Simulator Study Puts Numbers on the Dangers of Splitting One Ventilator Between Two Patients

When the COVID-19 pandemic overwhelmed intensive care units around the world, one of the most desperate ideas to emerge from the ventilator shortage was also one of the simplest: connect two patients to a single machine. The proposal, known as shared or split mechanical ventilation, circulated widely among clinicians and engineers, and multiple teams published circuit designs that could be assembled from off-the-shelf tubing and connectors. Yet the practice was never validated in real patients, and in March 2020 six major medical societies jointly warned against it. Now a team at The Ohio State University has done what ethics made impossible at the bedside: they put the idea to a rigorous, quantitative test using high-fidelity patient simulators, and their results show just how dangerous unregulated ventilator sharing can be—and how far a simple fix can go toward correcting it.

The study, published in Global Surgical Education, the journal of the Association for Surgical Education, was conceived as an example of translational simulation. Rather than using simulators purely to teach skills, the researchers treated the simulation laboratory as a controlled experimental platform for evaluating a crisis-care innovation that could not ethically or logistically be studied in living patients. The approach matters beyond this single device configuration. Professional societies cautioned against shared ventilation based on theoretical reasoning and bench-level data, but the magnitude of the risk had never been measured in a physiologically representative environment. The Ohio State team set out to fill that gap with reproducible numbers.

The experimental setup was deliberately built from components that any hospital could obtain. Two Laerdal SimMan 3G Plus patient simulators were connected in parallel to a single ventilator running in volume-control assist-control mode. The inspiratory and expiratory ports of the ventilator were split using T-connectors, with HEPA filters placed on each circuit to limit cross-infection and one-way valves positioned to prevent retrograde airflow from the stiffer lung unit to the more compliant one. The only custom components were four 3-D-printed adapters, designed at the university’s Center for Design and Manufacturing Excellence, to connect commercial flowmeters to standard 22-millimeter ventilator tubing. Similar off-the-shelf adapters, the authors note, are available in many markets.

The physiological core of the experiment was compliance mismatch. One simulator was programmed with normal lung compliance, while the other was set to markedly reduced compliance—33 percent of normal—reproducing the impedance imbalance that, in a parallel circuit, drives airflow preferentially toward the easier-to-inflate lung. This is the exact mechanism that the Society of Critical Care Medicine, the American Association for Respiratory Care, the American Society of Anesthesiologists, the Anesthesia Patient Safety Foundation, the American Association of Critical-Care Nurses, and the American College of Chest Physicians cited in their 2020 joint statement against multiple patients per ventilator. In a parallel circuit, air follows the path of least resistance, so the more compliant lung receives a disproportionate share of each breath while the stiffer lung may be severely under-ventilated.

The numbers the simulation produced are striking. At a set tidal volume of 900 milliliters, the normal-compliance simulator received approximately 436 milliliters per breath at baseline. When compliance was reduced to 33 percent, the delivered volume in that limb collapsed to just 141 milliliters—a drop of 295 milliliters, or 68 percent, and a volume the authors describe as incompatible with meaningful ventilation. At the higher setting of 1100 milliliters, the low-compliance limb fell from 487 to 414 milliliters, a 15 percent reduction. These measurements were highly stable: across seven consecutive breaths per condition, coefficients of variation stayed below 9 percent in all six experimental conditions, ranging from 1.4 to 8.9 percent.

The second hypothesis tested whether a pragmatic intervention could mitigate the maldistribution. The researchers fitted inline flowmeters with adjustable valves to the inspiratory tubing of each limb and gradually closed the valve on the higher-flow circuit until the floats in both flowmeters indicated comparable inspiratory flow rates. The effect was dramatic at the 900-milliliter setting: delivered volume in the low-compliance limb recovered from 141 to approximately 431 milliliters, restoring 99 percent of the normal-compliance baseline. At 1100 milliliters, the volume rose from 414 to 501 milliliters, slightly exceeding baseline—consistent, the authors explain, with manual titration to a matched float position rather than to a fixed volume target.

These results empirically confirm what the six societies had argued on theoretical grounds, and they do something the original joint statement could not: they quantify the risk. A patient whose lungs are three times stiffer than a neighbor’s could silently receive less than a third of the intended breath volume while the ventilator displays apparently reassuring numbers. The more compliant lung, meanwhile, faces elevated risks of volutrauma and barotrauma from overdistension. Pressure-controlled ventilation modes may limit barotrauma but do not solve the fundamental problem of unequal volume distribution when compliances diverge. The simulation also demonstrated that flow regulation, while a substantial improvement, does not achieve fully independent control of each patient’s ventilation.

That partial success carries real implications for crisis preparedness. The authors are careful not to suggest that ventilator sharing is safe, but their data show that if sharing were ever deemed necessary as an absolute last resort, it need not be entirely uncontrolled. Inline flowmeters with adjustable valves—a commercially available, inexpensive addition—can redistribute tidal volume toward individualized targets. Just as importantly, the equipment list, circuit configuration, and testing protocol described in the paper can be replicated at any institution with high-fidelity simulation capability, allowing hospitals to test configurations, identify failure modes, and pre-position equipment before a crisis arrives rather than improvising during one.

The study also doubles as an educational platform, which is central to the journal’s mission. The same circuit that generated the data converts directly into a mastery-learning scenario for surgical residents, critical-care fellows, and interprofessional ICU teams. Learners are tasked with recognizing tidal volume maldistribution between two mismatched simulated patients, titrating the flowmeter valves, and identifying the failure modes—unilateral hypoventilation, overdistension injury to the compliant lung, and the loss of independent control of PEEP, oxygen concentration, respiratory rate, and inspiratory time—that underlie the societies’ warning. A facilitator guide, predefined learning objectives, and a structured debriefing framework are provided in the supplementary appendix so other centers can adopt the scenario. The point, the authors emphasize, is not to teach trainees how to routinely split a ventilator but to let them observe, in a controlled environment, why guideline bodies advise against it.

Perhaps the study’s most important contribution is methodological. A 2021 survey of acute care hospitals found that while 96 percent used simulation for education, only 15 percent used it for research, 37 percent for testing, and 30 percent for error analysis. The Ohio State work illustrates how simulation can function as a clinical laboratory for crisis-driven innovations that occupy the gap between theoretical feasibility and clinical implementation—a role the authors argue is underutilized but critically important for disaster and pandemic preparedness. The team is candid about the limits of their pilot: flowmeters measure flow rather than tidal volume directly, the simulators cannot reproduce the spontaneous breathing efforts, secretions, and rapidly evolving lung mechanics of critically ill patients, and the circuit tested only a single assembly under static conditions. Future work, they suggest, should incorporate dynamic compliance changes, more than two simulators, and in situ testing of the complete workflow including team communication and troubleshooting. But as a proof of concept, the study shows that the questions raised in a pandemic’s darkest hours do not have to wait for the next one to be answered.

Subject of Research: Simulation-based evaluation of tidal volume maldistribution during shared mechanical ventilation

Article Title: High-fidelity simulation as a translational and educational platform for crisis ventilation: quantifying and mitigating tidal volume maldistribution during shared mechanical ventilation

Article References: Husain, S., Syed, A., Bockbrader, J., Dunlea, T., Winfield, S., & Vazquez, D. (2026). High-fidelity simulation as a translational and educational platform for crisis ventilation: quantifying and mitigating tidal volume maldistribution during shared mechanical ventilation. Global Surgical Education – Journal of the Association for Surgical Education, 5(1), Article 185. https://doi.org/10.1007/s44186-026-00589-1

Image Credits: AI Generated

DOI: 10.1007/s44186-026-00589-1

Keywords: shared mechanical ventilation, ventilator splitting, tidal volume, lung compliance, translational simulation, high-fidelity simulation, crisis preparedness, flowmeters, patient simulators, COVID-19, ventilator shortage, medical education