Determination of Ethanol in Breath and Estimation of Blood Alcohol Concentration with Alcolmeter S-D2 A.W. Jones and KÄ. Jönsson Departments of Alcohol Toxicology and Internal Medicine, University Hospital, 581 85 Linköping, Sweden. 1. Introduction The technology of breath-alcohol analysis can be subdivided into two broad categories; devices for pre-arrest screening of drunk drivers and more sophisticated instruments for quantitative evidential purposes (1,2). The Lion Alcolmeter S-D2 is a multipurpose hand-held device designed originally for use as a roadside screening test. However, its easy-to-read digital display and highly reproducible breath sampling system, has created many applications for this instrument in alcohol research, industrial chemistry, and clinical and emergency medicine. Although the Lion Alcolmeter S-D2 has been available for many years, hardly any studies have dealt with its precision and accuracy. When breath-alcohol analyzers are used for legal purposes the results are mostly reported as the concentration in the breath sample analyzed (BrAC) expressed in units of mg/l, jig/loo ml, or g/210 1. However, when breath-alcohol analyzers are used in research and clinical medicine, the BrAC is almost always translated into the presumed concentration in a specimen of blood (BAC) having units of g/1, mg/dl, or g% w/v. To make the conversion from BrAC into BAC, one assumes a constant blood:breath ratio of alcohol (usually 2100:1 or 2300:1), which is used to calibrate the instrument (3). The precision and accuracy of Lion Alcolmeter S-D2 was evaluated in experiments in-vitro with known strength air-alcohol standards and also invivo with volunteer subjects who ingested moderate amounts of alcohol. 2. Material and methods 2.1 Breath-alcohol instruments Four examples of the Lion Alcolmeter S-D2 breath-alcohol analyzer were provided by the manufacture (Lion Laboratories PLC, Barry, South Alcohol, Drugs and Traffic Safety - T92 Ed. by Utzelmann / Berghaus I Kroj Verlag TÜV Rheinland GmbH, Köln -1993 423
Glamorgan, Wales, UK). The Alcolmeter instruments were fitted with digital scales marked in mg/1 breath and readings could be made to two decimal places. After an initial control of the calibration, no further adjustments were made throughout the experiments. Another Alcolmeter S-D2 instrument was available and this had a digital scale marked in terms of mg/dl presumed blood-alcohol concentration based on a 2300:1 blood:breath ratio. 2.2 Experiments in vitro Air-alcohol mixtures were generated using a wet-bath simulator device manufactured by Guth Laboratories, Harrisburg, USA. The simulator operates at 34 C and the air/water partition coefficient of ethanol at this temperature was taken from the literature, being 0.393 x 10'3 (4). The concentration of ethanol in the equilibrated air-vapor emerging from the simulator was calculated as the product of the concentration of ethanol and the air/water partition coefficient. The simulator was charged with aqueous ethanol standards prepared from absolute ethanol by first making a 10% w/v stock solution and then volumetric dilution to give working standards at concentrations of 50, 100, 175, 380 and 640 mg/dl. The exact concentrations prepared were checked by headspace gas chromatography and were within ± 2% of the target values and often much better than this. The Lion Alcolmeter S-D2 instruments were fitted with mouth-pieces normally supplied. These tubes were joined to the outlet of the wet-bath simulator using a 2 cm long plastic tubing. The operator blew through the simulator to produce an equilibrated breath-sample at 34 C. When the airvapor standard had passed through the simulator for about 5 seconds, the READ button of the Lion Alcolmeter S-D2 was pressed capturing a sample for analysis. The 4 Alcolmeter S-D2 instruments were positioned side-by-side and exposed 10 times to each of the air-alcohol vapor mixtures. The same simulator charge was used for all 10 tests at each concentration of ethanol. The time interval between making the replicate tests varied from < 5 min to 15 min and was sometimes longer depending on practical circumstances. 23 Experiments in-vivo Ten male volunteers took part in these experiments. They drank a dose of ethanol equivalent to 0.80 g ethanol per kg body weight in 30 minutes. Ethanol was ingested in the morning, either after an overnight fast or after the volunteers had eaten a standard breakfast. An indwelling catheter was implanted in a cubital vein for taking blood samples at exactly 10, 20, 30, 424
45, 60, 90, 120, 150, 180, 240, 300, 360, and 420 min after the start of drinking. Blood was drawn into 5 ml Vacutainer tubes (Becton Dickinson, USA) containing sodium fluoride (20 mg) and 75 units heparin. The catheter tubing was flushed with heparin/saline between taking successive samples. The concentration of ethanol in blood was determined in duplicate by headspace gas chromatography as described in detail elsewhere (5). The standard deviation (SD) of this method was 0.6 mg/dl at a mean BAC of 80 mg/dl. 3. Results 3.1 Precision and accuracy of analysis in-vitro Table 1 summarizes the precision and accuracy of the Alcolmeter S-D2 when used to analyze air-alcohol vapor standards. The coefficients of variation (CV), as one index of precision, were between 2% and 2.5% at ethanol concentration up to 0.40 mg/1 and the mean response was within ± 0.02 mg/1 (± 1%) of the target values. When exposed to higher airvapor concentrations of alcohol, the CV was between 2% and 5% and accuracy was within ± 6.6% of the target values. The consecutive tests at 5-10 min intervals tended to decrease as the number of tests increased at concentrations of ethanol at 0.70 mg/1 and higher. The components of variance between and within instruments depended in part on the concentration of alcohol in the air-vapor samples analyzed. At 0.20 mg/1, the SD between instruments was 0.0056 mg/1 and within instruments the SD was 0.0028 mg/1. The pure analytical SD was 0.0037 mg/1. Table 1. Precision and accuracy of Alcolmeter S-D2 when used to analyze air-alcohol vapor standards produced from a breath simulator device. BrAC Target Value mg/1 Mean Response mg/1 (N = 10) SD mg/1 CV% Bias % 0.20 (46)1 0.201 0.005 2.5 +0.5 0.40 (92)1 0.404 0.009 2.2 + 1.0 0.70 (161)1 0.677 0.014 2.1-3.2 1.50 (343)1 1.40 0.049 3.5-6.6 2.50 (575)1 2.34 0.117 5.0-6.4 1 Approximate blood-alcohol concentration in mg/dl based on 2300:1 ratio. 425
3.2 Experiments in-vivo Figure 1 gives two examples of blood and breath ethanol profiles for a relatively fast and slow absorber of alcohol. T im e, min T im o, min Fig. 1. Concentration-time profiles of ethanol in venous blood analyzed by gas chromatography and values estimated using Lion Alcolmeter S-D2. The mean difference between duplicate determinations with Alcolmeter S- D2 (N = 109) was -0.11 ± 0.26 mg/dl (± SD) and this was not significantly different from zero. The SD of the differences was 2.81 mg/dl and the 95% range of individual differences was from -5.5 to +5.7 mg/dl. The precision of a single determination is given by 2.81/21being 1.98 mg/dl expressed in terms of the coexisting BAC. The mean difference between venous BAC and Alcolmeter S-D2 response (N = 109) was -0.37 ± 0.81 mg/dl (±SD) 426
and this was not significantly different from zero. The SD of the differences and the 95% limits of agreement were 8.45 mg/dl and -17 to + 16 mg/dl respectively. Figure 2 shows a blood/breath scatter plot of venous blood alcohol (x-variate) and breath alcohol (y-variate) analyzed with Lion Alcolmeter S-D2. The high correlation (r = 0.94), near zero intercept and slope close to unity confirms the reliability of this device and the lack of bias when used to estimate blood alcohol concentration. 120 100 O 05 E 80 OJ Q1 (/) 60 L_ <D E 40 o < 20 0 0 20 40 60 80 100 120 Venous Blood Alcohol, m g/dl Fig. 2. Scatter plot of venous blood alcohol and estimated BAC according to tests with Lion Alcolmeter S-D2. Note the near zero intercept and slope close to unity. 3 3 Variability between instruments Figure 3 compares the concentration-time profiles of ethanol in 2 subjects tested in rapid succession with each of 4 Alcolmeter S-D2 instruments. The SD of breath-alcohol measurements between instruments was 0.01 mg/dl which corresponds to 2.3 mg/dl BAC equivalent. 4. Discussion The uncertainty of any breath-alcohol test depends on analytical and physiological sources of variation. Physiological variations such as the way the subject exhales - pre-breathing maneuver - dominates the total 427
variability. Differences between instruments might depend on the care taken in calibration procedure. The relative contribution of analytical and biological sources of variation can be evaluated using analysis of variance after making replicate measurements in each subject with several different instruments. T im e, h T im #, h Fig. 3. Concentration-time profiles of ethanol in breath for 2 subjects tested with 4 different Alcolmeter S-D2 instruments. The in-vitro experiments showed that at high concentrations of alcohol in the breath (> 0.40 mg/1 or 92 mg/dl BAC) the Alcolmeter response tended to decrease slightly as the number of tests made at short intervals (5 min) increased. This observation has been reported elsewhere and is caused by a "fatigue effect" of the electrochemical cell (6). Presumably, the products of ethanol oxidation (acetaldehyde) requires time to diffuse away from the electrode surface interfering with the oxidation of ethanol in each new test. Experiments with human subjects showed that the Alcolmeter response 428
was mostly higher than venous BAC during the absorption-distribution phase of ethanol kinetics. In the post-absorptive phase the Alcolmeter response was almost always less than the venous BAC (4). This observation is not new and reflects the uneven distribution of alcohol in the vascular system and arterio-venous difference existing after subjects ingested a bolus dose of ethanol on an empty stomach. Breath-alcohol follows more closely the arterial blood concentration and this exceeds the venous BAC until equilibrium distribution of alcohol in all body fluids and tissue is complete. Moreover, the magnitude and direction of these BAC-BrAC differences depends also on the value of the blood/breath calibration factor used. In conclusion, the precision and accuracy of Lion Alcolmeter S-D2 was exceptionally good for a hand-held instrument with manual sampling procedure. The training and experience of the person operating the unit plays an important role in the precision of breath sampling. As expected, accuracy and precision in-vitro was better than in tests with human subjects. The stability of the calibration was remarkably good over long periods and no adjustments were deemed necessary. The calibration of the instruments should be checked periodically but making haphazard changes is not recommended. 5. References 1. Dubowski, KM. The technology of breath-alcohol analysis. US Department of Health and Human Services, DHHS publication No (ADM)92-1728, 1992, pp 1-38. 2. Jones, AW. The measurement of alcohol in blood and breath for legal purposes. In Human Metabolism of Alcohol, Vol. 1, edited by K E. Crow and R.D. Batt, CRC Press, 1989, pp 71-99. 3. Jones, AW. Blood and breath alcohol concentrations, Brit. Med. J. 305:955:1992. 4. Jones, AW. Variability of blood:breath alcohol ratio in-vivo. J. Stud. Ale. 1978;39:1931-1939. 5. Jones AW, Schuberth, J. Computer-aided headspace gas chromatography applied to blood-alcohol analysis: Importance of on-line process control. J Forens Sei 1989;34:1116-1127. 6. Jones AW. Evaluation of breath alcohol instruments: In-vitro experiments with Alcolmeter pocket model. Forens Sei Int 1978;12:1-9. 429