Measuring ethanol in blood and breath for legal purposes: variability between laboratories and between breath-test instruments.

We determined the concentrations of ethanol in nearly simultaneous specimens of venous blood (BAC) and end-expired breath (BrAC) after healthy volunteers drank moderate amounts of alcohol. BAC was measured at two laboratories and BrAC was analyzed with two instruments (Intoxilyzer 5000) from the same manufacturer. The mean difference in BAC between laboratories was 0.0105 mg/g (SD 0.0219); 95% of the differences ranged from -0.0333 to 0.0543 mg/g. The mean difference in BrAC between instruments was 0.0153 mg/L (SD 0.0136), and 95% of the differences ranged from -0.0119 to 0.0425 mg/L. The coefficient of variation (CV) between laboratories was 2.9% compared with 4.5% between breath-test instruments. Venous BAC (y) and BrAC (x) were highly correlated (r = 0.978). However, when the Intoxilyzer instruments indicated that BrAC had reached zero, the actual BAC was 0.135 mg/g, according to the average forensic laboratory reports. The Intoxilyzer 5000 breath analyzers used in this study seem to have a constant analytical bias.