The footprint is the square feet of floor space occupied by the ventilator. The appearance is its relative size and visual complexity.⁸ This is important because of the increasing number of devices now at the neonatal and pediatric critical bedside. I have seen episodes of care where the wrong device was inadvertently unplugged at the bedside of a critical patient. Visually, the room was a complex unorganized array of devices, hoses, tubing, cords, stands, monitors etc. Obviously this kind of complexity can lead serious clinical mishaps. Also, storage of ventilators when not in use is an important consideration in hospitals where storage space is constrained.

The ease of movement of the ventilator will matter considerably to those charged with moving them around. This includes the likelihood of tipping over while moving. I once had an equipment technician who tipped over a brand new mechanical ventilator on the first day the device was uncrated. She was trying to move two ventilators at a time. If equipped with batteries, compressors, and cylinders, it is possible to transport patients around the hospital while attached to their ventilator (Jacques et al. 2005).

Potential buyers should test the actual battery performance (as opposed to believing the stated performance data from the manufacturer). We have learned that batteries rarely perform as well as manufacturers claim. Also remember that battery performance deteriorates as the batteries age. So periodic retesting of battery performance is also recommended. Battery capability might be very important in hospitals with older utilities infrastructure where unplanned loss of electricity is more likely.

Most modern ventilators can be equipped with a built-in compressor that allows the ventilator to operate without an external supply of compressed air. This can be important in care environments where there is a chance of an unplanned loss of compressed air to the ventilator, such as buildings with older utilities infrastructure or hospitals undergoing remodeling because of the chance of accidental cutting of existing gas lines during demolition. Adding these compressors will add cost and weight. My view is that this feature is well worth the additional cost and weight, particularly in pediatric and neonatal environments where the patients typically have less reserve to tolerate physiologic insults than adults.

This refers to the ability of the ventilator to connect with the hospitals’ electronic monitoring and/or medical records systems. Almost all critical care ventilators now sold are equipped with serial RS-232 output connection (and other types) that can then be interfaced with the hospital systems or laptop computers. However, interface protocols may have compatibility issues with existing hospital systems, and work must be done to ensure compatibility before purchase. Also, some ventilators now have the capability to download stored patient-level ventilator measurements and settings onto removable media or laptop computers. This feature can be very useful, especially in retrospective analysis of troubling episodes of care. Additionally, some ventilators can download pulmonary graphic waveforms. We have found this feature very useful, especially in analyzing questionable ventilator performance or patient-ventilator dyssynchrony.⁹

These may be either a pressure-controlled, or a volume-controlled, or a dual-control mode. Dual-control modes allow for a setting-targeted tidal volume combined with a decelerating flow wave form typically seen with pressure-controlled breaths. In this mode, the peak airway pressure is allowed to vary within certain limits to obtain the desired tidal volume. Opinions are divergent and in my view the evidence base is inconclusive about the relative merits of volume- versus pressure-limiting and decelerating versus constant flow modes, as well as dual-control modes of ventilation (Abubakar and Keszler 2005; Campbell and Davis 2002; Cheema and Ahluwalia 2001; Chiumello et al. 2002; Clark et al. 2000; Claure and Bancalari 2008; D’Angio et al. 2005; Dembinski et al. 2004; Herrera et al. 2002; Keszler and Abubakar 2004; Kocis et al. 2001; Lista et al. 2004; Marraro 2003; McCallion et al. 2005; Piotrowski et al. 1997; Roth et al. 2004; Yang and Huang 2007). The interested reader is directed to the excellent review of this complex topic in neonates by (Keszler 2009).

In general, the conventional wisdom about the benefits of volume-targeted decelerating flow modes is thought to include (1) improved distribution of ventilation, (2) improved patient comfort and thus possibly less sedation required, (3) decreased patient-ventilator dyssynchrony, and (4) lower airway pressures (for the same tidal volume). The value of lowering airway pressure in neonates has been misconstrued and misrepresented over the years. It is clear to me that the risk of mechanical injury of the diseased lung is most closely related to volumetric overdistention of the lung (Brunherotti et al. 2003; Dreyfuss et al. 1988). This can occur at various pressures depending on the patient’s respiratory system compliance. Thus, it is perplexing to me that there is a perceived benefit of lowering airway pressures while giving the same tidal volume. In neonates, volume-targeted decelerating flow mode is thought to be a better form of protective lung strategy because of the possibility of rapidly changing lung compliance and airway resistance in infants. But this would depend greatly on how nimble the ventilator is at adjusting the flow and pressure and the airway in response to changes in the patient’s compliance and resistance (more on this later).

Some ventilators measure tidal volume at the proximal airway, and some measure tidal volume at the gas outlet from the ventilator. See the section on tidal volume accuracy below to better understand some of the issues with the location of these measurements.

There is also a debate about the relative merits of using inhaled versus exhaled tidal volume for the feedback to ventilator to run the volume-targeted modes. The relative merits of inhaled versus exhaled tidal volume measurement may seem an arcane subject, but the debate becomes more important when considering the airway leaks associated with the use of dual-control mode ventilation of infants and children with uncuffed endotracheal tubes. Leaks can make it more difficult for the ventilator to measure and control the true tidal volume. Some ventilator brands use exhaled tidal volume measurements to inform their dual-control systems, and others use inhaled tidal volume. Consider that some ventilators can measure and display both inhaled and exhaled tidal volumes and use these measures to represent a leak percentage:
where

VT_i = inhaled tidal volume

VT_e = exhaled tidal volume

It is important not to forget that the ventilator displays are often customizable and that the tidal volume being displayed on the ventilator screen may not be the one that the ventilator is using for controlling the dual-controlled modes on the device. It may be that in the presence of a leak, some of inhaled tidal volume enters the trachea and immediately exits the trachea by leaking around the endotracheal tube during the inspiratory time and thus does not actually affect expansion of the lungs. In theory, using inhaled tidal volume measurements for feedback control helps to ensure that the actual tidal volume is no larger than that measured and controlled by the ventilator, but it is possible that the effective tidal volume might be considerably less, because of leakage around the endotracheal tube. When using exhaled tidal volume measurements, some of the tidal volume in the lung may leak around the endotracheal tube during the expiratory time, rendering this measurement lower than the actual tidal volume or the measured inhaled tidal volume.

In dual-control modes, the ventilator has to adjust the flow and the allowable pressure to meet the volume targets for each breath. Some brands do this by using a running average of the activities of the last few breaths to set allowable limits for the next breath. Others actually measure conditions during the current breath and adjust flow and pressure to attempt to meet the tidal volume target. One of the risks of dual-control modes is the phenomenon of overshoot. Consider the ventilators that use a running average to determine the limits of flow and pressure needed to give the targeted tidal volume. If the patient has been agitated and “fighting” the ventilator for several breaths (reducing lung compliance), some brands will increase the allowable pressure to try to give the targeted tidal volume in the next breath. But if the patient then suddenly relaxes, the pulmonary compliance can suddenly improve, and this increased pressure can then result in too large a tidal volume. This is theoretically less likely in ventilators that do adjustments of flow and pressure within a single breath. There has been very little published testing comparing these two approaches.

In most currently available fifth generation critical care ventilators, breaths may be triggered (or started) and cycled (or ended) by either time or flow or pressure changes measured in the respiratory system. Essentially all critical care ventilators now offer various combinations of these variables in dual-control and pressure support mode configurations. The best combination of triggering or cycling variables to use on infants and children on ventilators is still not completely understood although there is a growing body of research in this area (Greenough et al. 2008). As practice evolves the use of flow triggering and cycling is increasing. Some brands of ventilators have adjustable flow cycling thresholds, while others have a preset level. See the section on work of breathing and patient-ventilator dyssynchrony. Other triggering methods have been tried over the years including triggering off of changes in thoracic impedance, abdominal wall movement, and diaphragmatic innervations (Kondili et al. 2009). None of these has been widely adopted. Neurally adjusted ventilatory assist uses the measurement of diaphragmatic activity to trigger the ventilator. It shows some promise (Bengtsson and Edberg 2010; Breatnach et al. 2010; Spahija et al. 2010) but much research needs to be done before the benefits of this technology can be established. This requires the insertion of a specially equipped gastric tube orally or nasally. It seems premature to consider this feature a major decision driver in the selection of a neonatal or pediatric ventilator.

This relatively recent addition to modes available on some ventilators remains largely untested in neonates and pediatric patients. Fundamentally this mode resembles pressure-controlled inverse ratio ventilation but is different in that it allows for spontaneous breathing on top of relatively high continuous airway pressure during prolonged inspiratory phase of the respiratory cycle. Whether or not this is an essential feature of mechanical ventilators for infants and children is unknown. The relatively small amount of published literature about APRV is divergent with some reports suggesting a salutary effect on patients (Habashi 2005; Kamath et al. 2010; Krishnan and Morrison 2007) and others finding no benefit (González et al. 2010). My own view is that it requires considerable additional training in ventilator management for physicians, nurses, and respiratory therapists, since the ventilator variables that affect oxygenation and minute ventilation are significantly altered in this mode. There is insufficient evidence at this time to support the inclusion of APRV in available modes as a major decision driver in selecting neonatal and pediatric ventilators.

It is reported that 70 % of intubated neonates have an airway leak >10 % (Bernstein et al. 1995). Many ventilators now have sophisticated algorithms for rapid adjustment of flow profiles in the patient-ventilator system to compensate for airway leaks, and there appear to be differences in performance of these leak compensation protocols between brands of ventilators (Mehta et al. 2001). Without leak compensation, ventilators that flow trigger are prone to excessive autocycling. In my experience these algorithms are important when assessing ventilator performance because there tend to be more and larger leaks in the neonatal-pediatric population because of the use of uncuffed endotracheal tubes. Bench testing of ventilator performance should include tests with and without the leak compensation activated and include the presence of a constant but adjustable leak. This is possible in some sophisticated lung models.

There is little consensus around a precise definition of response time, but it is usually described as the elapsed time (in milliseconds) from when the proximal airway flow (or pressure) changes enough to activate the trigger threshold until the time when sufficient flow has been added to the system to bring the flow (or pressure) signal back to baseline. Relatively short response times are necessary to improve patient-ventilator synchrony. There are clearly differences in response time between brands of ventilators, but there is also a lack of agreement on the exact definitions of response time (Ferreira et al. 2008; Heulitt et al. 2000; Salyer et al. 1992; Terado et al. 2008). Breaths may be triggered by changes in flow or pressure, movement of the abdominal wall, changes in thoracic impedance, or diaphragmatic electrical activity. The most widely used triggering variable for spontaneous breaths is flow. The conventional wisdom now holds that flow triggering is the best method of synchronized triggering, but little testing in humans has never been done to verify this assumption (Donn and Becker 2003). Testing of response time between devices being evaluated is recommended, and all other things being equal, ventilators should be selected with the shortest response time possible.

The growth in the number of monitoring devices and alarms in the critical care environment has made it increasingly challenging to discern and differentiate unimportant or false alarms from meaningful alarms (Chambrin et al. 1999; Gorges et al. 2009; Lawless 1994; Sabar and Zmora 1997). Thus, a common design for ventilator alarm packages has two levels of alarms, a lower level for alarms that are not necessarily life threatening and an upper level for critical alarms that require an immediate response/intervention. This is essential since a large proportion of alarms in the critical care environment are low priority or “nuisance” alarms. The alarm volume should be adjustable from low volume to high volume, so that the alarm level can be customized to the care environment. In neonatal units, lower volumes might be advisable to minimize noxious stimulation of the patient.

The design of some intensive care units provides for the connection of ventilators and monitors to a system that creates audible and/or visible alarms outside the patient rooms. This decreases (but does not eliminate) the likelihood that critical alarms can go unnoticed if no one is in the patient’s room. This is an essential feature, particularly for monitoring the ventilation of infants and children. Their inability to cooperate creates an increased likelihood of accidental disconnection from the ventilator, and their small size increases the likelihood of accidental extubation. Plus, rapid response to alarms is particularly important in this population because of increased sensitivity to physiologic insult.

Many ventilators now have a vast array of alarm capabilities. For the purposes of patient safety, the essential alarms for neonates and children include:

Accidental disconnection of the ventilator circuit is the most common threat of mishap with a mechanical ventilator. Some ventilators now have sophisticated algorithms for detection of disconnection that have rendered the alarms very sensitive and specific. However, testing of disconnection alarms under various conditions is advisable.

A significant challenge in ventilating infants and small children is the accuracy of tidal volume measurements. Many ventilators currently being used in infants and children were not specifically designed for use in this population. As such the precision and accuracy of their tidal volume measures are acceptable for the ranges of flows and volumes used in adults, but may not be sufficiently accurate for the measurement of very small tidal volumes in infants and children. As an example, one widely used ventilator in both adults and pediatric populations has a stated tidal volume accuracy of ±1 mL +10 % of reading. So if the tidal volume measurement on the ventilator is reading 20 mL, the actual tidal volume could be between 17 and 23 mL, which in reality is ±15 %, and still be within the stated accuracy of the instrument. This is right at the limits of what I would consider acceptable since careful control and limitation of tidal volumes is so important in preventing ventilator-induced lung injury. But if the tidal volume is reading 4 mL, the actual tidal volume could be between 2.6 and 5.4 mL, which is actually ±35 %. The use of very small tidal volumes (4–6 mL/kg) in very low birth weight infants (<750 g) is increasing, and as this happens, some ventilators are challenged to accurately deliver and display tidal volumes.

The large volume of gas in a ventilator circuit, relative to the very small tidal volumes being delivered to neonates and small children, makes measurement of tidal volume at the proximal airway seem the best configuration for accurate determinations. This has been repeatedly verified by my clinical observations. The preponderance of the literature supports this notion. It has been repeatedly reported that very small tidal volumes cannot be accurately measured with sensors that are back inside the ventilator, but instead, a proximal flow sensor is required (Cannon et al. 2000; Castle et al. 2002; DiBlasi et al. 2007; Neve et al. 2003; Salyer and Jackson 2005a; Chow et al. 2002; Roske et al. 1998). A single research lab has suggested that this is not necessarily true with one particular brand of ventilator (Heulitt et al. 2005, 2009), but their findings have not been duplicated by others. Thus, I recommend that any ventilator to be used on neonatal or small pediatric patients has the capacity to measure tidal volume at the proximal airway.

There are currently two proximal airway volume measurement technologies that are available on various ventilators: (1) disposable differential pressure pneumotachometer and (2) reusable heated wire anemometers. Under ideal conditions these sensors have similar accuracies, in the range of +10 %. However, in practice these two types of sensors have strengths and weaknesses. The heated wire sensors are prone to the accumulation of moisture from condensation of the heated humidified gas passing through them. It is necessary to periodically remove the water from the sensor and to occasionally change the sensor because of degraded performance related to the condensation. The disposable pneumotachometers are also prone to problems with accumulated water, in this case inside the pressure sensing lines that go back to the ventilator. This problem was serious enough that we stopped using them and went to the heated wire anemometer for all neonatal patients (the brand of ventilator we were using had the capacity to use either style of proximal flow sensor.

As stated above, tidal volume targeting is accomplished using different approaches to control algorithms depending on ventilator brand. Consider that spontaneous tidal volumes are highly variable in newborns. Paetow et al. showed that spontaneous tidal volumes were highly variable in both healthy infants and those with significant lung disease (Paetow et al. 1999). Within the same patient spontaneous tidal volumes ranged from 2.0 to 22 mL (Fig. 67.3). This high variability has implications for the ventilators that attempt to do volume-targeted decelerating flow-type (dual control) modes since the ventilators must rapidly change their flow profiles and pressure limits to try to adjust the targeted tidal volume in the face of this high variability.

One widely sold ventilator has a volume limitation feature such that in volume-targeted modes, a maximum allowable V _T for each breath can be set. This, combined with a volume target, creates a “bracketed” tidal volume range. In theory this ought to help minimize the risk of volumetric overdistention of the lung, but there is very little published testing about this.

Thus, the ability of the ventilator to maintain targeted tidal volumes within a specific range should be tested. We developed a bench test to analyze this capability in ventilators. See the section on bench testing.

The interface between the ventilator and the patient is the point where the performance of neonatal and pediatric ventilators is sometimes challenged. Poor response to the patients’ own respiratory cycle can lead to dyssynchrony and an increase in the imposed work of breathing. There are reported differences in the imposed work of breathing among certain ventilators (el-Khatib et al. 1994). It is thought that imposed work of breathing was related to trigger design and that flow triggering reduced the imposed work of breathing and/or patient-ventilator dyssynchrony (Branson et al. 1994; Dimitriou et al. 1998; Sassoon et al. 1994; Stayer et al. 2001; Uchiyama et al. 1995). The responsiveness of the ventilator to infant breathing patterns can be tested, and I discuss this in the section on bench testing.

The introduction of new ventilator technology introduces risk to the patient since there will be a period of transition when the clinical staff will need to be trained and become proficient at operating these new devices. During this period there is real risk to patients of operator-induced application errors. I have seen this repeatedly. We often commit insufficient resources and attention to training. The amount of training time required for new ventilators will be difficult to estimate, but when evaluating new products, the complexity of the interface and resultant training burden should be considered.