Reference Limits for Urine/Blood Ratios of Ethanol in Two Successive Voids from Drinking Drivers

Transcription

1 Reference Limits for Urine/Blood Ratios of Ethanol in Two Successive Voids from Drinking Drivers A.W. Jones Department of Forensic Toxicology, University Hospital, Link6ping, Sweden Abstract[ Specimens of venous whole blood and two successive urinary voids were collected from 450 individuals apprehended for driving under the influence of alcohol in Sweden. The first specimen of urine (UAC-1) was obtained as soon as possible after arrest, and the second void (UAC-2) was collected about 60 rain later (mean 66 rain, range ). A specimen of venous blood was drawn approximately 30 rain after the first urine sample was collected. Ethanol was determined in blood and urine by headspace gas chromatography, a method with high analytical precision (coefficient of variation -1%). The mean UAC for the first void was 2.60 g/l (range ) compared with 2.40 g/l (range ) in the second void. The mean concentration of alcohol in venous blood (BAC) was 1.97 g/l (range ). The concentrations of ethanol in the two voids of urine were highly correlated (r , residual standard deviation [SD] 0.22 g/l). The UAC and BAC results were also highly correlated; r = (residual SD 0.28 g/l) for the first void and r = (residual SD 0.21 g/l) for the second void. The concentration of ethanol in the first void (UAC-1) was higher than the second void (UAC-2) in 383 (87%) instances, decreasing by 0.23 g/uh on average. In 57 instances (13%), UAC-1 was less or equal to UAC-2 with a mean increase of 0.19 g/l. When BAC exceeded 0.5 g/l (N ), the mean UAC-1/BAC ratio was with 95 % reference limits of and 1.72, which agreed well with median (2.5th and 97.5th percentiles) of (0.938 and 1.79). For the second void, the mean UAC-2/BAC ratio was with 95% reference limits of and 1.45 and with a median (2.5th and 97.5th percentiles) of (0.997 and 1.46). These reference limits are appropriate to use when a person's venous BAC needs to be estimated with reasonable scientific certainty from the concentration determined in specimens of urine. Introduction Many review articles and books dealing with forensic aspects of ethanol mention that the average ratio of ethanol concentrations in urine and blood (UAC/BAC) is 1.3:1 (1-3). When the British government introduced their 1967 Road Safety Act, a blood-alcohol concentration (BAC) of 80 rag/100 ml (0.08 g%) was considered as equivalent to a urine-alcohol concentration (UAC) of 107 rag/100 ml (0.107 g%), which suggests a UAC/BAC ratio of 107/80 or 1.33:1 (4). This kind of legislation avoids the need to discuss variability in the UAC/BAC relationship for any individual suspect and a similar approach was followed when statutory breath-alcohol limits were introduced in the U.K. (35 ]~g/100 ml). In many U.S. states, threshold alcohol limits for driving are set at a blood-alcohol concentration of 0.08 g/100 ml or a breath-alcohol concentration of 0.08 g/210 L (blood/breath ratio 2100:1). Scant attention was given to the magnitude of variation in urine/blood and blood/breath ratios of ethanol between and within subjects, and any temporal changes during absorption, distribution, and elimination of ethanol were seemingly ignored (5). Several studies have shown that UAC/BAC ratios are often less than unity during the absorption stage of ethanol kinetics and increase to reach a value close to 1.3:1 during the early part of the post-absorptive phase and increase further as BAC decreases to approach zero (6-9). The UAC/BAC ratio also seems to depend on the concentration of ethanol in the samples with low concentrations giving high ratios, and therefore the amount of ethanol consumed needs consideration (7). The length of time that urine is stored in the bladder before voiding is also an important consideration because ethanol is continuously being removed from the blood by metabolism, but no oxidation of ethanol occurs in the bladder. This situation results in abnormally high UAC/BAC ratios being obtained for the first void. Likewise, a failure to empty completely the bladder on micturition is another factor that can skew the UAC/BAC ratio for the second void. The combined influences of many physiological and experimental variables means that whenever a measured UAC is translated into a presumed BAC, the result obtained is subject to considerable uncertainty (10). The magnitude of uncertainty in the UAC/BAC relationship for two successive voids collected from a large number of individuals suspected of drunk driving is the subject of this article. The results of this study will prove useful for clinical and forensic scientists whenever urine is the only specimen available for analysis of ethanol and when a request is made to interpret this information in terms of the person's likely BAC and impairment of performance and behavior. Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission. 33:]

2 Materials and Methods The blood and urine specimens used in this study were obtained from drinking drivers apprehended by the police in Sweden. All specimens were sent for analysis to the National Laboratory of Forensic Toxicology (Link~ping, Sweden). For the purpose of this study, cases were selected for evaluation if two successive urinary voids and a specimen of venous whole blood were available. The main defense challenge in Sweden is alleged consumption of alcohol after driving or after being involved in a traffic accident (11). In these situations, the UAC/BAC ratio and the change in UAC between two successive voids can furnish useful information to support or challenge this defense tactic (11,12). Drunk drivers in Sweden are apprehended in connection with (i) police sobriety check points, (ii) "tips" provided by other motorists who might have observed dangerous or erratic driving, (iii) after committing a moving traffic violation, and (iv) involvement in a traffic accident. Conducting field sobriety tests prior to administrating a roadside breath-alcohol screening test is not necessary in Sweden. If the preliminary breath test is positive, the suspect is taken to a police station where either an evidential breath-alcohol test is made or blood and urine specimens are collected for forensic analysis. If the driver is apprehended at the wheel and lacks the opportunity to claim consumption of alcohol after driving, then urine specimens are unnecessary. The times of collecting blood and urine could not be controlled exactly because of the different circumstances, such as the need for hospital treatment after a crash or the location of the traffic stop as well as other time elements. Police instructions require that the first urine void be taken as soon as possible after the driver is apprehended. The sample of venous blood is drawn about rain later when a physician or registered nurse arrives, and then the police attempt to secure a second sample of urine. The suspect is asked to empty the bladder completely on each void, and the total volume of urine collected is recorded. The urination is closely monitored by police to counteract attempts at specimen adulteration. A 10- ml aliquot from the pool of urine is transferred into a plastic tube containing NaF (100 rag) as preservative and then made airtight with a screw-on cap. Venous blood is taken under sterile conditions from an anticubital vein into Vacutainer tubes containing NaF (100 rag) and potassium oxalate (20 mg) as preservatives. Two tubes are filled in rapid succession. The tubes of blood and both urine specimens together with arrest forms are sent by express mail to the National Laboratory of Forensic Toxicology. The information written on the specimen tubes is compared with similar information on the accompanying forms, and any discrepancies are noted. The concentrations of ethanol in blood and urine are determined in duplicate by headspace gas chromatography as described in detail elsewhere (13). This method has a high analytical precision with a coefficient of variation of about 1%. Two technicians work independently with different sets of equipment and make duplicate determinations on blood and urine specimens. The means of duplicates were used in the calculation of UAC/BAC ratios, and the concentration units were grams per liter. The whole material consisted of 450 driving under the influence of alcohol (DUI) cases and the mean and standard deviation as well as median and 2.5th and 97.5th percentiles were used as estimates of central tendency and variability (14). The association between two variables, UAC-1 and UAC-2, for example, and between UAC and BAC was established by correlation-regression analysis, and the residual standard deviation (SD) indicates the random errors. Reference limits for the UAC/BAC ratios in the first and second voids were calculated as mean _+ (1.96 x SD) as well as by calculating the median and 2.5th and 97.5th percentiles. This latter method is independent of the shape of the distribution and can therefore be applied to variables that are not normally distributed (14,15). Confidence intervals (95%) for the UAC/BAC reference limits were calculated as described by Bland (14). Results The average time difference between collecting the two urine voids was 66 rain (range 30 to 130 rain) and venous blood was mostly taken rain after the first urinary void. Table I gives descriptive statistics for the whole material as well as selected sub-sets depending on the various conditions shown. Because the UAC/BAC ratios tended to increase as concentrations of ethanol in blood decreased, those cases with BAC less than 0.50 g/l (n = 21) were omitted from the main statistical analysis. Table I. Descriptive Statistics for Blood and Urine Alcohol Concentrations and Urine-to-Blood Ratios of Alcohol in First and Second Voids from Apprehended Drinking Drivers in Sweden Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) median median Sub-groups N BAC g/l UAC-1 g/l UAC-2 g/l UAC-1/BAC UAC-2/BAC All subjects (0.82) 2.60 (0.98) 2.40 (1.0) 1.38 (0.50) (0.19) 1.22 BAC > (0.75) 2.69 (0.89) 2.50 (0.91) 1.34 (0.19) (0.12) 1.22 BAC < (0.11) 0.59 (0.19) 0.40 (0.15) 2.21 (2.11) (0.70) 1.20 UAC-1 < UAC (0.88) 2.43 (1.12) 2.62 (1.12) 1.09 (0.18) (0.12) 1.22 UAC-I > UAC (0.80) 2.62 (0.95) 2.37 (0.97) 1.42 (0.52) (0.19) 1.22 UAC-1 > UAC (0.74) 2.72 (0.87) 2.46 (0.89) 1.38 (0.16) (0.11) 1.22 and BAC > 0.5 g/l 334

3 These probably reflect individuals who are approaching a zero blood-alcohol concentration. Furthermore, reference limits for UAC/BAC ratios were calculated separately when UAC-1 < UAC- 2 (n = 56), which might indicate that the absorption and distribution of alcohol in the body was incomplete. The remaining 429 cases were analysed in detail and used to determine reference limits for UAC/BAC ratios in first and second voids (Table II). A strong correlation (r = 0.973) was found between UAC-1 and UAC-2 in all 450 DUI cases and the residual SD was 0.22 g/l (Figure 1). The associations between UAC-1 and BAC and UAC- 2 and BAC are shown in Figure 2 as scatter plots. The correlation coefficients were highly significant (p < 0.001), and the residual SDs were 0.28 g/l for the first void and 0.21 g/l for the second void, indicating somewhat less random variation when a fresh urine specimen was obtained for analysis (second void). Figure 3 shows the UAC/BAC ratios plotted against the corresponding blood-alcohol concentration. A statistically significant negative correlation was found for the first void (r = -0.44) but not for the second void (r = ) where no trend was evident. This indicates that the value of the UAC/BAC ratio for freshly produced urine and provided that BAC is higher than 0.5 g/l is independent of the person's blood-alcohol concentration. 6.O".,." 5.0- r = I...-J... UAC-1 = ,953 UAC-2 i '"**~.,'" Resldu= so = o.=5 o/" I 9 :,~" i~ """ "" Table II gives mean, median, standard deviation, 95% reference intervals, and 2.5th and 97.5th percentiles for UAC/BAC ratios for both the first and second voids (iv = 429). The most likely UAC/BAC ratios for these drunk drivers are given by the mean or median of the distributions, being 1.33 for first void and 1.22 for the second void. These values could be used to gain an idea of what the person's BAC might be with a probability of 50%. However, because of analytical and physiological variations instead of an average or median value being used to estimate BAC, the 95% reference limits might be preferred. These limits were considerably narrower for the second void with 2.5% being above 1.45 and 2.5% below Because of the large sample size (N = 429), the 95% confidence intervals for these reference limits are fairly narrow (Table II). Dividing UAC in a second void by 1.47 (the upper confidence band) will give a highly conservative estimate of the person's venous BAC. By contrast, for a randomly collected first void, the UAC must be divided by 1.75 to achieve the same level of confidence in the estimated BAC. When only those cases with UAC-1 > UAC-2 are included (iv = 393), the mean and medians were about the same although the 95% reference limits were much narrower (data not shown). This stricter selection criteria probably eliminates all individuals that have not reached the post-absorptive phase of ethanol kinetics. According to Table I, the mean and median of the UAC/BAC ratios agree well for the second void, and the histogram and cumulative frequency plot show a very good fit to a Gaussian distribution (Figure 4) ,'.".,e ~.., ~ o e "".'""...'~ 9.~." ".. "~*. ~'.-"; *,.";"d, t 4.t.' 0.0 9,.,.., :0 UAC (g/l) (second void) Figure 1. Scatter plot and regression analysis of urine-alcohol concentration in first and second voids, where N = number of x-y pairs, r = correlation coefficient, and the dotted lines running parallel with the regression line are the 95% prediction intervals ( x residual standard deviation). Discussion Table II. Reference Limits and 95% Confidence Intervals for UAC/BAC Ratios of Ethanol for First and Second Voids Collected from Apprehended Drinking Drivers with BAC > 0.5 g/l* Reference limits UAC/BAC and (95% UAC/BAC 2.5thand97.5th Urine specimen mean (SD) CV% confidence intervals) median percentiles First void (0.192) ( ) and ( ) Second void (0.119) ( ) and ( ) * N = 429. For comparison non-parametric median and 2.5th and 97.5th percentiles are given. Urine has a long history as a biological specimen for the analysis of alcohol and the results corroborate any clinical signs and symptoms of drunkenness as noted by Erik M.P. Widmark in 1914 (16). Interest in measuring ethanol in body fluids for legal purposes accelerated in the 1930s when it was found that many traffic crashes and deaths on the highway were caused by drunk drivers (17,18). Evidence of drunkenness and unfitness to drive was obtained by recording various signs and symptoms of intoxication along with the concentration of ethanol in a specimen of urine (19). Urine and breath were the preferred specimen for forensic analysis of alcohol because obtaining blood raised the constitutional issue of self-incrimination, whereas the urine a person voided and the breath exhaled were considered waste products voluntarily discarded (17-19). Over the past century, a vast literature has accumulated on the UAC/BAC relationship in specimens collected from both living and dead subjects (20-25). In postmortem toxicology, the analysis and interpretation of ethanol concentrations measured in more than one body fluid (blood and urine or blood and vitreous humor) is virtually essential to boost confidence in the results (26). By contrast, in living subjects, despite severe penalties and sanc- 335

4 i Journal of Analytical Toxicology, Vol. 26, September 2002 tions for DUI, the use of more than one body fluid for the determination of ethanol is extremely rare. The methods for measuring ethanol in urine are accurate and precise, and the concentrations remain stable during long periods of storage at 4~ (27,28). Some potential problems with the use of urine for the analysis of ethanol were dispelled when it was shown that neither urine-creatinine, a biochemical marker for dilute specimens, nor drinking water before voiding had any negative impact on the UAC/BAC relationship (29,30). Another often cited criticism of urine-ethanol measurements relates to the phenomenon of urine retention, which has been suggested might lead to abnormally high UAC/BAC ratios (31). Although it is widely recognized that older men with prostrate enlargement have difficulties emptying the bladder completely on each void, it is also well-known that these individuals tend to urinate more frequently (32). Experiments are needed to establish the mean and variation of UAC/BAC ratios in a well-characterized patient material made up of men and women with problems in emptying the bladder completely on micturition. To the author's knowledge, these experiments have not yet been published. It seems reasonable to assume that the concentration of ethanol in any residual urine in the bladder, which was pro- duced when the BAC was high becomes diluted with freshly produced ureter urine that might contain a lower concentration of ethanol because of the metabolism of ethanol taking place in the liver. When the BAC eventually reaches zero, the residual urine in the bladder that might still contain ethanol starts to become diluted with alcohol-free urine. Accordingly, the first morning void after an evening's drinking could contain ethanol reflecting the BAC prevailing during the night. But the second morning void will be alcohol-free provided that the BAC has reached zero at the time of the first morning void, and one of the most urgent tasks on arising in the morning is to visit the bathroom (33). In the present large group of DUI suspects, the UAC/BAC ratios were somewhat higher for the first void (median 1.33), compared with a second void (median 1.22). The residence time of urine in the bladder before collecting the first void is unknown but is probably longer than the mean of 66 rain before the second void was collected. During the production and storage of urine in the bladder, a person's BAC does not remain static but decreases by 0.10 to 0.25 g/l/h in the post-absorptive state. The median UAC/BAC ratio of the second void was 1.22 probably representing freshly produced urine and a much shorter storage time in the bladder. If the water content of N=450 ~ j 9 g5.0 r= " 9 /... UAC = BAC..'~" ~.'" ~' Residual SD = 0.28 g/l... "~:'4rL~../" '~ 4,0 :";9 ~ 9 "".... %.', -g :,, 9... O...~., 9 9 "" "~" First void,.o... :S;; 3 0.0,,,, ~, 6.0.~ N=4~o./' ~. ~.o-i ~ = o.~7o.. : / / "B ~ UAC = o.o~, 1.10,AC. ~. / Residual SD = 0.21 g/l...;,,j,.dl...' "o 4.0-I,,,_. FT :." o." ",.'' 0 I /... 9,'~ I~'~;'. 9 ''' 9 2.ol O J..;,".."9 ~ 10 1 " '";" E. ' ISec~176 I '~:3 0.0~S'""" "'"",".,.,., Figure 2. Scatter plots and regression analysis of urine-alcohol concentration against blood-alcohol concentration for first and second voids, where N = number of x-y pairs, r = correlation coefficient, and the dotted lines running parallel with the regression line are the 95% prediction intervals ( x residual standard deviation). i IN= [r = ' " I UAC/BAC = BAC ~ ~ ;~..,... " ; " o ~0.5 " 0.0 o o 1:o 2:0 3:o 4:o 2.5" ~' 2.0- "~ 1.5- & O 1.0-,~ I N=429 r = I UAC/BAC = BAC.'. 'LL... ~ ",~_:i~.~l;z,;.~. 'Z. ;;5;' ~;:... I.~... ~...~..~...p...p.,,.,...*.... =... 1'o 21o 31o 41o Figure 3. Scatter plots showing the relationship between urine/blood ratios of ethanol for first (upper frame) and second voids (lower frame) and the underlying venous blood-ethanol concentration, where N = number of x-y pairs, r = correlation coefficient, and the dotted lines running parallel with the regression line are the 95% prediction intervals ( x residual standard deviation). s:o 336

5 urine is taken as 100% (v/v) compared with 80-85% (v/v) for whole blood, this gives a theoretical UAC/BAC ratio of 1.18:1 to 1.25:1 and these are the ratios expected if the urine was obtained by catheterization. After a period of continuous drinking, especially with a rising BAC, the diuretic action of ethanol ensures that the person urinates frequently, so even the first void from drunk drivers is a fair representation of the BAC existing 1-2 h earlier. Dividing the UAC of the first void by 1.33 gives an idea of the average BAC during the time that the urine was being produced and stored in the bladder. Accordingly, the concentration determined in the first urine sample collected after driving is a good representation of the BAC at the time of driving. This unique advantage of urine over other body fluids for forensic analysis, namely to give an indication of the BAC at an earlier point in time (i.e., the time of driving or crash), has surprisingly not been fully appreciated. The person's BAC at the time of o>, u_ lool,,o '0 t. / 1,o t /.,, /j UAC/BAC (second void) oo" oo I i o [ ' i i!. i 9 i UAC/BAC (second void) Figure 4. Frequency distribution (upper part) of urine/blood ratios of alcohol (N = 375) for the second urinary void when the BAC was above 0.5 g/l and when the concentration in first void (UAC-1) was higher than for the second void (UAC-2). The lower plot shows UAs ratios as a cumulative distribution indicating a good fit to a normal Gaussian curve with arrows showing median (50%) value of UAC/BAC ratio indicated. driving is often a hotly debated issue in DUI litigation. If a more conservative estimate of BAC at time of driving is needed, instead of dividing the UAC by the mean or median UAC/BAC ratios of 1.33, an upper 97.5th percentile could be used instead (Table II). Such a calculation, with reasonable scientific certainty, yields a minimum value for the person's BAC at the time of driving. If a second urinary void is collected, dividing by 1.22 gives a good estimate of BAC during the period of collection of urine in the bladder and dividing by 1.47 gives a result less than the true value with a probability of 40 to 1. The choice between mean and median values as measures of central tendency and 95% reference limits derived from SD or 2.5th and 97.5th percentiles is somewhat arbitrary considering that the distribution of UAC/BAC ratios, for the second void agreed well with a Gaussian curve. The first urinary void should never be discarded because the ethanol concentration gives valuable information about the person's BAC some time prior to obtaining the blood sample such as at the time of driving. As recognized by Heise (23), this is a unique advantage of urine over other body fluids used for analysis of alcohol. Other strong advocates of measuring alcohol in urine were Biasotti and Valentine (31) who presented an excellent review of forensic aspects of UAC and variability in UAC/BAC ratios. Their study and the conclusions therein deserve close attention from those interested in the use of urine as a biological specimen for alcohol analysis in traffic law enforcement and forensic science. Forensic toxicologists are often required to interpret the concentrations of illicit drug in urine in relation to the time of last use and occasionally also the amount taken or the likely blood or plasma concentration (34). For drugs other than ethanol, such questions are difficult or even impossible to answer with any degree of certainty. The physicochemical properties of ethanol and the high correlation between the concentrations in blood and urine (Figure 2) are unlike those of other drugs of abuse. The transfer of ethanol from blood to urine occurs by passive diffusion, and the concentration in the glomerular filtrate will mirror the concentration in plasma. Ethanol shows negligible binding to plasma proteins and distributes evenly into the total body water, making it a lot easier to interpret the concentration of ethanol in urine compared with all other abused drugs (7,8,31). The risk for post-sampling synthesis of ethanol should not be overlooked when urine-ethanol concentrations are interpreted (35,36). This problem can be avoided if the collection tubes contain at least 1% NaF and are stored in a refrigerator (+ 4~ pending analysis (37,38). These requirements are very important if information suggests that the donor was diabetic and therefore prone to excrete sugar in urine and/or was suffering from a urinary tract Candida infection (39). Without an enzyme inhibitor, glucose in the urine together with viable yeast or bacteria lead to production of ethanol if the samples are stored at room temperature for at least 24 h (38). Whenever postsampling synthesis of ethanol is suspected, a control analysis can be performed after allowing an aliquot to stand at room temperature in a closed container for a few days (26). An appreciable increase in the concentration of ethanol above that expected from analytical imprecision alone is evidence for 337

6 on-going production of ethanol. However, absence of such an increase is not proof that post-sampling synthesis of ethanol had not already occurred earlier and all fermentable glucose converted before making the control analysis. Herman Heise, a U.S. pioneer in use of chemical tests for alcohol influence (23) appreciated that the UAC/BAC relationship could also be used to support or challenge alleged consumption of alcohol after driving, which is a very common DUI defense tactic (11,12). The BAC existing between two successive voids can be estimated by dividing the UAC (second void) by a factor of 1.33, the presumed population average urine/blood ratio of ethanol. To avoid discussion and debate about a person's urine/blood ratio of ethanol when making these conversions, some jurisdictions have endorsed a threshold concentration of ethanol in urine as a punishable offence, such as 107 mg/dl the current legal limit in the U.K. In the state of Minnesota, a statutory UAC of 0.1 g per 60 ml urine for a single void was taken as being equivalent to a BAC of 0.1 g/100 ml and hence a UAC/ BAC ratio of 1.5:1 (100/60). The results from this study, which involved the analysis of blood and urine specimens from 450 individuals apprehended for alcohol-impaired driving, have given reference limits for the UAC/BAC ratios in first (random) and second urinary voids. These limits are appropriate for use in civil and criminal litigation when UAC needs to be converted into the expected BAC with reasonable scientific certainty. Dividing UAC by the 97.5th percentile of the UAC/BAC distribution gives a conservative estimate of the person's actual BAC during the collection period of urine in the bladder. The BAC obtained by this calculation, with a probability of 40:1, will not exceed the true value. A more conservative estimate can be obtained by use of the 99th percentile of the UAC/BAC distribution ratio (data not shown). Acknowledgment This work was supported in part by funding from the National Board of Forensic Medicine. References 1. R.N. Harger. Ethyl alcohol. In Toxicology, Mechanisms andanalytical Methods, C.P. Stewart and A. Stolman, Eds. Academic Press, New York, NY, 1961, pp R.N. Harger and R.B. Forney, St. Aliphatic alcohols. In Progress in Chemical Toxicology, Vol. 3, A. Stolman, Ed. Academic Press, New York, NY, 1967, pp J.C. Garriott. Medicolegal Aspects of Alcohol. Lawyers and Judges Publishing, Tuscon, AZ, Special Report, The Drinking Driver. British Medical Association, London, U.K., H.J. Walls and A.R. Brownlie. Drink, Drugs and Driving. Sweet and Maxwell, London, U.K., 1985, pp M. Staak, E. Springer, and H. Baum. Vergleichende Untersuchungen ~ber den Aussagewert des HarnalkohoI-Blutalkoholquotienten unter experimentellen Bedingungen sowie im Probenmaterial. Blutalkohol 13: (1967). 7. A.W. Jones. Ethanol distribution ratios between urine and capillary blood in controlled experiments and in apprehended drinking drivers. J. Forensic Sci. 37:21-34 (1992). 8. P. Zink and G. Reinhardt. Zur Theorie der,~thanolausscheidung im menschlichen Urin. Blutalkohol 8:1-15 (1971 ). 9. S. Hishida, M. Kinoshita, I. Ijiri, T. Okada, J. Adachi, and Y. Mizoi. Studies on the ratio between alcoholic concentrations in urine and blood. Jpn. J. Legal Med. 27: (1973). 10. C.L. Winek, K.L. Murphy, and T.A. Winek. The unreliability of using a urine ethanol concentration to predict a blood ethanol concentration. Forensic Sci. Int. 25: (1984). 11. A.W. Jones. Top-ten defense challenges among drinking drivers in Sweden. Meal. Sci. Law31" (1991). 12. R. Iffland. Nachtrunk und Harnprobe. Blutalkohol 36: (1999). 13. A.W. Jones and J. Schuberth. Computer-aided headspace gas chromatography applied to blood-alcohol analysis: importance of online process control. J. Forensic Sci. 34' (1989). 14. M. Bland. An Introduction to Medical Statistics, 3rd ed. Oxford University Press, Oxford, U.K., H.E. Solberg. Establishment and use of reference values. In fietz Textbook of Clinical Chemistry, ch. 13, 2nd ed., C.A. Burtis and E.R. Ashwood, Eds. W.B. Saunders, Philadelphia, PA, E.M.P. Widmark. Om alkoholens ~verg~ng i urinen saint en enkel kliniskt anv~nbar metod f~r diagnostocering af alkoholf6rekonst i kroppen. Uppsala L~'karef6reningens F6rhandlinagm 19: (1914). 17. A.W. Jones. Measuring alcohol in blood and breath for forensic purposes--a historical review. Forensic ScL Rev. 12: (2000). 18. M. Ladd and R.B. Gibson. The medicolegal aspects of the blood test to determine intoxication. The Iowa LawJ. 24:1-77 (1939). 19. R.L. Donigan. Chemical test case law; Legal aspects and constitutional issues involved in chemical tests to determine intoxication. Northwestern University, Evanston, IL, 1950, pp W.R. Miles. The comparative concentrations of alcohol in human blood and urine at intervals after ingestion. J. Pharmacol. Exp. Ther. 20: (1922). 21. H.W. Haggard, I.A. Greenberg, R.P. Carroll, and D.P. Miller. The use of the urine in the chemical test for intoxication. J. Am. Med. Assoc. 115: (1940). 22. F. Lundquist. The urinary excretion of ethanol in man. Acta Pharmacol. Toxicol. 18: (1961 ). 23. H. Heise. Concentrations of alcohol in samples of blood and urine taken at the same time. J. Forensic Sci. 12: (1967). 24. J.P. Payne, D.V. Foster, D.W. Hill, and D.G.L. Woods. Observations on interpretation of blood alcohol levels derived from analysis of urine. Br. Med. J. 3: (1967). 25. W.H.D. Morgan. Concentrations of alcohol in samples of blood and urine taken at the same time. J. Forensic Sci. Soc. 5:15-21 (1965). 26. A.W. Jones. Alcohol; postmortem. In Encyclopedia of Forensic Sciences, J.A. Siegel, P.J. Saukko, and G.C. Knupfer, Eds. Academic Press, London, U.K. 2000, pp P.M. Hayden, M.T. Layden, and M.D. Hickey. The stability of alcohol content in samples of blood and urine. IrishJ. Med. Sci. 146: (1977). 28. W. Neuteboom and P.G.M. Zweipfenning. The stability of the alcohol concentration in urine specimens. J. Anal. Toxicol. 13: (1989). 29. A.W. Jones. Lack of association between urinary creatinine and ethanol concentrations and urine/blood ratios of ethanol in two successive voids from drinking drivers. J. Anal. Toxicol. 22' (1998). 30. P. Bendtsen and A.W. Jones. Impact of water-induced diuresis on urine-ethanol profiles, urine-creatinine, and urine-osmolality. J. Anal. Toxicol. 23: (1999). 31. A.A. Biasotti and T.E. Valentine. Blood alcohol concentration determined from urine samples as a practical equivalent or alternative to blood and breath alcohol tests. J. Forensic Sci. 30' (1985). 338

7 32. M. Emberton and K. Anson. Acute urinary retention in men: an age old problem. Br. Med. J. 331]: (1999). 33. L. Kadehjian. Urine alcohol testing; valid and reliable. Syva Monitorg: 9-12 (1991}. 34. R.H. Liu and B.A. Goldberger. Handbook of Workplace Drug Testing, AACC Press, Washington, D.C., 1995, pp W.D. Alexander, P.D. Wills, N. Eldred, and R. Gower. Urinary ethanol levels and diabetes. Lancet i: 789 (1981). 36. J.J. Saady, A. Poklis, and H.R Dalton. Production of urinary ethanol after sample collection. J. Forensic Sci. 38: (1993). 37. RS. Lough and R. Fehn. Efficacy of 1% sodium fluoride as a preservative in urine samples containing glucose and Candida albicans. J. Forensic Sci. 38: (1993). 38. A.W. Jones, L. Hyl~n, E Svensson, and A. Helander. Storage of specimens at +4~ or addition of sodium fluoride (1%) prevents formation of ethanol in urine inoculated with Candida albicans. J. AnaL Toxicol. 23: (1999). 39. A.W. Jones, A. Eklund, and A. Helander. Misleading results of ethanol analysis in urine specimens from rape-victims suffering from diabetes. J. Clin. Forensic Med. 7: (2000). Manuscript received November 26, 2001 ; revision received March 26,