Objectives Neonatal ventilator softwares provide information on respiratory mechanics (RM). Integration of those values with clinical variables could improve ventilation management. We investigated accuracy and reproducibility of those variables in Assist Control (AC) and Synchronised Intermittent Mandatory Ventilation (SIMV) modes.

Methods Data (Ventilation pressures, Tidal Volume (VTe), Minute Ventilation (MV), Compliance (C), Resistance (R)) from a Babylog 8000 ventilator were collected during 9 minutes in 15 infants ventilated in AC. Variability of individual values was compared to that of the 1 and 3 minute(s) mean periods. For each period, C, R and VTe were calculated from one demonstrative respiratory loop reconstructed from continuous recordings. Nine newborns ventilated in SIMV were recorded in SIMV and AC.

Results In AC, VTe, MV, C, and R variability of individual values represents 12, 12, 13 and 20%, respectively, and improves to 5.5, 7.6, 7, and 10% using one minute average and 3.5, 5.6, 4.7, and 5.8% with three minutes average. Variability in SIMV has a similar pattern. Calculated values from loops are within 20% of ventilator values. SIMV values for mean pressures and VTe are different from AC. Difference is 24% for R and 10% for C (not significant).

Conclusions In both modes continuous ventilator RM data are difficult to integrate into clinical practice. Averaging those parameters allows for more reproducible values that could be used for trend monitoring. Non-assisted breaths interfere with calculation of RM values. RM assessment of patients on SIMV ventilation should be done with a brief switch to AC.