When a Ventilator Takes Autonomous Decisions without Seeking Approbation nor Warning Clinicians: A Case Series

BACKGROUND: Complexity and functions of automated medical devices used to support life (eg, ventilators, dialysis machines, monitors, insulin pump with continuous blood glucose monitoring system, etc.) increase over time. Until recently, devices were partially automated by very simple feedback loops, with no or few software dependence (such as the simplest home thermostat). For the last two decades, devices have been increasingly driven by complex algorithms devoted to improve patient’s treatment and monitoring as well as users experience. METHODS: We report the unexpected and inappropriate operation of two recent ventilators, associated to potential harmful consequences. We provide both a description of the clinical situations (five ICU patients, archetypal situations) and a test bench analysis. RESULTS: While set in volume mode, these ventilators activated an algorithm dedicated to limit airway pressure when an increase in airway resistance occurred. In such situations, a pressure-like mode was activated (with decelerating inspiratory flow and set pressure, with target of volume). The main consequences observed were that the tidal volume was no longer guaranteed or delivered and that the pressure limitation operated by the algorithm prevented the airway pressure from reaching the high-pressure alarm threshold. CONCLUSION: This led to the silent takeover of commands by the ventilator without clinicians or nurses being aware of it and without any warnings or alarms emission adapted to the severity of the event. Generally speaking, such an algorithm questions the place of automation and its limit when users are not aware of its presence as well as the need for regulation and additional tests before its implementation. Intensivists and respiratory care specialists should remain vigilant regarding the risk of rare but critical events related to unexpected functioning or insufficiently tested equipment during the pre-clinical development phases. They should not neglect misunderstood critical events without having performed sufficient investigations.