Forensic Science International 135 (2003) 206212

Urine/blood ratios of ethanol in deaths attributed to acute alcohol poisoning and chronic alcoholism
A.W. Jones*, P. Holmgren
Department of Forensic Toxicology, National Laboratory of Forensic Chemistry, University Hospital, 581 85 Linkoping, Sweden Accepted 2 May 2003

Abstract The concentrations of ethanol were determined in femoral venous blood (BAC) and urine (UAC) and the UAC/BAC ratios were evaluated for a large case series of forensic autopsies in which the primary cause of death was either acute alcohol poisoning (N 628) or chronic alcoholism (N 647). In alcohol poisoning deaths both UAC and BAC were higher by about 2 g/l compared with chronic alcoholism deaths. In acute alcohol poisoning deaths the minimum BAC was 0.74 g/l and the distribution of UAC/BAC ratios agreed well with the shape of a Gaussian curve with mean standard deviation (S.D.) and median (2.5th and 97.5th centiles) of 1:18 0:182 and 1.18 (0.87 and 1.53), respectively. In alcoholism deaths, when the BAC was above 0.74 g/l (N 457) the mean S:D: and median (2.5th and 97.5th centiles) UAC/BAC ratios were 1:30 0:29 and 1.26 (0.87 and 2.1), respectively. When the BAC was below 0.74 g/l (N 190), the mean and median UAC/BAC ratios were considerably higher, being 2.24 and 1.58, respectively. BAC and UAC were highly correlated in acute alcohol poisoning deaths (r 0:84, residual S:D: 0:47 g/l) and in chronic alcoholism deaths (r 0:95, residual S:D: 0:41 g/l). For both causes of death (N 1275), the correlation between BAC and UAC was r 0:95 and the residual S.D. was 0.46 g/l. The lower UAC/BAC ratio observed in acute alcohol poisoning deaths (mean and median 1.18:1) suggests that these individuals died before absorption and distribution of ethanol in all body uids were complete. The higher UAC/BAC ratio in chronic alcoholism (median 1.30:1) is closer to the value expected for complete absorption and distribution of ethanol in all body uids. # 2003 Elsevier Ireland Ltd. All rights reserved.
Keywords: Alcohol; Alcoholism; Autopsy; Bloodethanol; Death; Urine; Urine/blood ratio

1. Introduction Interpreting results of postmortem alcohol analysis is simplied if multiple specimens are submitted for toxicological evaluation, e.g. blood, urine and vitreous humor [14]. The risk of ethanol being synthesized postmortem is less in urine compared with blood owing to the negligible amounts of glucose excreted in the urine in people with normal carbohydrate metabolism [5]. However, for individuals with poorly controlled diabetes, the urine might be loaded with sugar, which increases the risk for postmortem production of ethanol [68]. Another advantage of analyzing urine is the fact that the UAC furnishes information about the
* Corresponding author. Tel.: 46-13-252-114; fax: 46-13-104-875. E-mail address: wayne.jones@rmv.se (A.W. Jones).

bloodethanol concentration during the time that the urine was being produced in the kidney and stored in the bladder some time prior to death [911]. In medical examiner cases it is not unusual to nd an elevated concentration of ethanol in urine even though the blood-ethanol concentration at autopsy is reported as negative (<0.1 g/l) [4]. This might be the case when a considerable time has elapsed after the end of drinking and the time of death and when the persons BAC decreases to below the limits of quantitation by the method of analysis used, owing to ongoing metabolic degradation of ethanol [12]. Oxidation of ethanol does not occur in the urinary bladder and reabsorption into the bloodstream appears to be negligible [1315]. An abnormally low UAC/BAC ratio is possible if the person died shortly after drinking alcohol, that is, during the absorption phase when the BAC was still rising [9,10]. In forensic medicine much useful information can be gleaned

0379-0738/$ see front matter # 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S0379-0738(03)00213-5

A.W. Jones, P. Holmgren / Forensic Science International 135 (2003) 206212

207

by comparing the concentrations of ethanol in blood and urine and calculating the urine/blood ratios of ethanol in both living and dead [4,9,16]. This paper reports UAC/BAC ratios of ethanol in a large material of forensic autopsies when the primary cause of death was attributed either to acute alcohol poisoning or chronic alcoholism. The UAC/BAC ratios were evaluated according to the persons age, gender and the underlying blood-ethanol concentration.

2. Materials and methods The Swedish National Board of Forensic Medicine (Rattsmedicinalverket) has created a forensic pathology database (RattsBase), which contains, among other things, information about the cause and circumstances of death in all forensic autopsies performed in Sweden [17]. This database along with the forensic toxicology database (ToxBase) was searched to nd instances when ethanol was the only drug present in femoral venous blood and urine (>0.1 g/l) and when the primary cause of death was either attributed to acute alcohol poisoning (N 628) or chronic alcoholism (N 647). The forensic pathologists reach their diagnosis from information gleaned from case les, police reports, hospital records, autopsy ndings and toxicology results, and decide on primary and secondary causes of death. The diagnosis chronic alcoholism came from knowledge about alcohol psychoses in the deceased, liver disorders such as gross accumulation of fat, hepatitis, cirrhosis, and jaundice as well as ascites uid, and also pancreatitis and cardiomyopthy. In reaching the conclusion acute alcohol poisoning the bloodethanol concentration was the major factor to consider as well as police reports about the circumstances of the death. Important evidence includes risk for asphyxia caused by compromised breathing or suffocation

by inhalation of vomit as veried by histological examination of the lungs. The blood specimens for toxicological analysis were always taken from a femoral vein and the blood and urine specimens were placed into containers containing potassium uoride (12%) as preservative prior to shipment to the National Laboratory of Forensic Toxicology (Linkoping, Sweden) for determination of ethanol by headspace gas chromatography [18]. This method of analysis is accurate, precise and highly specic for ethanol and is generally considered the gold standard for analysis of organic volatiles substances in body uids. The limit of quantitation for ethanol in routine postmortem casework was 0.1 g/l and if the concentrations were below this they were reported as negative. The analytical results were reported to two decimal places for concentrations less then 1.0 g/l and one decimal place was used at higher concentrations. The mean of duplicate determinations of ethanol in blood and urine after analysis on two different GC columns was used for computing UAC/BAC ratios and statistical analyses was done by traditional methods [19].

3. Results Bloodethanol concentrations in chronic alcoholism deaths ranged from 0.1 to 5.6 g/l (mean 1.87 g/l) and the corresponding UAC ranged from 0.1 to 6.2 g/l (mean 2.42 g/l). In acute alcohol poisoning deaths concentrations of ethanol were nearly 2 g/l higher, with mean BAC of 3.64 g/l (range 0.746.8 g/l) and mean UAC of 4.26 g/l (range 1.28.2 g/l). Fig. 1 shows that UAC/BAC ratios tend to decrease as BAC increases and this trend was especially evident in chronic alcoholism deaths with many values below 0.5 g/l. In the acute poisoning deaths the lowest BAC was 0.74 g/l,

Fig. 1. Relationships between mean UAC/BAC ratios of ethanol and the bloodethanol concentration in deaths attributed to chronic alcoholism (right) and acute alcohol poisoning (left).

208

A.W. Jones, P. Holmgren / Forensic Science International 135 (2003) 206212

Table 1 Gender and mean age, bloodethanol and urineethanol concentrations and urine/blood ratios of ethanol in deaths attributed to acute alcohol poisoning and chronic alcoholism and both causes together Cause of death Acute alcohol poisoning Chronic alcoholism N 628 647 Gender (%) Men 492 (78%) Women 136 (22%) Men 572 (88%) Women 75 (12%) Men 1064 (83%) Women 211 (17%) Age (S.D.)a 54 (10.9) 52 (10.9) 56 57 55 54 (10.0) (8.4) (10.0) (10.2) BAC (S.D.) (g/l) 3.59 (0.86) 3.82 (0.83) 1.87 1.84 2.67 3.12 (1.35) (1.26) (1.43) (1.38) UAC (S.D.) (g/l) 4.22 (0.99) 4.40 (0.90) 2.43 2.37 3.26 3.68 (1.56) (1.45) (1.60) (1.50) UAC/BAC (S.D.) 1.19 (0.17) 1.17 (0.21) 1.30 1.29 1.24 1.20 (0.29)b (0.26)b (0.24)c (0.23)c

All deaths
a b

1275

Standard deviation. UAC/BAC calculated when BAC exceeded 0.74 g/l (N 457) to match the acute poisoning deaths. c UAC/BAC calculated when BAC > 0:74 g/l.

Fig. 2. Frequency distributions of UAC/BAC ratios of ethanol in deaths attributed to acute alcohol poisoning (upper plot) and chronic alcoholism (lower plot).

A.W. Jones, P. Holmgren / Forensic Science International 135 (2003) 206212

209

which means that if UAC/BAC ratios are compared for the two causes of death the underlying BAC needs to be considered. Summary statistics including age, gender, bloodethanol and urineethanol concentrations and UAC/ BAC ratios are shown in Table 1 for each cause of death and both together. The percentage of men dominated over women in all categories although their mean age was similar; both sexes being in their mid-fties. However, mean BAC and UAC were about 0.2 g/l higher in women compared with men dying from acute alcohol poisoning but no such difference was observed in chronic alcoholism deaths (Table 1). The mean UAC/BAC ratios were signicantly higher (P < 0:001) when death was ascribed to chronic alcoholism (mean 1:30 0:29) compared with acute alcohol poisoning (mean 1:18 0:18) and the overall mean was 1:23 0:24 (BAC > 0:74 g/l). UAC/BAC ratios of ethanol were not correlated with the persons age in alcoholism deaths (r 0:04), acute poisonings (r 0:06) nor both causes combined (r 0:01). In the chronic alcoholism deaths the frequency distribution of UAC/BAC ratios were skewed towards higher values with a mean and median of 1.58 and 1.29, respectively (Fig. 2). The corresponding frequency distribution of UAC/ BAC ratios in acute alcohol poisonings was more like a Gaussian curve with few outliers present (Fig. 2). In deaths ascribed to chronic alcoholism, there were 180 cases when BAC was less than 0.74 g/l (mean 0.32 g/l) and UAC/BAC ratios were 2.24 (mean) and 2.46 (median), respectively. These values are considerable higher than the mean of 1.30:1 when the BAC was above 0.74 g/l (N 457) (Table 1). In the alcoholism deaths there were 80 individuals with UAC/ BAC ratios above 2.0 compared with only three such cases among the acute alcohol poisonings.

Fig. 3. Correlation between urine and bloodalcohol concentration in deaths attributed to acute alcohol poisoning (upper plot) and chronic alcoholism (lower plot), where N is number of xy pairs, r the Pearsons correlation coefcient, and residual S.D. is standard deviation of residuals.

Fig. 4. Correlation between urine and bloodalcohol concentration for both causes of death, where N is number of xy pairs, r the Pearsons correlation coefcient, and residual S.D. is standard deviation of residuals.

210

A.W. Jones, P. Holmgren / Forensic Science International 135 (2003) 206212

Table 2 Linear regression analysis of relationships between urineethanol (x-variate) and bloodethanol concentration (y-variate) in deaths attributed to acute alcohol poisoning, chronic alcoholism and both causes of death Conditions/source Acute alcohol poisoning Chronic alcoholism All deaths [28] [13] N 628 647 1275 109 435 Regression equationa (BAC a bUAC) BAC 0.45 0.75 UAC BAC 0.12 0.82 UAC BAC 0.11 0.86 UAC BAC 0.06 0.80 UAC BAC 0.06 0.81 UAC Correlation coefficient 0.84 0.95 0.95 Not given Not given Residual S.D.b (g/l) 0.47 0.41 0.46 0.31c 0.41c 95% limits of agreement (1.96S.D.) 0.92 0.80 0.90 0.61 0.80

Results from two other autopsy studies are included for comparison. a BAC a bUAC is the least squares regression equation, where a intercept and b regression coefcient. b Standard deviation. c Calculated from the cited articles.

Fig. 3 shows the scatter plots comparing BAC and UAC in deaths caused by acute alcohol poisoning (upper plot, r 0:84) and chronic alcoholism (lower plot, r 0:95). There was a wide scatter of the individual data points around the regression lines as reected in the residual S.D. of 0.47 and 0.41 g/l in acute alcohol poisonings and chronic alcoholism deaths respectively. Fig. 4 shows the corresponding scatter plot for both causes of death (r 0:947) and the residual S.D. was about the same, being 0.46 g/l. Table 2 gives regression statistics for these scatter plots (Figs. 3 and 4) and compares the values with two previous studies done in postmortem blood and urine specimens. Owing to the large amount of scatter, as reected by the residual S.D., if the BAC is estimated from UAC the result is subject to much uncertainty in any individual case. These regression equations can be used to derive the prediction interval for BAC at any desired UAC. The regression equation in Fig. 4 (BAC 0:11 0:85 UAC) has a residual S.D. of 0.46 g/l so if UAC is taken as 1.07 g/l, which is the legal limit for driving in UK, the expected mean BAC is 0.80 g/l but in 95 of 100 cases this BAC might range from 0.00 to 1.72 g/l (mean 0:80 1:96S.D.). Similarly, it can be shown for a persons BAC to be 0.80 g/l or more, this would require that the UAC at autopsy was at least 2.13 g/l.

4. Discussion Blood circulation stops at the moment of death and production of urine in the kidneys ends. The UAC/BAC found at autopsy should therefore be comparable with values for randomly timed specimens of urine and blood in living subjects (20). The UAC/BAC ratio expected for the newly formed ureter urine should theoretically be about 1.25, based on the average water content of whole blood ($80%) and urine ($100%) [4,16]. Finding a mean UAC/BAC ratio of 1.30:1 or more indicates the person had reached the postabsorptive phase of ethanol metabolism at the time of death [16]. We found statistically signicant differences between mean UAC/BAC ratios in chronic alcoholism deaths (1.30)

and acute alcohol poisoning deaths (1.18) (P < 0:001). This suggests that the deaths ascribed to acute alcohol poisoning occurred before alcohol was fully absorbed and distributed in all body uids and tissues, which suggests that the person had recently been drinking alcohol [9,10]. Finding abnormally high or low UAC/BAC ratios in medical examiner cases might have several possible explanations. If the bladder contained an alcohol-free pool of urine before the drinking started, the concentration of ethanol in newly formed urine becomes diluted when this is collected in the bladder [4,16]. Low UAC/BAC ratios are likely to be found if a large amount of alcohol was consumed shortly before death and the BAC curve was in the absorption phase before equilibration of ethanol in all body uids was completed [4]. Under these circumstances appreciable amounts of alcohol probably still remain in stomach and this should be veried by sending stomach contents for toxicological analysis [16]. High UAC/BAC ratios are found if urine is stored in the bladder for long period during the time when the postabsorptive elimination phase of ethanol kinetics has been reached. Studies have shown that BAC decreases by about 0.100.25 g/(l per h) in most people although there is no metabolism of ethanol in the urinary bladder [4,16]. After a traumatic event, such as a blow to the head, the person might be unconscious and survive many hours before death ensues [12]. During this period urine continues to be produced and collected in the bladder whereas the ethanol is progressively being cleared from the bloodstream [12,16]. One study showed that UAC/BAC ratios were higher when there was more urine in the bladder and thus indicating a longer period of storage or higher rates of production [21]. The total volume of urine in the bladder should be routinely measured at autopsy [16]. Another necessary assumption when UAC and BAC results are interpreted is that ethanol was neither synthesized nor destroyed in the body uids after death. Bacteria and yeasts can both produce and destroy ethanol after death [22]. Although 12% potassium uoride was added to specimens of blood and urine taken at autopsy this does not exclude the

A.W. Jones, P. Holmgren / Forensic Science International 135 (2003) 206212

211

possibility that some ethanol might have been synthesized between the time of death and the time of autopsy [3,4]. This problem is exaggerated if the cadaver is decomposed or putreed, thus increasing the risk for postmortem synthesis of ethanol in blood specimens [3,4,16]. Many others have investigated relationships between BAC and UAC in autopsy material, albeit not such a large case series as reported here. Moreover, our strict quality assurance requirement with only femoral venous blood samples and uoride preservative in the tubes boosts condence in the toxicological results. Indeed, problems caused by postmortem diffusion of ethanol from the gut after death, which is always a possible after recent drinking as in acute alcohol poisoning deaths, is minimized when femoral blood is sampled [23,24]. Comparing mean UAC/BAC ratios reported in different studies is complicated because much depends on the range of alcohol concentration in the specimens investigated as illustrated in Fig. 1. In this study, even when the same BAC range was considered (>0.74 g/l), we still found lower UAC/BAC ratios in acute alcohol poisonings (1.18:1) compared with chronic alcoholism deaths (1.30:1). Previous quantitative evaluations of UAC/BAC relationships either evaluated ratios and their variance or applied a linear regression analysis, whereas we used both methods. Winek et al. [25] looked at UAC/BAC ratios for a large number of forensic cases presumably meaning medical examiner or autopsy cases. The overall mean UAC/BAC ratio was 1.57:1 with a range from 0.07 to 21:1. When the concentration of ethanol in blood exceeded 1 g/l (N 714) the mean UAC/BAC ratio was 1.18:1 (range 0.073.4) compared with 1.93 (range 0.1121) when BAC was less then 1 g/l (N 258). The variances of the UAC/BAC ratios were not presented in the study. In a small case series of alcohol-related deaths Budd [26] found a mean UAC/BAC ratio of 1.5:1 (N 100) and a standard deviation of 0.7. He recommended subdividing UAC/BAC ratios into various concentration ranges if and when UAC was used to estimate BAC. In medical examiner cases (N 148), Kaye and Cardona [27] reported a mean UAC/BAC ratio of 1.28:1 (range 0.212.55) and recommended against trying to estimate BAC from UAC in any individual case, owing to the large errors that might result. Kuroda et al. [13] applied a linear regression analysis to evaluate relationships between UAC (x-variate) and BAC (yvariate) and found a wide scatter of data points (large residual S.D.) and consequently a wide prediction interval. They concluded that UAC was not helpful in assessing BAC in an individual case fatality but that UAC was useful to corroborate BAC. Heatley and Crane [28] reached similar conclusions based on a much smaller case series of coroner autopsies (N 109). These authors showed a good appreciation for statistical analysis of the data and reported an overall mean UAC/BAC ratio of 1.39 (S.D. 0.44). When BAC was above 0.8 g/l the mean UAC/BAC ratio was 1.32 compared with 1.71 when BAC was less then 0.8 g/l.

The UAC versus BAC relationship was also evaluated by linear regression analysis and results are compared with our data in (Table 2). The Pearsons correlation coefcient is not a good statistic for comparing two measurement such as UAC and BAC because the value depends on the variance of the independent variable here the UAC [19]. When regression analysis is used for purposes of prediction, that is, when UAC is used to estimate BAC much depends on legal requirements in terms of acceptable uncertainty [2931]. Often a key question concerns conformity or nonconformity with some threshold value such as the legal bloodalcohol limit for driving (e.g. 0.80 g/l in blood or 1.07 g/l in urine in UK). If a beyond any reasonable doubt answer is requested, an upper 99% condence limit for the UAC/BAC ratio might be used in the computation of BAC. However, if reasonable scientic certainty is sufcient then an upper 95% condence limit is probably acceptable. If the estimated BAC should more likely then not exceed 0.8 g/l then a median UAC/BAC ratio might be deemed appropriate. In forensic toxicology it is very important that both blood and urine specimens are obtained for analysis of ethanol to allow a proper interpretation of whether alcohol was involved in the death and to rule-out postmortem artifacts and in this way increase condence in the nal report to the police [3,4,16]. The UAC/BAC ratio also gives an indication about the status of alcohol absorption and elimination in the body at the time of death [10]. Finding a low UAC/BAC ratio along with other information, e.g. appreciable amounts of alcohol in the gastric contents (>5 g/l) or empty liquor bottles at the death-scene gives hints of recent consumption of alcohol [4]. By contrast, if the UAC/BAC ratio exceeds 1.3 this indicates that alcohol was already absorbed and distributed in all body uids at the time of death [20]. These matters are not trivial because in civil litigation, e.g. during insurance investigations much depends on the autopsy report and the toxicological ndings regarding BAC. When accidents in the work-place or on the roads are an issue the person might be accused of negligence or liability because of drinking on duty or before driving.

References
[1] B. Levine, M.L. Smith, J.E. Smialek, Y.H. Caplan, Interpretation of low postmortem concentrations of ethanol, J. Forensic Sci. 38 (1993) 663667. [2] H. Kalant, Interpretation of post-mortem ethanol concentrations, Aerospace Med. 39 (1968) 633637. [3] C.L. ONeal, A. Poklis, Post-mortem production of ethanol and factors that inuence interpretation: a critical review, Am. J. Forensic Med. Pathol. 17 (1996) 820. [4] A.W. Jones, Alcohol Post-Mortem, Encyclopaedia of Forensic Sciences, Academic Press, London, 2000, pp. 112126. [5] J. Fine, Glucose content of normal urine, Br. Med. J. i (1965) 12091214. [6] W.D. Alexander, P.D. Wills, N. Eldren, Urinary ethanol and diabetes mellitus, Diabet. Med. 5 (1988) 463464.

212

A.W. Jones, P. Holmgren / Forensic Science International 135 (2003) 206212 [19] M. Bland, An Introduction to Medical Statistics, third ed., Oxford University Press, Oxford, 2000. [20] A.W. Jones, Time-adjusted urine/blood ratios of ethanol in two successive voids from drinking drivers, J Anal. Toxicol. 27 (2003) 167168. [21] H.J. Mittmeyer, A.K. Blattert, UAK/BAK quotient unter Berucksichtigung der Harnblasenfullung, Beitr Gerichtl Med. 49 (1991) 263267. [22] J.E. Corry, Possible sources of ethanol ante- and postmortem its relationship to the biochemistry and microbiology of decomposition, J. Appl. Bacteriol. 44 (1978) 156. [23] H. Gifford, H.W. Turkel, Diffusion of alcohol through stomach wall after death, JAMA 161 (1956) 866868. [24] D.J. Pounder, D.R.W. Smith, Postmortem diffusion of alcohol from the stomach, Am. J. Forensic Med. Pathol. 16 (1995) 8996. [25] C.L. Winek, K.L. Murphy, T.A. Winek, The unreliability of using a urine ethanol concentration to predict a blood ethanol concentration, Forensic Sci. Int. 25 (1984) 277281. [26] R.D. Budd, Ethanol levels in post-mortem body uids, J. Chromatogr. 252 (1982) 315318. [27] S. Kaye, E. Cardona, Errors of converting a urine alcohol value into a blood alcohol value, Am. J. Clin. Pathol. 52 (1969) 577584. [28] M.K. Heatley, J. Crane, The relationship between blood and urine alcohol concentrations at autopsy, Med. Sci. Law 29 (1989) 209217. [29] A.W. Jones, P. Holmgren, Uncertainty in estimating blood alcohol by analysis of vitreous humor, J. Clin. Pathol. 54 (2001) 699702. [30] H. Andersson, The use of uncertainty estimates of test results in comparisons with acceptance limits, Accred. Qual. Assur. 7 (2002) 228233. [31] J.Y. Christensen, F. Holst, E. Olsen, P.T. Wilrich, Rules for stating when a limiting value is exceeded, Accred. Qual. Assur. 7 (2002) 2835.

[7] H.A. Sulkowski, A.H.B. Wu, Y.S. McCarter, In-vitro production of ethanol in urine by fermentation, J. Forensic Sci. 40 (1995) 990993. [8] J.J. Saady, A. Poklis, H.P. Dalton, Production of urinary ethanol after sample collection, J. Forensic Sci. 38 (1993) 14671471. [9] A.W. Jones, Ethanol distribution ratios between urine and capillary blood in controlled experiments and in apprehended drinking drivers, J. Forensic Sci. 37 (1992) 2134. [10] B. Levine, J.E. Smialek, Status of alcohol absorption in drinking drivers killed in trafc accidents, J. Forensic Sci. 45 (2000) 36. [11] A.W. Jones, Reference limits for urine/blood ratios of ethanol concentration in two successive voids from drinking drivers, J. Anal. Toxicol. 26 (2002) 333339. [12] A.R. Alha, V. Tamminen, Fatal cases with elevated urine alcohol but without alcohol in blood, J. Forensic Med. 11 (1964) 35. [13] N. Kuroda, K. Williams, D.J. Pounder, Estimating blood alcohol from urinary alcohol at autopsy, Am. J. Forensic Med. Pathol. 16 (1995) 219222. [14] H.W. Haggard, L.A. Greenberg, R.P. Carroll, D.P. Miller, The use of urine in the chemical test for intoxicationpossible errors and their avoidance, JAMA 115 (1940) 16801683. [15] E. Weinig, P. Zink, G. Reinhardt, Uber die forensische Bedeutung der Alkoholkonzentration im Urin, Blutalkohol 7 (1970) 307315. [16] D.J. Pounder, A.W. Jones, Measuring alcohol postmortem, In: S. Karch (Ed.), Drug Abuse Handbook, CRC Press, Boca Raton, 1998, pp. 356374. [17] H. Druid, P. Holmgren, P. Lowenhielm, Computer-assisted systems for forensic pathology and forensic toxicology, J. Forensic Sci. 41 (1996) 830836. [18] A.W. Jones, J. Schuberth, Computer-aided headspace gas chromatography applied to blood-alcohol analysis: importance of online process control, J. Forensic Sci. 34 (1989) 11161127.