Precision of continuous neonatal ventilator respiratory mechanics is improved with selected optimal respiratory cycles

Given their high apparent variability, bedside continuous respiratory mechanics (RM) parameters [excepting tidal volume (V _T)] remain infrequently used for adjustment of neonatal ventilatory settings. RM parameters provided by ventilator (VRC) from ten recordings of newborns [10 min in synchronised intermittent mandatory ventilation and 10 min in assist/control (A/C)] were compared to those computed from visually selected assisted leak-free optimal respiratory cycles (SRC). Mean values, variability and ability to distinguish patients were compared between VRC and SRC. Dynamic resistances were more correlated (r ² = 0.95) than compliances (r ² = 0.42). V _Ts were correlated only in A/C (r ² = 0.78). C20/C was significantly higher in VRC (1.81 ± 0.67) than in SRC (1.23 ± 0.36) and frequently out of neonatal reference range. In A/C ventilation, V _T was higher in VRC (5.6 ± 1.8 ml/kg) than in SRC (4.8 ± 1.0 ml/kg) (p < 0.05). Displayed V _Ts do not reflect those found in optimal assisted breaths and therefore have incomplete value in assessing adequacy of ventilator settings. The variability of RM parameters provided by the ventilator is large, and coefficients of variation were significantly lower with optimal respiratory cycles (for resistance, compliance, V _T and C20/C; 27%, 26%, 18%, 24% in SRC and 36%, 35%, 40% and 33% in VRC). Selecting optimal cycles yields RM with two to three times higher discriminating power between patients. Conclusion: Current ventilator’s RM parameters have limited clinical use. Using optimal breaths to calculate RM parameters improves precision and discriminating power. For integration to ventilatory care, automation of this selection must be implemented first.