Direct measurement

Blood samples for BAC analysis are typically obtained by taking a venous blood sample from the arm. A variety of methods exist for determining blood-alcohol concentration in a blood sample.^[8] Forensic laboratories typically use headspace-gas chromatography combined with mass spectrometry or flame ionization detection,^[9] as this method is accurate and efficient.^[8] Hospitals typically use enzyme multiplied immunoassay, which measures the co-enzyme NADH. This method is more subject to error but may be performed rapidly in parallel with other blood sample measurements.^[10]

By breathalyzer

Novelty "Breathalyser 'pint'" beer glass, about 2 inches tall, dating from around the time of the introduction of breathalyzers in the United Kingdom, in 1967

The amount of alcohol on the breath can be measured, without requiring drawing blood, by blowing into a breathalyzer, resulting in a breath alcohol content (BrAC). The BrAC specifically correlates with the concentration of alcohol in arterial blood, satisfying the equation BAC_arterial = BrAC × 2251 ± 46. Its correlation with the standard BAC found by drawing venous blood is less strong.^[11] Jurisdictions vary in the statutory conversion factor from BrAC to BAC, from 2000 to 2400. Many factors may affect the accuracy of a breathalyzer test,^[12] but they are the most common method for measuring alcohol concentrations in most jurisdictions.^[13]

By intake

Blood alcohol content can be estimated by a model developed by Swedish professor Erik Widmark in the 1920s.^[14] The model corresponds to a pharmacokinetic single-compartment model with instantaneous absorption and zero-order kinetics for elimination. The model is most accurate when used to estimate BAC a few hours after drinking a single dose of alcohol in a fasted state, and can be within 20% CV of the true value.^[15][16] It is less accurate for BAC levels below 0.2 g/L (alcohol is not eliminated as quickly as predicted) and consumption with food (overestimating the peak BAC and time to return to zero).^[17][18] The equation varies depending on the units and approximations used, but in its simplest form is given by:

${\displaystyle EBAC={\frac {A}{V_{d}}}-\beta \times T}$

where:

• EBAC is the estimated blood alcohol concentration (in g/L)
• A is the mass of alcohol consumed (g).
• T is the amount time during which alcohol was present in the blood (usually time since consumption began), in hours.
• β is the rate at which alcohol is eliminated, averaging around 0.15 g/L/hr
• V_d is the volume of distribution (L); typically body weight (kg) multiplied by 0.71 L/kg for men and 0.58 L/kg for women

Examples:

• A 80 kg man drinks 2 US standard drinks (3 oz) of 40% ABV vodka, containing 14 grams of ethanol each (28 g total). After two hours:

$${\displaystyle EBAC=28/(0.71\cdot 80)-(0.148\cdot 2)\approx 0.197{\text{g/L}}=0.0197\%{\text{BAC}}}$$

• A 70 kg woman drinks 63 g of 40% ABV vodka, containing 21 grams of ethanol. After two hours:

$${\displaystyle EBAC=21/(0.58\cdot 70)-(0.156\cdot 2)\approx 0.205{\text{g/L}}=0.0205\%{\text{BAC}}}$$

The volume of distribution V_d contributes about 15% of the uncertainty to Widmark's equation^[19] and has been the subject of much research. It corresponds to the volume of the blood in the body.^[14] In his research, Widmark used units of mass (g/kg) for EBAC, thus he calculated the apparent mass of distribution M_d or mass of blood in kilograms. He fitted an equation ${\displaystyle M_{d}=\rho _{m}W}$ of the body weight W in kg, finding an average rho-factor of 0.68 for men and 0.55 for women. This ρ_m has units of dose per body weight (g/kg) divided by concentration (g/kg) and is therefore dimensionless. However, modern calculations use weight/volume concentrations (g/L) for EBAC, so Widmark's rho-factors must be adjusted for the density of blood, 1.055 g/mL. This ${\displaystyle \rho _{v}=V_{d}/W}$ has units of dose per body weight (g/kg) divided by concentration (g/L blood) - calculation gives values of 0.64 L/kg for men and 0.52 L/kg for women, lower than the original.^[18] Newer studies have updated these values to population-average ρ_v of 0.71 L/kg for men and 0.58 L/kg for women. But individual V_d values may vary significantly - the 95% range for ρ_v is 0.58-0.83 L/kg for males and 0.43-0.73 L/kg for females.^[20] A more accurate method for calculating V_d is to use total body water (TBW) - experiments have confirmed that alcohol distributes almost exactly in proportion to TBW. TBW may be calculated using body composition analysis or estimated using anthropometric formulas based on age, height, and weight. V_d is then given by ${\displaystyle TBW_{\text{kg}}/F_{\text{water}}}$, where ${\displaystyle F_{\text{water}}}$ is the water content of blood, approximately 0.825 w/v for men and 0.838 w/v for women.^[21]

The elimination rate from the blood, β, is perhaps the more important parameter, contributing 60% of the uncertainty to Widmark's equation.^[19] Similarly to ρ, its value depends on the units used for blood.^[18] β varies 58% by occasion and 42% between subjects; it is thus difficult to determine β precisely, and more practical to use a mean and a range of values. The mean values for 164 men and 156 women were 0.148 g/L/h and 0.156 g/L/h respectively. Although statistically significant, the difference between sexes is small compared to the overall uncertainty, so Jones recommends using the value 0.15 for the mean and the range 0.10 - 0.25 g/L/h for forensic purposes, for all subjects.^[22] Explanations for the gender difference are quite varied and include liver size, secondary effects of the volume of distribution, and sex-specific hormones.^[23] Elaborating on the secondary effects, zero-order kinetics are not an adequate model for ethanol elimination; the elimination rate is better described by Michaelis–Menten kinetics. M-M kinetics are approximately zero-order above a BAC of 0.15-0.20 g/L, but below this value alcohol is eliminated more slowly and the elimination rate more closely follows first-order kinetics. This change in behavior was not noticed by Widmark because he could not analyze low BAC levels.^[18] A 2023 study using a more complex two-compartment model with M-M elimination kinetics, with data from 60 men and 12 women, found statistically small effects of gender on maximal elimination rate and excluded them from the final model. Eating food in proximity to drinking increases elimination rate significantly.^[24]

In terms of fluid ounces of alcohol consumed and weight in pounds, Widmark's formula can be simply approximated as^[14]

${\displaystyle EBAC=8\times {\text{fl oz}}/{\text{weight in pounds}}-\beta \times T}$

for a man or

${\displaystyle EBAC=10\times {\text{fl oz}}/{\text{weight in pounds}}-\beta \times T}$

for a woman, where EBAC and β factors are given as g/dL (% BAC), such as a β factor of 0.0015% BAC per hour.^[14]

By standard drinks

United States standard drinks of beer, malt liquor, wine, and spirits compared. Each contains about 14 grams or 17.7 mL of ethanol.

The examples above define a standard drink as 0.6 fluid ounces (14 g or 17.7 mL) of ethanol, whereas other definitions exist, for example 10 grams of ethanol.

By training

If individuals are asked to estimate their BAC, then given accurate feedback via a breathalyzer, and this procedure is repeated a number of times during a drinking session, studies show that these individuals can learn to discriminate their BAC, to within a mean error of 9 mg/100 mL (0.009% BAC).^[26] The ability is robust to different types of alcohol, different drink quantities, and drinks with unknown levels of alcohol. Trained individuals can even drink alcoholic drinks so as to adjust or maintain their BAC at a desired level.^[27] Training the ability does not appear to require any information or procedure besides breathalyzer feedback, although most studies have provided information such as intoxication symptoms at different BAC levels. Subjects continue to retain the ability one month after training.^[28]

Other methods

Vitreous (eye) fluid provides an accurate account of blood alcohol concentration in cadavers.^[29]

Extrapolation

Retrograde extrapolation is the mathematical process by which someone's blood alcohol concentration at the time of driving is estimated by projecting backwards from a later chemical test. This involves estimating the absorption and elimination of alcohol in the interim between driving and testing. The rate of elimination in the average person is commonly estimated at 0.015 to 0.020 grams per deciliter per hour (g/dL/h),^[41] although again this can vary from person to person and in a given person from one moment to another. Metabolism can be affected by numerous factors, including such things as body temperature, the type of alcoholic beverage consumed, and the amount and type of food consumed.

In an increasing number of states, laws have been enacted to facilitate this speculative task: the blood alcohol content at the time of driving is legally presumed to be the same as when later tested. There are usually time limits put on this presumption, commonly two or three hours, and the defendant is permitted to offer evidence to rebut this presumption.

Forward extrapolation can also be attempted. If the amount of alcohol consumed is known, along with such variables as the weight and sex of the subject and period and rate of consumption, the blood alcohol level can be estimated by extrapolating forward. Although subject to the same infirmities as retrograde extrapolation—guessing based upon averages and unknown variables—this can be relevant in estimating BAC when driving and/or corroborating or contradicting the results of a later chemical test.

Citations

General and cited references

• Carnegie Library of Pittsburgh. Science and Technology Department. The Handy Science Answer Book. Pittsburgh: The Carnegie Library, 1997. ISBN 978-0-7876-1013-5.
• Perham, Nick; Moore, Simon C.; Shepherd, Jonathan; Cusens, Bryany (2007). "Identifying drunkenness in the night-time economy". Addiction. 102 (3): 377–80. doi:10.1111/j.1360-0443.2006.01699.x. PMID 17298644.
• Taylor, L., and S. Oberman. Drunk Driving Defense, 6th edition. New York: Aspen Law and Business, 2006. ISBN 978-0-7355-5429-0.