A review of the first 2000 incidents reported to the Australian Incide
nt Monitoring Study found 317 incidents which involved problems with v
entilation. The major portion (47%) were disconnections; 61% of these
were detected by a monitor. Monitor detection was by a low circuit pre
ssure alarm in 37% but this alarm failed to warn of non-ventilation in
12 incidents (in 6 because it was not switched ''on'' and in 6 becaus
e of a failure to detect the disconnection). Failure of detection was
usually with ventilator bellows descending in expiration. Complete fai
lure to ventilate occurred in 143 incidents, most commonly because of
a disconnection. Disconnection was associated, in one-third of the cas
es, with interference to the anaesthetic circuit by a third party and
in nearly half with surgery on the head and neck. Leaks affected venti
lation in 129 incidents, but in only 19 was ventilation totally lost;
leaks associated with seal failure of the absorber were common. Miscon
nections occurred in 36 incidents, most commonly involving the scaveng
ing system. The frequency of a complete failure to check an anaestheti
c machine was greater when an induction room was involved than when on
ly the operating theatre was the site of the incident. These incidents
suggest that meticulous checking and monitoring for failure of ventil
ation, preferably using at least two separate, self-activating systems
is highly desirable. The Australian and New Zealand College of Anaest
hetists' policy on low circuit pressure alarms, oximetry and capnograp
hy is vindicated by these reports.