Abstract
OBJECTIVE
The Food and Drug Administration's (FDA) checkout recommendations are to be applied, at least daily, on anesthesia machines. Devices included in the checkout are crucial to the safe operation of the machine. Remaining devices do not need to be checked as often. However, they should be checked at some appropriate interval. We calculated the reliability of the non-FDA checkout devices, to predict failure rates of these devices.
METHODS
The study included 36 North American Drager Narkomed anesthesia machines that were in service a total of 109,410 days at a university hospital. Reliability (i.e., statistical) analyses were done using in-house data recorded by biomedical technicians during 3359 monthly machine inspections. If any one device was operating outside of the limits of the manufacturer's specifications, the machine was labeled as having failed its inspection.
RESULTS
Assumptions of the failure analysis were satisfied. The mean time between failures equaled 1351 days or 3.7 years (lower 95% confidence bound 1123 days or 3.1 years). The mean daily probability of one or more devices failing to perform within the manufacturer's specification between inspections equaled 1.1, 2.2%, and 4.4% for inspections every 1, 2, or 4 months, respectively. Probabilities that a machine would fail on demand between inspections were approximately equal to the mean daily probabilities of machine failure. The mean number of days per year that a machine would be operating with a device that is outside of the limits of its specifications equaled 4, 8, and 12, days, respectively. The mean daily probabilities of failure and the probabilities of failure on demand were both directly proportional to the inspection interval.
CONCLUSIONS
Our failure analysis provides a rational basis for choosing an appropriate inspection interval for anesthesia machines' devices that are not included in the FDA's checkout recommendations.
Collapse