Noninvasive Ventilation with ICU Ventilators: A Note of Caution

Abstract & Commentary

By David J. Pierson, MD, Editor, Professor, Pulmonary and Critical Care Medicine, Harborview Medical Center, University of Washington, Seattle, is Editor for Critical Care Alert.

Synopsis: When noninvasive ventilation is administered using ICU ventilators that offer this feature, air leaks in the circuit cause triggering delay, increased patient work, and delayed cycling. Activating the special noninvasive ventilation feature improves these problems on some of the ventilators tested (but variably so), and actually made them worse on others.

Source: Vignaux L, et al. Intensive Care Med. 2007 (e-pub prior to print): DOI 10.1007/s00134-007-0713-0.

Vignaux and colleagues at the University Hospital in Geneva performed a laboratory study of the performance of ICU ventilators featuring a noninvasive ventilation (NIV) mode, with and without an induced air leak. As a model representing a patient being ventilated with NIV, they used a PVC head equipped with upper airways and a trachea, the latter connected to a 2-chamber lung model. A driving ventilator ventilated one chamber using airway pressure release mode, while the other chamber mimicked spontaneous breathing and was connected to the model's trachea. The investigators applied NIV via a standard oronasal mask, and produced a controlled air leak using a 3-way stopcock placed between the NIV ventilator circuit's Y-piece and the mask.

Eight ICU ventilators currently available in Europe that feature NIV mode and are marketed for this use were tested under similar conditions. Triggering and pressurization were tested under conditions of normal mechanics in the test lung, and cycling was tested under conditions of normal, obstructive, and restrictive mechanics. The tests were performed using pressure support mode with no leak, pressure support plus leak, and pressure support mode plus leak with the NIV feature activated. With the experimental setup used, the leaked volume under the latter two conditions was essentially 100% of the baseline tidal volume and "patient" minute ventilation.

With most of the ventilators tested, leaks led to an increase in trigger delay and workload, a decrease in pressurization, and delayed cycling. The NIV mode partially or totally corrected these problems—with most but not all of the ventilators—with marked differences between the different machines. On some of the ventilators the NIV mode actually made the leak-induced dysfunction worse.


Air leaks around the mask are a major problem with NIV. They are unpleasant for patients and can lead to poor acceptance of this life-saving therapy, and they can interfere with its effectiveness even when tolerated by the patient. Ventilators such as the Respironics BiPAP and others of similar design that were developed for home care and other non-ICU use compensate well for leaks, as several studies have demonstrated. However, these and other studies have also shown that standard ICU ventilators, which were designed for closed-circuit use with intubated patients who have no or minimal leaks, are poorly suited for NIV because of their lack of leak compensation and can be dangerous in this application, especially in pressure support mode.

Thus, the availability of better leak compensation via the new NIV mode features on an increasing number of ICU ventilators (most of these still not available in the US) would seem to be a real advance. But not so fast: this study shows that—at least with the NIV simulation model used and under the conditions applied—these ventilators vary a great deal in their ability to deliver on what they promise.

One additional point made by Vignaux et al in their discussion is worthy of repeating. The use of new modes and other ventilator features tends to expand beyond their intended specific applications into clinical circumstances in which they have not been investigated and may be dangerous. Using the new NIV mode in intubated patients could be hazardous, in two respects. First, if an intubated patient were switched to NIV mode from another mode in which the ventilator had performed its automated self-tests of circuit resistance and compliance in order to determine correction factors for its flow and pressure sensors, its function might be altered. Second, because NIV modes are designed to deal with large system leaks and to compensate for them, the ventilator may not alarm in the presence of such leaks as it does in other modes that assume a closed circuit with minimal leak; in such circumstances an intubated patient could develop a clinically important air leak that would not trigger the intended alarm.