate in infants who were stabilised with early nCPAP in the delivery room. Linear regression analysis is shown.

To identify early predictors of nCPAP failure, parameters of gas exchange were calculated from the first blood gas at admission to the NICU. The predictive values of PaO2/FiO2, AaDO2 and a/APO2 calculated from the blood gas at NICU admission were not superior to FiO2. Elevated PaCO2 was a poor predictor of subsequent nCPAP failure (table 2).

Table 2

Prediction of nCPAP failure at NICU admission

Median age at nCPAP failure was 5.6 (3.3–19.3) h. Effects of different intubation thresholds (FiO2, PCO2) on the age at intubation and on the rate of intubation are given in table 3 and in the online supplementary figure.

Table 3

Effects of different intubation criteria on age at intubation and rate of unnecessary intubations


Infants with early nCPAP success had excellent outcomes with very low rates of intraventricular haemorrhage and BPD, consistent with previous reports.7 18 Therefore, this group will most likely not benefit from intubation to deliver prophylactic surfactant. In contrast, delayed surfactant therapy might cause harm in infants who fail early nCPAP.19,,22 Therefore, early identification of the group with subsequent nCPAP failure would help to overcome the potential negative effects of prolonged and futile nCPAP treatment through the use of selective early intubation.

Initial stabilisation with nCPAP in the delivery room is strongly associated with gestational age.7 18 More immature infants were more likely to need intubation at delivery. However, some very immature infants were successfully stabilised with nCPAP without subsequent failure. We speculate that lung recruitment prior to nCPAP may have contributed to this success.1 23

The rate of maternal risk factors for severe respiratory distress syndrome (RDS), such as premature rupture of membranes or lack of prenatal steroid treatment, was similar in the responder and failure groups. Similar to the indices of gas exchange at admission to the NICU, they do not predict subsequent nCPAP response well enough to be used as predictors in clinical practice.7 Other predictors of RDS severity, such as lamellar body counts in gastric aspirates, are currently under investigation.24 However, so far most clinicians use arbitrary thresholds of FiO2 or PCO2 to select infants for intubation.3 5 19

In our cohort of infants intubated late with high thresholds, we calculated the possible effects of lower intubation thresholds on intubation rate and timing of intubation. The rate of unnecessary intubation increased only marginally if the threshold was reduced from FiO2≥0.6 to FiO2≥0.35–0.45. Most infants with early nCPAP success require no or very little supplemental oxygen. Therefore, a high threshold such as FiO2≥0.6 seems to postpone the time of intubation with little effect on the subsequent rate of intubation. However, increased FiO2 requirement may be an indicator of more severe RDS and earlier surfactant treatment may prevent further lung injury and complications, such as air leaks.

However, it remains unclear if reducing the time from birth to intubation from 5.6 h (FiO2≥0.6) to, for example, 3.8 h (FiO2≥0.4), results in a clinically relevant benefit. A meta-analysis of four trials assessing early versus late surfactant treatment in mechanically ventilated preterm infants suggests a reduction in key clinical outcomes including air leaks, BPD and mortality with the use of early surfactant.25 It suggests that in mechanically ventilated preterm infants surfactant should be given early.25

In contrast, earlier intubation of more mature spontaneously breathing infants at a threshold FiO2>0.4 compared with expectant management was not associated with more favourable short term outcomes.26 However, timing of surfactant delivery might be more important in less mature infants.

Verder et al21 compared selective early versus late surfactant therapy followed by rapid extubation in infants of <30 weeks GA. Less subsequent need for mechanical ventilation was found, but the trial was stopped early and data on long term outcomes were inconclusive. Rojas et al19 compared early surfactant therapy (age <1 h) with selective intubation at high thresholds in infants of 27–31 weeks GA with mild RDS. Less need for subsequent mechanical ventilation and fewer air leaks were reported in the group intubated early.

More recently, delivery room management of extremely preterm infants was the subject of several randomised trials. The COIN trial compared nCPAP with mechanical ventilation in infants able to breathe at 5 min of age.3 The CURPAP trial compared nCPAP with prophylactic intubation, surfactant therapy and rapid extubation,16 while the SUPPORT trial compared nCPAP with prophylactic surfactant therapy followed by mechanical ventilation.5 All trials suggest that nCPAP as first line respiratory support results in no differences in relevant short term outcomes, but may prevent mechanical ventilation and eventually preserve pulmonary function.27 Preliminary results of a Vermont Oxford network trial support these observations.6

The association of air leaks and early nCPAP reported in the COIN trial3 and the trial by Rojas et al,19 but not in the CURPAP16 or SUPPORT trials,5 might be explained by different thresholds for selective intubation in the nCPAP groups. In the COIN trial3 and Rojas trial,19 high thresholds for nCPAP failure were applied (FiO2>0.6, PCO2>60 mm Hg and apnoeas vs FiO2>0.75, PCO2>65 mm Hg and apnoeas), while in the CURPAP16 trial infants in the nCPAP group were intubated for FiO2>0.4, PCO2>65 mm Hg, pH<7.2 or apnoeas and in the SUPPORT trial5 for FiO2>0.5, PCO2>65 mm Hg or haemodynamic instability. These lower thresholds reduced the median age at intubation and surfactant treatment (4.0 h in the CURPAP trial16 vs 6.6 h in the COIN trial3), eventually resulting in more favourable outcomes. These data also support the findings of a stratified meta-analysis suggesting that a low FiO2 threshold (≤0.45) for intubation and surfactant treatment results in a lower incidence of air leaks.28

In conclusion, there are no clinically adequate predictors of early nasal CPAP failure at time of admission to the NICU. A threshold FiO2 of ≥0.35–0.45 as compared to ≥0.6 for selective intubation would shorten the time until surfactant delivery in infants with nCPAP failure without a relevant increase in intubation rate. An individualised approach with initial early nCPAP but with prompt intubation and surfactant treatment at a low FiO2 threshold may be the best approach in very low birthweight infants. However, only a prospective randomised trial can clarify if such an approach results in clinically better outcomes.


The authors would like to thank Mike-Andrew Westhoff for his review of the manuscript.



  • Competing interests None.

  • Ethics approval Study approval was obtained from the local ethics committee of the University of Ulm (no. 144/09).

  • Provenance and peer review Not commissioned; externally peer reviewed.