Ventilation characteristics of cardiopulmonary resuscitation manikins: a mixed-methods bench study.

Background: Sufficient ventilation during cardiopulmonary resuscitation (CPR) is increasingly recognised as a critical determinant of patient outcomes. The effectiveness of ventilation training, however, likely depends on the anatomical and mechanical fidelity of CPR manikins. While chest compression mechanics of manikins have been studied extensively, ventilation-related anatomy, airflow pathways, and respiratory mechanics have not been evaluated systematically. Methods: This study analysed 10 adult CPR manikins from three manufacturers. Airway and lung replicas were documented using photography, video laryngoscopy, and bronchoscopy. Expiratory tidal volume and air leakage were measured during volume-controlled mechanical ventilation via facemask, supraglottic airway devices and endotracheal tube. Mechanics of the respiratory system were described by the peak and plateau airway pressures and respiratory compliance. Results: Manikins differed substantially in airway design and expiratory airflow design. Three distinct expiratory mechanisms were identified, of which only one permitted exhalation through the airway and thus quantification of expiratory tidal volumes. Air leakage varied across manikins and airway devices, including during ventilation via advanced airway devices. Respiratory mechanics showed marked variability, with peak inspiratory pressures ranging from 21 ± 1 to 56 ± 1 mbar and compliances from 7 ± 0 to 51 ± 4 mL/mbar. Several models exhibited substantial leakage via the gastric channel of the i-gel® supraglottic airway. Conclusion: CPR manikins exhibited substantial variability in ventilation-related anatomy, airflow pathways, leakage and respiratory mechanics. None of the assessed manikins consistently approximated human ventilation characteristics across airway devices, underscoring the need for local evaluation of manikin-airway device combinations to optimise ventilation fidelity in CPR training.