Accuracy and precision of breath-alcohol measurements for a random subject in the postabsorptive state.

The accuracy of estimates of blood-alcohol concentration based on measurements of breath-alcohol concentration in a randomly selected subject by a random quantitative evidential breath-alcohol analyzer is evaluated with respect to the breath analyzer itself, its calibration, and the biological variables of the subject being tested. There are no suitable experimental data for rigorous determination of the overall accuracy, so I estimate it from the CV of the available data. I find that the uncertainty in these breath-analyzer readings for a random subject in the postabsorptive state is at least +/- 15%, +/- 19%, or +/- 27%, depending on whether +/- 2 CV, the experimental range, or +/- 3 CV, respectively, is used to express the overall uncertainty. Over 90% of this uncertainty is due to biological variables of the subject, and at least 23% of subjects will have their actual blood-alcohol concentration overestimated. Manufacturers' specifications for the accuracy and precision of these instruments are inconsistent with the experimental values reported in the literature and I recommend that an appropriate amount of uncertainty be reflected in the results from these breath analyzers, especially when they are used for law-enforcement purposes.