This retrospective cohort study compared the short- and mid-term neonatal morbidity and mortality of a new LR strategy combining individualized increase in PEEP and SI. This LR policy is retrospectively associated with a decrease in i) endotracheal intubation in the DR, ii) rate and duration of mechanical ventilation in the first 72 hours, iii) surfactant use and iv) postnatal corticosteroids consumption. These results were more pronounced in infants < 29 weeks GA. Moreover, this strategy was not associated with any significant increase in neonatal morbidity or mortality.
Our significant decrease in endotracheal intubation rate in the DR is in accordance with other retrospective studies [20,21] and 3 RCTs [22–24] on SI. However, in our study as in all retrospective studies in the DR, it is difficult to ascertain whether this decrease was related to our individualized LR maneuvers or to a longer time allowed to the infants to stabilize without intubation. Nevertheless, in favor of our LR maneuvers we observed a decrease intubation rate in the DR with all physicians, not only the “early intubators”. Nonetheless, only an RCT could add any definitive evidence on this subject. Laryngeal closure is described as the main cause of airway obstruction [25] impeding NIV in the DR in about 26% of VLGAI [26]. In our 2016 cohort, apneic airway obstructions were also observed but only 18% of our VLGAIs were intubated for this reason in the DR, that is, on average in our medical team, twice a year per physician.
Although the total MV duration did not reach a statistically significant decrease, pulmonary morbidity was reduced, as witnessed by a lessened surfactant and postnatal corticosteroids consumption. However, early-rescue HFO, the INSURE procedure, an increased caffeine dosing in the case of moderate apnea, and sedation-analgesia as needed were practiced more often during 2016 and, thus, might have contributed to some pulmonary protection.
A more-than-halved BPD rate was observed in both subgroups but without reaching significance except for a multivariate analysis in the whole population of VLGAIs. We propose two explaining hypotheses. First, our relatively low prevalence of BDP makes a significant decrease more difficult to achieve. Second, the multifactorial origin of BPD [27] might make any effect of our very early intervention difficult to show without a large multicentric RCT.
Our data showed an increased trend in mortality and high grade IVH. The large multicentric “Sustained Aeration of Infants Lungs” (SAIL) trial comparing SI with conventional ventilation in the DR was recently cancelled (after 426 infants analyzed) because of a significant excess of deaths at less than 48 hours of age in the intervention arm without any reduction in BPD rates in preterm infants < 27 weeks GA [28]. Our speculative explanation of this harmful finding could be that the most immature infants may require an individualized and gentler support than a 15 s SI from the outset, explaining the initial persistent bradycardia observed in the SI group of the SAIL trial, and reflecting potential lung overdistention. In fact, our available data in preterm infants < 27 weeks GA show that only 19% required an SI = 15 s with a PEEP = 15 cmH2O. In our case, the mortality increase was related to an outbreak of first week late-onset septic shock during 2016 (i.e., 6 septic shock cases, including 2 complicated by high grade IVH) instead of 2 during the first period while the overall rate of culture-proven sepsis remained steady (8 then 9 cases per year). We speculate that this mortality increase, although nonsignificant, might be related to some delay in sepsis management among medical and nurse teams unaccustomed to early NIV.
Our results are in accordance with all other retrospective studies [20,29,21,30] assessing SI. Only one [20] showed a decrease in BPD and severe IVH, but these results could be related to other interventions in the DR such as an individualized approach to intubation, more CPAP use, and novel thermoregulation interventions. Four RCTs showed no benefits at all [31,32,22,33]. However, according to the meta-analysis of Bruschettini et al. [10], a decreased duration of MV was found (mean difference: –5.37 days; 95% CI: –6.31 to –4.43), but without differences in the rate of BPD nor combined mortality/BPD. Whether our positive results could be explained only by the adjunction of the possibly more protective dynamic PEEP LR manoeuver [34] should be addressed in a specific RCT.
Two studies using a prophylactic approach with a fixed duration of SI [33,35] have shown a nonsignificant increase in the rate of pneumothorax from 1 to 6% (OR = 4.57; 95% CI: 0.97–21.5; p = 0.06) [33] and from 0 to 3% (p = 0.08) [35]. This side-effect was not found in our study, despite the application of repeated SIs associated with much higher PEEPs. To explain this discrepancy, we postulate that our individualized approach might play a fundamental role. The rate of treated PDA was slightly but significantly increased in the meta-analysis of Schmölzer et al (RR = 1.27 (1.05–1.54)) [36], but our data do not support this observation.
A nonsignificant 20% increase in the rate of cesarean sections was observed during 2016 (p = 0.10). Paradoxically, our results show immediate benefits of the LR strategy, despite the absence of improved clearance of lung fluid that occurs during vaginal delivery [37]. Until recently, one of the described mechanisms of airway liquid clearance at birth was Na+ uptake across the airway epithelium. However, this cellular mechanism develops only in late gestation [38], and is too slow to clear the volume of liquid to be expelled within seconds to minutes after birth from airways. Therefore, the airway liquid clearance after birth is thought to result from an increase in the transepithelial pressure gradient, occurring during inspiration, which is incompletely effective in premature neonates [7]. Thus, our results might validate that personalized LR maneuvers helped to clear the fluid-filled airways and initiate gas exchange.
This study shares the limitations of monocentric retrospective cohort studies. Thus, the steady high rate of IUGR observed in our population (~24%) could account for a greater rate of morbidities. Given the changes that occurred in our neonatal management, confounding factors were introduced such as lowering the FiO2 threshold for surfactant administration, volume guarantee ventilation, early-rescue HFO, permissive hypercapnia, individualized doubling of the caffeine maintenance dose, and individualized sedation-analgesia. Moreover, the multiple secondary statistical analyses made in this study limit the value of significant results. Therefore, the beneficial pulmonary outcomes must be interpreted cautiously except probably for the intubation in the DR which could not be impacted by the subsequent management changes.
In conclusion, this retrospective study, shows the feasibility of an individualized LR strategy based on a stepwise increase in PEEP and SI with potentially beneficial short-term neonatal outcomes, especially in the ELGAI. A large RCT is needed to confirm these results.