Marchese et al. studied the performances of five ventilators designed to ventilate patients of all ages and sizes from neonates to adults [1]. They concluded that all ventilators were able to equal the performance of a specific neonatal ventilator (Babylog 8000 plus). Although we agree with those conclusions, we would like to comment on two points raised by this manuscript: trigger characteristics and the reproducibility of the tests.

Trigger delays of 100 ms are reported. This is quite long for neonates for whom inspiratory time is frequently set at 300 ms and may be responsible for asynchrony. We performed a similar test on three respirators with a resistance of 50 cmH2O/L/s, compliance of 2 mL/cmH2O, and a P 0.1 of −2 cmH2O on our bench test composed of a 5601i Adult/Infant PneuView (Michigan Instruments Inc.) test lung with pressures and flow rates independently recorded at the T-piece with a Biopac MP150 system connected to a computer. The pressure to trigger was 0.08–0.14 cmH2O and the time to trigger ranged between 45 and 60 ms (Fig. 1). Our results are similar to other reports [2]. The discrepancy in the trigger delay may be due to the definition of breathing effort. To standardize breathing effort, negative pressure produced in 100 ms is usually used (P 0.1) [2]. In adults, a P 0.1 of −2 and −4 cmH2O correspond to normal and strong effort, respectively [3]. For children, Harikumar et al. measured a maximal P 0.1 of −2.4 (0.9–6.1) cmH2O [4]. A breathing effort of −2 cmH2O probably corresponds to a strong effort for neonates.

Fig. 1
figure 1

Pressure to trigger and time to trigger of three bench-tested ventilators. Parameters: resistance of 50 cmH2O/L/s, compliance of 2 mL/cmH2O, and a P 0.1 of −2 cmH2O

Full size image

The advantages of evaluating the performances of ventilators on a bench test are both (1) to reproduce clinical conditions that we may have difficulties in observing and analyzing in clinical practice and (2) to strictly control those conditions so as to obtain data reproducibility. Marchese et al. reported standard deviations of around 50% for the measured values in several tests. Such differences are uncommon when working on a bench test as we theoretically control all parameters (compliance, resistance, ventilator settings, breathing effort). We do not have much experience with the IngMar ASL5000 lung simulator and it may be that this simulator has some unknown limits, explaining such a lack of reproducibility.