key: cord-0880067-krly70iu authors: Guerin, C.; Cour, M.; Stevic, N.; Degivry, F.; L'Her, E.; Louis, B.; Argaud, L. title: Sharing ventilators in the Covid-19 pandemics. A bench study date: 2020-11-04 journal: nan DOI: 10.1101/2020.11.02.20224774 sha: 6b654239b95583396295a753d0d12717966d4cd7 doc_id: 880067 cord_uid: krly70iu COVID-19 pandemics sets the healthcare system to a shortage of ventilators. We aimed at assessing tidal volume (VT) delivery and air recirculation during expiration when one ventilator is divided into 2 patients. The study was performed in a research laboratory in a medical ICU of a University hospital. An ICU-dedicated (V500) and a lower-level ventilator (Elisee 350) were attached to two test-lungs (QuickLung) through a dedicated flow-splitter. A 50 mL/cmH2O Compliance (C) and 5 cmH2O/L/s Resistance (R) were set in both A and B lungs (step1), C50R20 in A / C20R20 in B (step 2), C20R20 in A / C10R20 in B (step 3), and C50R20 in A / C20R5 in B (step 4). Each ventilator was set in volume and pressure control mode to deliver 0.8L VT. We assessed VT from a pneumotachograph placed immediately before each lung, rebreathed volume, and expiratory resistance (circuit and valve). Values are median (1st-3rd quartiles) and compared between ventilators by non-parametric tests. Between Elisee 350 and V500 in volume control VT in A/B patients were 0.381/0.387 vs. 0.412/0.433L in step 1, 0.501/0.270 vs. 0.492/0.370L in step 2, 0.509/0.237 vs. 0.496/0.332L in step 3, and 0.496/0.281 vs. 0.480/0.329L in step 4. In pressure control the corresponding values were 0.373/0.336 vs. 0.430/0.414L, 0.416/0.185/0.322/0.234L, 0.193/0.108 vs. 0.176/0.092L and 0.422/0.201 vs. 0.481/0.329L, respectively (P<0.001 between ventilators at each step for each volume). Rebreathed air volume ranged between 0.7 to 37.8 ml and negatively correlated with expiratory resistance in steps 2 and 3. The lower-level ventilator performed closely to the ICU-dedicated ventilator. Due to dependence of VT to C pressure control should be used to maintain adequate VT at least in one patient when C and/or R changes abruptly and monitoring of VT should be done carefully. Increasing expiratory resistance should reduce rebreathed volume. During the COVID-19 pandemic, a risk of a shortage of ICU ventilators was claimed very early [1] . As the poliomyelitis pandemic prompted the caregivers to discover tracheotomy, iron lung, and mechanical ventilation, the current COVID-19 pandemic prompted innovative solutions [2] . They include ventilator multipliers, portable and open-source designs of ventilators [3] , and frugal ventilators [4] . Sharing ventilation provides ventilatory support to two or more patients with the same ventilator [5] . This approach raised ethical issues [6] due to the many technical problems to solve from sharing the same ventilator with patients with different respiratory mechanics and, hence different requirements [7] . For multiplex ventilation, with no means of independently controlling positive end-expiratory pressure (PEEP) and tidal volume (VT), patients sharing the same ventilator should have respiratory mechanics as similar as possible. In this case, in volume control ventilation and pressure control ventilation mode, each patient is expected to receive half of the set VT. Any decrease in compliance and/or increase in resistance in one patient will decrease VT in each mode [8, 9] . For the other patient with unchanged compliance and resistance, VT will depend on a part on mechanical ventilation mode. The degree to which differences in mechanics cause differences in tidal volume depends both on how the mechanics differ and on the mode of ventilation (ie volume versus pressure control), but in general, changes in one patient cause changes in the other patient regardless of the model [8] . Moreover, air may recirculate during inspiration and expiration from one patient to the other, with the risk of CO2 retention and cross-transmission of infection. The patient with the shortest inspiratory time constant, i.e. with the lowest product of resistance by compliance, breathed out faster (earlier) while the other was still filling in. However, the role of expiratory resistance (circuit and ventilator valve) on rebreathed air has not been previously addressed. Nevertheless, the feasibility and safety of ventilator sharing have been reported recently in a few patients highly selected, deeply monitored, and for a few hours [10] [11] [12] . All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224774 doi: medRxiv preprint Because ventilator sharing is still experimental and not completely investigated we designed a bench study where 2 or 3 patients with different respiratory mechanics were attached to the same ventilator. We compared a high-performance ICU ventilator and a ventilator used for patient transportation. We assessed VT delivery, rebreathed volume, and expiratory resistance. Two ventilators were tested: the Elisée 350 (ResMed, Saint-Priest, France) a turbine-driven ventilator used for patient transportation, and in stepdown-units and the V500 ICU ventilator (Drager, Lubeck, Germany). They were attached to two QuickLung tests (IngMar Medical, Inc., Pittsburgh, PA) equipped with resistance (R) of 5, 20, and 50 cmH2O/L/s and compliance (C) of 10, 20, and 50 mL/cmH2O in the first part of the experiment. In the second part a third lung test (SelfTestLung, Draeger, Lubeck, Germany) of 10 mL/cmH2O C and 20 cmH2O/L/s ± was added (Table 1) Two designs were used for each ventilator in both volume and pressure control ventilation (Table 1) . Two Lungs design 1 C50-R5 C50-R5 none 2 C50-R20 C20-R20 none 3 C20-R20 C10-R20 none 4 C50-R20 C20-R5 none C: compliance in ml/cmH2O, R: resistance in cmH2O/L/s In the first part, the two QuickLung models were arranged in parallel (Fig 1) and 4 steps performed. In step 1a similar C50R5 was applied to both patients, to replicate type L COVID-19-related acute respiratory distress syndrome (ARDS) [13] . The 3 other steps had contrasted time constants: step 2 (C50R20 vs. C20R20), step 3 (C20R20 vs. C10R20), and step 4 (C50R20 vs. C20R5) ( Table 1 ). In volume control, VT was set to 0.800 L to deliver 0.4 L to each patient, i.e. 6 mL/kg VT for the average predicted body weight in patients with ARDS in the Lung Safe study [14] , PEEP to 15 cmH2O, a value chosen to stretch the ventilator, and inspired oxygen fraction (FIO2) to 21%. In pressure control, ventilator was set to get a VT of 0.8 L in step1. This setting was kept for the other steps. The other ventilator settings are displayed in In pressure control, the pressure was adjusted to deliver a VT of 0.800 L and settings were unaltered because the steps of the experiment (Table 1) were thought to reflect the time course of asymmetrical disturbance that may occur between patients attached to the same ventilator, i.e. sudden loss of lung volume in one patient (pneumothorax, atelectasis), and hence the ventilator has no reason to be set differently before the situation occurs. Such a double circuit design is also suitable at the time of patient selection for parallel ventilation. The safety guard is to share the single ventilator between patients with respiratory mechanics as close as possible [7, 19] . We wanted to explore how much VT would depart from each patient to the other when the respiratory mechanics markedly differ between them. In the second design, a third lung with a fixed C20-R20 was added in parallel to the previous C-R (Table 1) Collected data were analyzed off-line by using in-house software specifically developed for the present study (Matlab R2019b, MathWorks, Inc.). VT was obtained by integration of the flow signal over the inflation time in each lung, which can be different between each lung (see below) and different from the machine inflation time Taking care of this an important in design, like our present one, that accommodates time constants differences between lungs. The rebreathed volume was computed as the amount of air that flowed from one patient to the other(s) during inspiration and expiration ( Figure 2 ). All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224774 doi: medRxiv preprint At the time of the ventilator end of insufflation, ̇= 0, and ̇= −̇. PalvA is equal to Paw minus PresA (which is the product of resistance through lung A to ̇ and similarly PalvB is equal to Paw plus PresB (which is the product of resistance through lung B to ̇) . It comes that PalvB>Paw>PalvA. At the time of patient A end of inspiration, ̇= 0, ̇=̇, ̇< 0. Therefore, PalvA=Paw and PalvB is equal to Paw minus PresB (which is the product of the resistance through lung B to ̇) . PalvB is then greater than Paw, which is equal to PalvA. The perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224774 doi: medRxiv preprint On each breath, the instantaneous expiratory resistance was determined as the ratio of the pressure drop between Paw and atmosphere to the corresponding flow as previously described [21] . For this computation, we discarded flows lower than 0.01 L/S to avoid extreme values corresponding to the closing of the valve. Therefore, the instantaneous expiratory resistance was determined in roughly 400 instances in each breath. We used the minimal value of the instantaneous resistance in each condition. The primary end-point was the value of VT and secondary end-points were rebreathed volume and minimal instantaneous expiratory resistance. Values are presented as median (1st-3rd quartiles) and compared across ventilators for each C-R condition by using a non-parametric test. The statistically significant level was set to P-value < 0.05. The analysis was performed by using R software Version 3.5.2 (R: A Language and Environment for Statistical Computing, R Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2018). As expected for the step1, VT was equally delivered to patients A and B in both Volume and perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224774 doi: medRxiv preprint (orange), and C (grey) in volume and pressure control ventilation during the 4 steps for each ventilator in the design with two (panels A and B) and three patients (panels C and D). Step1 = C50-R5 for each patient, Step2 = C50-R20 in A and C20-R20 in B, Step3 = C20-R20 in A and C10-R20 in B, Step4 = C50-R20 in A and C20-R5 in B. C25-R20 for patient C at each step. Bars are median and quartiles omitted for clarity. The horizontal broken black lines indicate the target tidal volume the ventilator should deliver to each patient. *P<0.05 as compared to V500. In step 2 C was 2.5 times greater in patient A than in B and R was similar in both of them but 4 times greater than in step 1. In volume control, VT was greater in patient A than in patient B by a factor of 1.9 with Elisée 350 and 1.33 with V500 ( Figure 3 . Therefore, VT delivery was 20% greater in patient A and 33% lower in patient B than expected with Elisée, these values being of 23% and 8%, respectively, with V500. A similar figure was observed in step 3 where C was 2 times greater in patient A than in patient B but 2.5 and 5 times lower, respectively, as compared to step 1, and R was similar in both lungs and similar to in step 3 ( Figure 3 ). The same was true for step 4 ( Figure 2 ). Contrary to step 1, in steps 2-4 VT to patient A was greater with Elisée 350 than with V500 and the opposite was true for VT to lung B (Figure 3 ). In pressure control with the asymmetrical design, VT in a given patient changed as a result of both the overall VT decrease due to the greater impedance and the difference between time constants. In step 2, VT was greater in patient A than in B by a factor of 2.2 with Elisée 350 and of 1.4 with V500 ( Figure 3 ). However, patient A accurately received the target VT (+4%), whilst patient B had VT reduced by 54% as compared to the target VT with Elisée 350. By contrast, with V500 in patient A VT was under-delivered by 20% as compared to the target VT whilst All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224774 doi: medRxiv preprint patient B received VT reduced by 43% from the target VT (P<0.001 between ventilators). The same picture was observed in steps 3 and 4. In the triple circuit, VT delivered to patient C in volume control was very small according to its low C, with lower values with Elisée 350 than V500 ( Figure 3C ). In step 1 VT to patient C was Figure 3D ). However, in the asymmetrical 2-4 steps, VT delivered to that patient remained stable and in line with its low C preserving it from overdistension ( Figure 3D ). In the double circuit, the rebreathed volume ranged from 0.7 to 37.8 ml ( Table 3 ). The direction of the significant differences between ventilators varies across conditions. The amount of rebreathed volume tended to be lower in pressure control than in volume control. The picture was essentially the same in the triple circuit. The rebreathed volume to patient C was very small, as expected from its very low C. The volume of the rebreathed volume was small as compared to the volume of the expiratory circuit (607 mL). All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in Values are median (1st-3rd quartiles) in ml *P<0.05 versus V500 The minimal expiratory resistance was different between ventilators in most instances in Figure 4 . The differences between ventilators were statistically significant in every comparison without any systematic and consistent direction. We found that: 1) the target VT was achieved by the tested ventilators in volume and pressure control when they faced two symmetrical lungs, 2) asymmetrical C-R changed VT distribution between patients, 3) the risk for air recirculation from one patient to the other was related to the difference in lung time constants and also to the ventilator, 4) the performance of the two ventilators was close. The shortage in ventilators to support COVID-19 patients in acute respiratory failure results from the imbalance between an acute enormous demand and a limited supply. The response of the healthcare system was a dramatic increase in the number of ICU beds in a very short period but the number of ventilators available was an issue. Trying to share ventilators is typical behavior of the "we have to do something" concept [22] to provide a fair allocation of resources [23] [24] [25] in the COVID-19 pandemics. Started before the current COVID-19 pandemic [26, 27] , it was stressed that shared ventilation cannot support its use in mass causality because VT was too much variable across C-R All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224774 doi: medRxiv preprint Page 15 / 27 conditions and largely dependent on changes in C [27] . Since then, the current COVID-19 pandemic prompted additional bench studies to extend these previous results and proposed solutions to try to overwhelm some related issues [8, 11, [28] [29] [30] (Table 4 ). All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224774 doi: medRxiv preprint We found that pressure control should be the mode of choice because it preserves VT in the least injured lung while volume control sets the healthier lung to overdistension and the worst lung to hypoventilation. By comparison to the step 1 the introduction of asymmetry in pressure control does not compromise the target VT of the lung whose R-C set is not too much modified (lung A steps 2 and 4 in design 2 lungs, lung C all steps in design 3 lungs). In steps 2-4 adjusting the pressure preset to restore the targeted VT in each lung will increase VT in the preserved lung. Therefore, technical innovations have been proposed to individualize ventilator settings in each patient, such as set inspiratory pressure, PEEP, and FIO2. These innovations include a one-way flow control valve at inspiratory and expiratory limbs in each patient [12] , a fixed pressure resistor regulator added at the inspiratory limb [31] , a variable flow restrictor at the inspiratory limb, and a one-way valve at the expiratory limb [32] , a flow restrictor on-way valve at the outlet of the ventilator [9] , and bag-in-the box [29] . It should be noted that even though some of the interventions described above have been tested in a few patients [12] , the experience is limited, they are complex to use and may generate further severe problems, as in case of an acute change in respiratory mechanics or gas exchange in one or two patients if the staff is not well trained. The present study brings up new findings by testing two ventilators of different categories and measured air recirculation and expiratory resistance of the ventilator valve. Even though most of the differences in VT between ventilators were statistically significant the clinical significance of them was irrelevant, meaning that the performance of the lower-level ventilator was close to that of the ICU ventilator. Therefore, the present findings suggest that the shortage of ICU-dedicated ventilators can be overcome by using safely lower-level ventilators like the one we tested. The amount of rebreathing air from one patient to the other is another issue of shared ventilation. We quantified this in the present study both during inspiration and expiration ( Figure 2 ). Even though the recirculating air is lower than the anatomical dead space (including the endotracheal tube volume) and arises at the end of the inspiration of the patient it would not induce CO2 retention but sets the patient at an increased risk of cross-infection. Rebreathing can be explained by a difference of "plateau pressure" between the two patients as depicted in figure 2 . We expected that pressure control should prevent rebreathing from the following considerations. In terms of pressurization pressure control mode tends to favor fast and high pressurization finishing with a plateau while the volume control mode with its volume target tends to use less fast pressurization with a continuous increase of the pressure. In other words, we can expect that the pressure control mode is, at the end of the inspiration phase closer to an equilibrium situation for the plateau pressures than the volume control mode. The difference in rebreathing observed between the two ventilators is not simple to explain. Nevertheless, the difference observed in minimal expiratory resistance may also suggest that the opening speed of expiratory valves may be different between the ventilators. A delay to open the valve should favor gas recirculation between the two lungs at the start of the expiratory phase. The clinical implication of present findings is that if one ventilator is dedicated to two or three patients, pressure control mode should be preferred and VT and end-tidal CO2 of each patient should be closely monitored. Those patients should be managed by skilled caregivers, which would require an extensive education schedule in particular in those ICUs where pressure control is not routinely used. They would be able to understand and alert the clinical team of important changes in each of these patients. An efficient filter upstream of the expiratory valve is mandatory. More complex implementations mentioned above have been achieved but may also generate problems. The All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224774 doi: medRxiv preprint recirculation of the air supposes not only some difference in pressure between the two patients but also an expiratory circuit common to the two patients (including the expiratory valve of the ventilator) sufficiently resistive. We recently found that expiratory resistance differs between ICU ventilators [21] . Therefore, adding a poorly resistive valve at the beginning of the expiratory circuit should deemphasize the phenomenon of air recirculation. Nevertheless, the presence of such a valve could affect the capacities of the ventilator to maintain the PEEP. Preclinical studies should be considered to increase our experience and knowledge in sharing ventilation in the dawn of the second wave of COVID-19 pandemic that is hitting the planet The lower-level ventilator performed closely to the ICU-dedicated ventilator. Due to dependence of VT to C pressure control should be used to maintain adequate VT at least in one patient when C and/or R changes abruptly and monitoring of VT should be done carefully. All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted November 4, 2020. ; https://doi.org/10.1101/2020.11.02.20224774 doi: medRxiv preprint The Toughest Triage -Allocating Ventilators in a Pandemic Fast and frugal innovations in response to the COVID-19 pandemic Easier access to mechanical ventilation worldwide: an urgent need for low income countries, especially in face of the growing COVID-19 crisis Makers for Life. MakAir, a COVID-19-born ventilation machine, combining 3D printing design Positive-pressure ventilation equipment for mass casualty respiratory failure COVID-19: instruments for the allocation of mechanical ventilators-a narrative review Ventilator Sharing: The Good, the Bad, and the Ugly Multiplex Ventilation: A Simulation-Based Study of Ventilating 2 Patients With a Single Ventilator Individualized mechanical ventilation in a shared ventilator setting: limits, safety and technical details Ventilator Sharing during an Acute Shortage Caused by the COVID-19 Pandemic Simultaneous ventilation of two simulated ARDS patients in COVID-19 pandemic Differential Ventilation Using Flow Control Valves as a Potential Bridge to Full Ventilatory COVID-19 pneumonia: different respiratory treatments for different phenotypes? Intensive Care Med Patterns of Care, and Mortality for Patients With Acute Respiratory Distress Syndrome in Intensive Care Units All rights reserved. No reuse allowed without permission. perpetuity preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for this this version posted Covid-19 Does Not Lead to a "Typical" Acute Respiratory Distress Syndrome Ventilation management and clinical outcomes in invasively ventilated patients with COVID-19 (PRoVENT-COVID): a national, multicentre, observational cohort study. The lancet Respiratory medicine Clinical features, ventilatory management, and outcome of ARDS caused by COVID-19 are similar to other causes of ARDS Pathophysiology of COVID-19-associated acute respiratory distress syndrome: a multicentre prospective observational study. The lancet Respiratory medicine Bench assessment of expiratory valve resistance of current ICU ventilators in dynamic conditions. Respiratory Care. 2020. 22. Zagury-Orly I, Schwartzstein RM. Covid-19 -A Reminder to Reason Brown MJ, Goodwin J. Allocating Medical Resources in the Time of Covid-19 Allocating Medical Resources in the Time of Covid-19 A single ventilator for multiple simulated patients to meet disaster surge Use of a single ventilator to support 4 patients: laboratory evaluation of a limited concept Shared Ventilation in the Era of COVID-19: A Theoretical Consideration of the Dangers and Potential Solutions Personalized Ventilation to Multiple Patients Using a Single Ventilator: Description and Proof of Concept One ventilator for two patients: feasibility and considerations of a last resort solution in case of equipment shortage Regulating inspiratory pressure to individualise tidal volumes in a simulated two-patient, one-ventilator system A simulated single ventilator/dual patient ventilation strategy for acute respiratory distress syndrome during the COVID The authors would like to thank Michelin France for providing us with the flow-splitter, Maud Grammatica MD and Sylvain Guibert for their help to get the flow-splitter.