Comparison of Breath and Blood Alcohol Concentrations in a Controlled Drinking Study

In this work, 114 volunteers were dosed with 80-proof liquor to produce peak blood- or breath- alcohol concentration of 0.040 to 0.080 g/100mL blood or g/210L breath. This was followed by a 30-minute deprivation period before simultaneous blood and breath samples were collected and the alcohol concentration quantified. Blood alcohol concentration was determined by gas chromatography with flame ionization detection and breath alcohol concentration by a dual-sensor Intox EC/IR II instrument. Paired Student t-tests showed that differences between paired blood and breath alcohol results differed significantly. Results from these two measurement methods are highly correlated and, on average, measured blood alcohol concentration was 11.3% greater than breath alcohol concentration. There were ten instances of breath alcohol concentration being greater than the corresponding blood alcohol concentration, and the average difference between these two values was 0.0059 g/100mL. Agreement plots of coupled blood and breath alcohol concentration revealed a mean bias of 0.00754 g/100mL and 95% limits of agreement at -0.00705 and 0.0221 g/100mL. Once breath alcohol concentration values were truncated to the hundredths place as required by Wisconsin state statute, only three participants had greater breath alcohol concentration than corresponding blood alcohol concentration, with an average difference between these values of 0.008 g/100mL. Agreement plots with truncated breath alcohol concentration values gave a mean bias of 0.0120 g/100mL and 95% limits of agreement at -0.00344 and 0.0275 g/100mL. Data showed that typically, blood samples had greater alcohol concentrations than corresponding breath values. Differences were exacerbated by Wisconsin's statutory requirement that reported breath alcohol measurements be truncated to the hundredths place whereas blood has no corresponding mandate.