Evaluation of a Pneumatic Vest to Treat Symptoms of ARDS Caused by COVID-19

With the onset of the COVID-19 (SARS-CoV-2) pandemic in 2019, patients admitted to intensive care units (ICUs) with acute respiratory distress syndrome (ARDS) worldwide have substantially increased from previous years [1–3]. This dramatic increase in ARDS patients has overwhelmed many hospitals across the globe. Pulmonary inflammation and subsequent edema due to ARDS prohibits proper oxygenation and causes patients to exhibit low partial pressure of oxygen in their bloodstream. Roughly 25% of patients admitted to hospitals with COVID-19 develop ARDS and are transferred to ICUs [4]. The mortality rate of patients with ARDS caused by COVID-19 treated with conventional therapy may be upwards of 23% but drastically varies between studies [5,6].

ARDS patients admitted to hospital ICUs are typically placed on mechanical ventilators if their condition is severe enough. Even on mechanical ventilation, some patients with ARDS cannot recover from hypoxemia since the mechanical ventilator can only provide oxygen to the lungs and depends on the pulmonary system to uptake this oxygen [7]. Ventilation through alveoli is decreased, wherever there is inflammation while perfusion remains normal, leading to a decrease in the ventilation–perfusion ratio. This is referred to as a ventilation–perfusion mismatch (V/Q mismatch) and leads to lower overall blood oxygenation levels [8]. The common metric used to determine a V/Q mismatch is the P/F ratio, the partial pressure of oxygen in the arterial blood (⁠$PaO2$⁠) normalized by the fraction of inspired oxygen (⁠$FiO2$⁠) [9]. ARDS is graded in severity of mild, moderate, and severe with P/F ratios of 200 to 300, 100 to 200, and less than 100 mmHg respectively [10].

There are low-cost, conventional techniques utilized to manage the pathophysiology of ARDS experienced by mechanically ventilated patients. The first conventional method is to increase the $FiO2$ delivered to the patient and/or raise the mean airway pressure through higher positive end expiratory pressure. This increase in mean airway pressure has been shown to increase ventilation of inflamed alveoli, or typically termed, recruiting alveoli to participate in the gas exchange process [11]. Another method is to “prone” the patient, meaning to flip the patient from a supine position to a prone position. Since the systolic pulmonary artery pressure of humans is around 20 to 25 mmHg, the pressure potential energy produced by gravity can significantly affect where perfusion is highest [12]. When someone is in the prone position, perfusion pressure is decreased in the posterior portions of the lungs and increased in the anterior portions of the lungs. Proning allows hospital staff to manipulate this pressure difference caused by gravity to increase perfusion to areas where there is increased ventilation and decrease perfusion where there is decreased ventilation. However, the proning maneuver can take up to seven hospital staff to safely carry out and increases the risk of complications [13,14].

More expensive, nonconventional solutions for overcoming a V/Q mismatch are the Rotoprone and Extracorporeal Membrane Oxygenation (ECMO) machines [15,16]. The Rotoprone machine is a rentable device that can rotate patients to the ideal position instead of having hospital staff members perform a proning maneuver. These machines can cost about US$1000 to rent every day and are in limited supply [13]. Furthermore, Rotoprone machines are tedious, still requiring multiple hospital staff members to place a patient into one safely. Rotoprone machines also limit access to the patient for physical examination, blood draws, and other procedures. ECMO systems take blood from patients, oxygenate the blood, and then return it to the patient. In essence, performing the vital function of the lungs. The average cost for an ECMO procedure was investigated in 2006 to be about US$73,000, not including the pre- and post-ECMO procedures [17].

Another more current research technique for improving the P/F ratio in patients with ARDS is to increase chest wall elastance (stiffness) [18]. This primitive technique is to simply place weights onto patients' chests while they are in the supine position [19]. The resultant effect of this method is an increase in mean airway pressure of the patients while mitigating overdistension which has been shown to increase the alveolar recruitment [20]. The change in chest wall elastance from this method is only local to where the weight is placed and, therefore, is not efficient at decreasing the overall distension of patients' lungs.

With this new technique in mind, the V/Q Vest was developed to capture and enhance the effects of the chest weight experiments by decreasing static lung compliance. Figure 1 shows a picture of the V/Q Vest on a researcher. The V/Q Vest has many benefits over previous methods for improving ARDS patients' condition. The V/Q Vest can be more easily applied to patients than performing a proning maneuver, takes minimal training to control and monitor, is extremely cost-effective to manufacture (about US$200), and can be manufactured on a larger scale than nonconventional treatments for a V/Q mismatch. It is hypothesized that the V/Q Vest can be used as a surrogate for other anterior chest wall compression devices and proning by adjusting the pressure the V/Q Vest applies to the chest wall of patients. It is also hypothesized that the V/Q Vest could be used to determine how a patient will respond to proning. The V/Q Vest will be evaluated based on patient performance compared with proning these same patients.

The V/Q Vest was designed to alter lung mechanics and potentially impact oxygenation, similar to the effects of weights being placed on patients' chests. To do this, V/Q Vest uses air-tight bladders that can be inflated to the desired pressure. This internal pressure then causes the bladders of the vest to expand, and with the restrictive elastic band that is wrapped around the patients, imparting pressure onto the thorax and upper abdomen of patients. The following section has been presented in previous work but is restated here for clarity to show how the V/Q Vest applies pressure to patients [21].

The V/Q Vest was designed to accommodate a variety of patient sizes by facilitating use on patients with chest sizes ranging from 36 to 52 in. in circumference. The V/Q Vest was constructed from multiple layers of 200 denier thermoplastic polyurethane-coated nylon that create independently controlled airtight bladders. The polyurethane-coated nylon sheets were cut to shape using a Universal Laser Systems PLS6.150D Laser Cutter. The cut polyurethane sheets were sealed together using an H-306 tabletop impulse sealer from ULINE Inc to create the bladders. The impulse sealer melts the polyurethane layers of the fabric together forming airtight seals.

As the stiffness of the restrictive element increases, the ratio of pressure imparted on the patient and the internal pressure of the bladder grow closer together. With higher stiffness, the bladder walls also do not separate as much (h approaches 0). This causes the membrane stress that detracts from the pressure imparted to the patient (⁠$σmem$⁠) to diminish to zero. The V/Q Vest sacrifices a high stiffness restricting band for a lower stiffness element that can stretch to accommodate a large variety of patient sizes as demanded by COVID-19 patient demographics [22]. Another reason for the V/Q Vest to use a restrictive elastic band is to minimize the varying $Ppat$ throughout the breathing cycle. If a high stiffness restrictive element was used, the pressure imparted on the patient would drastically vary as the thoracic cavity expands and contracts.

The inclusion criteria for this study were patients with age > 18 years, presence of ARDS due to COVID-19, currently intubated patients on ventilator support, and authorized representative's ability to provide informed consent. The presentation of ARDS was defined by the 2012 Berlin Criteria. Nine patients with ARDS caused by COVID-19 gave informed consent to participate in this study. Both institutional review boards at Emory University and the Georgia Institute of Technology approved this study. All the patients were admitted to the Acute Respiratory Intensive Care Unit at the Emory University Hospital. The relevant demographics of these nine patients are shown in Table 1.

Since the tidal volume and PEEP are controlled by the mechanical ventilator, the only effect on $Cstat$ comes from the plateau pressure. It was expected that the vest treatment or proning would decrease lung compliance by stiffening the thorax expansion of patients. P/F ratios here were calculated by simply dividing $PaO2$ by $FiO2$⁠.

Two studies, the main study (n = 6) and the substudy (n = 3), are shown in this work. Figure 3 shows an overview of the study flow. The goal of the main study was primarily to see if the V/Q Vest with all the bladders inflated to the same pressure could:

• improve patients' condition similarly to proning or other treatments

• be used to predict whether patients would respond well to proning

• decrease patients' static lung compliance

The first phase of the main study was the control for all six of the patients. All patients were placed in the supine position and on mechanical ventilation. Their vitals were recorded for an hour after their vitals stabilized. During the second through the fourth phase of the main study, patients 1–6 had the V/Q Vest applied while still on mechanical ventilation. The hospital staff increased the V/Q Vest pressure between these three 1-h long trials (10 mmHg for the first hour, 20 mmHg for the second hour, and 40 mmHg for the third hour). Every bladder of the V/Q Vest in this main study was inflated to the same pressure. At the end of each hour, their vitals were recorded before engaging in the next V/Q Vest treatment pressure. There were two V/Q Vest failures where pressure was not able to be held for the duration of the study and these two patients were excluded from the study. The last phase of the main study was to remove the V/Q Vest from the patient and prone the patient while still on mechanical ventilation only if proning was clinically feasible. Patients 3, 4, 7, and 9 were not able to be safely proned due to clinical reasons or were not proned since proning was done and minimal effects were seen.

The substudy was performed to determine how the location of the V/Q Vest pressure affects:

• patients' response

• patients' static lung compliance

Patients 7–9 were subjects in this substudy. This substudy had the same control and proning phase as the main study. The middle four phases were using the V/Q Vest to apply pressure to specific locations of the patient while still on mechanical ventilation. For the first location, only the anterior bladders were inflated to 30 mmHg. The second was only inflating the posterior bladders to 30 mmHg. The third was only inflating the anterior or posterior chambers to 40 mmHg depending on which side had the better response in the first two stages. For the last stage of the substudy, the anterior or posterior-superior bladders used in stage three were deflated to 30 mmHg while the rest of the bladders used in stage three remained at 40 mmHg.

The first iteration of the V/Q Vest was used only in the main study of the work presented here.

The first iteration of the V/Q Vest had eight independent bladders, but for the main study, these eight bladders were always inflated to the same pressure. This iteration of the V/Q Vest had issues with the welds tearing around the interior corners of the anterior and posterior bladders shown in Fig. 4. As a result, the V/Q Vest treatment was immediately stopped, and these patients' data were excluded from the analysis. We also observed that the lower bladders on the anterior of the vest were not applying as much pressure as the upper bladders. This is thought to happen because the cross-sectional area of the lower bladders is much smaller. These two issues with the first iteration of the V/Q Vest are why the design was modified.

An electro-pneumatic controller was fabricated to autonomously regulate the internal pressure of the independent bladders that form the V/Q Vest. The controller senses and regulates the internal pressure of all the bladders at 100 Hz. The pressure of each bladder was controlled using two solenoid valves in series, one for inflation and one for deflation. A proportional-integral-derivative (PID) controller was implemented to drive the internal pressure of the bladders to a specified internal pressure set by the hospital staff. A graphical user interface (GUI) was developed to simplify the control of the V/Q Vest for the hospital staff. The GUI allowed the hospital staff to independently control each bladder and monitor the internal pressure of the V/Q Vest in real-time. For further safety of the patient, an emergency stop was implemented that cuts power to the controller and vents all the bladders leading to rapid deflation.

Table 2 shows the patients' $FiO2$ and $PaO2$ levels in mmHg during each phase of the main study. Patients' $FiO2$ was held constant throughout every trial. Four of the six patients exhibited an increase in $PaO2$ levels while wearing the V/Q Vest. It is surmised that patient 2 did not see an increase in $PaO2$ levels from the V/Q Vest due to a high body mass index (BMI) or excess soft tissue. It is hypothesized that excess thoracic tissue disperses the pressure applied to their chest wall away from the lungs.

While $PaO2$ is a good measure to ensure that patients are getting the oxygenation needed to remain in stable condition, this measure alone does not reflect the patients' efficiency at infusing oxygen from their environment into their bloodstream, so the P/F ratio is analyzed. Figure 5(a) shows the P/F ratios for each patient in the main study. Healthy P/F ratios are between 400 and 500 at sea level on atmospheric $FiO2$⁠. The P/F ratios for all but two participants increased while wearing the V/Q Vest. For patients 5 and 6, proning was more effective at raising their P/F ratios than wearing the V/Q Vest. Patient 2 was morbidly obese and patient 5 had an extensive number of comorbidities. These factors are hypothesized to have led to the decreased effectiveness of the V/Q Vests.

A correlation between the V/Q Vest pressure and the change in P/F ratio compared to the control trial was analyzed with a linear correlation. The resulting correlation coefficient was found to be 0.94 (p = 0.22). Indicating that correlation between the V/Q vest pressure and the increase in patients' P/F ratio is uncertain. This result is accepted since there is evident intersubject variability due to varying severity of ARDS and comorbidities. However, comparing each participant on their own shows that four of the six patients experienced increases in P/F ratios due to the V/Q Vest. In two out of these four patients, the increase in P/F ratios due to the V/Q Vest treatment was higher than their P/F ratios experienced while proned.

The last metric analyzed in this study was the static lung compliance of the patients. Figure 5(b) shows the static lung compliance of patients 1–6. Static lung compliance is a measure for determining the sensitivity of the lung to alterations in pressure and is a factor in how hard the person must breathe under their own power. Average lung compliance for patients with ARDS is found to be less than 50 mL/$cmH2O$ and a healthy adult would have static lung compliance above 150 mL/$cmH2O$ [23–25]. All the patients in the main study had static lung compliances less than 50 mL/$cmH2O$⁠. With the V/Q Vest applied, static lung compliance was observed to decrease depending on the amount of pressure applied to the patients' chest wall only for a few patients. Again, no correlation was found between V/Q Vest pressure and static lung compliance when grouping all patients together. However, when comparing each patient individually, the V/Q Vest (inflated to at least one of the three pressures used in this study) decreased static lung compliance in every patient. The pressure required to reduce static lung compliance varied between participants, which may be why a correlation between static lung compliance and vest pressure was not found when grouping all patients together.

The mechanical ventilation was controlled to keep the tidal volume and PEEP constant for each patient throughout the study. This would suggest that decreasing lung compliance using the V/Q Vest, proning, or placing weights on patients' chest causes patients' plateau pressure to increase. Typically, this increase in plateau pressure would increase alveolus distension, but with the V/Q Vest, the alveolus overdistension is likely reduced due to the pressure the vest imparts on patients' thoracic cavities. Therefore, it is hypothesized that the V/Q Vest may also be useful in reducing the risk of lung injury that is caused by an increase of driving pressure from mechanical ventilation [26].

The second iteration of the V/Q Vest was used in the substudy of the work presented here.

The second iteration of the V/Q Vest only had four independent bladders. This iteration of the V/Q Vest still had upper and lower chambers on both the anterior and posterior of the vest, but the upper and lower chambers were connected, resulting in only two independent bladders on both the anterior and posterior of the V/Q Vest. This design was instituted to avoid interior corners that caused failures in the first iteration of the V/Q Vest. Figure 6 shows the second iteration of the V/Q vest. The second iteration of the V/Q Vest used the same electro-pneumatic controller from the first iteration. Only four of the eight electro-pneumatic control systems were used, and an updated GUI was implemented. The second iteration also had separate pneumatic connectors for the pressure sensors. It was observed from the first iteration that having the pressure sensors connected to the tubes that inflated and deflated the bladders resulted in pressure instabilities. As airflow passed by the tubes connected to the pressure sensors, the static pressure measured by the sensors dropped causing instabilities in the PID control algorithm. These extra connectors were implemented to minimize the airflow past the pressure sensors so that they would more precisely sense the static pressure of the bladders of the V/Q Vest.

Table 3 shows the patients' $FiO2$ and $PaO2$ levels in mmHg during each phase of the substudy. Patients' $FiO2$ was held constant throughout every trial. Two of the three patients experienced an increase in $PaO2$ and P/F ratios while wearing the V/Q Vest. Patients 7 and 9 were proned before the study with no apparent response, but the V/Q Vest was successful in increasing their P/F ratios. Patient 8 was able to be proned and their increase in P/F ratio while proned was higher compared with the V/Q Vest treatment. For patient 8, the V/Q Vest did not help the hospital staff correctly predict the response from the patient while proned. Table 4 and Fig. 7(a) show the changes in the P/F ratios compared to the control trial with respect to where pressure was applied. No evident correlation was apparent in this small sample size substudy. However, the substudy did concur with the evidence from the main study that static lung compliance was decreased during the V/Q Vest treatment as shown in Fig. 7(b). Only patient 7 did not experience a decrease in static lung compliance for anterior V/Q Vest treatment.

After our studies with the original electro-pneumatically controlled vests, a redesign is to be implemented before continuing with the rest of the clinical pretrials. With many malfunctions of the V/Q Vests used in the studies, a new more robust design needs to be implemented and produced on a larger scale for continued clinical trials and commercialization. Comments about how the V/Q Vest was donned resulted in an updated design of the V/Q Vest where there is only an anterior portion. This change will have additional benefits by speeding up manufacturing and further decreasing the cost of the V/Q Vest. It is not apparent that there is a need for a posterior section of the V/Q Vest. The substudy shows that there is no apparent difference between applying pressure to only the anterior of the patient and anterior along with the posterior of the patient. The posterior ribs are more rigid than the anterior ribs, so it is thought that this pressure imparted on the posterior has negligible effects on the patient. For these reasons, the posterior of the V/Q Vest is to be discarded going forward.

Another reason for this redesign is to further lower the cost of the device. For this third iteration of the V/Q Vest, the electro-pneumatic controller will also be discarded and replaced with modified sphygmomanometers (blood pressure cuff devices). These sphygmomanometers are more reliable than the electro-pneumatic controller and are significantly cheaper. Sphygmomanometers that are calibrated can measure pressures between 10 mmHg and 300 mmHg with a precision of ±5 mmHg [27]. For the third iteration of the V/Q Vest, the manufacturing process will also become more cost-effective. For the third iteration, the V/Q Vest welds will be made using a radio frequency (RF) welding process, which scales the manufacturing process better than welding by hand. The iteration of the V/Q Vest system is projected to reduce the overall cost of manufacturing from over US$2000 to less than US$200. This new iteration will be used on the next twenty patients that will be enrolled in this pilot study over the next year.

After this pilot study is finished with the third iteration of the V/Q Vest, a future study will investigate whether the V/Q Vest effectiveness is consistent between patient sex and BMI. With the current shape of the vest, it is hypothesized that soft tissue, such as breast tissue, disperses the pressure imparted on users' chest wall which caused less of an impact seen in the female patient and obese patients in this pilot study. This future study will investigate this hypothesis and may lead to another V/Q Vest design iteration before proceeding to clinical trials.

The V/Q Vest improved P/F ratios and improved hypoxemia among four out of the six patients who participated in the main study. Two patients had higher P/F ratios with the V/Q Vest compared to proning. Conversely, two patients showed lower P/F ratios with the V/Q Vest compared to proning. The remaining two patients in the main study were not able to be proned; however, the V/Q Vest was able to be used and increased one of these patients' P/F ratios. The substudy presented in this work was used to determine whether the location of pressure applied by the V/Q Vest affects patient responses. The substudy showed that there was neither statistically nor clinically significant difference in patient responses depending on the location of applied pressure. For this reason and for decreasing the cost of the V/Q Vest going forward, a new design was discussed that is projected to decrease the cost of each V/Q Vest system by a factor of ten. The results from both studies presented here show that the V/Q Vest did decrease the static lung compliance of all patients. It is hypothesized that decreasing static lung compliance while holding tidal volume and PEEP constant increases the plateau pressure of the patients, which may help in alveolus recruitment. It is also hypothesized that the V/Q Vest reduces overdistension of alveolus which could decrease the risk of lung injury imposed by higher plateau pressure treatments.

More clinical testing will be conducted to differentiate the differences in effects of the V/Q Vest, weights placed on patients' chests, and proning. The V/Q Vest was successful at determining which patients would benefit from proning in all of the main study cases except for patient 5 based on analyzing the P/F ratios of patients. One important takeaway from this work is the fact that the V/Q Vest can help improve ventilation-perfusion for patients that cannot be safely proned, which may decrease the mortality rate of ARDS.

The authors would like to thank Dr. Roberta Kaplow at the Emory University Hospital for their help in conducting the preclinical research. The authors would also like to thank Harold Solomon with the Venture Lab at Georgia Tech and Mike Fisher at GCMI for their crucial support throughout this research.

• Emory University School of Medicine I3 grant (Funder ID: 10.13039/100007623).

• Georgia Tech EVPR COVID-19 Rapid Response Seed Grant (Funder ID: 10.13039/100006778).

• Georgia Research Alliance (GRA) (Funder ID: 10.13039/100008065).

 
• $Amem$ =

  cross-sectional area of the membrane of the bladder

 
• $Aint$ =

  cross-sectional area of the bladder of the V/Q Vest

 
• $Apat$ =

  contact area between the patient and the bladders of the V/Q Vest

 
• $Cstat$ =

  static lung compliance in mL/$cmH2O$—inverse of the stiffness of the thoracic cavity

 
• $FiO2$ =

  fraction of inspired oxygen—amount of oxygen in the air supplied to patient

 
• $Pint$ =

  internal bladder pressure of the V/Q Vest

 
• $Pplt$ =

  plateau pressure in $cmH2O$—air pressure that mechanical ventilation applies to the alveoli and small airways

 
• $PaO2$ =

  partial pressure of oxygen in arterial blood in mmHg

 
• PEEP =

  positive end-expiratory pressure in $cmH2O$—pressure of the airways at the end of the exhalation

 
• $P/F$ =

  partial pressure of oxygen in arterial blood normalized by fraction of inspired oxygen

 
• V/Q =

  ventilation-perfusion—the ratio between oxygen supplied by alveoli to the blood flow through capillaries

 
• $Vtid$ =

  tidal volume in mL—volume of air moved in and out of lungs

 
• $Ppat$ =

  pressure imparted on the patient by the bladders of the V/Q Vest

 
• $σmem$ =

  membrane stress of the bladder that contributes to pressure imparted on the patient