Forensic Science International 161 (2006) 151–157

www.elsevier.com/locate/forsciint

Headspace gas chromatographic determination of ethanol:

The use of factorial design to study effects of blood
storage and headspace conditions on ethanol
stability and acetaldehyde formation in
whole blood and plasma
Lena Kristoffersen *, Liv-Ellen Stormyhr, Anne Smith-Kielland
Norwegian Institute of Public Health, Division of Forensic Toxicology and Drug Abuse, P.O. Box 4404 Nydalen, NO-0403 Oslo, Norway
Received 27 November 2005; received in revised form 2 February 2006; accepted 3 March 2006

Abstract
Our headspace gas chromatographic flame ionization detection (HS-GC-FID) method for ethanol determination showed slightly, but
consistently, low ethanol concentrations in whole blood (blood) in proficiency testing programs (QC-samples). Ethanol and acetaldehyde were
determined using HS-GC-FID with capillary columns, headspace equilibration temperature (HS-T8) of 70 8C and 20 min equilibration time (HS-
EqT). Full factorial designs were used to study the variables HS-T8 (508–70 8C), HS-EqT (15–25 min), ethanol concentration (0.20–1.20 g/kg)
and storage at room temperature (0–6 days) with three sample-sets; plasma, hemolyzed blood and non-hemolyzed blood. A decrease in the
ethanol concentration in blood was seen as a nearly equivalent increase in the acetaldehyde concentration. This effect was not observed in plasma,
indicating chemical oxidation of ethanol to acetaldehyde in the presence of red blood cells. The variables showed different magnitude of effects in
hemolyzed and non-hemolyzed blood. A decrease in ethanol concentration was seen even after a few days of storage and also when changing the
HS-T8 from 50 to 70 8C. The formation of acetaldehyde was dependent on all the variables and combinations of these (interactions) and HS-T8
was involved in all the significant interaction effects. Favorable instrumental conditions were found to be HS-T8 of 50 8C and HS-EqT of
15–25 min. The ethanol concentrations obtained for the range 0.04–2.5 g/kg after analyzing authentic forensic blood samples with a HS-T8 of
50 8C were statistically significantly higher than at 70 8C (+0.0154 g/kg, p < 0.0001, n = 180). In conclusion, chemical oxidation of ethanol to
acetaldehyde in the presence of red blood cells has been shown to contribute to lowered ethanol concentrations in blood samples. Storage
conditions before analysis and the headspace equilibration temperature during analysis were important for the determination of blood ethanol
concentrations.
# 2006 Elsevier Ireland Ltd. All rights reserved.

Keywords: Ethanol; Acetaldehyde; Chemical oxidation; Headspace gas chromatography; Factorial design

1. Introduction
The Norwegian Institute of Public Health, Division of
Forensic Toxicology and Drug Abuse receives forensic blood
and urine samples for determination of ethanol. Introduction of
two new HS-GC-FID methods for determination of ethanol in
blood and urine showed slightly low blood ethanol concentra-
tions in proficiency testing programs. This was seen over time
* Corresponding author.
E-mail address: lena.kristoffersen@fhi.no (L. Kristoffersen).
0379-0738/$ – see front matter # 2006 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.forsciint.2006.03.034
as a small, but distinct drop in our results compared to the mean
of the participating laboratories. It was most pronounced at low
blood ethanol concentrations. The same effect was not observed
in urine (Fig. 1).
Several studies have described the stability of ethanol in
biological samples [1–6]. Brown and Neylan [4] performed a
25 factorial design experiment that included the variables
storage time, storage temperature, concentration of preserva-
tive, ethanol concentration and type of container. They
concluded that only storage time, storage temperature and
concentration of sodium fluoride (NaF) preservative were of
importance. Addition of 0.5% (w/v) NaF completely inhibited
152 L. Kristoffersen et al. / Forensic Science International 161 (2006) 151–157

Fig. 1. Ethanol in blood and urine QC-samples analyzed with a HS-T8 of 70 8C. Results are reported as Bias Index Score (BIS)* [17] which is the difference of the
laboratory measurement from the consensus mean. A BIS value of 300 would results for a measurement 3 S.D. from the consensus mean. Blood ethanol results from
June, July and August have an ethanol concentration lower than 0.6 g/kg and a late delivery of 7–10 days.

the growth of micro-organisms. The loss of ethanol from
preserved blood was due to an extremely temperature
dependent, non-enzymatic ethanol oxidation reaction which
were independent of the ethanol concentration over a wide
range. It was proposed that oxyhemoglobin was involved in the
chemical oxidation of ethanol to acetaldehyde [4,5]. However,
acetaldehyde was not measured in this study. Chen et al. [6]
showed that the rise in acetaldehyde in the presence of ethanol
was dependent on the concentration of oxygenated hemoglo-
bin and concluded that oxyhemoglobin contributes to the
oxidation of ethanol to acetaldehyde. Furthermore, the
acetaldehyde formation was not inhibited by fluoride [6]
indicating that the ethanol loss was not a result of
microbiological activity. Neither could salting out effects
explain the observations, since NaF concentrations up to 0.8%
(w/v) has been shown to be non-significant for ethanol
determinations [7,8]. Acetaldehyde accumulation has been
reported during long HS-EqT of ethanol positive blood
samples at a HS-T8 of 50 8C or higher [9,10]. Acetaldehyde
can also be observed after deproteinization of ethanol positive
blood samples [11]. In order to prevent ethanol oxidation,
chemical inhibitors can be used to eliminate oxyhemoglobin
[5,12–14].
Although ethanol stability has been studied, there is limited
information available concerning non-enzymatic ethanol
oxidation during sample storage and headspace analysis of
different sample matrices. In this paper, we present factorial
designs with the variables HS-T8, HS-EqT, ethanol concen-
tration and storage time and their effects on the responses
ethanol and acetaldehyde concentrations. Separate factorial
designs were used for three sample sets; plasma, hemolyzed
blood and non-hemolyzed blood. To confirm the relevance of
the results to routine analysis, authentic blood and QC-
samples were analyzed using headspace temperatures of both
50 and 70 8C.
2. Materials and methods
2.1. Reagents and solutions
Ethanol 99.8%, 1-propanol 99.5%, tert-butanol 99.5%,
acetaldehyde 99.9% (all analytical grade) and sodium
dithionite were obtained from Merck (Darmstadt, Germany).
Distilled water (<5 mS/cm) was used for all solutions. Internal
standard solutions at concentrations of 0.08 mg/ml 1-propanol
(Rtx-BAC1) and 0.032 mg/ml tert-butanol (Rtx-BAC2),
respectively, were prepared. In the experiments with a chemical
agent to inhibit the oxidation of ethanol, a tert-butanol internal
standard solution containing 0.1 M sodium dithionite was used.
The aqueous standards were prepared in-house and stored
(4 8C) in air-tight glass ampoules and volumetric flasks for
ethanol and acetaldehyde, respectively. The standards were
prepared for the ethanol and acetaldehyde concentration ranges
0.04–4.80 g/kg (n = 7) and 0.015–0.29 g/kg (n = 6), respec-
tively.
2.2. Instrumentation and quantification
A 100 mL aliquot of sample was diluted with 1000 mL
internal standard solution (Hamilton dual syringe diluter
Micro lab1 530B, Hamilton, Bonaduz, Switzerland) in 20 mL
HS-vials (Apodan, Copenhagen, Denmark). Thus, deprotei-
nization which could give rise to acetaldehyde [11], was not
part of the procedure. The samples were analyzed on a HS-GC-
FID system with a Turbomatrix 110 HS-autosampler, an
Autosystem XL GC FID and an interface 600 series link, all
from Perkin-Elmer (Norwalk, USA). Totalchrom Client Server
(Perkin-Elmer) was used for instrument control and data
collection. Two capillary columns were used: Rtx-BAC2
(30 m, 0.32 mm i.d, 1.2 mm d.f.) in the factorial design
experiments and Rtx-BAC1 (30 m, 0.32 mm i.d., 1.8 mm d.f.)
 L. Kristoffersen et al. / Forensic Science International 161 (2006) 151–157 153

for the method comparisons (Restek Corporation, Bellefonte,
USA). HS-GC-FID methods: The HS-T8 was 70 8C (recom-
mended by Restek Corporation [15]) or 50 8C and HS-EqT was
20 min, pressurization time 2.0 min, injection time 0.01 min
Rtx-BAC1 (0.02 min Rtx-BAC2), needle and transfer line
temperature 140 8C and 160 8C, respectively. The GC injector,
oven and detector temperature were 200 8C, 45 8C Rtx-BAC1
(40 8C Rtx-BAC2) and 250 8C, respectively. Carrier gas
pressure was 30 psi in the HS-autosampler and 23 psi in the
GC.
Ethanol was quantified with a quadratic curve fit and
acetaldehyde with a linear curve fit using internal standard. Due
to differences in specific weights, the theoretical concentrations
of aqueous standards (specific weight 1.000 g/mL) were
divided by the specific weight of blood (1.055 g/mL) to give
correct concentrations (g/kg) in the blood samples.
2.3. Samples
2.3.1. Factorial design samples
Blood from a single person who had not consumed ethanol
for the last 36 h before sampling was used. Freshly drawn blood
(non-hemolyzed), blood and plasma that had been frozen
(hemolyzed) at 20 8C and thawed, were spiked to the
concentration 0.2, 0.7 and 1.2 g/kg ethanol in 5 mL Vacutai-
ner1 vials containing 20 mg NaF (0.4%, w/v) and sodium
heparin 75 USP units (BD Vacutainer Systems, Plymouth,
England). Samples were prepared in duplicates and analyzed at
day 0 and 6, respectively.
2.3.2. Authentic samples
Forensic blood samples from suspected drunken drivers
were received in Vacutainer1 vials (see above).
2.3.3. QC-samples
The ethanol QC blood and urine samples were obtained from
Heath Control-UKNEQAS & Toxicology schemes (Cardiff
Bioanalytical Services Ltd., United Kingdom) covering low to
toxic concentrations of ethanol. All samples were stored at 4 8C
prior to analysis.
2.4. Experimental design
Higher order interaction effects were used to determine
statistically significance (95% confidence level) in the full
factorial designs [16]. One replicate was performed for each

Table 1
Experimental variables, levels, design table and results for the 24 full factorial design for ethanol and acetaldehyde determination in hemolyzed and freshly drawn
blood
Variables Coded Center
( 1) (0) (+1)
HS-Temperature (8C) HS-T8 50 60 70
HS-Equilibration time (min) HS-EqT 15 20 25
Ethanol concentration (g/kg) Ethanol 0.2 0.7 1.2
Storage time (days) Storage 0 3 6

Exp. no. Variables Responses

Hemolyzed Freshly
blood drawn blood
HS-T8 HS-EqT Ethanol Storage Ethanol Acetaldehyde Ethanol Acetaldehyde
(8C) (min) (g/kg) (days) (g/kg) (g/kg) (g/kg) (g/kg)
1 50 15 0.2 0 0.1891 0.0056a 0.1885 0.0000a
2 70 15 0.2 0 0.1792 0.0172 0.1849 0.0122
3 50 25 0.2 0 0.1814 0.0053a 0.1886 0.0000a
4 70 25 0.2 0 0.1840 0.0238 0.1870 0.0186
5 50 15 1.2 0 1.1762 0.0058a 1.1572 0.0024a
6 70 15 1.2 0 1.1277 0.0223 1.1216 0.0161
7 50 25 1.2 0 1.1330 0.0059a 1.1305 0.0026a
8 70 25 1.2 0 1.1489 0.0324 1.1385 0.0265
9 50 15 0.2 6 0.1722 0.0235 0.1890 0.0054a
10 70 15 0.2 6 0.1623 0.0403 0.1809 0.0167
11 50 25 0.2 6 0.1620 0.0262 0.1804 0.0068a
12 70 25 0.2 6 0.1531 0.0461 0.1732 0.0208
13 50 15 1.2 6 1.1414 0.0249 1.1365 0.0056a
14 70 15 1.2 6 1.0886 0.0471 1.1146 0.0220
15 50 25 1.2 6 1.1168 0.0292 1.1189 0.0084a
16 70 25 1.2 6 1.1007 0.0565 1.0996 0.0262
Center sampleb 60 20 0.7 3 0.6385 0.0234 0.6414 0.0080a
Center sampleb 60 20 0.7 3 0.6276 0.0244 0.6489 0.0079a
Center sampleb 60 20 0.7 3 0.6397 0.0238 0.6461 0.0070a
a
Results below LOQ.
b
For the center sample the value of every variable is set at its mid-level (halfway between low ( 1) and high (+1) level).
154 L. Kristoffersen et al. / Forensic Science International 161 (2006) 151–157

design sample. The experimental error was obtained by 3. Results

analyzing three center samples. Unscrambler 9.1 software
(Camo Process AS, Oslo, Norway) was used for design and
evaluation of the chemometric studies.
2.5. Validation
Ethanol: The methods were validated using 50 8C as HS-T8
for spiked blood samples. Limits of detection (LOD) and
quantification (LOQ) were determined by analyzing five
different ethanol negative blood samples on 10 successive
assays and were set to the mean of the negative samples plus 3
and 10 standard deviations respectively, of a blood control
sample spiked to a concentration near LOQ. Repeatability was
determined analyzing in one assay spiked blood (n = 10; 1.0 g/
kg). Between assay precision was determined by analyzing the
same samples in 10 different assays, one replicate in each assay.
Accuracy was calculated as the percent deviation of the
measured concentration from the corresponding theoretical
concentration. Acetaldehyde: LOD, LOQ and between assay
precision were determined by analyzing an ethanol negative
blood sample and a spiked aqueous acetaldehyde control
sample (0.026 g/kg) with the factor settings for each experi-
ment number (Table 1). LOD and LOQ were otherwise
determined as for ethanol.
2.6. Matrix and method comparison
Matrix comparisons were carried out with QC blood and
urine samples. The samples were analyzed with a HS-T8 set at
70 8C (see above). The samples had identical storage conditions
before the analysis and were analyzed on the same instrument
in the same sequence.
Method comparisons were carried out with identical
instrument settings with the exception of HS-T8 that was set
at 50 8C and 70 8C, respectively. Both authentic blood and QC
blood samples were analyzed. From the sample an aliquot was
drawn for each method and both were analyzed on the same
instrument in the same sequence.
Three sample sets were studied. Since oxyhemoglobin in the
red blood cells was considered essential for the chemical
oxidation of ethanol [6], the first sample set was plasma that
was free from red blood cells, the second set was completely
hemolyzed blood (frozen and thawed) where all the red blood
cells were disrupted and the third set was non-hemolyzed blood
(freshly drawn) where the red blood cells were mainly intact.
3.1. 23 Factorial design with ethanol spiked plasma
The factor levels were equal to those for blood (Table 1), but
storage time was excluded. Neither HS-T8 nor HS-EqT was
found to be statistically significant for ethanol concentration and
no acetaldehyde was detected in any samples (results not shown).
3.2. 24 Factorial design with hemolyzed ethanol spiked
blood
The factorial design (Table 1), shows the levels of variables
and how the two responses, ethanol and acetaldehyde
concentrations, were affected when the levels of the variables
were changed simultaneously. There was not detected ethanol
or acetaldehyde in the blood blank samples analyzed with each
exp. no. (not shown).
3.2.1. Ethanol response
Increased storage time and increased HS-T8 showed
significant negative main effects on ethanol concentration,
i.e. a decrease in ethanol concentration was seen when these
design variables were changed from low ( 1) to high (+1) level
(Tables 1 and 2).
3.2.2. Acetaldehyde response
All the main effects and three interaction effects (HS-
T8  HS-EqT, HS-T8  ethanol concentration and HS-T8 
storage time) were significant (Table 2). HS-T8 was involved in
all the significant interaction effects, indicating the importance

Table 2
Results of the 24 full factorial design with hemolyzed and freshly drawn ethanol spiked samples
Variables Hemolyzed blood Freshly drawn blood
Ethanol Acetaldehyde Ethanol Acetaldehyde
p-Value Effect (g/kg) p-Value Effect (g/kg) p-Value Effect (g/kg) p-Value Effect (g/kg)
HS-T8 (A) 0.0403 0.0160 0.0000 0.0199 0.0403 0.0112 0.0000 0.0160
HS-EqT (B) 0.2754(NS) 0.0071 0.0004 0.0048 0.1423(NS) 0.0071 0.0055 0.0037
Ethanol (C) 0.0000 +0.956 0.0005 0.0045 0.0000 +0.943 0.0058 0.0036
Storage (D) 0.0049 0.0278 0.0000 0.0219 0.0242 0.0130 0.0032 0.0042
AB 0.0565(NS) 0.0026 0.0031 0.1908(NS) 0.0223 0.0026
AC 0.1651(NS) 0.0024 0.0023 0.1970(NS) 0.0571(NS)
AD 0.3501(NS) 0.0341 0.0016 0.4981(NS) 0.2121(NS)
BC 0.8031(NS) 0.1050(NS) 0.4232(NS) 0.4011(NS)
BD 0.8860(NS) 0.2736(NS) 0.2593(NS) 0.4908(NS)
CD 0.2954(NS) 0.1785(NS) 0.1655(NS) 0.5224(NS)
Effects with p > 0.05 was not significant (NS).
 L. Kristoffersen et al. / Forensic Science International 161 (2006) 151–157 155

Fig. 2. HS-T8 and HS-EqT interaction effect for acetaldehyde in hemolyzed blood samples. Variations in HS-T8 is presented on the x-axis, where as HS-EqT is seen
on the y-axis, the ethanol concentration and storage time were constant (0.2 g/kg and storage time 0 days). Each line in the plot represents a concentration of
acetaldehyde. The same interaction plot was seen for freshly drawn blood samples, however the acetaldehyde concentrations were lower.

of HS-T8 on the formation of acetaldehyde. Increased storage
time resulted in an increase in acetaldehyde concentration
which was of the same magnitude (+0.0219 g/kg) as the
decrease in ethanol concentration ( 0.0278 g/kg) (Table 2).
The interaction between HS-T8 and HS-EqT (Fig. 2) showed
that the acetaldehyde formation was nearly independent of HS-
EqT at 50 8C. At 70 8C more acetaldehyde was formed and it
was dependent on HS-EqT showing an increase in acetaldehyde
concentration with increasing HS-EqT. The interaction
between HS-T8 and ethanol concentration showed that the
acetaldehyde concentration was low and nearly independent of
ethanol concentration at 50 8C. At higher HS-T8 the
acetaldehyde formation was higher and dependent on ethanol
concentration. However, the increased acetaldehyde formation
was not proportional to the increase in ethanol concentration
(from 0.2 to 1.2 g ethanol/kg) (results not shown).
3.3. 24 Factorial design with freshly drawn ethanol spiked
blood
A full factorial design for freshly drawn blood spiked with
ethanol was carried out with the same levels of the design
variables as for the hemolyzed samples (Table 1).
the storage effect for both ethanol and acetaldehyde was smaller
for freshly drawn samples when compared to hemolyzed
samples (Table 2). Thus, the acetaldehyde concentrations
observed in freshly drawn samples were generally lower than
for hemolyzed samples.
3.4. Chemical inhibitor
Sodium dithionite was added to the internal standard
solution in an attempt to reduce oxyhemoglobin to hemoglobin.
However, it disturbed the determination of acetone performed
with the same method. Therefore no chemical inhibitor was
added to the final method (results not shown).
3.5. Validation
Ethanol: LOD and LOQ were 0.010 g/kg and 0.04 g/kg,
respectively. Repeatability, expressed as relative standard
deviation, was <1.3%, between assay reproducibility was
<2.9% and accuracy was < 6.7% deviation from theoretical
value. Acetaldehyde: LOD and LOQ were 0.003 g/kg and
0.01 g/kg, respectively. Between assay reproducibility was
5.1% with an accuracy of 18%.

3.3.1. Ethanol response 3.6. Matrix and method comparison

Storage time and HS-T8 showed significant negative main
effects. The effect of increasing the HS-T8 from 50 to 70 8C was
a decrease in ethanol concentration ( 0.0112 g/kg) of the same
magnitude as observed for hemolyzed samples ( 0.0160 g/kg)
(Table 2).
3.3.2. Acetaldehyde response
All the main effects and one interaction effect (HS-T8  HS-
EqT) were significant (Table 2 and Fig. 2). The magnitude of
Matrix comparison: In the QC-samples lower ethanol
concentrations compared to consensus mean were found in
blood, but not in urine (Fig. 1).
Method comparison: The mean concentrations of the
authentic blood samples (Table 3) for the range 0.04–2.50 g/
kg were 1.3395 g/kg (70 8C) and 1.3549 g/kg (50 8C),
respectively. The difference (average) of +0.0154 g/kg was
statistically significant ( p < 0.0001, Student’s paired t-test). At
156 L. Kristoffersen et al. / Forensic Science International 161 (2006) 151–157
Table 3
HS-GC-FID determination of ethanol in authentic blood samples (n = 180)
analyzed with identical instrument settings with the exception of HS-T8 that
was set at 50 and 70 8C, respectively
Ethanol (g/kg) n Mean difference p-value (Student’s
concentration range 1 S.D. (50–70 8C) paired t-test)
0–0.5 27 0.0068  0.0047 <0.001
0.5–1.0 21 0.0130  0.0069 <0.001
1.0–1.5 47 0.0186  0.0199 <0.001
1.5–2.0 43 0.0185  0.0173 <0.001
2–2.5 30 0.0183  0.0410 0.021
>2.5 12 0.0103  0.0304 0.263
higher ethanol concentrations the methods showed no
statistically significant difference. The results of the QC blood
samples were higher at HS-T8 of 50 8C than at 70 8C, but still
somewhat lower than the consensus mean (results not shown).
4. Discussion
Ethanol blood concentrations analyzed with a HS-T8 of 70 8C
showed systematically low ethanol concentrations in proficiency
testing samples. A similar effect was not observed in urine
samples. Blood was therefore investigated in a full factorial
design setup. Such a setup is a powerful tool in studying complex
interaction effects and differences between sample sets. In the
present study it was shown how storage (freezing, room
temperature) and various settings of the headspace equilibration
affected ethanol concentration and acetaldehyde formation in
hemolyzed and non-hemolyzed (freshly drawn) blood.
It has been shown that a decrease in blood ethanol
concentrations resulted in an increase in acetaldehyde concen-
trations in the same order of magnitude. Control experiments
showed that there was no formation of acetaldehyde in plasma
spiked with ethanol. Neither ethanol nor acetaldehyde was
detected in non-spiked blood and the sodium fluoride added to
the blood made microbial activity unlikely [4,6]. Thus, it was
concluded that the detected acetaldehyde was a result of a
chemical oxidation of ethanol to acetaldehyde in the presence of
red blood cells. The acetaldehyde formation was nearly
independent of the ethanol concentration explaining the
observation that the greatest relative deviations were observed
at low ethanol levels. Similar results were obtained analyzing
ethanol positive authentic blood samples showing approximately
the same mean differences for the studied concentration ranges
(Table 3). The results were in accordance with a chemical
oxidation of ethanol catalyzed by the oxyhemoglobin (limiting
factor) as described in literature [4–6].
The factorial designs showed that the headspace temperature
was of importance, while equilibration time was less critical.
Decreasing the temperature from 70 to 50 8C gave a significant
increase in ethanol concentrations in both non-hemolyzed and
hemolyzed blood. The increase was similar in the two groups
and low HS-T8 should therefore be preferred.
The decrease in ethanol concentration due to storage at room
temperature was more extensive in hemolyzed than in non-
hemolyzed blood. Non-hemolyzed blood may be subjected to
gradual hemolysis during storage making the oxyhemoglobin
in the red blood cells slowly available for oxidation. These
findings verify further the findings by Chen et al. [6] that
oxyhemoglobin contribute to the non-enzymatic ethanol
oxidation under ‘‘favorable’’ conditions.
Proficiency testing samples analyzed at HS-T8 of 50 8C were
higher than at 70 8C, but still somewhat lower than the
consensus mean (n  80). The explanation may by a late
delivery of samples (Fig. 1) and in addition, ethanol loss by
growth of micro-organisms in these QC-samples cannot be
ruled out, since the concentration of preservative was low
(typically 0.1% preservative).
5. Conclusion
This work demonstrates by the use of factorial designs that the
decrease in blood ethanol concentration due to chemical
oxidation resulted in a nearly equivalent increase in acetaldehyde
concentration. Furthermore, this non-enzymatic oxidation of
ethanol was dependent on both storage conditions before analysis
and the headspace equilibration temperature during analysis.
Although the differences may seem small, it is important in
forensic science where legal limits demand high degree of
accuracy in the determination of blood ethanol concentrations.
Acknowledgments
The authors are indebted to the laboratory staff for technical
assistance and to the National Laboratory of Forensic
Chemistry in Linköping (Sweden) for useful help in introdu-
cing our ethanol method.
References
[1] T. Winek, C.L. Winek, W.W. Wahba, The effect of storage at various
temperatures on blood alcohol concentration, Forensic Sci. Int. 78 (1996)
179–185.
[2] F. Tagliaro, G. Lubli, Chromatographic methods for blood alcohol deter-
mination, J. Chromatogr. B 580 (1992) 161–190.
[3] F. Musshoff, Chromatographic methods for the determination of markers
of chronic and acute alcohol consumption, J. Chromatogr. B 781 (2002)
457–480.
[4] C.A. Brown, D. Neylan, The stability of ethanol in stored blood. Part 1.
Important variables and interpretation of results, Anal. Chim. Acta 66
(1973) 271–283.
[5] K.W. Smalldon, G.A. Brown, The stability of ethanol in stored blood.
Part 2. The mechanism of ethanol oxidation, Anal. Chim. Acta 66 (1973)
285–290.
[6] H.-M. Chen, W.W. Lin, K.H. Ferguson, B. Keyes Scott, C.M. Peterson,
Studies of the oxidation of ethanol to acetaldehyde by oxyhemoglobin
using fluorigenic high-performance liquid chromatography, Alcohol Clin.
Exp. Res. 18 (1994) 1202–1206.
[7] A.W. Jones, Salting-out effect of sodium fluoride and its influence on the
analysis of ethanol by headspace gas chromatography, J. Anal. Toxicol. 18
(1994) 292–293.
[8] A.A. Solanky, Effect of different concentrations of sodium fluoride on
blood alcohol determination by headspace gas chromatography using the
internal standard method, J. Anal. Toxicol. 18 (1994) 63.
[9] P.K. Wilkinson, J.G. Wagner, A.J. Sedman, Sensitive headspace gas-
chromatographic method for the determination of ethanol utilizing capil-
lary blood samples, Anal. Chem. 47 (1975) 1506–1510.
 L. Kristoffersen et al. / Forensic Science International 161 (2006) 151–157
[10] M. Chiarotti, N. de Giovanni, Acetaldehyde accumulation during head-
space gas-chromatographic determination of ethanol, Forensic Sci. Int. 20
(1982) 21–25.
[11] A.R. Stowell, R.M. Greenway, R.D. Batt, Acetaldehyde formation during
deproteinization of human blood samples containing ethanol, Biochem.
Med. 18 (1977) 392–401.
[12] D.S. Christmore, R.C. Kelly, L.A. Doshier, Improved recovery and
stability of ethanol in automated headspace analysis, J. Forensic Sci.
29 (1984) 1038–1044.
[13] K. Wantanabe-Suzuki, H. Seno, A. Ishii, T. Kumazawa, O. Suzuki, Ultra-
sensitive method for determination of ethanol in whole blood by head-
space capillary gas chromatography with cryogenic oven trapping, J.
Chromatogr. B. 727 (1999) 89–94.
157
[14] D.G. McCarver-May, L. Durisin, An accurate, automated, simultaneous gas
chromatographic headspace measurement of whole blood ethanol and
acetaldehyde for human in vivo studies, J. Anal. Toxicol. 21 (1997) 134–141.
[15] Restek Corporation (Bellefonte, US). Dual-column confirmational GC
analysis of blood alcohols using the Rtx1-BAC1 and Rtx1-BAC2 col-
umns optimized for the Perkin-Elmer HS 40 Headspace autosampler.
Application note 59598.
[16] K.H. Esbensen, Multivariate data analysis—in practice. An introduction to
multivariate data analysis and experimental design, Camo Process AS,
Oslo, Norway, 2001.
[17] D.G. Bullock, C.E. Wilde, External quality assessment of urinary preg-
nancy oestrogen assay: further experience in the United Kingdom, Ann.
Clin. Biochem. 22 (1985) 273–282.