REMOTE BREATH ALCOHOL TESTING. W. R. Pi cton. D W I Alcohol Testing Ltd

Transcription

1 AN EVALUATION OF THE REMOTE BREATH ALCOHOL TESTING INDIUM ENCAPSULATION SYSTEM FOR W. R. Pi cton D W I Alcohol Testing Ltd Edmonton, Canada While it has obvious advantages over cumbersome blood alcohol procedures, one serious short-coming to breath alcohol testing is that because of police control the system does not appear to be fair. Forrester(l) described several areas of controversy in breath testing systems and detailed several criticism s by defense counsel of police control of breath testing. There seems l i t t l e doubt that a system which would provide an accused driver with the opportunity to have a sample of his own breath for independent analysis would be an improvement. Such a provision might be expected to confirm properly conducted breath te s ts, eliminate sloppy testin g, and greatly improve the fairness with which the system is perceived. In order to promote fairness in breath testin g, the Parliament of Canada passed legislation which included provision that an accused driver be offered a breath sample for independent analysis. This provision, passed in 1969, has not been proclaimed into force pending the development and testing of a suitable breath alcohol preservation procedure. Penton and Forrester(2) introduced a method of remote testing of breath for subsequent alcohol analysis. This method employs an indium encapsulation procedure and has had several improvements since introduction in It is used as a primary breath test procedure in some ju r is 648

2 d iction s(3). procedure was The indium encapsulation breath sampling examined wi.th respect to storage capability, environmental changes, and accuracy. A report(4) of this investigation was submitted to the Breath Test Committee of the Canadian Society of Forensic Science in The system u tilizes an indium tube which is held in a heated Field Sampler* and receives a sample of deep lung breath. The Field Sampler crimps the tube into three equal segments (capsules), each containing approximately 0.25 ml of breath. The trapped sample can be released and analyzed in a Gas Chromatographic Intoximeter (GCI). Tubes are supplied in kits* which also contain a disposable mouthpiece and a filte r. Simulated breath samples provided by Alcoholic Breath 2 Simulators, Mk 11, were used to calibrate equipment, test accuracy and precision of breath measurements, and provide a r tific ia l breath samples. Alcohol water concentrations 2 were prepared as indicated in the Breathalyzer Manual. Concentrations of these solutions are described by convention in terms o f an equivalent Blood Alcohol Concentration (BAC) at 34 C rather than the actual concentration of the al cohol-water solution from which they are generated. A "true" or "target" value was determined from the results of trip licate analyses of the a 1 coho 1-water solution by a modified Widmark procedure(5). GCI measurements of vapours from solutions prepared volumetrically were corrected when a subsequent chemical analysis indicated such action to be necessary. 500mL aliquots from a 5L parent solution equivalent to 152 mg/looml BAC were used to check calibration of the breath testing equipment. A new GCI Mk IV was operated in accordance with the instruction manual. No timing adjustments were necessary. Three figure digital results were obtained. A simultaneous recorder tracing provided a permanent record of the 1 Intoximeter Inc, St. Louis, Mo. 2 Smith & Wesson, Springfield, Mass. 649

3 650 D BACs. Results from a Breathalyzer Model 900, were estimated to the nearest 1 mg/ml. All Breathalyzer mea s u r e ments were taken from the "zero" line. When actual breath was tested, 3 mg/looml were subtracted from the resultant BAC in order to simulate a uniform "start" line. Indium sampling procedures as outlined by Jacobs and Goodson(6) were followed. Indium tubes were warmed for at least 5 minutes in a Field Sampler maintained at 5 0 C. Four calibration solutions equivalent to 45,90,97, and 138 mg/looml were used to derive a linear regression equation to relate the BACs obtained from indium analysis to the BACs obtained directly from the GCI without the use of indium encapsulation. The same Field Sampler was used to crimp all the capsules which were subsequently stored and was one of the three used in the following Field Sampler error e x p e r iment. Errors associated with individual capsules or a particular Field Sampler were examined. Aliquots of a 2L parent solution equivalent to 98 mg/looml BAC were used to calibrate the GCI, provide samples for encapsulation in 3 different Field Samplers, and provide samples for direct GCI analysis on each of 3 successive days. Two tubes were used each day in each Field Sampler. A testing sequence designed to eliminate sequential error was followed. GCI analysis of each separate capsule took place within an hour of encapsulation. Indium results were converted to BAC using the previously developed regression equation. Breath samples were provided by 27 male and female persons ages 19 to 39 yrs who weighed from 100 to 220 lbs. Testing began at least lk hrs after the end of alcoholic beverage consumption. Cubital vein blood was drawn by a qualified Medical Laboratory Technologist from an arm previously swabbed with 0.1% aqueous benalkonium chloride into loml Vacutainers containing thymol(7mg) and NaF (10mg). Chemical analysis of the refrigerated blood occured within 24 hrs of collection. Breath and blood 3 Becton, Dickinson & Co, Mississauga, Ont, Canada

4 samples were obtained from drinking subjects in a definite sequence of either Breathalyzer-blood-GCI-indium or Breathalyzer-blood-indiurn depending on the testing day. The longer sequence averaged 7.5 minutes (s = 2.0) while the shorter averaged only 3.1 minutes (s*1.2). Error due to body alcohol elimination is very small under these tight time schedules. One capsule was analyzed within 24 hrs(day 0) and the remaining two capsules from each indium tube were analyzed 14 to 17 days (Day 14) after collection. One of these two capsules was stored at ambient temperature (20-25 C) while the other was either subjected to environmental change or else sent to a distant laboratory for analysis. EXPERIMENTAL RESULTS Simulator solutions prepared in 50 mg/looml increments from 50 to 408 mg/looml were used to establish the accuracy and precision with which the GCI could measure BAC. Ten pairs of chemical(y) and GCI( X) analyses yeilded the following sets of regression equations: Y =. 949X ay1 = 3.4 X = 1.05Y Owl = 3.6 The correlation coefficien t, r =.999 In the range 50 to 200 mg/looml the following regression equations obtain: Y =. 994X -.1 ayl = 2.5 X = 1.00Y +.3 axl = 2.5 There is more accuracy and less variability over the narrower range of BACs. Over both ranges, the close agreement between GCI and chemical analysis indicates that the GCI, the Alcoholic Breath Simulator, and the chemical analysis are all under s ta tistic a l control. The accuracy and precision of the GCI measurement of simulated breath samples were studied with BACs at the concentration chosen for calibration. Ten samples generated from a 500mL aliquot of the 5L parent solution (152mg /looml) were analyzed on each of 6 days. Daily preparation of the GCI included regulation of the gas to a 651

5 constant pressure and the equilibration of the GCI at constant temperature for at least 30 minutes prior to use. (Jacobs and Goodson(6) did not "shut down" their GCI between days). The average GCI measurement was x = 153 mg/ looml, s = 2.6. No measurement varied from the target value by more than 5%. These results agree with the previously determined standard error of estimate, Oyl = 2.5, over the range of 50 to 200 mg/looml. Jacobs and Goodson reported a precision of 3.0 mg/10oml was obtained with a GCI,Mkll. Vapours from the four calibration solutions used to relate the indium capsule to the GCI response produced the following regression equation: Y = 1.13X + 4 where Y = chemical analysis and X = indium capsule BAC obtained by GCI analysis, r =.998. In ter and intra differences in Field Samplers were investigated with a random block experimental design of the type described by N atrella(7). There was no s ig n if i cant difference (a =.05) in the results obtained from capsules in the same Field Sampler but there were differences (a =.05) between Field Samplers. A daily effect also occured but this variation was thought to be due to a re-calibration of the GCI and not to the Field Samplers. A GCI is calibrated by adjusting the response to standard alcohol vapours. There can be variation in the sample size between GCIs or even in the same GCI with new 0 rings. These variations are of no importance and are not a source of error when breath is directly analyzed but become a source of error when a GCI is used to analyze capsules. The relationship between GCI measurement of direct breath samples and encapsulated samples can be determined with a regression equation. Once an equation is derived for a particular Field Sampler and a particular GCI then any change in the volume of the sample analyzed or to other parameters which effect quantitative gas chromatographic analysis will effect the result. The volume of the capsules is fixed for a particular Field Sampler 652

6 but the volume of breath analyzed by a GCI will vary if certain parameters are changed. Regression equations or other methods used to relate capsule and GCI measurements should be confirmed from time to time and especially whenever GCI re-calibration is performed. Differences in the sample volume of capsules crimped by different Field Samplers could be controlled by relating the volume of each Field Sampler to the volume crimped by a standard Field Sampler. The standard Field Sampler could be used to establish and check the regression equation for a particular GCI. A correction factor relating each Field Sampler to the standard Field Sampler would then relate each Field Sampler to the particular GCI. Jacobs and Goodson(6) established a relationship between a Field Sampler and their GCI by using standard alcohol vapours in each Field Sampler near the time an unknown breath sampie was obtained. Such a procedure invoives more of the r e la tiv e ly expensive indium. Regardless of which method is employed, the use of a correction factor will improve the accuracy of the method. Results from blood and breath samples collected at nearly the same time are shown in figure 1 & 2. There are several ways to compare methods of alcohol analysis. Percentage difference between methods is derived by: 1 % difference = ^ ^ Z Y-X y x 100 where Y = blood BAC i = 1 X = breath BAC, and N = number of pairs of tests. The error of estimate is a measure of the variability of Y associated with a particular X value, averaged over all values of X where Owl is the standard error of estimate and (Y*-Y) is the amount of squared error. 653

7 240 BLOOD ANALYSIS «' ' 0ML BREATHALYZER (MG / 100ML) FIGURE 1. NEAR-SIMULTAHEOUS SAMPLES OF BLOOD AND BREATH ANALYZED BY CHEMICAL AND BREATHALYZER PROCEDURES. 654

8 BLOOD ANALYSIS (MG / 100M L) INTOXIMETER (DIRECT) <***'»o om u FIGURE 2. I4EAR-SIMULTANEOUS SAMPLES OF BLOOD AND BREATH ANALYZED BY CHEMICAL AND GCI PROCEDURES. 655

9 A correlation coefficien t (r) at or near +1 denotes positive association of the X and Y values. The percentage of measurements within ± 20 mg/looml is a practical measure which can be used to predict in easily understandable terms, the variability associated with a particular measurement. Table 1 compares BACs obtained by blood and two breath methods. The BACs determined by blood analysis average about 9% greater than those obtained from GCI or Breathalyzer measurements. TABLE 1 - Comparison of BACs measured by blood alcohol analysis and near-simultaneous Breathalyzer and GCI analysi s No. of pairs of samples (N) Y = BAC from X = BAC from bl ood Breathalyzer blood GCI Y-X + 9.1% + 9.3% Regress i on equation Y on X Y = 1.11X-1 Y =. 95X+10 X on Y X = 0.87Y+4 X = 1.01Y-7 Error of estimate ay1 = 8.3 ayx = 8.8 Correlati on coefficien t (r) No. of pairs where Y-X is within ± 20 mq/looml 98% 94% The relationship between BACs obtained from indium capsules after 14 Day storage and from a Breathalyzer test taken at the time of encapsulation (Fig 3) is of importance in Canada. Breathalyzers are the instrument of choice by most police forces in that country. Table 2 shows that 90% BACs obtained from stored indium were within ± 20 mg/looml of Breathalyzer tests performed at the time of encapsulation. 656

10 BREATHALYZER (MQ / 100ML) INDIUM AFTER 14 DAYS AT ROOM TEMPERATURE FIGURE 3. NEAR-SIMULTANEOUS BREATH SAMPLES MEASURED IMMEDIATELY WITH A BREATHALYZER AND AFTER M DAYS OF INDIUM STORAGE AT ROOMTEMPERATURE WITH A GCI.

11 TABLE 2 - Relationship between BACs over the range 20 to 170 mg/looml measured by direct breath analysis and breath samples analyzed after 14 Days storage. No. of pairs BREATHALYZER of samples (N) GCI Y = BAC from Breathalyzer GCI X = BAC from Indi um Indi um Reg res s i on e a u a t i o n Y =. 86X+16 Y =. 99X+6 Error of,,» (oy 1) estimate Correlati on coefficien t (r) No. of pairs where Y-X is within ±20 mg/looml 90% 94% 98% of the indium analyzed after 14 days agreed within ±20 mg/looml with the analysis of capsules from the same tube made on Day 0. 97% of the indium analyzed on Day 0 agreed within ±20 mg/looml with direct Breathalyzer tests. 94% of the indium analyzed on Day 0 agreed within ±20 mg/ looml with directly sampled breath analyzed with the GCI. The third capsule from each tube was either subjected to environmental stress or analyzed at a distant laboratory. Capsules were immersed in boiling water, or frozen for 13 days at -19 C, or refrigerated for 13 days at +3 C. When analyzed on Day 14 all these treated capsules produced BACs within ±20 mg/looml of those from the same tube analyzed on Day 0. Encapsulated simulated breath samples which were heated to 150 C for 34 hrs tended to have lesser BACs than capsules from the same tube after both were stored for 17 days. Near-simultaneous Breathalyzer and Field Sampler tests were obtained from 10 persons. One capsule from the 658

12 Field Sampler was analyzed on Day 0. One of the remaining two capsules after 14 Day storage was analyzed in the GCI MklV used throughout the experimental work. The other capsule was analyzed in a GCI Mkll located about 3000 km away at the Ottawa laboratory of the Royal Canadian Mounted Police. The analyses were made on the same day in both locations. The Field Sampler which was used to encapsulate the 10 breath samples was sent to Ottawa so that the GCI Mkll could be calibrated for indium analysis. The results of this testing are shown in Table 3. BACs obtained from Breathalyzer and indium encapsulation were all measured within ±20 mg/looml of one another. TABLE 3 - BACs (mg/looml) from 10 subjects whose breath was sampled at the same time and place but analyzed in different places. BREATHALYZER INDIUM Day 0 Day 0 Day 14 Edmonton Edmonton Edmon ton Ottawa DISCUSSION - The 9% greater measurement obtained from d i rect blood analysis is consistant with similar bloodbreath comparisons by other workers during the past several years(8)(9). Morales(lO) found 97% of 14 Day indium capsule results to be within ±20 mg/looml of direct GCI analysis but he used the mean of 3 capsules in his as - 659

13 sessment. In this present study, BACs from individual capsules are reported. 94% fell within ±20 mg/looml of direct GCI analysis. When the difference in reporting is considered both studies found good agreement between Day 0 and Day 14 re suits. Harris(3) found evidence of a 13% difference between capsules from the same tube but this figure includes inter Field Sampler error and the analyses were made after at least 90 days storage. No significant difference (a=.05) between capsules was found under the conditions of our study, (ie: Field Sampler error id en tified and a much shorter storage time.) Both the Breathalyzer and the GCI used in the present study are of a kind approved for use by police forces in Canada. Experimental error always mitigates against identical results by two different analytical methods. We compared BACs obtained from Breathalyzer and GCI measurement with those obtained from Breathalyzer and Day 14 indium measurement. 90% of the Breathalyzer vs Day 14 indium and 91% of the Breathalyzer vs GCI results were within ±20 mg/looml of one another. Regression equations, error of estimate, and correlation coefficien ts displayed sim ilarity (Table 4). The variation due to indium encapsulation does not appear to be any greater than the variation which occurs when two different approved instruments are used to measure the breath from the same person. If direct analysis and 14 Day indium results can be expected 95% of the time within ±20 mg/looml of each other in the BAC range of forensic interest then they can be expected to be farther apart only 1 time in 20. If two direct breath tests and two encapsulated samples were to be taken then the probability that both of the encapsulated samples differ by more than ±20 mg/looml from the direct results due to chance alone would be 1/20 x 1/20 = Two indium tests which a re both more than ±20/100 ml different from the direct method would indicate a real difference in BAC measurement. One indium test within ±20 mg/looml of the direct tests would be consistant with the 660

14 direct tests. TABLE 4 - Relationships between Breathalyzer-indium Day 14 and Breathalyzer-GCI pairs. No. of pairs INDIUM DAY 14 GCI of samples (N) Y = BAC from X = BAC from Regression Breathalyzer Indium Day 14 Breathalyzer GCI equation Y =.86X+16 Y =.82 X+13 Error of /, \ (o y 1) estimate Correlation coefficien t (r) No. of pairs where X-Y is within ±20 mq/looml 90% 91% REFERENCES (1) Forres ter,m. R., Proc. 7th Int'l Conf. A le., Drugs, Traf. Safety, Melbourne, 1977, pp (2) Penton, J.R. and Forrester,M.R., Konferenzbericht der 5, Internati onal en Konferenz liber Alkohol und Verkelrssicherheit, Freiburg, 1969, ppll: (3) H arris,l.s., Hodnett,C.N., Mullen,J.B., Proc. 6th Int'l Conf. A le., Drugs, Traf. Safety, Toronto, 1974, pp (4) Picton, W.R., Oman,D.A., Donath.D.S., "A Laboratory Evaluation of the Indium Encapsulation System of Breath Alcohol Preservation and Analysis, May 1978", Can. Soc. Forensic S c i., Ottawa, Ont. (5) Smith,H.W., J. Lab. Clin. Med. 38, 1951, pp (6) Jacobs,W.B., Goodson, L.H., "Preservation and Analysis of Breath Alcohol Samples", MRI Project No B, Midwest Research Institute, Kansas City, Mo. 661

15 (7) Natrel1 a,m.g., "Experimental S ta tistic s", National Bureau of Standards Handbook 91, 1966, U.S. Dept of Commerce, NTIS, Springfield, Va. (8) Harger,R.N., "Recently Published Analytical Methods for Determining Alcohol in Body Materials", No. PB , 1974, NTIS, Springfield, Va. (9) Jones,A.W., Wright,B.M., Jones,T.P., Proc. 6th Int'l Conf. A le., Drugs, Traf. Safety, Toronto, 1974 pp (10) Morales,D.R., "Sum m ary of A ctivities Relating to Evaluation of Instruments and Related Accessories For Breath Alcohol Analysis, Report No. 2", California State Dept, of Health,