Quantitative Comparison of Ventilation Parameters of Different Approaches to Ventilator Splitting and Multiplexing

Abstract

CONTEXT:

Amid the COVID-19 pandemic, this study delves into ventilator shortages, exploring simple split ventilation (SSV), simple differential ventilation (SDV), and differential multiventilation (DMV). The knowledge gap centers on understanding their performance and safety implications.

HYPOTHESIS:

Our hypothesis posits that SSV, SDV, and DMV offer solutions to the ventilator crisis. Rigorous testing was anticipated to unveil advantages and limitations, aiding the development of effective ventilation approaches.

METHODS AND MODELS:

Using a specialized test bed, SSV, SDV, and DMV were compared. Simulated lungs in a controlled setting facilitated measurements with sensors. Statistical analysis honed in on parameters like peak inspiratory pressure (PIP) and positive end-expiratory pressure.

RESULTS:

Setting target PIP at 15 cm H₂O for lung 1 and 12.5 cm H₂O for lung 2, SSV revealed a PIP of 15.67 ± 0.2 cm H₂O for both lungs, with tidal volume (Vt) at 152.9 ± 9 mL. In SDV, lung 1 had a PIP of 25.69 ± 0.2 cm H₂O, lung 2 at 24.73 ± 0.2 cm H₂O, and Vts of 464.3 ± 0.9 mL and 453.1 ± 10 mL, respectively. DMV trials showed lung 1’s PIP at 13.97 ± 0.06 cm H₂O, lung 2 at 12.30 ± 0.04 cm H₂O, with Vts of 125.8 ± 0.004 mL and 104.4 ± 0.003 mL, respectively.

INTERPRETATION AND CONCLUSIONS:

This study enriches understanding of ventilator sharing strategy, emphasizing the need for careful selection. DMV, offering individualization while maintaining circuit continuity, stands out. Findings lay the foundation for robust multiplexing strategies, enhancing ventilator management in crises.

KEY POINTS

Question: What are the outcomes of several ventilator sharing strategies of connecting multiple patients to a single mechanical ventilator: simple split ventilation, simple differential ventilation, and differential multiventilation (DMV) in terms of control of peak inspiratory pressure and positive end-expiratory pressure and patient safety?

Findings: In a comparative analysis of several ventilator splitting and multiplexing strategies, DMV achieved target pressures within specific ranges, while other strategies exhibited variable outcomes, some of which raise patient safety concerns.

Meaning: Rigorous evaluation of ventilator sharing strategies is crucial as some strategies result in unpredictable behaviors in certain conditions.

During the first months of the COVID-19 pandemic, many healthcare resources were overwhelmed including hospital ventilator capacity. Simultaneously, the pandemic disrupted the supply chain of ventilator components resulting in a shortage of available ventilators for purchase. Consequently, a global effort arose to design and build capabilities to expand mechanical ventilation capability with locally sourced materials, resources, engineering support, and manufacturing facilities. Interdisciplinary groups were rapidly formed around the world to devise new ways to increase and expand ventilator capacity. This global effort directly led to the rapid development of low-cost mechanical ventilators and related methods to increase ventilator capacity such as ventilator splitting to connect multiple patients to a single ventilator. Rudimentary ventilator splitting was trialed in some hospitals despite multiple professional medical bodies making statements against such practices due to safety concerns like air trapping and cross contamination (1, 2). Following these statements, several groups developed and proposed alternative techniques of ventilator splitting and multiplexing that might improve patient safety (1, 3–7).

While many ventilator sharing strategies were proposed, many of these designs were not rigorously tested to evaluate and compare their performance. We and others observed that some of these strategies, while intended to solve some patient safety issues, had the potential to create new safety concerns, including cross-contamination, inadequate ventilation, and increased risk of ventilator-induced lung injury and other complications (1, 8). We undertook a comparative study to quantitatively describe the resultant ventilator circuit parameters of different strategies. The objective of this study is to compare three identified strategies. By employing the same mechanical ventilator and simulated lungs across experiments, we gather empirical data seeking to provide factual comparisons into the differing resultant behavior outcomes of these strategies.

OVERVIEW OF APPROACHES TO VENTILATOR SPLITTING AND MULTIPLEXING

Several ventilator sharing strategies have been newly proposed since the beginning of the pandemic. While the exact equipment and their arrangements vary with each proposal, the strategies generally fall into a few conceptual approaches. The strategies fall into roughly two major categories: ventilator splitting and ventilator multiplexing. Ventilator splitting strategies are the simplest involving unsophisticated splitting or branching of the gas path between multiple patients. Ventilator multiplexing strategies attempt to create multiple different gas paths of different physical characteristics (e.g., different airway pressures). As no previously devised nomenclature exists to discuss the differences between these strategies, we have used a descriptive nomenclature to identify the general strategies namely simple split ventilation (SSV), simple differential ventilation (SDV), and differential multiventilation (DMV).

The first category of approaches, SSV (schematic shown in eFig. 1, http://links.lww.com/CCX/B357) (9), originally proposed by Neyman and Irvin (3), involves simply introducing a T-piece or wye at the inspiratory and expiratory ports of the ventilator. This strategy is the least sophisticated approach and relies only on the ventilator to produce and manage the pressure and flow in both circuits. Therefore, key ventilation parameters such as peak inspiratory pressure (PIP) and positive end-expiratory pressure (PEEP) can only be modified by changing the ventilator settings. Consequently, SSV creates several patient safety issues such as uneven ventilation support and difficult PEEP control that have been enumerated by numerous professional medical bodies and articulated widely in the COVID-19 literature (1–6, 8).

In the second category of strategies (SDV; schematic shown in Fig. 1) (10), the circuits attached to the ventilator have additional circuit individualization equipment placed in the circuit to modify the pressure and/or flow of the individual circuit with the goal of individualizing ventilation in each circuit. Typically, various mechanical restrictors or valves, such as inline PEEP valves in our study, are used to modify key ventilator parameters such as PIP and PEEP. Tuning the valve on the inspiratory limb governs the PIP of the adjacent lung, while the expiratory limb’s valve adjustment controls the PEEP. While SDV allows modification of the circuit dynamics, it also has the potential to obstruct the so-called bias flow, which many ventilators use to detect problems such as insufficient replenishment of oxygen and removal of carbon dioxide within the circuit, ensuring the predictable operation of the ventilator (1, 11).

Figure 1.

Schematic illustration of simple differential ventilation and differential multiventilation (1). A, Ventilator (inspiratory port). B, Inline positive end-expiratory pressure valve. C, Check valve. D, Heat and moisture exchangers/high efficiency particulate air filter. E, Lung 1 (patient 1). F, Lung 2 (patient 2). G, Bypass circuit. H, Inspiratory limb. I, Circuit wye. J, Expiratory limb. K, Ventilator (expiratory port).

Finally, in the third category of strategies, DMV, an extra “bypass” circuit is introduced together with the circuit individualization equipment seen in SDV (schematic shown in Fig. 1; photo shown in eFig. 2, http://links.lww.com/CCX/B357) (1, 9). This bypass circuit provides a common pathway for the bias flow as well as the ventilation functions enabled by this flow, while still allowing individualization of parameters for patient circuits. The control of PIP and PEEP is maintained similar to SDV.

Each of the systems above requires components and patient monitoring equipment to be in place. More information about the options for different components of the system as well as safe patient monitoring have been described in detail elsewhere (1).

METHODS

We evaluated three distinct circuit designs of SSV, SDV, and DMV alongside traditional mechanical ventilation, all of which used T-pieces to ensure continuous circuit connectivity. SSV featured T-pieces without any additional devices, adjusting target pressure settings solely through ventilator modifications. SDV incorporated circuit individualization equipment in the inspiratory and expiratory limbs of each individual circuit, acting as flow restrictors to personalize circuit characteristics, as they have been widely adopted in various designs (1, 12). We considered both the ventilator and the circuit individualization equipment settings to determine target pressures for each simulated lung. Installation of the equipment in the expiratory limb added to PEEP, while the inspiratory limb subtracted from PIP for each simulated lung (1). DMV similarly employed the use of installed circuit individualization equipment, resembling the SDV design.

Patient individualization, another key aspect assessed, involved the circuit individualization equipment that also serve the important function of one-way valves (1). This function prevents complex flow dynamics, avoiding issues like Co₂-rebreathing or cross-contamination from rebreathing of expired gas by one or both patients. For this reason, cross-contamination risk was assessed by checking for backflows in the circuit. Additionally, we evaluated the interaction of patient characteristics (e.g., respiratory compliance) or ventilator settings on predictable ventilator behavior and effects. This was evaluated by analyzing the responding parameters of a simulated lung while progressively changing the independent parameters of the other simulated lung such as the target PIP, target PEEP, and lung compliance (C_Lung).

The effect of external variables on the test results was minimized by using two identical simulated lungs (SmartLung Adult 1L; IMT Analytics, Buchs, Sankt Gallen, Switzerland). They were set to a C_Lung of 30 mL/mbar (29.42 mL/cm H₂O) unless otherwise specified. The ventilator (Evita XL; Dräger, Lübeck, Germany) always operated in pressure control mode, which is the suggested mode for ventilator splitting setups (1, 3, 10). The detailed ventilator settings, data recording setup, and procedures are elaborated in Detailed Methods section of the Supplemental Digital Content (http://links.lww.com/CCX/B357) (9).

Experimental airway pressures and airflow (Fig. 2) were assessed for PIP and PEEP control by adjusting ventilator and circuit individualization equipment. The test aimed to evaluate each design across diverse target settings and systematically detect interactions. “Target PIP” and “Target PEEP” were set as independent parameters to define intended PIP and PEEP for each lung. Airway pressure and flow rate, which was used to calculate tidal volume (Vt), were measured at various circuit regions.

Figure 2.

Sample graphs plotted with the data collected from a ventilator splitting circuit. A, Airway pressures. B, Flow rate. Exp Port = expiratory port, Ins Port = inspiratory port.

During the first set of experiments (Fig. 3) to compare the circuit designs, the target PIP for the simulated lungs was varied, setting lung 1 at 15 cm H₂O, and lung 2 at 12.5 cm H₂O. The target PEEP was kept at 6 cm H₂O for both lungs, exception for the traditional where only a single simulated lung (lung 1) was used. In the case of traditional experiments, the target pressures were adjusted solely by modifying the ventilator settings.

Figure 3.

Experimental runs with various circuit designs including traditional (Trad) single lung ventilation, simple split ventilation (SSV), simple differential ventilation (SDV), and differential multiventilation (DMV). All simulated lungs were set at a consistent lung compliance (= 30 mL/mbar). Simulated lungs shared the same target positive end-expiratory pressure (PEEP) of 6 cm H₂O. Target peak inspiratory pressures (PIPs) are 15 and 12.5 cm H₂O for lung 1 and lung 2, respectively. A, Measured airway pressures of lungs 1 and 2, inspiratory port (Ins Port) and expiratory port (Exp Port) of the ventilator, and the measured pressure difference (Pres Diff) between the two ports. Dashed horizontal lines denote the target PIPs and PEEP. B, Measured tidal volume for each lung per respiration. Total of five respirations per capture segment.

Additional experiments explored DMV and SDV. Data collection for SDV was hindered by repeated breakage of ventilator ports. In the DMV trials, the first set (Fig. 4A) incrementally raised target PIP of lung 1 from 10 to 12.5 and finally to 15 cm H₂O to match the target PIP of lung 2 (15 cm H₂O), maintaining other settings. The second set (Fig. 4B) reduced target PEEP of lung 1 from 12 to 9 and ultimately to 6 cm H₂O to equal the target PEEP of lung 2, keeping other settings stable. The final set (Fig. 5) increased lung 1 compliance from 10 to 20 and finally to 30 mL/mbar (9.807, 19.61, and 29.42 mL/cm H₂O), while keeping lung 2 constant at 30 mL/mbar (29.42 mL/cm H₂O).

Figure 4.

Response of simulated lungs connected to differential multiventilation. Lung 1 settings are only adjusted. Ventilator and lung 2 settings are consistent. A, Changing target peak inspiratory pressure (PIP) of lung 1 by adjusting the circuit individualization equipment at the inspiratory limb of lung 1. B, Changing target positive end-expiratory pressure (PEEP) of lung 1 by adjusting the circuit individualization equipment at the expiratory limb of lung 1.

Figure 5.

Response of simulated lungs connected to differential multiventilation. Changing compliance of lung 1 (C_Lung1) (mL/mbar). Ventilator and lung 2 settings are consistent. PEEP = positive end-expiratory pressure, PIP = peak inspiratory pressure.

RESULTS

In the circuit design compare set of experiments, as shown in Figure 3, traditional achieved a PIP of 14.41 ± 0.06 cm H₂O and Vt of 146.0 ± 1 mL, with five respiratory cycles captured. With this Vt, the minute ventilation was calculated as 2.920 L. Additionally, the maximum difference in pressure between the inspiratory and expiratory ports of the ventilator was 0.2319 cm H₂O.

In SSV trials, both lung 1 and lung 2 demonstrated a PIP of 15.67 ± 0.2 cm H₂O, with a maximum difference of 0.8843 cm H₂O in airway pressure between the two lungs. Similarly, the Vt was observed at 152.9 ± 9 mL for both lungs 1 and 2, with a maximum difference in Vt of 16.66 mL between the lungs. This calculated the minute ventilation as 3.058 L. Five respiratory cycles were captured for both simulated lungs with SSV. Last, the maximum difference in pressure between the inspiratory and expiratory ports of the ventilator was 0.8843 cm H₂O.

In SDV trials, one failure of breath delivery was observed for both lungs so only four respiratory cycles were observed instead of five. Excluding the failed breath, lung 1 exhibited a PIP of 25.69 ± 0.2 cm H₂O, whereas lung 2 demonstrated a PIP of 24.73 ± 0.2 cm H₂O. The Vt for lung 1 was measured at 464.3 ± 0.9 and for lung 2 at 453.1 ± 10 mL for the delivered. The minute ventilations for lung 1 and lung 2 were calculated as 7.510 and 7.368 L, respectively, including the failed breath. The maximum pressure difference between the inspiratory and expiratory ports of the ventilator was 14.43 cm H₂O at PIP.

In DMV trials, lung 1 received a PIP of 13.97 ± 0.06 cm H₂O, whereas lung 2 received a PIP of 12.30 ± 0.04 cm H₂O. Consequently, a Vt of 125.8 ± 0.004 mL was measured for lung 1 and 104.4 ± 0.003 mL for lung 2. This corresponds to minute ventilation values of 2.516 L and 2088 mL for lung 1 and lung 2, respectively. Additionally, the maximum observed pressure difference between the inspiratory and expiratory ports of the ventilator was 0.2936 cm H₂O.

In the additional DMV trials (Figs. 4 and 5), the first set, as shown in Figure 4A, measured PIPs for lung 1 of 10.04 ± 0.09, 12.59 ± 0.05, and 14.57 ± 0.2 cm H₂O, and corresponding escalating Vts of 36.05 ± 0.3, 67.55 ± 0.7, and 140.7 ± 0.5 mL. During these experiments, lung 2 maintained consistent PIP and Vt measurements at 14.66 ± 0.1 cm H₂O and 136.5 ± 2 mL, respectively.

In the second DMV set, as shown in Figure 4B, the measurements of lung 1 PEEPs and Vts of 12.03 ± 0.07 cm H₂O and 44.94 ± 0.2 mL, 8.49 ± 0.1 cm H₂O and 80.83 ± 0.3 mL, and 5.382 ± 0.2 cm H₂O and 140.7 ± 0.5 mL for respective target PEEPs of 12.00, 9.00, and 6.00 cm H₂O were recorded. Concurrently, PEEP and Vt for lung 2 were measured at 6.373 ± 0.1 cm H₂O and 134.0 ± 4 mL during these experiments.

In the third DMV set, demonstrated in Figure 5, the PIP and PEEP of both lungs were recorded as 14.92 ± 0.2 and 6.225 ± 0.4 cm H₂O respectively. The Vts for lung 1 were measured at 30.68 ± 0.5, 56.51 ± 0.6, and 132.1 ± 0.4 mL for C_Lung of 10, 20, and 30 mL/mbar, respectively. Throughout these experiments, a Vt of 115.9 ± 3 mL was measured at lung 2.

DISCUSSION

In this study, three circuit designs were assessed across six key categories (Table 1). SSV showed potential for multiventilation but had limitations. It aligned with prescribed PIP and PEEP settings for lung 1 but lacked individual pressure control due to the absence of circuit individualization equipment. Both lungs exhibited identical readings, and safety devices against cross-contamination were absent, making it unsuitable for clinical use. However, breath delivery was uninterrupted, and the pressure difference between the ventilator’s ports stayed below the alarm limit. In summary, SSV resembled a traditional mechanical ventilation system but lacked additional controllability, serving primarily for ventilator splitting.

TABLE 1.

Observed Traits of the Experimented Ventilator Multiplexing Strategies

Circuit Design	Predictable Control	Cross-Contamination Prevention	Patient Individualization	Compliance with Ventilator Alarm Limits	Compliance to Set Breath Rate	Overventilation Avoidance
Simple split ventilation				O	O
Simple differential ventilation	O	O
Differential multiventilation	O	O	O	O	O	O

O represents the observed trait corresponding to the respective circuit design.

SDV displayed uncontrollable behavior, experiencing a breath loss due to a pressure difference exceeding the alarm limit of 5.10 cm H₂O between inspiratory and expiratory ports. This activated the emergency pressure relief mechanism of the ventilator to equalize pressures (13). This issue persisted due to unresolved pressure resistance from circuit individualization equipment. Removing this equipment near lung 2 reinstated the breath rate but killed individual adjustability, rendering the system inadequate. Notably, the circuit individualization equipment serves as a patient safety device, and its removal compromises safety (1).

In addition, the SDV design led to overventilation, with both lungs’ PIP significantly exceeding the target values. In typical ventilator operations (traditional design), maintaining consistent pressure from the inspiratory to the expiratory port is expected. Therefore, monitoring airway pressure solely at the expiratory port ensures acceptable operation (14). However, in our experiments, high PIP values resulted from the ventilator striving to meet the set PIP at the expiratory port without considering additional pressure resistance from the circuit individualization equipment. This issue could be resolved by establishing a “bypass” between the ports by completely removing the circuit individualization equipment for one lung pathway, eliminating overventilation. However, this sacrifices individual adjustability and cross-contamination prevention. Our observations suggest the SDV design should not be further investigated to be used for patient treatment.

The DMV design successfully demonstrated its functionality by providing stable patient independence and predictable controllability during ventilator multiplexing. The changes in PIP and PEEP of the simulated lung were responsive to adjustments in target pressures, while the other simulated lung remained unaffected (Fig. 4). Additionally, changes in compliance of one simulated lung did not result in noticeable changes in the other (Fig. 5). Any minor discrepancies in Vt between simulated lungs 1 and 2 in Figure 5, despite identical settings, can be disregarded as they may be attributed to external factors such as different orientations of the simulated lungs affecting flow restriction to each lung (15).

Importantly, the issues identified in the other two designs, such as overventilation, breath loss, and exceedance of ventilator alarm limits, were not observed with the DMV design. Furthermore, the patient individualization requirements were met as the DMV design installed circuit individualization equipment at all pathways to the simulated lungs. In conclusion, the DMV design fulfilled all the ventilator multiplexing requirements outlined in this study.

LIMITATIONS

While our study aimed to provide insights into the performance of different ventilator splitting and multiplexing strategies, it is important to acknowledge several limitations that may have impacted the results and generalizability of our findings.

Our experiments employed simulated lungs, potentially limiting the extrapolation of our findings to real-world patient scenarios due to the diverse characteristics, diseases, and responses seen in actual patients (11). Additionally, we isolated the effects of individual strategies by using simplified circuit configurations for controlled experiments. Yet, real-world ventilator circuits involve additional components like suction ports and filters, introducing variables absent in our study. These unconsidered factors should be acknowledged when applying our findings to clinical settings.

Our study focused on evaluating the performance characteristics of different ventilator splitting and multiplexing strategies, primarily assessing the control of PIP and PEEP. However, key ventilatory variables such as Fio₂, oxygen saturation, Po₂, and Pco₂ were not explored. These unexamined variables, vital for indicating oxygen delivery, oxygen, and carbon dioxide levels in the blood, could significantly influence physiologic responses and vary patient outcomes. Therefore, the omission of these parameters may limit the applicability of the findings in real-world clinical settings and underscores the need for further investigations for a comprehensive understanding of ventilator sharing strategies.

Finally, pressure sensors (HSCDRRN001PDAA5; Honeywell, Golden Valley, MN) were bounded to a tolerance of ± 1.406 cm H₂O, while the flow sensors (SFM3300-D; Sensirion AG, Staefa, Switzerland) had a tolerance of ± 5% of the measured value (16, 17). These error ranges introduce a degree of uncertainty in the measurements of pressure and flow, which may affect the accuracy of the calculated parameters such as PIP, PEEP, and Vt. Despite the potential for slight measurement variations, it is likely that the recorded values are close approximations to the true values.

Despite the limitations, our study contributes to the understanding of ventilator sharing strategies and their implications for patient care. These findings can serve as a foundation for further research and clinical investigations, helping guide the development of more effective and individualized ventilator management approaches in the future. While our study provides important insights into the performance of different ventilator multiplexing strategies, it is essential to validate these findings in clinical settings. Real-world clinical validation studies are necessary to assess the feasibility, safety, and efficacy of these strategies in diverse patient populations and healthcare environments.

CONCLUSIONS

In this study, we compared three ventilator splitting and multiplexing strategies: SSV, SDV, and DMV, aiming to quantitatively describe their effects on ventilator circuit parameters and evaluate their performance in terms of PIP, PEEP, and patient individualization. SSV, the simplest approach, led to shared settings and compromised patient safety. SDV, incorporating circuit individualization equipment, showed improved PIP control but exhibited over-pressure and volume effects. DMV, with an additional bypass circuit, achieved target pressures within a narrow range, demonstrating better control and safety than SSV and SDV. The study underscores the importance of robust ventilator splitting strategies for patient safety and ventilation adequacy. Considering individual patient needs and achieving predictable PIP and PEEP control is crucial in strategy selection. DMV, offering individualization while maintaining circuit continuity, appears promising in addressing challenges posed by SSV and SDV. Overall, the study emphasizes the need for careful strategy selection, with future advancements and research expected to enhance ventilator technology and improve sharing strategies during healthcare resource challenges like the COVID-19 pandemic.

ACKNOWLEDGMENTS

We express our sincere gratitude to Mitchell Weber for his technical support. We also acknowledge the contributions of our research team members and colleagues.