Evaluation of breath-alcohol instruments. III. Controlled field trial with Alcolmeter pocket model.

A breath-alcohol screening device, Alcolmeter pocket model, was evaluated in a controlled field trial with policeman operating the instruments. The results of tests made with subjects before they drank alcohol were always zero. The standard deviation (S.D.) of breath alcohol determinations increased with increase in the concentration of alcohol in the sample, being 0.036 mg/ml at a mean blood-ethanol concentration of 0.53 mg/ml. The S.D. varied among subjects tested (from 0.022 to 0.053 mg/ml) as well as among the instruments used (from 0.023 to 0.054 mg/ml). The breath test results were on average less than the actual blood-ethanol concentrations when a 2100: 1 blood/breath ratio was used to calibrate the Alcolmeter device. Blood ethanol (x) and Alcolmeter readings (y) were highly correlated (r = 0.95 +/- 0.018) and the regression equation was y = -0.017 + 0.95x. At a mean blood-ethanol concentration of 0.50 mg/ml, the Alcolmeter instrument will indicate 0.46 mg/ml on average. The standard error estimate was 0.085 mg/ml, being 17% of the mean Alcolmeter reading and this corresponds to 95% confidence limits of +/- 0.17 mg/ml. The results of this study show that Alcolmeter pocket-model is a useful device for breath-alcohol screening purposes at a blood-alcohol level of 0.50 mg/ml. A blood/breath ratio of 2300 should be used to calibrate the Alcolmeter device.