Journal of AnalyticalToxicology,Vol. 20, May/June1996
Comparison of Ethanol Concentrations in Blood, Serum,
and Blood Cells for ForensicApplication*
R.C. Charlebois 1, M.R. Corbett 2,f, and J.G. Wigmore 2
158 Landfair Crescent, Scarborough, Ontario, Canada M lJ 3A7 and 2Centre of Forensic 5ciences, 25 Grosvenor Street, Toronto,
Ontario, Canada M7A 2G8
Abstract I
Ethanol concentrations in serum (SAC) and whole blood (BAC)
were determined for 235 subjects by a headspace gas
chromatographic method. The SAC:BAC ratios ranged between
1.04 and 1.26. The mean was 1.14, and the normal distribution
had a standard deviation (SD) of 0.041. When a conversion
(division) factor for SAC to BAC of 1.18 (mean + 1 SD) was used,
84% of estimated BACs were less than that measured; the
remaining estimates differed by less than 7 mg/dL. An SAC greater
than 100 mg/dL reliably indicated a BAC of more than 80 mg/dL.
Ethanol concentrations in blood cells (CAC) were similarly
determined for 167 of these subjects. The CAC:BAC ratios ranged
from 0.66 to 1.00 and had a mean of 0.865 and a negatively
skewed normal distribution with an SD of 0.065. When a
conversion (division) factor for CAC to BAC of 0.93 (mean + 1 SD)
was used, 89% of estimated BACs were less than that measured;
the remaining estimates differed by less than 8 mg/dL. A CAC
greater than 80 rag/d/reliably indicated a BAC of more than 80
mg/dL. The CAC is useful in forensic practice when either blood or
serum is not available.
Introduction
In Canada, blood samples taken for medical purposes from
drivers injured in motor vehicle collisions may be submitted to
a forensic laboratoryfor the determination of ethanol (alcohol).
Typicallythese samples have been processed by hospital technologists to separate serum from blood cells for clinical analyses, and occasionallysuch analyses use all of the serum. To be
meaningful to a court of law, the serum ethanol concentration
(SAC) or blood cell ethanol concentration (CAC) determined
must be converted to a whole blood ethanol concentration
(BAC),which is specified in the criminal code. Therefore, reliable scientificstudies of the ethanol concentrations in blood and
*Presented al the Joint Congressof The International Associalion of Forensic Toxicologists
(TIAFT) and the Societyof ForensicToxicologists (SOFT)on November 3, 1994 in Tampa,
Florida by M.R. Corbett.
'~Authorto whom correspondenceshou[d be addressed.
derived serum and blood cell samples are necessary for the required numerical conversion.
The SACs and CACshave been compared with the BACs in
the literature (Table I) (1-16); however, such studies have
been conducted hitherto on smaller groups of subjects. In
this work, the SACsand BACswere measured in paired samples from 235 subjects; the CACswere also measured for 167
of these subjects.
Experimental
Blood collection and preparation
Blood samples were collected from police officers who consumed ethanolic beverages during training courses on Breathalyzer| instruments. The subjects drank beer (5% ethanol,
v/v) or spirits (diluted to approximately 20% ethanol, v/v)
starting 1/2h after a lunch meal and continuing for I to 11/4h.
After 11/4h from cessation of drinking, two blood samples
were drawn from the cubital vein within i rain using the Vacutainer| technique. One sample was drawn into a gray-topVacutainer XF947 (10-mL tube that contained 100 mg sodium fluoride and 20 mg potassium oxalate); the second sample was
drawn into a red/gray-top Vacutainer SST (13-mL tube that
had an inert gel and clot activator for separating serum and
blood cells).
After collection, the tubes were inverted several times for the
purpose of mixing the samples and the chemicals. The
red/gray-top tubes were allowed to stand for 30 rain for complete blood coagulation and then spun at 2400 rpm (700 x g)
for 10 rain.
The packed blood cells were mechanically homogenized
using a Polytron| to facilitate volumetric sampling.
Determination of ethanol
Ethanol concentrations were measured using a previouslydescribed headspace gas chromatographic (GC) method (17). The
instrument calibration was verifiedor conducted using aqueous
ethanol standards that contained 1% sodium fluoride (wAl)and
Reproduction(photocopyin8) of editorialcontentof thisjournalis prohibitedwithoutpublisher'spermission.
171
Journal of Analytical Toxicology, Vol. 20, May/June 1996
Table I. Literature Reports of Ethanol Concentrations in Blood, Serum, Plasma, and Blood Cells*
BAC range
Sample AC:BAC
N
nt
Sample
155
211
134
14
50
165
19
25
1
1
1
1
1
I
8
2
serum
serum
serum
serum
serum
serum
serum
serum
40-398
12-522
22-155
74-195
40-442
-50-I 50+
-150
106-376
1.08
1.16
1.15
1.15
1.14
1.15
1.12
1.17
0.058
0.13
0.02
0.015
0.019
0.061
0.034
0.74-1.29
0.88-1.59
1.1 0-1.25
1.12-1.18
1.09-1.18
1.11-1.23
ACA/GC
GCD
GCH
GCD,ADH
GCD
CHEM,ADH
CHEM,FER
WID
17
4
20
5
10
4
14
2
1
I
5
1
plasma
plasma
plasma
plasma
plasma
plasma
7-136
121-277
21-167
117-I 94
111-199
1.10
1.11
1.18
1.15
1.16
1.16
0.017
0.012
0.057
0.031
0.049
1.03-1.24
1.10-1.13
1.10-1.35
1.12-I .20
1.10-1.21
GCH
GCD
NC
GC
pWlD
CH EM
161
106
20
5
1
0.76
0.863
0.771
0.805
0.06
0.06
0.065
0.015
0,649-0.884
0.784-0.821
I
I
(mg/dL)
cells
cells
cells
cells
Mean
106-281
SDN
Range
Method
Ref.
15,16
1
2
3
4
5
6
14
7
8
9
10
11
12
5
13
9
10
pWID
'Abbreviations: N, number of subjects; nt, numberof times BAC sampled; BAC, blood ethanol concentration;AC, ethanol concentration; SD N, standarddeviation for subjects; Ref.,
reference; ACA, DuPont automatedclinical analyzer;GC, gas chromatography;GCO, direct-injection gas chromatography;GCH, headspacegas chromatography;ADH, alcohol
dehydrogenasemethod; CHEM, chemical oxidation method; FER,ferment method; WlD, Widmark method; NC, Nickolls-modified Cavett method; pWlD, photometricWidmark
method.
Table II. Relative Ethanol Concentrations in Blood, Serum, and Blood Cells*
Ralio
Mean~) Min
SAC:BAC 1.14
CAC:BAC 0.871
PORII
0.997
Max Median SD* %CV~ SEMi
1.04 1.26 1.14 0.041 3.6
0.669 1.00 0.874 0.061 7.0
0.757 1.260 0.996 0.088 8.8
N Skewness Kurtosis %_<,~•
0.0027 235 0.33
0.0050 144 -0.85
0.0073 144 0.040
0.28
1.32
0.48
70.6
75.4
72.5
% _<~•
% _<i~•
94.9
94.0
91.6
I00
98.8
99.4
% <_~+ls % _<~+2s %_<i~+3s
86.0
88.6
85.6
96.6
99.4
96.4
I00
100
99.4
* Abbreviations:8AC, blood ethanol concentration;SAC serum ethanol concentration;CAC, blood cell ethanol concentration
, SD = Standarddeviation.
r CV = Coefficient of variation.
wSEM = Standarderror of the mean.
II POR = Product of ratios; (SAC:BAC) x (CAC:BAC).
50
0.20
40
0.15
30
o
c
0.10
20
0.05
10
0
70
140
BAC (mg/dL)
Figure 1. Distribution of subject blood ethanol concentrations (BAC).
172
210
0.5% sodium citrate (w/v). All samples were
analyzed in duplicate, and the mean results
were used for statistical analysis.
The coefficient of variation (CV) for the
intra-assay duplication (n = 2) of the ethanol
measurements was 0.55% for aqueous standards (mean concentration, 101.7 mg/dL),
0.79% for serum, 0.92% for blood, and
1.11% for blood cell samples; the proportion of these measurements within plus or
minus 2 CVs ranged from 95.4 to 97.7%. No
duplications had variabilities exceeding 5%.
The CV for the interassay replication (n =
100) was 1.91% for an aqueous ethanol
standard (mean concentration, 77.5 mg/dL);
96.0% of these measurements were within
plus or minus 2 CVs.
Blood samples were analyzed for ethanol
within 1 day of collection. Serum and blood
cell samples were similarly analyzed or
Journal of Analytical Toxicology, Vol. 20, May/June 1996
250
l/
I
N= 235
Ra = 0.986
SAC = 1.130 x BAC + 0 . 9 3 6 , / /
200
"~
I
I
/
A
150
I00
-
99
-/
50
0
0
~ ~ ~ . , ~ . ~
9
9
!
lY//I//II////////I/II//I/////I//I/I///A
I
I
I
I
50
100
150
200
250
BAC ( m g / d L )
200
I
l
I
N= 167
R2 = 0.946
CAC = 0.866 x BAC - 0.061
B
150
\
100
r,&
.<
I
:;5
~
'
'
'
9
50
_
/.,'
L:
i
"
"
6
WI//I/////////./A
0
0
50
9100
150
200
BAC (mg/dk)
Figure2. Scatterplot of ethanol concentrations in (A) serum (SAC)with blood (BAC) and (B) blood cells (CAC) with blood.
stored under refrigeration (4~ and analyzed from time to
time up to 2 years later. The delay in analysis of samples has the
well-known potential to cause concentration changes; however,
this was not detected in the ratios for serum samples. Changes
were discernible only for the group of 23 blood cell samples
that were analyzed after 2 years: The ratio was lowered by
5.3% (corresponding to 3.7 mg/dL), and these were excluded
from the final data analysis presented in Table II.
173
Journal of Analytical Toxicology, Vol. 20, May/June 1996
60
0.25
A
50
0.20
40
~.
c
0
&
2
a.
0.15
30
O.lO
20
0.05
10
1.02
1.08
1.14
1.20
1.26
SAC:BAC
25
B
0.15
20
~I=
0
15
0.10
0
c
&
0
A-
10
o.o5
0.66
0.78
0.90
1.02
0.25
~.
tO
C
30
&
2
a.
0.15
20
c
0110
IO
0.05
0.70
0.85
1.00
1.15
1 30
POR
Figure 3. Distribution of ratios of ethanol concentrations in (A) serum
(SAC)to blood (BAC);(B) blood cells (CAC)to BAC;and (C) Distribution
of the product of the ratios (POR)SAC: BAC and CAC:BAC.
Results
Ethanol concentrations
The BACs for 235 subjects ranged between 14 and 202
mg/dL, and the mean was 81 mg/dL; the SACsranged between
16 and 233 mg/dL, and the mean was 92 mg/dL. The CACsfor
167 of these subjects ranged between 12 and 153 mg/dL; the
mean was 68 mg/dL. The distribution of BAC measurements is
174
Ratio of SAC to BAC
The SAC:BACratios for 235 subjects ranged between 1.04
and 1.26, had a mean of 1.14 (Table II), and had a normal
(Gaussian) distribution (Figure 3A) with a standard deviation
(SD) of 0.041. A plot of the expected value for a standard
normal variable with the ratio found agreement with a normal
distribution by adherence to a straight line.
The mean SAC:BACratio was independent of BAC;however,
the slope of SACwith BACwas variable between groups of data.
For example,the slope was 1.19 using BACsof 50 • 20 mg/dL
and 1.12 using 80 + 20 mg/dL. In contrast, the mean SAC:BAC
ratio was constant at 1.14 for these two groups of data (TableIII).
When the mean of 1.14 was used, the differencebetween the
estimated and measured BACsdid not exceed 10 mg/dL for 235
subjects. Using higher ratios decreasedthe size and proportion
of overestimates, whereas underestimates were increased in
both aspects: When SAC:BACratios of the mean (1.14), 1.16,
1.18 (mean + 1 SD), 1.20, and 1.22 (mean + 2 SD) were used,
the largest overestimates were 10, 8, 6, 4, and 3 mg/dL, respectively, for a measured BAC of 94 mg/dL. The largest underestimates were 11, 14, 17, 20, and 23 mg/dL, respectively,
and the proportion of BACs underestimated were 51.9, 68.5,
83.8, 91.1, and 96.2%, respectively.
Ratio of CAC to BAC
CAC:BAC
0.20
presented in Figure 1.
The scatter-plot and least-squares linear regression for
ethanol concentrations in (A) serum to blood and (B) blood
cells to blood are presented in Figure 2.
The CAC:BACratios for 167 subjects ranged between 0.664
and 1.00 and had a mean of 0.865 (SD, 0.065) and a negatively skewed normal distribution (Figure 3B). The CAC:BAC
ratio ranged between 0.669 and 1.00 and had a mean of 0.87]
and an SD of 0.061 (Table II), excluding ratios for 23 subject
samples that had a 2-year delay in measurement.
The difference between the estimated and measured BACs
did not exceed 20 mg/dL for 167 subjects using the mean.
When CAC:BACratios of the mean (0.87), 0.90, 0.93 (mean +
1 SD), 0.96, and 1.00 (mean + 2 SD) were used, the argest
overestimates were 20, 13, 7, 4, and 0 mg/dL, respectively,for
a measured BACof 94 mg/dL.The largest underestimates were
20, 23, 25, 27, and 30 mg/dL, respectively,and the proportion
of BACsunderestimated were 49.7, 68.8, 88.6, 96.4, and 100%,
respectively.
Product of SAC:BAC and CAC:BAC ratios
The product of SAC:BAC and CAC:BAC ratios (POR) was
computed, that is, (SAC:BAC)• (CAC:BAC).The POR for 167
subjects ranged between 0.712 and 1.26 and had a mean of
0.989 (SD, 0.091) and an approximate normal distribution
(Figure 3C). The POR ranged between 0.757 and 1.26 and had
a mean of 0.997 and a more normal distribution (in skewness)
with an SD of 0.088, excluding ratios for 23 subject samples
that had a 2-year delay in measurement.
Dependence of ratios on measurement delay and BAC
The mean and range of grouped SAC:BACratios were inde-
Journal of Analytical Toxicology, Vol. 20, May/June 1996
Table III. Average and Mean SAC:BAC Ratio for BAC Range and Method of Data Analysis*
Method
BAC Range
(mg/dt)
N
30 to 70
60 to 100
> 100
13 to 202
81
156
37
235
BACav
SD
(mg/dL)
58.3
79.3
124.5
80.7
2.45
3.32
5.09
3.50
LSLR
CV
(%)
ANOVA
Slope• SE
Mean • SEM
SD
3.67
3.66
3.59
3.80
1.187 • 0.031
1,118 + 0.025
1.115 + 0.034
1.130 + 0.0087
1.145 _+0.0047
1.145 • 0.0033
1.137 + 0.0064
1.143 _+0.0027
0,042
0.041
0.039
0.041
(%)
3.67
3.60
3.40
3.62
* Abbreviations: SAC, serum ethanol concentration; BAC, blood ethanol concentration; LSLR, least-squares linear regression; ANOVA, analysis of variance; SD, standard deviation;
CV, coefficient of variation; SEM, standard error of the mean.
Table IV. Dependence of Relative Ethanol Concentrations on Measurement Delay
SAC*:BAC~
Delay
N
Mean
0d
1-2 d
3-9 d
38
64
42
1-3 m
4-6
2y
CAC*:BAC
PORw
Range
N
Mean
Range
N
Mean
Range
1.14
1.15
1.15
1.07-1.23
1.08-1.26
1.09-1.24
0
46
36
0,867
0,902
0.677-1.00
0.766-0.988
0
61
31
1.00
1.04
0.783-1.26
0.893-1.19
32
35
1.12
1.14
1.04-1.22
1.05-1.26
22
40
0.872
0.846
0.808-0.950
0.669-0.925
31
21
0.974
0.962
0.834-1.10
0.757-1.10
24
1.13
1.06-1.22
23
0.825
0.664-0.959
23
0.935
0.712-1.11
* SAC = Serum ethanol concentration.
BAC = Blood ethanol concentration.
CAC = Blood cell ethanol concentration.
SPOR = Product of ratios; (SAC:BAC) x (CAC:BAC).
pendent of measurement delay for at least 2 years (Table IV).
The mean CAC:BACratio decreased from 0.867 to 0.825 for the
2-year delay,and the range limiting ratios also decreased from
0.677 and 1.00 to 0.664 and 0.959. Similarly,the mean POR decreased from 1.00 to 0.935, and the range limiting ratios decreased from 0.783 and 1.26 to 0.712 and 1.11.
The ratios were also independent of BAC.When linear leastsquares analysis was used, the regression equations for (a)
SAC:BAC, (b) CAC:BAC,and (c) POR with BAC had slopes and
y-intercepts of (a) -6.8E-5 • 1.0E--4 and 1.148 • 0.008, (b)
4.0E-5 • 2.0E-4 and 0.861 • 0.017, and (c) -1.1E-4 • 2.9E-4
and 0.998 • 0.024. The coefficients of determinations were
1.9E-3, 2.3E-4, and 9.5E-4, respectively.
Discussion
The range of mean SAC:BACratios reported in the literature
(Table I) was 1.08-1.18 and supports the mean obtained in
this study of 1.14. Although Payne et al. (9) found a mean of
1.18, their study included an extreme ratio of 1.35, which, if
excluded, would lower the mean to 1.17 (N = 19), and all
remaining ratios would be within plus or minus 1.7 SDs. The
1.35 ratio would correspond to plus 4.3 SDs and be inconsistent with the number of subjects studied.
Variation in the SAC:BACratio can arise from sample preparation, method of ethanol measurement, subject condition,
low ethanol concentrations, and technique of data analysis.
For example, Hodgson and Shajani (8) found that samples
prepared by protein precipitation with trichloroacetic acid had
a plasma ethanol concentration (PAC) lowered by 6%.
By using an alcohol dehydrogenase method of ethanol measurement, Illchmann-Christ (5) found a mean SAC:BAC of
1.13, whereas the use of a chemical oxidation method gave a
mean of 1.17. Details related to quality assurance between
methods were not reported; therefore, this difference may be
merely due to calibration differences and not true differences
in the potential to get equally accurate results by different
methods. By using a DuPont Automated Clinical Analyzer~
(ACA| for 155 serum samples and a GC method for the corresponding blood samples, Wells and Barnhill, Jr. (15,16) found
a mean SAC:BACof 1.08 (SD, 0.058), whereas their use of the
ACAon 41 of these samples for both serum and blood gave a
mean of 1.16 (SD, 0.084).
Several subject conditions were found to be associated with
lower w/w ratios of serum to blood water content, and, therefore, lower SAC:BACratios would be expected. The SAC:BAC
ratio may be lower in subjects with hypovolemicshock (18,19),
dialysis (20), anemia (21), pregnancy (22), hyperlipidemia (7),
and those subjects who were ambulatory sick outpatients (23).
Jones et al. (7) excluded concentrations of lower than 18
mg/dL from data analysis. In this work, the range limiting ratios occurred with BAC (or SAC) measurements greater than
48 mg/dL and were not distorted by the measurements at
lower concentrations.
175
Journal of Analytical Toxicology, Vol. 20, May/June 1996
Data analysis that computes a mean SAC:BAC, or (using
least-squares linear regression) the slope of SAC with BAC,
can provide different numbers. For example, the mean
SAC:BAC in this study was 1.142, whereas the slope of SAC
with BACwas 1.130. By scrutinizing data reported by Shajani
et al. (3), we found a mean SAC:BACof 1.149 and a slope of
1.144, yet both are reported as 1.14. Greater differences can
occur. For example, two groups of the data in this study at
BACsof 50 • 20 and 80 • 20 mg/dL have identical mean values
of 1.145, yet the slopes differed by 0.069 (1.187 and 1.118,
respectively).
SAC:BACratios reported since 1960 range between 1.03 and
1.26 because of acceptable analytical and intersubject variabilities. Recently, both Rainey (1) and Wells and Barnhill, Jr.
(15,16) have reported ratios outside this range, in addition to
the 1.35 measurement by Payne et al. (9).
Rainey (1) found that the SAC:BACratios ranged between
0.88 and 1.59 and had a mean of 1.16 and a logarithmic-normal
distribution. This range is not reliable when used in forensic
practice because these measurements had relatively poor precision (CV of 7.4% at 90 mg/dL), which the author warned
would result in a wider range than "use of the more precise
forensic techniques". In this work, the intra-assayprecision was
an order of magnitude better (CV of 0.79% at 92 mg/dL for
serum) and was similar to that reported by Jones et al. (7), and
the SAC:BACratios ranged between 1.04 and 1.26 and had a
normal distribution.
Wells and Barnhill, Jr. (15,16) found that the SAC:BACratios
ranged between 0.75 and 1.29 and had a mean of 1.08 (SD,
0.058) and a normal distribution. The SAC was determined in
a clinical laboratory using an ACA, and the BAC was determined in a forensic laboratory using GC. In their study, two
SAC:BAC ratios exceeded 1.18 (both were 1.29) and involved
two of the three highest measurements of SAC (410 and 393
mg/dL). Considering also the absence of other ratios exceeding
1.18 and the probability (less than .002) of finding a ratio that
follows a normal distribution at greater than or equal to 3.6
SDs from the mean, the usefulness of these two outlying ratios
is uncertain.
Wells and Barnhill, Jr. (15,16) also found five SAC:BAC
ratios less than 1, which is incompatible with the biological
variability of the relative water content of serum and blood
samples and, hence, with ethanol partitioning. Specifically,
these ratios and SACs (measured in milligrams per deciliter)
were 0.747 and 62, 0.964 and 80, 0.815 and 110, 0.929 and
144, and 0.979 and 274. Gadsen, Sr. (24) found using the
DuPont ACA ethanol method that "the higher the sample
hemoglobin concentration the greater would be the effect in
decreasing the apparent ethanol concentration" because of a
spectral (ultraviolet) interference. He found that this interference for "hemolyzed" clinical samples decreased as the
actual SAC increased and was eliminated when protein-free
supernatant (serum or plasma) samples were prepared using
trichtoroacetic acid.
Analyzing the data of Wells and Barnhill, Jr. (16) after
excluding SAC:BAC ratios less than unity and greater than
1.18, we found that their mean SAC:BACratio remained 1.08,
and there was a reduced SD of 0.038 (comparedwith 0.058) for
176
148 subjects. Their study found that the mean SAC:BACratio
in practices using typical clinical (ACA)and forensic (GC)
methods was lower than when using a single method.
Delay in measurement of refrigerated samples did not
change the mean or range of SAC:BAC ratios beyond the
inherent intersubject and analytical variabilities (Table IV).
The SAC:BACratio in this study was independent of BAC
over the range of the measurements, which agrees with the
findings of Gruner (11). The SAC:BAC ratio has also been
shown to be independent of the ethanol pharmacokinetic phase
(7,8,11,25).
To assess the reliability of estimating a BAC using a SAC
measurement, the differences were examined. Because the
SAC:BACratios were normally distributed, using a conversion
factor higher than the mean will increase the proportion of
underestimated BACs, increase the magnitude of underestimates, and decrease the magnitude of overestimates. For
example, when a conversion factor for SAC:BACof 1.18 (mean
+ 1 SD) was used, 83.8% of BACswere underestimated, by 17
mg/dL at most, whereas the largest overestimate was 6 mg/dL.
Using a conversion factor of 1.18 for SAC:BAC in forensic
practice involving the SAC of a person accused of operating a
motor vehicle with an excess BAC can provide a significant
advantage to the accused while retaining acceptable analytical
accuracy.
Winek and Carfagna (4) found that serum and plasma samples had the same ethanol concentration; the SAC:PACratio for
50 subjects ranged between 0.98 and 1.04 and had a mean of
1.00. Therefore, our findings on serum samples are applicable
to plasma samples.
In contrast to serum and plasma, CAC:BAC ratios
(5,9,10,13) and relative water content (7,22,23) have been
sparsely investigated. Schleyer (13) studied the alcohol content of blood clots that were homogenized by rolling and
dabbing with filter paper, followed by centrifuging. By using
a photometric Widmark method, he estimated a mean
CAC:BAC of 0.863 (SD, 0.06) for 106 serum and blood cell
samples; those with BACsless than 100 mg/100g (106 mg/dL)
have been excluded because of the poor precision of the
method for these concentrations. Jones et al. (7) measured
the water content of blood and blood cells for eight specimens
from two subjects, and assuming average fluid densities to
convert w/w measurements to w/v, the estimated mean
CAC:BAC was 0.88. This value is similar to the estimated
mean CAC:BACof 0.85 (N = 15) and 0.84 (N = 20) from the
studies by Hinckers (22,23). These estimates support the
measured mean CAC:BACof 0.871 obtained in this work for
144 subjects.
Payne and co-workers determined lower mean CAC:BAC
ratios (0.771 [9] and 0.805 [10]), which were similar to an
estimate from a study by Illchmann-Christ (0.79 [5]). Unfortunately, the latter study, although noting that CAC:BACratios
depend on sample preparation, did not list measurements or
provide a sufficient description of experimental procedure.
The equilibrium distribution of ethanol between biological
fluids is considered to occur in proportion to their water content (26,27). When a blood sample is separated into serum
and blood cell fractions, the increase in water content of serum
Journal of Analytical Toxicology, Vol. 20, May/June1996
occurs with a decrease in water of blood cells. Therefore, the
relative increase in ethanol content of serum (SAC:BAC)would
occur with a proportional decrease in ethanol content of blood
cells (CAC:BAC)if the separation is independent of the hematocrit. Winekand Carfagna (4) found the SAC:BACratios for 50
subjects were independent of the hematocrit; their finding is
supported by the data of Wells and Barnhill, Jr. (16). Accordingly, the POR would be unity if no significant change in water
or ethanol content has occurred.
In this work the mean POR for 167 subjects was 0.989 (SD,
0.091) and ranged from 0.712 to 1.26. There was also a decrease
in the mean and range limiting ratios of the POR and CAC:BAC
for measurements with a 2-year delay, yet there was no discernible change in those of SAC:BAC.This discrepancy can be
explained by the oxidation (loss) of ethanol to acetaldehyde
from the reduction of oxyhemoglobin to methemoglobin in
erythrocytes (28-33), which are only separated into the blood
cell fraction. When the PORs for the 23 subjects with the substantial (2 year) delay were excluded, the mean POR for 144
subjects was 0.997 (SD, 0.088), ranged from 0.757 to 1.26,
and had a more normal distribution (in skewness). The SD
found was similar to an estimate of 0.079 using Gaussian propagation of relative variabilities.
We computed the PORs for 25 subjects studied by Payne
and co-workers (9,10) and found the PORshad a mean of 0.912
(SD, 0.069) and ranged from 0.779 to 1.03. Because their mean
SAC:BACwas 1.17, which was similar to our mean of 1.14, the
departure from unity of the POR indicates their mean CAC:BAC
of 0.778 may be low by approximately 10% (1/0.912). This departure can account for the difference between their findings
and ours for the CAC:BACratio.
Further scrutiny of the studies by Payne and co-workers
found their subject with the extreme SAC:BACratio of 1.35 and
a CAC:BACratio of 0.682 had a POR of 0.919. However,the subjects with the closest CAC:BACratios (e.g., 0.649, 0.673, and
0.706) had consistently lower PORs (0.803, 0.793, and 0.779;
mean, 0.791) than that subject, which suggests the SAC:BAC
ratio of 1.35 is inconsistently high by approximately 16%
(0.919/0.791).
The variability in ethanol measurements in this work increased progressively for the series of an aqueous standard,
serum, blood, and blood ceils from 0.55 to 1.11%, which
agrees with the trend expected from considering the homogeneity and viscosity of those samples. The variability (CV) in
CAC:BACratios was approximately twice that of SAC:BACratios (7.0% compared with 3.6%) and includes additional contributions from volumetric sampling and in vitro oxidation of
ethanol by erythrocytes. The relative variability in the
CAC:BACratios in this work is consistent with the findings of
0.065 by Payne et al. (9) and 0.06 by both Illchmann-Christ (5)
and Schleyer (13).
To assess the reliability of estimating a BAC using a CAC
measurement, we examinedthe differences.When a conversion
factor for CAC:BACof 0.93 (mean + 1 SD) was used, 88.6% of
BACs were underestimated, at most by 25 mg/dL, and the
largest overestimate was 7 mg/dL. Thus, the measurement of
ethanol in blood cells can be forensicallyuseful when sufficient
blood or serum is not available.
Acknowledgments
M. Corbett expressesappreciation to Ronald A. Hallett for his
comments on the manuscript and to David J. Wells for providing data associatedwith his presentation at the TIAFT--SOFT
1994 meeting in Tampa (15).
References
I. P.M. Rainey. Relation between serum and whole-blood ethanol
concentrations. Clin. Chem. 39:2288-92 (1993).
2. E.A. Hak, B.J. Gerlitz, P.M. Demont, and W.D. Bowthorpe. Determination of serum alcohol:blood alcohol ratios. Can. Soc.
Forensic Sci. J. 28:123-26 (1995).
3. N.K. Shajani, W. Godolphin, and B.A. Image. Blood alcohol
analysis: comparison of whole blood analysis by gas chromatography with serum analysis by enzymatic method. Can. Soc.
Forensic Sci. J. 22(4): 317-20 (1989).
4. C.L. Winek and M. Caffagna. Comparison of plasma, serum, and
whole blood ethanol concentrations. J. AnaI.Toxicol. 11:267-68
(1987).
5. A. IIIchmann-Christ. Untersuchungen uber die relationen yon
blutkuchen-vollbult-(serum-) alkoholwerten. CA study on the relationship between the alcohol content of blood clots, whole
blood and serum"). Deutsche. Z. Ges. Gerichtl. Med. 49:113-29
(1959).
6. G. Dotzauer, H. Redetzki, K. Johannsmeier, and T. Bucher. Erprobung einer spezifishen fermentmethode aus mikrobestimmung von athylalkohol. Deutsche. Z. Ges. Gerichtl. Med. 41:
15-29 (1952).
7. A.W. Jones, R.G. Hahn, and H.P. Stalberg. Distribution of ethanol
and water between plasma and whole blood; inter- and intraindividual variations after administration of ethanol by intravenous infusion. Scand. J. Clin. Lab. Invest. 50:775-80 (1990).
8. B.T. Hodgson and N.K. Shajani. Distribution of ethanol: plasma
to whole blood ratios. Can. Soc. Forensic Sci. J. 18(2): 73-77
(1985).
9. J.P. Payne, D.W. Hill, and D.G.L. Wood. Distribution of ethanol
between plasma and erythrocytes in whole blood. Nature 217:
963-64 (I 968).
10. J.P. Payne, D.V. Foster, D.W. Hill, and D.G.L. Wood. Observations
on interpretation of blood alcohol levels derived from analysis of
urine. Br. Med. J. 3:819-23 (1967).
11. O. Gruner. Die verteilung des alkohols im blut. ("The distribution
of alcohol in blood"). Deutsche. Z. Ges. Gerichtl. Med. 46:10-I 7
(1957).
12. E.M.P. Widmark. Principles and Applications of Medicolegal Alcohol Determination. R.C. Baselt, Ed., Biomedical Publications,
Davis, CA, 1981, p 23.
13. F. Schleyer. Noch einmal: alkoholbestimmung am blutkuchen.
("Alcohol determination in blood clots: once again"). Blutalkohol
6:286-87 (I 969).
14. H. Dencks. Blutalkoholbestimmung aus blutigem serum. ("Blood
alcohol analysis of bloody serum"). Deutsche. Z. Gericht. Med.
36:181-86 (1942).
15. D.J. Wells and T.B. Barnhill, Jr. Comparison of hospital laboratory
serum alcohol obtained by an enzymatic method with whole
blood values determined by gas chromatography. Presentedat the
Joint Congress of The International Association of Forensic Toxicologists (TIAFT) and the Society of Forensic Toxicologists (SOFT)
on November 3, 1994 in Tampa, Florida by D.J. Wells.
16. D.J. Wells, Ph.D. College of Medicine, Dept. of Pathology, University of South Alabama Medical Center, Mobile, Alabama, personal communication.
17. J.G. Wigmore and M.J. Ward. The incidence of ethanol and ace177
Journal of Analytical Toxicology, Vol. 20, May/June1996
18.
19.
20.
21.
22.
23.
24.
25.
178
tone in the blood and urine of victims of sexual assault. Can. Soc.
Forensic Sci. J. 19(1 ): 49-58 (1986).
H.F. Brettel and B. Maske. Zur alkoholbestimmung bei blutentnahme im schockzustand. ("Blood alcohol determination in conditions of shock"). Blutalkohol 8:360-73 (1971).
H.F. Brettel. Die alkoholbegutachtung bei traumatisierten und
narkotisierten ("The expert opinion of alcohol in individuals
with trauma and under anesthetic"). Blutalkohol 11 (1): 1-10
(1974).
O. Gruner, N. Bilzer, and J. Walle. Blutalkoholkurve und Widmark-werte bei dialyseabhangigen patienten. ("Blood alcohol
curve and Widmark values in patients dependent on dialysis").
Blutalkoho117: 371-78 (1980).
H.F. Brettel. Blutalkohol und blutwassergehalt: methodik-theorie-praxis ("Blood alchol and blood water: method-theory-practice"). Verlag Max Schmidt-Romhild, LObeck, Germany, 1972.
H.J. Hinckers. Der wasser-(alkohol-) verteilungsfaktor: serum/vollblut wahrend der schwangerschaft ("The water (alcohol) distribution factor: serum/whole blood during pregnancy"). Blutalkohol
8:342-49 (1971 ).
H.J. Hinckers. Die wasserkonzentration des vollblutes, der erythrozyten und des serums und deren bedeutung fur den wasser(alkohol-) verteilungsfaktor: serum/vollblutl ("The water content
of whole blood, erythrocyte and serum and its importance for the
water (alcohol) distribution ratio: serum/whole blood").
Blutalkohol 5:436-41 (1968).
R.H. Gadsden, St. Study of forensic and clinical source
hemoglobin interference with the DuPont ACA ethanol method.
Ann. Clin. Lab. Sci. 16(5): 399-406 (1986).
J.V. Karger. Der diffusionsausgleich des alkohols im blut. ("The dif-
26.
27.
28.
29.
30.
31.
32.
33.
fusion equalization of alcohol in the blood"). Blotalkohol 6:
47-52 (1969).
Y.H. Caplan. Blood, urine, and other fluid and tissue specimens
for alcohol analyses. In Medicolegal Aspects of Alcohol Determination in Biological Specimens, J.C. Garriott, Ed., Lawyers &
Judges Publishing, Tucson, AZ, 1993.
T.G. Schwar. Distribution of alcohol. In Alcohol, Drugs and Road
Traffic. W.E. Cooper, T.G. Schwar, and L.S. Smith, Eds., Juta and
Co., Cape Town, South Africa, 1979.
G.A. Brown, D. Neylan, W.J. Reynolds, and K.W. SmalldOn. The
stability of ethanol in stored blood. Part I. Important variables and
interpretation of results. Anal. Chim. Acta 66:271-83 (1973).
K.W. Smalldon and G.A. Brown. The stability of ethanol in stored
blood. Part II. The mechanism of ethanol oxidation. Anal. Chim.
Acta 66:285-90 (1973).
K.W. Smalldon. Ethanol oxidation by human en/throcytes. Nature
245:266-67 (1973).
P.K. Wilkinson, J.G. Wagner, and A.J. Sedman. Sensitive headspace gas chromatographic method for the determination of
ethanol utilizing capillary blood samples. Anal. Chem. 47:
1506-10 (1975).
R.B. Chang, W.A. Smith, E. Walkin, and RC. Reynolds. The stability of ethyl alcohol in forensic blood specimens. J. Anal. Toxicol. 8:66-67 (1984).
D.S. Christmore, R.C. Kelly, and L.A. Doshier. Improved recovery
and stability of ethanol in automated headspace analysis.
]. Forensic Sci. 29:1038-44 (1984).
Manuscript received March 10, 1995;
revision received August 10, 1995.