The Science Journal of the American Association for Respiratory Care

2010 OPEN FORUM Abstracts

REFERENCE DATA FOR DETERMINING VENTILATOR ALARM LIMITS

Rory Mullin, Robert L. Chatburn; Respiratory Institute, Cleveland Clinic, Cleveland, OH

BACKGROUND: Few studies are available regarding ventilator alarm settings and no rational approach to developing intelligent alarms has been described. A review of the literature revealed one study that reports 23% of ICU alarms as effective (Anesth Analg 2009;108:1546 –52). The purpose of our study was to determine the inherent variability of three alarm parameters as a potential rationale for setting limits based on expected percentage of alarm events. The three alarm parameters considered were minute ventilation (MV), peak inspiratory pressure (PIP) and tidal volume (TV). METHODS: We conducted a chart review of patients in surgical, medical, neurological, and cardio-thoracic intensive care units. Modes of ventilation included were pressure control (PC) and volume control. Pressure support was not included. Measured MV was recorded for all modes. Peak inspiratory pressure (PIP) was recorded for VC. Tidal volume was recorded for PC. Patients intubated for less than one day were excluded. RESULTS: Data sets for 1,243 ventilator checks were recorded, 882 in pressure control and 361 in volume control. Data are expressed as (mean, ± standard deviation). PIP was 28 ± 7 cm H2O. TV was 8.6 ± 2.2 mL/kg and mean TV was the same for VC. MV (L/min) was overall 9.6 ± 3.0: for PC 9.7 ± 3.2: for VC 9.3 ± 2.5. The Figure shows percentile plots with horizontal lines at 10th, 25th, 50th, 75th, and 90th percentiles. CONCLUSIONS: The main finding of this study is that actual values for minute ventilation, peak inspiratory pressure and tidal volume are highly variable, with significant portions at extreme values. It also suggested that surveillance of TV used in both PC and VC may be warranted. Current textbooks recommend fixed values for some alarm limits (e.g., ± 10 cm H2O for PIP) and fixed proportions for others (e.g. ± 25% for MV and TV). However our data would suggest that these limits may reduce safety for some extreme values while increasing nuisance events for other values. An alternative approach might be to reference the alarm limits to the current value of the parameter such that extreme values have tighter limits. Further research is needed to identify optimization algorithms (i.e., minimize both harmful and nuisance events) for intelligent targeting schemes to automatically set alarms during mechanical ventilation. Sponsored Research - None