Abstract
BACKGROUND: Many pediatric and neonatal ICU patients receive nitric oxide (NO), with some also requiring magnetic resonance imaging (MRI) scans. MRI-compatible NO delivery devices are not always available. We describe and bench test a method of delivering NO during MRI using standard equipment in which a NO delivery device was positioned in the MRI control room with the NO blender component connected to oxygen and set to 80 ppm and delivering flow via 12 m of tubing to a MRI-compatible ventilator, set up inside the MRI scanner magnet room.
METHODS: For our bench test, the ventilator was set up normally and connected to an infant test lung to simulate several patients of differing weight (ie, 4 kg, 10 kg, 20 kg). The NO blender delivered flows of 2–10 L/min to the ventilator to achieve a range of NO and oxygen concentrations monitored via extended tubing. The measured values were compared to calculated values.
RESULTS: A range of NO concentrations (12–41 ppm) and FIO2 values (0.67–0.97) were achieved during the bench testing. The additional flow increased delivered peak inspiratory pressure and PEEP by 1–5 cm H2O. Calculated values were within acceptable ranges and were used to create a lookup table.
CONCLUSIONS: In clinical use, this system can safely generate a range of NO flows of 15–42 ppm with an accompanying FIO2 range of 0.34–0.98.
Introduction
Inhaled nitric oxide (INO) can be used as a pulmonary vasodilator in critically ill neonates and children.1-6 We regularly provide INO therapy to some of our neonatal and pediatric ICU patients using the common INOmax DSIR Plus NO delivery device (Mallinckrodt Pharmaceuticals, Staines-Upon-Thames, Surrey, United Kingdom). Some patients receiving INO, such as those suffering or at risk of hypoxic-ischemic encephalopathy, may also require magnetic resonance imaging (MRI) scans for diagnosis and prognosis. A version of the INOmax DSIR Plus that is compatible with MRI does exist in the United States but not in other countries. The standard DSIR Plus is not MRI compatible, so a judiciously modified or alternative system is required to provide INO in the MRI environment for sites outside the United States and for sites in the United States without access to the MRI-compatible DSIR Plus or to a MRI device such as the Embrace Neonatal MRI System (Aspect Imaging, Nashville, Tennessee), which can be taken to the patient in the neonatal ICU. In addition, current or future users of alternative devices that are not MRI-compatible also require a solution for providing NO during MRI. We have successfully used a system to deliver INO to neonates while undergoing MRI. Our approach is based on using the INOblender component of the DSIR Plus in combination with the MRI-compatible babyPAC 100 pediatric ventilator (Smiths Medical, Minneapolis, Minnesota). The babyPAC ventilator is designed for ventilating infant and pediatric patients < 20 kg and is well suited to this application as it is a simple transport ventilator capable of generating a constant bias flow (nominally 10 L/min). The aims of this study were to describe and bench test a method of delivering NO to ventilated neonates during MRI, establish its effect on ventilator performance, and create a lookup table to facilitate its clinical use.
Quick Look
Current Knowledge
Some ventilated neonates and infants receiving nitric oxide (NO) may require magnetic resonance imaging (MRI) scans without interruption of NO delivery. Not all units will have access to MRI-compatible NO delivery systems.
What This Paper Contributes to Our Knowledge
The blender component of a non-MRI-compatible NO delivery system together with an MRI-compatible ventilator can provide clinically appropriate NO and oxygen levels to ventilated patients < 20 kg while undergoing an MRI scan.
Methods
Setup
This study consisted of bench testing performed within the neonatal and pediatric ICUs and clinical application in the pediatric ICU, neonatal ICU, and MRI suite at Royal Children’s Hospital in Parkville, Victoria, Australia. Departmental funding was used to support the study. The MRI environment consists of 4 safety zones (I–IV) with only MRI-compatible devices allowed in Zone IV, which is the area with the strongest magnetic field.7 Non-MRI-compatible equipment is restricted to Zones I and II and are only allowed to enter Zone III (ie, the MRI control room) if necessary and adequately supervised. The general setup of our equipment consists of the DSIR Plus positioned safely away from the magnetic field in Zone III. The INOblender component of the DSIR Plus is then used to deliver a flow of NO via a long length of tubing to the babyPAC ventilator, which is placed within the strongest magnetic field area in Zone IV (ie, the MRI scanner magnet room). Extended tubing is used to monitor NO, NO2, and FIO2. For bench testing, properly maintained and serviced babyPAC ventilators were set up as for normal MRI use (Fig. 1) with a standard nondisposable circuit (W7623, Smiths Medical) and an extension kit (W196-004, Smiths Medical) to lengthen the circuit. Ports for administering NO and monitoring were added (Fig. 1A, T-piece adapter [#1948, Intersurgical, Wokingham, Berkshire, United Kingdom] with a 3-way tap [BD Connecta, #394600, Becton Dickinson Infusion Therapy Ab, Helsingborg, Sweden]).
Setup. Connections specific for testing and not required during clinical use are highlighted in orange. See text for full details. (A) T-piece adaptors with 3-way tap; (B) male-to-male luer adapter; (C) tubing-to-female luer adapter; (D) oxygen tubing; (E) AVEA flow sensor; (F) 5-mm endotracheal tube adapter; (G) 22m/15f–22m/15f connector; (H) extended monitoring line; (I) oxygen tubing. The babyPAC was connected to wall air and oxygen for delivering FIO2 < 0.50 and to only wall oxygen for higher FIO2 values. INOblender was set to 80 ppm (NOset). Measurement sites and the respective measured parameters are indicated in parentheses (see text for details). Note that [FNOth] is a theoretical, calculated value relevant to the position indicated. The respiratory monitor was connected at the circuit Y to measure tidal volume.
Our test settings consisted of the babyPAC being placed in the CMV + active PEEP mode (pressure controlled continuous mandatory ventilation, providing a nominal bias flow (Fb) of 10 L/min) and using a volume-targeted ventilation strategy with PEEP set to as close to 5 cm H2O as possible and peak inspiratory pressure (PIP) adjusted to target measured exhaled tidal volumes of 6 mL/kg. Pressures were read from the babyPAC’s pressure gauge. The ventilator circuit was connected to an infant test lung (5601i, Michigan Instruments, Grand Rapids, Michigan) with resistance and compliance set in each of 3 groups of measurements to approximately simulate nominal 4 kg, 10 kg, and 20 kg patients (with target tidal volumes of 24, 60, and 120 mL, respectively). The age-based test lung compliances, in order of decreasing absolute severity, were set to 0.0015 L/cm H2O, 0.004 L/cm H2O, and 0.008 L/cm H2O representing moderate-severely diseased lungs for the 4-kg, 10-kg, and 20-kg simulations, respectively, based on the range available on the test lung and other studies.8-11 This represents our relevant patient group including neonates (ie, 4-kg simulation) with meconium aspiration syndrome, persistent pulmonary hypertension of the newborn, respiratory distress syndrome, or chronic lung disease, as well as representing moderate-severe respiratory distress syndrome in the 10-kg and 20-kg simulations. To be realistic and match clinical practice using specific fittings, resistances were constructed by connecting the ends of 2 endotracheal tube adaptors (3.0-mm, 4.0-mm, and 5.5-mm adaptors for the 4-kg, 10-kg, and 20-kg simulations, respectively) to each other with a short length of tubing.12 Tests were conducted at the clinically relevant FIO2 values of 0.60, 0.80, and 1.00 and at 30 breaths/min (inspiratory time of 1 s and expiratory time of 1 s) for the 4-kg simulation and 20 breaths/min (inspiratory time of 1 s and expiratory time of 2 s) for the 10-kg and 20-kg simulations. Flow and exhaled tidal volumes were measured using the NM3 Respiratory Profile Monitor (Philips Respironics, Andover, Massachusetts) positioned between the test lung and the babyPAC circuit Y-piece. Gas compensation on the NM3 monitor was set to 21% O2 and 79% N2 and automatically calibrated flow sensors were used (#9765-00 for the 4-kg and 10-kg simulations with accuracy allowing for oxygen and humidity offsets the greater of –6% to +3% or ± 2 mL; or #9767-00 for the 20-kg simulation with accuracy allowing for oxygen and humidity offsets of –6% to +3%).
A calibrated DSIR Plus was fitted with an 800 parts per million (ppm) source cylinder of NO (NOcyl). The fitted INOblender was connected to wall oxygen (FIO2in) and set to 80 ppm (NOset) and delivered a NO-O2 mixture into the babyPAC circuit via ∼ 12 m of oxygen tubing (Fig. 1D; 2034, Teleflex Medical Australia, Moorebank, New South Wales, Australia) connected to the T-piece with adaptors (Fig. 1B, male-to-male luer adapter [893.00, Vygon, Ecouen, France]; Fig. 1C, tubing-to-female luer adapter [801.00, Vygon]). The additional flow of NO-O2 into the babyPAC circuit was set on the INOblender flow meter at a number of flows between 2 L/min and 10 L/min as read from the top of the ball (FNOset) to achieve a range of delivered NO and oxygen concentrations. The flow read from the flow meter is expected to be less than the theoretical output flow of the INOblender (FNOth) because NO is added to the flow of O2 in the INOblender after the flow meter. This flow of NO-O2 from the INOblender (FNO) as well as the ventilator Fb are measured using hotwire anemometers (flow sensors; 16465, Carefusion, Yorba Linda, California) (Fig. 1E) that were zeroed before use and connected to separate AVEA ventilators (Carefusion). Whereas the NM3 device was used to monitor tidal volumes, the AVEA ventilators were used solely to obtain Fb and FNO measurements. The AVEA ventilators are unable to reliably measure tidal volumes when used in this way (ie, when the measured flow is not synchronous to their own inspiratory and expiratory cycles), but we used them in the absence of other available dedicated flow measurement devices. The flow sensor was connected to the outlet of the INOblender using a 22m/15f–22m/15f connector (1422, Hudson RCI, Teleflex Medical, North Carolina), a 5-mm endotracheal tube adapter, and ∼ 5 cm of oxygen tubing (Fig. 1F, G, I). Flows were measured when stable during the expiratory phase. Examples of Fb waveforms at different inspiratory pressures and rates are shown in Supplementary Figure 1 (see the supplementary materials at http://www.rcjournal.com). Delivered FIO2, NO, and NO2 were measured by the DSIR Plus using 4 lengths (∼ 12 m) of standard monitoring line (M1090916-02, Mallinckrodt Pharmaceuticals, Staines-Upon-Thames, Surrey, United Kingdom) connected together with female-to-female luer adaptors (892.00, Vygon) (Fig. 1H).
Procedure and Measurements
Initially, and before any tubing was added to the outlet of the INOblender, the INOblender flow (FNOset) was set to 2 L/min on the flow meter. The babyPAC was set to deliver a FIO2 of 0.60 (FIO2set) and was set to ventilate as described above. PEEP was initially set to 5 cm H2O with PIP adjusted to achieve as close as possible to 6 mL/kg exhaled tidal volumes before adding the additional FNO. A flow sensor was connected to the outlet of the INOblender to measure the delivered flow of NO before connecting the 12 m of oxygen tubing to the INOblender outlet (FNOpre) to determine whether adding the 12 m of tubing affected the delivered flow. After connecting the 12 m of oxygen tubing, the FNO was measured again and the flow, as indicated by the position of the INOblender flow meter ball, was rechecked and recorded. The PIP and PEEP of the babyPAC were then checked to determine whether the additional flow had affected them; they were adjusted back to their original values if required: PEEP was reset to the lowest possible value ≥ 5 cm H2O, and PIP was adjusted to reestablish as close as possible to 6 mL/kg exhaled tidal volumes. The babyPAC Fb was then recorded as were the delivered oxygen, NO, and NO2 levels. The babyPAC FIO2set was then sequentially changed to 0.80 and 1.00, and the Fb and delivered oxygen, NO, and NO2 levels were recorded for each FIO2set. This procedure was repeated for INOblender flows of 4 L/min, 6 L/min, 8 L/min, and 10 L/min. FNO values were recorded once for each INOblender set flow at the initial FIO2 setting. The initial PIP and PEEP were recorded from the babyPAC, as were the PIP and PEEP settings required to continue to deliver 6 mL/kg in the presence of additional FNO.
Measured values of delivered NO and FIO2 were compared to theoretical values calculated using a principle formula (Formula 1) to test the accuracy of delivery. In addition, 2 related formulas (Formula 2, Formula 3) were derived from the principle formula after making various levels of simplifications to have a practical tool and to test these simplifications. The formulas were determined from first principles using the broad concepts of gas dilution and conservation of mass. The principle formulas, using all measured flows, for the theoretically delivered FIO2 (FIO2th) and NO (NOth) were determined as follows:
Note that 
is 1.00 and NOset is 80 ppm.
In practice, the flows (FNO and Fb) are unlikely to be measured; the INOblender flow meter setting would be used as the NO flow, and the bias flow would be assumed to be the nominal 10 L/min. Using the additional formulas that describe the INOblender FIO2 and flow outputs:
Now modified versions of the primary formulas can be determined using the NO flow set on the INOblender with the measured babyPac bias flow rather than only using measured values:
This can be further simplified and made more practical by assuming that the babyPac bias flow (Fb) is the nominal 10 L/min (Formula 3). All FIO2 values used in the formulas are between 0 and 1.
The safety of the described method was also assessed by confirming that additional flow did not cause an unsafe buildup of pressure when the babyPAC was turned off.
Results
Results of the bench tests demonstrate that a range of NO levels and FIO2 values were achieved and the babyPAC continued to ventilate properly (Table 1). PIPs and PEEPs of 25 cm H2O and 5 cm H2O, 18 cm H2O and 5 cm H2O, and 18 cm H2O and 5 cm H2O were required for the 4-kg, 10-kg, and 20-kg simulations, respectively, to achieve 6 mL/kg tidal volumes (24.3 mL, 58.8 mL, and 122 mL, respectively) in the absence of additional NO flow. Once the additional NO flow was added, delivered PIP and PEEP increased by 1–5 cm H2O and were unable to be compensated for in all cases with the exception of the tests at 2 L/min in the 4-kg simulation (Table 1). The babyPAC’s bias flow was between 10.8 L/min and 11.5 L/min and was slightly biphasic in nature, with a minimum during inspiration (< 0.4 L/min lower than during expiration). Examples of bias flow waveforms are provided in Supplementary Figure 1 (see the supplementary materials at http://www.rcjournal.com). The bias flow also showed a small effect with changes in FIO2 and was typically lowest with a FIO2 of 1.00. Examples of the flow and pressure waveform measured at the Y-piece of the test lung are presented in Supplementary Figure 2 (see the supplementary materials at http://www.rcjournal.com). When the babyPAC was turned off with the additional NO flow still on, an increase in pressure was not observed, indicating that the additional flow of NO safely vented freely through the expiratory valve of the babyPAC.
Bench Test Results
It was noteworthy that the measured INOblender outlet flows (FNOpre and FNO) were consistently higher than FNOset set on the flow meter and the theoretically calculated total INOblender flow (FNOth) by up to 4.2 L/min. Initially, adding the extended length of tubing to the INOblender outlet caused the flow displayed by the ball to be reduced, particularly at flows > 6 L/min, by between 0.1 L/min and 0.9 L/min. The actual flow was not affected in the same way, however, with the only noticeable changes in measured flow occurring at flows of ≥ 8 L/min, where the measured flow was 1.1 L/min higher after the tubing was added.
The DSIR Plus was able to sample through the extended monitoring lines without any appreciable change in measurement response time. As shown in Table 1 and Supplementary Table 1 (see the supplementary materials at http://www.rcjournal.com), a range of NO concentrations (12–41 ppm) and FIO2 values (0.67–0.97) were achieved across the 3 simulated patient sizes. Delivered FIO2 varied from that set on the babyPAC by between 0.18 (above) and 0.06 (below). NO2 levels were always < 0.7 ppm. An additional flow of 4 L/min produced the clinically typical NO levels of ∼ 20 ppm with corresponding FIO2 values ranging between 0.71 and 0.96 in all 3 simulated patient sizes. Measured FIO2 and NO were compared to expected, calculated values (Supplementary Table 1; see the supplementary materials at http://www.rcjournal.com). The formulas performed satisfactorily given the inherent measurement and delivery errors, with measured FIO2 typically within 0.03 of the expected calculated FIO2 across all versions of the formula, and expected calculated NO levels tended to be higher than measured levels, but errors were within 5 ppm across all formulas.
Discussion
There are a small number of papers describing the temporary trialing of NO administered via face mask while performing cardiac13-15 or pulmonary16 MRI for investigative purposes, but we found none that describe the particular method and application we have used, whereby a ventilated neonate or child from a neonatal or pediatric ICU who is already receiving INO can be transported to and from MRI and have a MRI while continuing to receive INO therapy using the same system. We have demonstrated a novel system capable of doing this, a system that can generate a range of clinically relevant NO concentrations while providing adequate ventilation for patients up to ∼ 20 kg. These findings support our practical experience where we have successfully transported patients to MRI using this setup.
As it is likely to be an uncommon setup, it is important to have documented guidelines such as those provided here to assist with its proper use. In practice, the system simply involves the connection of extended delivery and monitoring lines from the INOblender to the ventilator circuit, and then following the setup provided in Figure 1 and the settings guide provided in the lookup table (Table 2) to achieve delivery targets, as well as adjusting ventilator pressures and fine-tuning flows as needed. Users unfamiliar with the connections and equipment involved should seek guidance, and users should be careful to ensure proper setup and function by utilizing monitoring and alarms as well as the attainment of the expected values as a guide. Both the DSIR Plus and babyPAC ventilator or similar devices are familiar equipment in neonatal and pediatric ICUs, so this novel application could be widely implemented or modified to suit combinations of other similar equipment. All that is required is a MRI-compatible ventilator with a continuous bias flow (ie, to generate constant NO levels when the DSIR Plus’s injector module is not used) and a constant flow of NO.
Lookup Table
While users in the United States may currently have access to the MRI-compatible DSIR Plus, for users outside the United States who do not have it, or for sites in the United States without access to it, or for sites without access to newer portable neonatal MRI scanners as an option, alternative means of delivering NO in MRI are required. In addition, potential changes in the future supply of INO may result in new manufacturers that do not offer MRI-compatible devices. This would mean that users may benefit from the specifics or some of the general principles detailed here if working with non-MRI-compatible devices in the future. Using ventilators without continuous bias flow is possible, but this would generate varying NO and oxygen concentrations that could not be properly monitored or readily calculated and may be clinically inappropriate. Further study is required to assess the suitability of any particular ventilator and combination of settings.
The additional NO flow into the babyPAC did cause initial increases in ventilation pressures. This was anticipated as a result of the design of the babyPac, which incorporates a diaphragm expiratory valve,17,18 the resistance of which increases the pressure when additional flow is present. These increases in pressure manifested as an increase in PEEP for all flows except 2 L/min in the 4-kg simulation, while the PIP required adjusting upward to maintain the targeted tidal volumes. For higher NO or bias flows, or for larger patients, a higher PEEP may need to be used if clinically suitable. The situations when this increase could not be largely tolerated or compensated for are likely to be uncommon in practice (eg, in larger patients or when particularly high NO levels are required). The addition of the NO flow into the babyPAC circuit also altered the delivered FIO2 (FIO2 offset) because of the presence of oxygen with a FIO2 of 0.90 with the NO from the INOblender. A FIO2 range of 0.67–0.97 was achieved across all simulations and should be all that is clinically necessary. The delivered FIO2 tended to be higher than that set on the babyPAC (if set FIO2 is < 0.90), but this could be modified by changing the set FIO2 on the babyPAC. It should also be noted that a FIO2 of 1.00 cannot be achieved as is always the case when NO is added.
A number of small performance details of the babyPAC and INOblender were highlighted in this study. The babyPAC’s bias flow varies slightly with pressure (∼ 0.25 L/min/10 cm H2O), and the bias flow decreases slightly at a FIO2 of 1.00.17 These issues, however, do not make a significant difference in the babyPAC’s performance. The flow meter ball of the INOblender also displays a flow less than actually delivered because NO is added after the oxygen flows through the flow meter. The theoretical total actual flow of NO and oxygen delivered from the INOblender (FNOth) equals the oxygen flow set on the ball (FNOset) plus an amount of NO flow coming from the NO cylinder (∼ 0.11 × FNOset). In addition, as stated in the user manual, the flow meter is not back pressure-compensated, so it will display a lower flow than is actually flowing when pressure is applied to the gas outlet.19 We saw this when the connection of tubing to the outlet of the INOblender decreased the flow indicated by the ball but the actual flow (measured independently) was unchanged, especially at flows < 8 L/min. All considered, these and other factors result in the accuracy of the INOblender flow meter (unstated in the manual)19 to be low and for it to underestimate flow, but it is still an adequate measure of additional NO flow for the purposes detailed here. In practice, this means that, for typical NO levels, the NO flow should ideally be set on the INOblender before the tubing is added or initially set slightly below the target (by 0.2–0.4 L/min) if the tubing is already on and then adjusted on the basis of monitored values to achieve the required NO and FIO2.
Errors in the expected calculated values of NO and FIO2 compared to measured values were acceptably small using all versions of the formulas (Supplementary Tables 1 and 2; see the supplementary materials at http://www.rcjournal.com), with FIO2 within 0.05 and NO within 5 ppm of expected values across all simulations, which is important to establish the utility of the formulas for the lookup table. This is in spite of but within the system’s inherent delivery and measurement errors (ie, the DSIR Plus and its components: INOblender NO ± 20%, DSIR Plus NO measurement ± 10% + 0.5 ppm, DSIR Plus O2 measurement ± 0.03; the babyPAC: delivered FIO2 > ± 0.05; and the AVEA flow sensors: ± 10%), and the precision that the babyPAC and INOblender can be physically set. In addition, the fact that the flow sensors were only zeroed and used in dry gas rather than heated and humidified gas did not appear to cause additional errors. It is of note that formula 2, based on measured bias flow and the INOblender flow read from the ball, resulted in the smallest NO errors, whereas formula 1 and formula 3 had larger NO errors, possibly reflecting competing influences of variations in the types of flow and errors inherent in the measurements. For FIO2 calculations, all 3 formulas performed similarly. The larger errors when the babyPAC was set to a FIO2 of 1.00 also reflect the fact that the babyPAC commonly only delivered a FIO2 of ∼ 0.95 when set to 1.00. Overall, formula 3 performed well and is the most practical to use because it does not require any additional equipment or measurements to be taken as it assumes the bias flow to be 10 L/min and uses the INOblender flow simply read from the ball. If a user were able to measure the NO or bias flows, formula 1 or formula 2 could be used to improve theoretical accuracy, although we have found that formula 3 is satisfactory in practice, and the formulas ultimately just provide initial settings that can be subsequently refined by observing the monitored values of NO and FIO2 and then manually adjusting the flows and or FIO2 to achieve the desired NO and FIO2 levels or clinical response.
We constructed a lookup table (Table 2) using formula 3 to assist users in utilizing this system to achieve required patient NO and O2 levels. We verified this lookup table by (1) generating all the presented combinations of settings in the 10-kg and 20-kg simulations described in the methods with a PIP of 20 cm H2O and PEEP of 5 cm H2O, (2) measuring the resultant NO and O2 levels, and (3) confirming that measured NO was within ± 5 ppm and FIO2 values were within ± 0.05 of expected values. A potential limitation may be that we only tested at a limited number of ventilator settings, but the formulas used performed well across volume- and pressure-targeted ventilation strategies (Supplementary Tables 1 and 2; see the supplementary materials at http://www.rcjournal.com), variations in bias flow, different PIPs and PEEPs, a range of tidal volumes, and at different breathing frequencies. This is because the underlying principle and the related formulas fundamentally rely only on the bias flow, and our results indicate that this is not appreciably affected by ventilation. In addition, we have successfully used the lookup table in practice and found it to generate delivered FIO2 values and NO levels within the verified level of accuracy (ie, NO within ± 5 ppm, FIO2 values within ± 0.05 of expected values) irrespective of ventilation settings. The lookup table is used to determine the required additional flow of NO and the babyPAC FIO2 setting based on the desired NO and FIO2 levels. For example, to achieve delivery of 20 ppm and FIO2 of 0.60, an additional flow of 3 L/min is required from the INOblender with the babyPAC set to a FIO2 of 0.50. As discussed above, if the tubing was already connected to the INOblender outlet when the flow is first set, it would be prudent to initially set it to just below 3 L/min and then make any adjustments on the basis of the measured NO and FIO2.
Some practical considerations should be borne in mind when using this system. To avoid a period of reduced NO delivery when first transferring the patient onto this system from their primary ventilator, it may be necessary to prime the NO delivery tubing (ie, 12 m of oxygen tubing) with NO and oxygen from the INOblender. Alternate means of delivering NO or ventilating the patient should always be available as a general backup and will probably be required at various transition points on the trip between the neonatal or pediatric ICU and MRI (such as when NO delivery into the babyPAC is interrupted when tubes are temporarily disconnected to be put through the wall tube/hole into the magnet room). When adjusting NO or oxygen levels using the flow, it is important to be mindful of the interaction between delivered NO and FIO2(ie, NO concentration can be increased by increasing flow but FIO2 will also change). Adjusting the delivered NO level by reducing the INOblender setting (ie, reducing NOset from 80 ppm) rather than by reducing the flow will avoid pressure changes but will still cause FIO2 changes. Our current setup is limited to a minimum INO of ∼ 15 ppm because we did not test INOblender flows < 2 L/min because the INOblender flow meter is physically obstructed by its front panel at flows below ∼ 2 L/min. Delivering INO < 15 ppm would require reducing the flow as best as possible in the presence of the obstructed view or reducing the set NO on the INOblender below our standard of 80 ppm. If FIO2 is increased by increasing the INOblender flow, NO can be held constant by reducing the NO level set on the blender level (eg from 80 to 70 ppm). If possible, adjusting the babyPAC’s FIO2 setting directly avoids these interactions. Monitoring and alarms should be utilized to ensure adequate NO and oxygen delivery. Users must also be aware that the delivered PIP and PEEP pressures will increase with any increases in the INOblender flow. Pressure increases can be compensated for by reducing the PEEP or PIP as necessary; eg, for additional flows of 2–4 L/min, PIP and PEEP may increase by up to 5 cm H2O (typically 1–3 cm H2O).
An alternative means of delivering NO is to set a constant flow of NO on the INOblender and adjust the INOblender NO concentration (NOset) rather than the method described above and altering the flow of NO. We believe that the method described in this study, where the NO concentration is held constant (80 ppm) and the flow is adjusted as required, is the better approach despite the alterations in ventilation pressures that can be generated by changing flows. This approach is more intuitive, and the flow generally only has to be adjusted once at the outset; in addition, using the highest possible NO concentration only requires the lowest flow and therefore results in the smallest impact on PEEP and PIP and the smallest change in delivered FIO2 (with the babyPAC still largely controlling FIO2), and the range of achievable NO covers a more practical range (ie, 15–32 ppm with FIO2 values 0.34–0.98 with flows of 2–6 L/min). By comparison, if a set flow is chosen, the choice of flow is still limited to a similar narrow range because only low NO concentrations can be obtained with low flows, and high flows increase the minimum achievable FIO2 values and cause larger changes in ventilation pressures.
There are some unresolved safety issues that any user should bear in mind when using this system. No NO scavenging is provided, and circuit kinks could build up pressure in parts of the circuit or interfere with NO delivery, so ventilator pressure monitoring and alarms should be utilized. This bench study is also potentially limited because we only tested at 2 breathing frequencies and a limited range of PIP and PEEP settings, although the PEEP did vary due to its natural increase with the additional flow in the tests when it could not be fully compensated for back to 5 cm H2O, and PIP varied between 18 and 29 cm H2O. Together, these changes resulted in test pressures that covered a significant portion of the clinically relevant range. In addition, previous testing and the performance details of the babyPAC indicate that there would only be a slight but insignificant change in bias flow at different pressures and rates (Supplementary Figure 1; see the supplementary materials at http://www.rcjournal.com). Tidal volumes ranged from ∼ 24 mL to 122 mL and did not influence performance. Furthermore, we only tested the lung models at babyPAC FIO2 values of ≥ 0.60. This allowed us to focus on the clinically relevant range and to simplify the comparisons; in addition, the babyPAC’s User Manual17 already describes the effect of FIO2. We performed additional tests with the 4-kg simulation across the range of FIO2 values using pressure control ventilation on a second babyPAC, with results consistent with those in Table 1 and Supplementary Table 1 (see the supplementary materials at http://www.rcjournal.com) and without any FIO2 effect detected apart from the previously demonstrated and expected slight decrease in bias flow that occurs at a FIO2 of 1.00. Finally, the lookup table was verified across a range of FIO2 values and nominal patient sizes. Ultimately, the system was tested and performed successfully over a range of patient sizes, ventilation pressures, FIO2 values, and volumes, with no appreciable changes in the bias flow, which, in conjunction with the set NO flow and FIO2, is the main determinant of the achieved NO and FIO2 levels.
Conclusions
We have described and verified the performance of a system that allows NO to be delivered to a neonate or small child weighing < 20 kg during MRI. When used clinically, this system safely generates a NO range of 15–42 ppm with an accompanying FIO2 range of 0.34–0.98. We have successfully transported patients to MRI using this setup.
Footnotes
- Correspondence: Bradley G Carter PhD, Paediatric Intensive Care Unit, Royal Children’s Hospital, Flemington Road, Parkville, Victoria 3052, Australia. E-mail: bradley.carter{at}rch.org.au
Supplementary material related to this paper is available at http://www.rcjournal.com.
The authors have disclosed no conflicts of interest.
- Copyright © 2021 by Daedalus Enterprises
