Secondary Logo

Journal Logo

Evaluation of the Augmented Infant Resuscitator: A Monitoring Device for Neonatal Bag-Valve-Mask Resuscitation

Bennett, Desmond J. BA*; Itagaki, Taiga MD*,†; Chenelle, Christopher T. BA*; Bittner, Edward A. MD, PhD; Kacmarek, Robert M. PhD, RRT*,†

doi: 10.1213/ANE.0000000000002432
Pediatric Anesthesiology: Original Laboratory Research Report
Free
SDC

BACKGROUND: Annually, 6 million newborns require bag-valve-mask resuscitation, and providing live feedback has the potential to improve the quality of resuscitation. The Augmented Infant Resuscitator (AIR), a real-time feedback device, has been designed to identify leaks, obstructions, and inappropriate breath rates during bag-valve-mask resuscitation. However, its function has not been evaluated.

METHODS: The resistance of the AIR was measured by attaching it between a ventilator and a ventilator tester. To test the device’s reliability in training and clinical-use settings, it was placed in-line between a ventilation bag or ventilator and a neonatal manikin and a clinical lung model simulator. The lung model simulator simulated neonates of 3 sizes (2, 4, and 6 kg). Leaks, obstructions, and respiratory rate alterations were introduced.

RESULTS: At a flow of 5 L/min, the pressure drop across the AIR was only 0.38 cm H2O, and the device had almost no effect on ventilator breath parameters. During the manikin trials, it was able to detect all leaks and obstructions, correctly displaying an alarm 100% of the time. During the simulated clinical trials, the AIR performed best on the 6-kg neonatal model, followed by the 4-kg model, and finally the 2-kg model. Over all 3 clinical models, the prototype displayed the correct indicator 73.5% of the time, and when doing so, took 1.6 ± 0.9 seconds.

CONCLUSIONS: The AIR is a promising innovation that has the potential to improve neonatal resuscitation. It introduces only marginal resistance and performs well on neonatal manikins, but its firmware should be improved before clinical use.

From the *Department of Respiratory Care, Massachusetts General Hospital, Boston, Massachusetts

Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts.

Published ahead of print August 31, 2017.

Accepted for publication July 12, 2017.

Funding: This study was primarily supported by Respiratory Care Department funds at Massachusetts General Hospital. However, a small grant was received from the Consortium for Affordable Medical Technologies (CAMTech) as part of the Massachusetts General Hospital Global Health program.

Conflicts of Interest: See Disclosures at the end of the article.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.

Reprints will not be available from the authors.

Address correspondence to Robert M. Kacmarek, PhD, RRT, Department of Respiratory Care, Massachusetts General Hospital, 55 Fruit St, Boston, MA 02114. Address e-mail to rkacmarek@mgh.harvard.edu.

Intrapartum-related death of neonates is a leading cause of newborn mortality, responsible for approximately 23% of neonatal deaths.1 Improved training of providers during and after delivery has the potential to combat this phenomenon.1,2 This is particularly true in low- and middle-income regions, such as Africa and South Asia, where rates of intrapartum-related deaths are disproportionately high compared to industrialized countries.2–5 It is estimated that total neonatal mortality is nearly 14 times higher in Sub-Saharan Africa and South Asia than in industrialized countries. Respiratory problems are a major contributor to neonatal mortality, largely due to shortages of health care workers and proper training and equipment.2,3 Rates of attendance at births by trained personnel in these areas are among the lowest in the world, and accordingly, rates of intrapartum-related deaths are among the highest.2,3

Approximately 5%–10% of live births, or 10 million babies annually, require some level of breathing assistance at birth, while 3%–6%, or 6 million annually, require bag-valve-mask (BVM) resuscitation.3,6,7 However, <50% of women in Sub-Saharan African give birth in the presence of personnel capable of resuscitation.8 This insufficiency, coupled with the scarcity of neonatal intensive care units (NICUs) in low-resource settings (LRSs), may be contributing to the increased rates of intrapartum-related deaths. While NICUs in LRSs have improved (more beds, more staff, and patient databases) since the early 2000s,9 there exists much room for improvement, especially with respect to resuscitation. Programs such as Helping Babies Breathe10 have introduced educational resuscitation curriculums for use in LRSs, but long-term adequacy of proper resuscitation remains a problem. It has been shown that, even among health care professionals trained in neonatal BVM resuscitation, a significant number of providers use improper technique or show considerable deterioration in technique over time.11,12 Increased training is both cost-effective and successful in decreasing the rates of intrapartum-related deaths in low-resource areas,13–15 but there is significant progress to be made.5

The Augmented Infant Resuscitator (AIR),16 developed by clinicians and researchers at Mbarara University in Uganda and Massachusetts General Hospital and Massachusetts Institute of Technology in Boston, is a novel, low-cost device designed to facilitate proper technique of BVM resuscitation through both real-time feedback and data storage for post hoc review by trainers. The real-time feedback is based on measurements of flow, pressure, and rate of ventilation delivered through the BVM, and the device uses algorithms to inform the provider, through a simple interface, of an air leak, obstruction, or improper ventilation rate. The device also records usage data, allowing trainers and providers to review the records of these parameters and determine what variables contributed to specific training or actual resuscitation events. The AIR’s creators intend to implement the device in Uganda, where <10% of all babies are attended by staff trained in neonatal resuscitation.7 By helping to optimize neonatal BVM resuscitation technique of NICU physicians, nurses, and other care providers, this device has the potential to prevent morbidity from delays in adequate ventilation and the potential to reduce unnecessary intrapartum-related deaths resulting from suboptimal practice. However, its function has yet to be independently verified. The goal of this study was to assess the AIR for accuracy and effectiveness in a simulated setting of neonatal BVM resuscitation.

Back to Top | Article Outline

METHODS

The Device

Figure 1.

Figure 1.

The AIR is a simple device installed between a ventilation bag and a resuscitation mask (Figure 1A) and is designed to detect leaks and obstructions, as well as determine respiratory rate. The range for “OK” respiratory rate is currently set between 30 and 60 breaths per minute (bpm). The device tested in this study was a fourth-generation prototype, which is not yet commercially available; it can be seen in Figure 1B. For each indicator, the AIR displays a constant green light when there are no problems. When a leak, obstruction, or inappropriate respiratory rate is detected, the respective indicator flares red (Figure 1C). The thresholds for leak and obstruction are based on measured airway resistance and compliance. In the training mode, the AIR identifies a leak if compliance >4.0 mL/cm H2O and an obstruction if compliance <0.10 mL/cm H2O or resistance >90 cm H2O/L/s. The compliance and resistance settings for the AIR during the clinical mode are as follows: leak if compliance >8.5 mL/cm H2O or resistance <30 cm H2O/L/s and obstruction if compliance <1.6 mL/cm H2O or resistance >80 cm H2O/L/s.

Back to Top | Article Outline

The Evaluation

Figure 2.

Figure 2.

The evaluation was performed in 3 parts (Figure 2). First, we determined the resistance of the AIR and how it affected breath parameters when placed in-line in a ventilator circuit. In the second part, we used a neonatal manikin (a manikin used to practice neonatal resuscitation) to evaluate the device under normal conditions of neonatal ventilation, and under conditions of leak and obstruction. In the third part, we used simulated clinical models (a computerized lung model was used to simulate infants of varying sizes and lung mechanics) to evaluate the device’s clinical firmware.

Back to Top | Article Outline

Part 1: Evaluating Resistance to Flow and Effect on Breath Parameters

The effect of the AIR on flow resistance was tested using 2 methods. First, the device was attached in-line to a ventilator testing device (PTS 2000; Mallinckrodt, Dublin, Ireland). Pressure sensors were placed immediately before and after the AIR. Using a DAC1 dry air compressor (Siemens, Munich, Germany) and a flowmeter, continuous flow was delivered stepwise through the system from 0 to 15 liters per minute (lpm), pausing at each increment for 10 seconds. The PTS 2000 recorded the pressure change and flow rate at all increments. Based on standards set forth by the International Organization for Standardization, which state that resuscitation devices should not cause >0.5 kPa (≈ 5 cm H2O) drop in pressure at a flow rate of 5 lpm,17 we considered a pressure drop of 5 cm H2O or less at a flow rate of 5 lpm acceptable. Four trials were conducted. Second, using a Covidien PB980 ventilator (Covidien, Dublin, Ireland) in neonatal mode, the ASL 5000 Breathing Simulator (version 3.5; IngMar Medical, Pittsburgh, PA) was ventilated with and without the AIR in-line, and inspiratory time, expiratory time, tidal volume (VT), peak pressure, peak flow, positive end expiratory pressure (PEEP), and respiratory rate of the delivered breaths were compared. The ASL 5000 was preprogrammed to simulate neonates of 3 different sizes at specific settings18–20 (Table 1). Each neonatal model was ventilated for 5 minutes.

Table 1.

Table 1.

The ventilator was used in volume control mode and set to establish a delivered VT of 6 mL/kg as measured by the ASL 5000. A ventilator was used instead of a BVM to ensure accuracy and consistency of the delivered breaths. PEEP was set at 0 cm H2O, and inspiratory-to-expiratory time ratio was set as close to 1:2 as possible to achieve the desired VT for each model.

Back to Top | Article Outline

Part 2: Evaluating Accuracy and Reliability Using NeoNatalie Manikin

Since the AIR was designed to reinforce correct BVM resuscitation during resuscitation training, we initially evaluated the AIR using a NeoNatalie resuscitation manikin (Laerdal Global Health, Stavanger, Norway; Supplemental Digital Content, Figure S1A, http://links.lww.com/AA/B978). For this part of the evaluation, the AIR was programmed to its “training” firmware. The training firmware is designed for neonatal manikin lung mechanics only. During the NeoNatalie trials, an automated “bag-squeezer” was developed that accurately and consistently squeezed a 220-mL ventilation bag at a set rate and volume (device validation can be found in Supplemental Digital Content 1–3, http://links.lww.com/AA/B977, http://links.lww.com/AA/B978, http://links.lww.com/AA/B979). In addition to the bag-squeezer, NeoNatalie was also ventilated with a Crossvent 2i+ transport ventilator (Bio-Med Devices, Guilford, CT). This simple ventilator was used because modern ICU ventilators contain leak compensation algorithms, which would have made our setup unrealistic when leak was purposely introduced. If leak compensation is deactivated on ICU ventilators, auto-triggering occurs. Both the ventilator and bag-squeezer were set to achieve a VT of 40 mL (approximate NeoNatalie VT), a rate of 50 bpm, and an inspiratory-to-expiratory ratio of approximately 1:2. To create a system leak, we attached a small, perforated insert in the trachea of NeoNatalie. To simulate an obstruction, we placed a small-diameter insert in NeoNatalie’s trachea that markedly restricted airflow into the lungs of NeoNatalie (Supplemental Digital Content, Figure S1B, http://links.lww.com/AA/B979). The leak and obstruction inserts could only be placed individually, so a combined leak + obstruction scenario could not be conducted with NeoNatalie. Five trials each were run for normal conditions, leak only, and obstruction only scenarios.

Back to Top | Article Outline

Part 3: Evaluating Accuracy and Reliability Using Simulated Clinical Models

To test the accuracy and reliability of the AIR using simulated neonatal clinical models, the AIR’s programming was switched to its clinical firmware. The AIR was attached in-line between the Crossvent 2i+ transport ventilator and the bag-squeezer to the ASL 5000 breathing simulator. Both the bag-squeezer and Crossvent were set to ventilate each of the 3 neonatal models, which were established using the ASL 5000 (Table 1). Five trials were completed on each neonatal model, and each trial consisted of 5 stages as follows:

  1. Normal conditions: With no leak or obstruction, breaths were delivered by the ventilator and bag-squeezer at rates of 60, 50, or 40 bpm, for models 1, 2, and 3, respectively. Each run lasted for 15 seconds.
  2. Conditions of leak only: Each model was evaluated at 4-leak levels corresponding to no leak (leak 0), 25% of VT (leak 1), 50% of VT (leak 2), and 75% of VT (leak 3) using a 3-way stopcock system.21 Leak was established according to the following sequence: L0 → L1 → L2 → L3 → L1 → L3 → L2 → L1 → L0 → L2 → L0 → L3 → L0.
  3. Conditions of obstruction only: Partial obstruction and full obstruction were simulated using the IngMar Parabolic Restrictor Ring (IngMar Medical, Pittsburgh, PA) at settings of Rp200 and Rp500, respectively. Pressure drop across the resistors: flow 1.0 lpm, Rp500 4.70 cm H2O, Rp200 0.75 cm H2O; flow 3.0 lpm, Rp500 43.5 cm H2O, Rp200 6.50 cm H2O; and flow 5.0 lpm, Rp500 102.0 cm H2O, Rp200 18.9 cm H2O. The ventilator provided breaths for 15 seconds for each obstruction type.
  4. Conditions of leak and obstruction together: Leak was established with partial obstruction and full obstruction according to the following sequence: L0 → L1 → L2 → L3 → L1 → L3 → L2 → L1 → L0 → L2 → L0 → L3 → L0.
  5. Conditions of altered breath rates: Breaths were delivered by the Crossvent transport ventilator beginning at a rate of 5 bpm and increasing (in intervals of 10) to 85 bpm. From 28 to 32 and 58 to 62, breath rate was increased stepwise by 1 bpm to test the AIR’s lower (30 bpm) and upper (60 bpm) limits. The ventilator delivered the set breath rate for 15 seconds before moving on to the next interval. The bag-squeezer was not used during the rate trials because it is only capable of ventilating at rates of 40, 50, and 60 bpm.

Throughout the study, a stopwatch was used to determine the amount of time the AIR took to display or remove an indicator after each scenario was changed.

Back to Top | Article Outline

Statistical Analysis

Data were collected by the lung simulator’s software (ASL software version 3.5; IngMar Medical, Pittsburgh, PA) when analyzing breath parameters, and manually when addressing the AIR’s indicators and response time for both the NeoNatalie and clinical simulation trials. Results are expressed as either a percentage or mean values ± standard deviation. Multivariate logistic regression was used to compare proportions of success for both the NeoNatalie and clinical simulation trials while accounting for differences in scenario and alarm type. When statistical differences in the proportions were identified, post hoc comparisons were performed with Bonferroni correction. Means were compared using 2-way t tests. Statistical analysis was conducted using STATA Statistical Software Version 12 (StataCorp LLC, College Station, TX). A value of P < .05 was considered statistically significant. We report all results, but only discuss differences that were both statistically significant (P < .05) and clinically important (>10% difference).

Back to Top | Article Outline

RESULTS

Part 1: Evaluating Resistance to Flow

The largest pressure drop across the AIR was 2.01 cm H2O, which occurred at a flow rate of 15.00 lpm. Average pressure drop across the AIR was 0.730 ± 0.599 cm H2O (range, 0.030–2.01). The resulting maximum resistance of the AIR was 8.039 cm H2O/L/s, and average resistance was 4.750 ± 1.931 cm H2O/L/s (range, 0.000–8.039; Supplemental Digital Content, Figure S2, http://links.lww.com/AA/B979).

When attached in-line between the ventilator and the ASL 5000 test lung, the AIR had little effect on ventilator breath parameters. For all models, there were no changes that were both statistically significant and clinically important (Supplemental Digital Content 1, Table S2, http://links.lww.com/AA/B977).

Back to Top | Article Outline

Part 2: Evaluating Accuracy and Reliability Using NeoNatalie Manikin

During all scenarios tested on NeoNatalie (normal, leak only, and obstruction only), the AIR displayed the correct indicator during 100% of trials for both the ventilator and the bag-squeezer.

Back to Top | Article Outline

Part 3: Evaluating Accuracy and Reliability Using Simulated Clinical Models

Normal Conditions.

Under normal conditions (no leak and no obstruction), the AIR functioned perfectly in Model 3-Ventilator, properly displaying the correct light for all 15 seconds of all 5 trials; the AIR was almost perfect in Model 3-Bag-Squeezer (1 leak indicator false alarm). However, in Model 2-Ventilator, Model 2-Bag-Squeezer, and Model 1-Ventilator, the AIR’s “obstruction” indicator was on for all 15 seconds of all 5 “normal conditions” trials. In Model 1-Bag-Squeezer, the “leak” indicator was on for all 15 seconds of all 5 trials.

Back to Top | Article Outline

Conditions of Leak, Obstruction, and Leak ± Obstruction.

Bag-Squeezer:
Table 2.

Table 2.

Table 3.

Table 3.

Table 4.

Table 4.

Across all 3 indicators, the overall success rates of Model 1 (74.9%), Model 2 (76.5%), and Model 3 (77.9%) did not significantly differ from each other (P = .556). However, the leak indicator, specifically, functioned more effectively in Model 1 (88.1%) than in Model 2 (68.6%) and Model 3 (67.6%) (P < .001). Model 2 and Model 3 leak indicator functionality did not differ (P = 1.000). The obstruction indicator, specifically, did not differ between any of the models (P = .122). The rate indicator functioned worse in Model 1 (79.5%) compared to Model 2 (100%) and Model 3 (100%), but a pairwise statistical comparison could not be conducted because at least 1 variable had a 100% success rate. Detailed results for each of the 3 models can be found in Tables 2–4.

Back to Top | Article Outline
Ventilator:

Across all 3 indicators, Model 3’s overall performance (83.3%) was significantly better than that of Model 2 (71.1%) and Model 1 (58.3%) (P < .001). Model 2 was also significantly better overall than Model 1 (P < .001). Model 1’s leak indicator (35.7%), specifically, was significantly lower than that of Model 2 (55.7%) and Model 3 (78.6%) (P < .001). Model 1’s obstruction indicator (66.7%) did not differ from that of Model 2 (55.7%) or Model 3 (71.4%) (P = .121), but Model 3’s obstruction indicator was significantly higher than that of Model 2 (P = .005). The rate indicator functioned worse in Model 1 (72.4%) compared to that in Model 2 (100%) and Model 3 (100%), but a pairwise statistical comparison could not be conducted because at least 1 variable had a 100% success rate. Detailed results for each of the 3 models can be found in Tables 2–4.

Back to Top | Article Outline

Altered Breath Rates.

The AIR showed the correct rate display significantly more frequently in Model 2 (100%) than in Model 1 (33.7%) and Model 3 (78.9%) (P < .01; Supplemental Digital Content, Table S3, http://links.lww.com/AA/B977). Model 3 was also significantly higher than Model 1 (P < .01).

Back to Top | Article Outline

Overall.

On average, the leak indicator took 1.7 ± 1.0 seconds to indicate correctly, while the obstruction indicator took 1.4 ± 0.6 seconds. When displaying an inappropriate leak indicator, the AIR took 2.7 ± 0.7 seconds to display, and when displaying an inappropriate obstruction indicator, the AIR took 2.3 ± 0.9 seconds. Overall, the prototype displayed the correct indicator 73.5% of the time, and when doing so, took 1.6 ± 0.9 seconds.

Back to Top | Article Outline

DISCUSSION

The main findings of this study can be summarized as follows: (1) the AIR’s performance, although acceptable overall, varied among all models and ventilation types, performing best on the NeoNatalie manikin; (2) in the lung model scenarios, the AIR performed poorest in the simulated 2- and 4-kg infant models; (3) the AIR imposed relatively little resistance when introduced into the ventilatory circuit; and (4) ventilator breath parameters did not change significantly when the AIR was in-line.

Back to Top | Article Outline

AIR’s Performance on NeoNatalie

The AIR’s training firmware is designed specifically for NeoNatalie, which is why all 3 indicators were correct for the NeoNatalie trials. However, activating the training mode requires a computer, which is impractical in the field. Future versions of the device should enable users to select either training mode or clinical mode and within the clinical mode, select a body weight. This would allow for more precise detection of leaks, obstructions, and improper breath rates, and thus make the device more accurate and versatile.

Back to Top | Article Outline

AIR’s Performance on Simulated Clinical Models

The AIR performed remarkably well given that it is currently tailored primarily for training purposes with a NeoNatalie manikin, whose lung mechanics greatly differ from the clinical models. The AIR’s “clinical firmware” is still in its developmental stages, yet it managed to display the correct indicator in 73.5% of cases. However, success did vary between scenarios and models; it performed poorest on the 2- and 4-kg infant models.

Back to Top | Article Outline

Normal Conditions.

The AIR’s variation in performance can be explained by its mechanism of operation. The AIR uses sensors to measure flow and pressure, and it integrates flow to obtain volume for each breath. It then fits the measured flow, pressure, and volume data to a system impedance model, and thereby identifies a leak or obstruction. An airway with high resistance and/or low compliance corresponds to an obstruction, whereas a high compliance and/or low resistance corresponds to a leak. Between these extremes exists an OK middle ground in which neither indicator flashes. Model 3’s resistance and compliance values likely fell within the OK range of the AIR programming. Models 1 and 2 have higher resistances and lower compliances, which shift them out of the AIR’s OK range. For this reason, Model 2-Ventilator, Model 2-Bag-Squeezer, and Model 1-Ventilator all displayed the obstruction indicator even in the absence of an obstruction.

Model 1-Bag-Squeezer, however, displayed the leak indicator under normal conditions. This phenomenon cannot be explained by the AIR’s firmware; in fact, it may be arbitrary. We ran poststudy tests and found that the AIR device has a repeatable, demonstrated sensitivity to orientation. At varying orientations, gravity affects the AIR’s membrane pressure sensors differently, so the measured pressure and flow are different if the device is oriented upwards, sideways, or downwards. For high flow and/or high pressure, the sensitivity does not have a significant impact. For low flows or pressures, like those in Model 1, the orientation effect could be significant.

Back to Top | Article Outline

Conditions of Leak, Obstruction, and Leak ± Obstruction.

The AIR functioned best in Model 3, likely because its mechanics fell within the AIR’s OK range, making leak and obstruction detection more accurate. Because Models 1 and 2 were not within range even during normal conditions, the detection of an introduced leak or obstruction was more difficult. However, the leak indicator in Model 1-Bag-Squeezer performed significantly better than the leak indicators of Models 2 and 3-Bag-Squeezer. This finding is difficult to explain, but better calibration of the device for specific weights and flow types should eliminate such variability. Moreover, the percent correct we reported for Models 1 and 2 may be overestimates of the prototype’s true capability. Across the ventilator and Bag-Squeezer, the AIR was correct for 930 of 1260 tests (73.8%) in Model 2, and 249 of those “corrects” were because the obstruction indicator flashed while there was an obstruction. However, the obstruction indicator flashed even during normal conditions, so its presence during an obstruction scenario might be misleading. The rate indicator in Model 1’s tests is also noteworthy because Model 1 was the only model in which the rate indicator faulted during the leak + obstruction scenario. The characteristics of Model 1’s breaths (short duration and low flow) result in fewer data points for the AIR to collect, so the device has trouble reconstructing the entire waveform of every breath. As a result, rate can be measured incorrectly.

Back to Top | Article Outline

Conditions of Altered Breath Rates.

The AIR performed best on Model 2. In Model 1, flow rates were low, which likely affected the AIR’s identification of each breath. In Model 2, flow and pressure were high enough that the AIR could detect each breath and accurately calculate the rate. The poor performance on Model 3 is somewhat perplexing. With higher pressures and flows, the AIR’s rate indicator should have been just as accurate as Model 2. The answer could be tied to auto-PEEP. At 85 bpm, auto-PEEP was about 2.2 cm H2O in Model 3, whereas in Model 2, it was only 0.90 cm H2O. The AIR has minimal PEEP compensation ability and malfunctions if PEEP is too high, so a nearly 145% increase between Model 2 and Model 3 could render the rate indicator inaccurate.

Back to Top | Article Outline

AIR’s Effect on Resistance and Breath Parameters

The results of this study indicate that the AIR marginally affects the resistance of a ventilatory circuit, imposing an average resistance of 5.625 cm H2O/L/s at a flow rate of 10 lpm. Manczur et al22 reported that standard neonatal endotracheal tubes at the same flow rate impose anywhere from 4.6 to 81.2 cm H2O/L/s of resistance, depending on the size of the tube’s inner diameter (2.5–6.0 mm). Our results demonstrate that at a flow rate of 10 lpm, the AIR imposes a resistance similar to that of a standard 5.0-mm neonatal endotracheal tube. Furthermore, the Program for Appropriate Technology in Health reported that the mean inspiratory pressure drop imposed by a NeoNatalie resuscitation bag was 0.73 cm H2O at a flow of 5 lpm.23 At the same flow rate, the AIR imposed a mean pressure drop of only 0.38 cm H2O. Therefore, the 2 together impose a combined pressure drop of just over 1 cm H2O, easily passing the International Organization for Standardization standard of 5 cm H2O. The AIR had an even smaller effect on breath parameters. Across all models, 17 of the 27 parameters evaluated had a <1% change with the AIR in-line. Therefore, providers should not have to provide any compensational effort because there is little, if any, change in VT or pressure.

Back to Top | Article Outline

Clinical Implications

With high rates of intrapartum deaths in LRSs, it is imperative that providers are properly trained. Moreover, providing real-time feedback during resuscitation can likely prevent deterioration in technique. The AIR device could be used for both: first to train care caregivers and eventually to provide real-time feedback during actual BVM resuscitation. However, the device must see significant improvements in its clinical performance before it can be used beyond practice with manikins.

Currently, each prototype is produced individually and with relatively inexpensive materials. Production by a major company could be a gateway to the aforementioned upgrades, making the AIR more cost-effective and versatile. We suggest adding a switch to toggle between neonate body weights. Neonate lung mechanics vary with body weight,24 and “normal” lung mechanics for a large neonate might read as an obstruction for a smaller neonate, which is exactly what we observed. Having multiple body weight settings would eliminate such mistakes and likely result in increased accuracy. The next generation of the AIR should be tested with a representative range of patients (eg, age, weight, physiology) and therapy types (self-inflating bag, flow-inflating bag, t-piece, ICU ventilators during mask ventilation, and after intubation). Initial testing should use an airway simulator under laboratory conditions.

Back to Top | Article Outline

Limitations

There are 4 major limitations to this study. First, the present study tested a prototype device, not the market-ready product. We hope, however, that our findings affect the device’s future development. Second, neonatal models with birth weight under 2 kg were not tested. Third, this study did not evaluate user-device interaction. Additional evaluation of the ability of clinicians to properly use the device is necessary. Finally, this study used a model test lung, not patients, which raises the question of how the device would perform in vivo. To fully validate the AIR’s functionality, clinical studies should be conducted with a version ready for clinical use.

Back to Top | Article Outline

CONCLUSIONS

In summary, the AIR is a promising innovation that could greatly impact neonate mortality. Its presence in-line has little effect on ventilatory circuits, and it performed well on a NeoNatalie manikin, for which it is currently optimized. However, the device’s infrastructure and programming should be improved before it is used clinically. Whether this device has the potential to improve resuscitation technique and whether the device is effective in preventing neonatal death have not been determined but present exciting opportunities for future research and development.

Back to Top | Article Outline

ACKNOWLEDGMENTS

The authors would like to thank Jim Wright (Principal Engineer, BLT Solutions LLC, Boston, MA) and Kevin Cedrone, PhD (Global Health, Massachusetts General Hospital, Boston, MA) for serving as technology liaisons and Dr Kristian Olson, MD (Pediatrics, Massachusetts General Hospital, Boston, MA) as a clinical liaison.

Back to Top | Article Outline

DISCLOSURES

Name: Desmond J. Bennett, BA.

Contribution: This author helped design the study, collect and analyze the data, and compose the manuscript.

Conflicts of Interest: None.

Name: Taiga Itagaki, MD.

Contribution: This author helped design the study, analyze the data, and compose the manuscript.

Conflicts of Interest: None.

Name: Christopher T. Chenelle, BA.

Contribution: This author helped design the study, analyze the data, and compose the manuscript.

Conflicts of Interest: None.

Name: Edward A. Bittner, MD, PhD.

Contribution: This author helped analyze the statistical data and compose the manuscript.

Conflicts of Interest: None.

Name: Robert M. Kacmarek, PhD, RRT.

Contribution: This author helped design the study, collect and analyze the data, and compose the manuscript.

Conflicts of Interest: R. M. Kacmarek has received research grants from Venner Medical and Covidien and is a consultant for Covidien, Orange Medical, and Teleflex.

This manuscript was handled by: James A. DiNardo, MD, FAAP.

Back to Top | Article Outline

REFERENCES

1. Lawn JE, Cousens S, Zupan J; Lancet Neonatal Survival Steering Team. 4 million neonatal deaths: when? where? why? Lancet. 2005;365:891–900.
2. UNICEFThe State of the World’s Children 2009: Maternal and Newborn Health. 2009.New York, NY: UNICEF.
3. World Health Organization. The World Health Report: 2005: Make Every Mother and Child Count. 2005.Geneva, Switzerland: World Health Organization.
4. Lawn JE, Lee AC, Kinney M, et al. Two million intrapartum-related stillbirths and neonatal deaths: where, why, and what can be done? Int J Gynaecol Obstet. 2009;107(suppl 1):S5–S18, S19.
5. Lee AC, Cousens S, Wall SN, et al. Neonatal resuscitation and immediate newborn assessment and stimulation for the prevention of neonatal deaths: a systematic review, meta-analysis and Delphi estimation of mortality effect. BMC Public Health. 2011;11(suppl 3):S12.
6. Zhu XY, Fang HQ, Zeng SP, et al. The impact of the Neonatal Resuscitation Program Guidelines (NRPG) on the neonatal mortality in a hospital in Zhuhai, China. Singapore Med J. 1997;38:485–487.
7. Wall SN, Lee AC, Niermeyer S, et al. Neonatal resuscitation in low-resource settings: what, who, and how to overcome challenges to scale up? Int J Gynaecol Obstet. 2009;107(suppl 1):S47–S62, S63S64.
8. Berkelhamer SK, Kamath-Rayne BD, Niermeyer S. Neonatal resuscitation in low-resource settings. Clin Perinatol. 2016;43:573–591.
9. Hedstrom A, Ryman T, Otai C, et al. Demographics, clinical characteristics and neonatal outcomes in a rural Ugandan NICU. BMC Pregnancy Childbirth. 2014;14:327.
10. Bang A, Patel A, Bellad R, et al. Helping Babies Breathe (HBB) training: what happens to knowledge and skills over time? BMC Pregnancy Childbirth. 2016;16:364.
11. Patel J, Posencheg M, Ades A. Proficiency and retention of neonatal resuscitation skills by pediatric residents. Pediatrics. 2012;130:515–521.
12. Carlo WA, Wright LL, Chomba E, et al. Educational impact of the neonatal resuscitation program in low-risk delivery centers in a developing country. J Pediatr. 2009;154:504–508.e5.
13. Adam T, Lim SS, Mehta S, et al. Cost effectiveness analysis of strategies for maternal and neonatal health in developing countries. BMJ. 2005;331:1107.
14. Darmstadt GL, Bhutta ZA, Cousens S, et al.; Lancet Neonatal Survival Steering Team. Evidence-based, cost-effective interventions: how many newborn babies can we save? Lancet. 2005;365:977–988.
15. Lawn JE, Kerber K, Enweronu-Laryea C, Massee Bateman O. Newborn survival in low resource settings–are we delivering? BJOG. 2009;116(suppl 1):49–59.
16. Cedrone K, Mielcarz C, Data S, Olson K. Augmented Infant Resuscitator (AIR). 2013.MIT IDEAS Global Challenge.
17. International Organization for Standardization. Lung Ventilators—Part 4: Particular Requirements for Operator-Powered Resuscitators. ISO 10651–4:2002(E). 2002.Geneva, Switzerland: International Organization for Standardization.
18. Marchese AD, Chipman D, de la Oliva P, Kacmarek RM. Adult ICU ventilators to provide neonatal ventilation: a lung simulator study. Intensive Care Med. 2009;35:631–638.
19. Oto J, Chenelle CT, Marchese AD, Kacmarek RM. A comparison of leak compensation during pediatric noninvasive ventilation: a lung model study. Respir Care. 2014;59:241–251.
20. Gerhardt T, Hehre D, Feller R, Reifenberg L, Bancalari E. Pulmonary mechanics in normal infants and young children during first 5 years of life. Pediatr Pulmonol. 1987;3:309–316.
21. Itagaki T, Chenelle C, Bennett D, Fisher D, Kacmarek R. Effects of leak compensation on patient-ventilator synchrony during premature/neonatal invasive and noninvasive ventilation: a lung model study. Respir Care. 2017;62:22–33.
22. Manczur T, Greenough A, Nicholson GP, Rafferty GF. Resistance of pediatric and neonatal endotracheal tubes: influence of flow rate, size, and shape. Crit Care Med. 2000;28:1595–1598.
23. Program for Appropriate Technology in Health (PATH). A Catalyst For Global Health. Testing of an Operator-Powered Neonatal Resuscitator. 2012.
24. Gagliardi L, Rusconi F. Respiratory rate and body mass in the first three years of life. The working party on respiratory rate. Arch Dis Child. 1997;76:151–154.

Supplemental Digital Content

Back to Top | Article Outline
Copyright © 2017 International Anesthesia Research Society