Patients may not receive prescribed oxygen because the oxygen face mask becomes displaced. Using video, we have observed the position of the face mask in 20 postoperative patients and recorded the timing and the events associated with mask displacement. Correct placement of the mask was confirmed at the start of the 8-h study period from 22:00 on the first night after operation. The mask remained on continuously and positioned correctly in only one patient. In the other 19 patients, it was removed 64 times (range 1-10 times per patient). The mask was removed 45 times for nursing tasks such as mouth care and temperature measurement and these represented 70% of the total number of times that the mask was not in position. Other reasons for removal were vomiting, retching and removal by the patient. The mask remained off a median time of 6 min 55 s per episode (range 46 s to 7 h 46 min 57 s) and per patient a median of 1 h 6 min 48 s (range 1 min to 7 h 46 min 57 s). Mask removal resulted in an average decrease in oxygen saturation of 4%. Oxygen by mask at 4 litre min-1 maintained an average saturation > or = 95% in most, but not all, of the patients.

We have studied simultaneously the pharmacokinetics of flumazenil and midazolam in 12 healthy Chinese children, aged 5-9 yr, undergoing circumcision. Two hours before operation each patient received midazolam 0.5 mg kg-1 orally for premedication and 0.5 mg kg-1 i.v. during induction. Six minutes after cessation of anaesthesia, a bolus of flumazenil 10 micrograms kg-1 was given i.v., followed by an infusion of flumazenil at 5 micrograms kg-1 min-1 which was maintained until the child could identify himself. Midazolam data were consistent with a three-compartment model with a mean (SD) elimination half-life of 107 (30) min, total body clearance of 15.4 (3.2) ml min-1 kg-1 and apparent volume of distribution at steady state of 1.9 (0.6) litre kg-1. Flumazenil data were best interpreted by a monoexponential function, with a mean terminal elimination half-life of 35.3 (13.8) min, a total plasma clearance of 20.6 (6.9) ml min-1 kg-1 and apparent volume of distribution at steady state of 1.0 (0.2) litre kg-1. No unchanged midazolam was detected in the 24-h urine sample, but 5.8-13.8% of the flumazenil dose was recovered unchanged. At the time of self identification, 4.5 (1.4) min after flumazenil administration, the mean plasma concentrations of midazolam and flumazenil were 163.1 (43.7) and 29.9 (16.1) ng ml-1, respectively.

SummaryBlood pressure measurements associated with malapplication of a finger cuff were compared with contemporaneous intra-arterial pressure data in seven volunteers to determine the influence of cuff application on the accuracy of the Finapres 2300e. Systolic readings in all cuff positions differed from arterial line data by more than the recommended standard and tight and loose cuff applications under and over-read respectively, in all subjects. The results show the Finapres to be sensitive to small degrees of finger cuff malapplication which contribute to the bias on direct arterial comparison and limit the reliability ofthe instrument in clinical practice.