Evaluation of breath alcohol instruments. II. In vivo experiments with alcolmeter pocket model

The precision and accuracy of an Alcolmeter Pocket Model breath alcohol instrument have been investigated in experiments with human subjects under controlled conditions. The instrument response was zero in all tests with breath samples from alcohol-free subjects. The standard deviations of ethanol determinations in breath were +/-0.0722 mg/ml during ethanol absorption and +/-0.416 mg/ml during ethanol elimination. The standard deviation during the elimination phase increased with ethanol concentration in the sample, being +/-0.0416 mg/ml on average at a mean concentration of 0.420 mg/ml, corresponding to a coefficient of variation of 9.9%. The blood alcohol estimates using the Alcolmeter were somewhat too high during active absorption of ethanol, and too low during elimination, when a constant blood-breath alcohol ratio of 2100:1 was used to calibrate the instrument. During the elimination phase of ethanol kinetics and at a mean blood alcohol concentration of 0.50 mg/ml, the mean Alcolmeter result was 0.456 +/- 0.169 mg/ml with 95% confidence, i.e. varying between 0.287 and 0.625 mg/ml 95 times out of 100 tests at this critical blood alcohol level