An observational, prospective, multicenter study on rescue high-frequency oscillatory ventilation in neonates failing with conventional ventilation

Background To achieve gas exchange goals and mitigate lung injury, infants who fail with conventional ventilation (CV) are generally switched to high-frequency oscillatory ventilation (HFOV). Although preferred in many neonatal intensive care units (NICUs), research on this type of rescue HFOV has not been reported recently. Methods An online registry database for a multicenter, prospective study was set to evaluate factors affecting the response of newborn infants to rescue HFOV treatment. The study population consisted of 372 infants with CV failure after at least 4 hours of treatment in 23 participating NICUs. Patients were grouped according to their final outcome as survived (Group S) or as died or received extracorporeal membrane oxygenation (ECMO) (Group D/E). Patients' demographic characteristics and underlying diseases in addition to their ventilator settings, arterial blood gas (ABG) analysis results at 0, 1, 4, and 24 hours, type of device, ventilation duration, and complications were compared between groups. Results HFOV as rescue treatment was successful in 58.1% of patients. Demographic and treatment parameters were not different between groups, except that infants in Group D/E had lower birthweight (BW) (1655 +/- 1091 vs. 1858 +/- 1027 g, p = 0.006), a higher initial FiO2 setting (83% vs. 72%, p < 0.001), and a higher rate of nitric oxide exposure (21.8% vs. 11.1%, p = 0.004) in comparison to infants who survived (Group S). The initial cut-offs for a successful response on ABG were defined as pH > 7.065 (OR: 19.74, 95% CI 4.83-80.6, p < 0.001), HCO3 > 16.35 mmol/L (OR: 1.06, 95% CI 1.01-1.1, p = 0.006), and lactate level < 3.75 mmol/L (OR: 1.09% 95 CI 1.01-1.16, p = 0.006). Rescue HFOV duration was associated with retinopathy of prematurity (p = 0.005) and moderate or severe chronic lung disease (p < 0.001), but not with patent ductus arteriosus or intraventricular hemorrhage, in survivors (p > 0.05). Conclusion Rescue HFOV as defined for this population was successful in more than half of the patients with CV failure. Although the response was not associated with gestational age, underlying disease, device used, or initial MV settings, it seemed to be more effective in patients with higher BW and those not requiring nitric oxide. Initial pH, HCO3, and lactate levels on ABG may be used as predictors of a response to rescue HFOV.