proficiency testing (external quality assessment) of drug detection in oral fluid
TRANSCRIPT
Proficiency testing (external quality assessment)
of drug detection in oral fluid
Joe Clarke a,*, John F. Wilson b
a Altrix Healthcare plc, Birchwood Science Park, Warrington WA3 7BP, UKb Department of Pharmacology, Therapeutics and Toxicology, Wales College of Medicine,
Cardiff University, Heath Park, Cardiff CF14 4XN, UK
Received 20 September 2004; received in revised form 17 November 2004; accepted 17 November 2004
Available online 6 April 2005
www.elsevier.com/locate/forsciint
Forensic Science International 150 (2005) 161–164
Abstract
Eighteen external quality assessment (proficiency testing) samples were prepared from client specimens collected with the
Intercept1 oral fluid collection device and by spiking drug-free oral fluid. Samples were circulated in pairs at quarterly intervals
to 13 UK and USA based laboratories for analysis by a panel of OraSure micro-plate Intercept1 enzyme immunoassay kits and
hyphenated mass spectrophotometric techniques. During the survey, there was a single case of non-specificity in a false report
for methadone. The major errors were of lack of sensitivity relative to the concentration thresholds specified for the
immunoassays. The sensitivity for overall ‘present’/‘not found’ reports calculated as true positives/(true positives + false
negatives) were for the amfetamine specific assay 50%, methyl-amfetamines 93%, barbiturates 64%, cannabinoids 73%, cocaine
and metabolites 100%, benzodiazepines 69%, methadone 95%, opiates 79% (opiates excluding oxycodone 93%), phencyclidine
93% and human gamma-globulin 97%. A small number of the sensitivity errors were attributable to errors in chromatographic
confirmation techniques.
# 2005 Elsevier Ireland Ltd. All rights reserved.
Keywords: Proficiency; Oral fluid; Amfetamines; Cannabinoids; Cocaine; Opiates; Human gamma-globulin
1. Introduction
There is a growing interest and utilisation of alternate
biological fluids than urine for drug testing. Oral fluid testing
has seen a rapid increase in use over recent years and the
publication of Proposed Revisions to Mandatory Guidelines
for Federal Workplace Drug Testing Programs [1] highlight
the growing acceptance of this testing medium. An increas-
ing number of laboratories worldwide are now routinely
analysing oral fluid samples collected from donors in a
combination of workplace, criminal justice and rehabilita-
* Corresponding author. Tel.: +44 1925 848900;
fax: +44 1925 848949.
E-mail address: [email protected] (J. Clarke).
0379-0738/$ – see front matter # 2005 Elsevier Ireland Ltd. All rights r
doi:10.1016/j.forsciint.2004.11.025
tion settings. It is important that results provided by analy-
tical laboratories testing the samples are error-free since, like
urine samples from similar settings, they can be used to
make clinical and/or legal decisions. Quality assurance must
be a key principle of any analytical laboratory and the
assessment of laboratory performance through external
quality assessment (proficiency testing or EQA) schemes
is an important element of a complete quality system [1–3].
In 2002, Cardiff Bioanalytical Services Ltd. (Cardiff,
UK) was commissioned by Altrix Healthcare plc (Warring-
ton, UK) to design and instigate an in-house EQA scheme.
The scheme was to be offered to laboratories analysing
samples collected using the Intercept1 oral fluid collection
device (OraSure Technologies, Inc., Bethlehem, PA, USA)
that were subsequently analysed by enzyme immunoassays
eserved.
J. Clarke, J.F. Wilson / Forensic Science International 150 (2005) 161–164162
Table 1
Analyte concentrations distributed by the oral fluid EQA scheme
between May 2002 and May 2004 and mean coefficient of variation
(CV) of measurements by scheme participants
Analyte Concentrations
(ng/ml)
CV% (range)
d-Amfetamine 10, 56, 283 16 (14–17)
Methylamfetamine 60, 61, 764 28 (6–54)
Pseudoephedrine 197 –
manufactured by the same organisation. Two samples would
be sent in a blind fashion every calendar quarter and
participants would report either qualitative results only or
a combination of qualitative and quantitative results, the
latter from a combined method of chromatography and mass
spectrometry. Thirteen laboratories from the USA and UK
have participated in the scheme and the results reported for
the 18 samples circulated between May 2002 and May 2004
are detailed and discussed in this paper.
Secobarbital 30, 31, 50 2Phenobarbital 61 –
D9-Tetrahydrocannabinol 0.6, 1.1, 4.0 42 (35–50)
Cocaine 3, 4, 6, 16, 45 34 (11–52)
Benzoylecgonine 6, 6, 9, 15, 17, 24 35 (14–84)
Diazepam 149 23
Temazepam 2 30
Methadone 6, 11, 22, 48 28 (14–48)
(d)Propoxyphene 405 –
6-Acetylmorphine 11, 13, 18, 57, 258 54 (41-61)
Morphine 15, 16, 16, 21, 63 60 (17–161)
Codeine 2, 3, 3, 7 49 (19–79)
Dihydrocodeine 12, 20, 29 –
Oxycodone 25 –
Phencyclidine 2.0, 4.0 32
Thioridazine (mesoridazine) 650 (500) –
2. Materials and methods
2.1. Preparation of oral fluid samples
Samples were prepared from oral fluid collected from
two sources. The first source of testing matrix was samples
that had been collected from clients by one of a number of
UK testing centres. The samples were collected using the
Intercept1 oral fluid collection device. The residual sample
remaining after routine laboratory analysis was selected for
use from those cases that had tested positive for the required
drug compounds. The sampling process involves placing a
collection pad impregnated with a mixture of common salts
between the sample donor’s lower gum and cheek and
leaving in place for at least 2 min. The collection pad
encourages movement by osmotic means of oral fluids
including oral mucosal transudate (OMT) from the surfaces
inside the mouth into the pad. The pad is then removed from
the mouth and inserted into a dedicated transport vial
containing 0.8 ml of a preservative (chlorhexidine digluco-
nate, Tween 20, deionised water) which minimises any
degradation of substances that might have been collected
[4,5].
Drug concentrations were determined in the donor sam-
ples at the test centre by GC–MS and samples stored at
�20 8C. The second source of matrix was oral fluid collected
by non-stimulated expectoration into a sterile tube by a drug-
free volunteer. The fluid was centrifuged and supernatant
diluted 1 + 3 (v/v) by Intercept1 negative calibrator for oral
fluid testing. EQA samples were prepared by dilution of the
client specimens with drug-free matrix to give the desired
concentration of drugs, by augmentation and addition of
certain analytes to the client specimens by spiking weighed-
in concentrations of drugs, and by spiking drugs into drug-
free matrix. All samples were heat treated at 60 8C for 1.5 h
and 1.7 ml volumes distributed in liquid form in plastic
screw-topped tubes (Cat no. 72.609 Sarstedt Ltd., Leicester,
UK) to participating laboratories by courier service.
The analytes included during the survey and their con-
centration is presented in Table 1. D-Amfetamine was spiked
into two samples and the third high concentration was a
client specimen. Methylamfetamine, pseudoephedrine,
secobarbital, phenobarbital, temazepam, (d)propoxyphene,
oxycodone, phencyclidine and thioridazine with its meta-
bolite mesoridazine were all spiked into samples. Cannabi-
noids were present in two client specimens and D9-
tetrahydrocannabinol was spiked at 4 ng/ml into drug-free
matrix in a third sample. Metabolised cocaine was present
in five client samples and benzoylecgonine alone was
spiked into a further sample at the highest concentration
circulated. Metabolites from diazepam use were present in
one client sample and from methadone treatment in four
samples. Five samples from heroin users were included.
Two of the latter additionally contained dihydrocodeine,
and four samples contained low concentrations of codeine.
The concentration of morphine in one sample was aug-
mented by spiking to obtain a concentration above the
opiate cut-off. A further sample was positive for opiates
from use of dihydrocodeine alone and contained 20 ng/ml
dihydrocodeine.
2.2. Target compounds
Laboratories participating in the EQA programme
reported the drugs detected in the samples within a spe-
cified 4-week deadline using some or all of a common
panel of OraSure micro-plate Intercept1 enzyme immu-
noassay (EIA) kits. The micro-plate assays are competitive
immunoassays intended for use in the qualitative determi-
nation of drugs and metabolites in oral fluid collected with
the Intercept1 oral fluid collection device. The drug
groups and the cut-off concentrations applied are given
in Table 2. One laboratory used immunoassays from an
alternative supplier (Cozart Bioscience Ltd., Abingdon,
UK). The Intercept1 oral fluid EIA cut-off concentration
is, in several cases, 25% of that specified by the US
J. Clarke, J.F. Wilson / Forensic Science International 150 (2005) 161–164 163
Table 2
Cut-off concentrations and target analytes applied in OraSure
Technologies, Inc., micro-plate Intercept1 enzyme immunoassay
kits
Test group Cut-off
concentration
(ng/ml)
Calibrator
compound
Amfetamine specific 100 d-Amfetamine
Methyl-amfetamines 40 d-Methamfetamine
Barbiturates 20 Secobarbital
Cannabinoids 1 D9-Tetrahydrocannabinol
Cocaine and metabolites 5 Benzoylecgonine
Benzodiazepines 1 Nordazepam
Methadone 5 Methadone
Opiates 10 Morphine
Phencyclidine 1 Phencyclidine
Human IgG 500 Human IgG
Table 3
Sensitivity and specificity of reports from the oral fluid EQA scheme
Test group Sensitivity (%) Specificity (%)
Amfetamine specific 50 100
Methyl-amfetamines 93 100
Barbiturates 64 100
Cannabinoids 73 100
Cocaine and metabolites 100 100
Benzodiazepines 69 100
Methadone 95 99
Opiates 79 100
Phencyclidine 93 100
Human IgG 97 No data
Department of Health and Human Services (HHS) [1] to
account for the approximate 1 + 3 dilution that occurs on
extraction of the Intercept1 collection pad with the pre-
servative in the transport vial [5,6]. Additional compounds
to those specified by HHS are included in the panel to more
closely resemble the existing urine testing programmes
and the requirements for clinical testing and monitoring.
The panel of tests also included a specific test for the
presence of human gamma-globulin (IgG) at a concentra-
tion above a threshold of 500 ng/ml in order to monitor
sample integrity and to exclude non-human and other
substituted fluids [7].
Where positive immunoassay results occurred, a num-
ber of laboratories used GC–MS, GC–MS–MS and LC–
MS–MS techniques to identify and quantify the drug
compounds concerned. In addition to reporting the indi-
vidual test results, laboratories combined their immu-
noassay and chromatographic data to determine an over-
all ‘present’ or ‘not found’ report for the 10 specified
analyte groups (Table 1) and for individual compounds
identified.
2.3. Data selection and statistical analysis
The overall ‘present’/’not found’ reports were analysed
for the 10 target groups listed in Table 1. Data were included
in the analysis where either an above threshold concentration
of drug group or no drug was present. Data for drug groups
present in samples at below threshold concentrations were
excluded. Individual results from the 18 samples were
classified as being true positive (TP), true negative (TN),
false positive (FP) or false negative (FN) for the 10 target
groups. In eight cases, where a laboratory had reported an
uncertain result for a test compound present in a sample, the
result was classed as FN. Six uncertain results for analytes
not present were excluded. Sensitivity was calculated
as 100 � TP/(TP + FN) and specificity as 100 � TN/
(TN + FP).
3. Results
Between 6 and 13 (mean 11) laboratories reported on
each distribution of samples. The sensitivity and specificity
for the nine analyte groups and for the human IgG integrity
test are presented in Table 3. The mean and range of
coefficients of variation for the chromatographic measure-
ments reported by scheme participants are given in Table 1.
A clear relationship between precision and drug concentra-
tion was not evident.
There were three cases of false negative reports for
specific compounds as a result of failures in confirmation
techniques. On two occasions, a laboratory failed to detect
morphine with a morphine-specific immunoassay following
a positive opiate group test. In a similar situation, a labora-
tory reported zero measured concentrations for 6-acetyl
morphine (6-AM), morphine and codeine by GC–MS fol-
lowing a positive opiate immunoassay test when all three
compounds were present in the sample.
4. Discussion
A single false positive report was made during the survey.
It was an unconfirmed immunoassay report for methadone in
the sample spiked with (d)propoxyphene. No lack of spe-
cificity was observed in samples containing pseudoephe-
drine or thioridazine and its metabolite. All other reporting
errors were false negative or uncertain reports where an
above threshold concentration of analyte was present.
The lowest sensitivity was observed for the specific
amfetamine assay where 50% of laboratories failed to detect
amfetamine at over twice the threshold concentration. The
93% sensitivity for methyl-amfetamine was the result of two
laboratory errors. One was an immunoassay negative at
61 ng/ml whilst, in the second case, a laboratory detected
methyl-amfetamine in the same sample by immunoassay but
quantified the level by GC–MS as being below threshold at
30 ng/ml and therefore returned a report of ‘not found’. The
target barbiturate compound, secobarbital, was missed by
more than half the participants on two occasions when
spiked at 150% of the barbiturate group cut-off. On the
J. Clarke, J.F. Wilson / Forensic Science International 150 (2005) 161–164164
second occasion, the concentration was verified by chroma-
tography. Sensitivity to phenobarbital was 100% at a con-
centration of 61 ng/ml. The 73% sensitivity for cannabinoids
was largely the result half the participants returning negative
reports for the sample containing the close to cut-off con-
centration of 1.1 ng/ml. The laboratory employing the
Cozart immunoassay missed spiked D9-tetrahydrocannabi-
nol at 4 ng/ml. In the two benzodiazepine-containing sam-
ples, insensitivity was caused by temazepam spiked at 2 ng/
ml. However, the cross-reactivity of the benzodiazepine
Intercept1 micro-plate EIA to this compound is 55%, giving
an effective concentration of only 1.1 ng/ml relative to the
cut-off of 1 ng/ml. Insensitivity to opiates was the result of
negative reports in samples containing dihydrocodeine
metabolites alone or spiked with oxycodone. The reported
cross-reactivity of the Opiates Intercept1 micro-plate EIA to
dihydrocodeine is 185% (10 ng/ml) but that of oxycodone is
only 0.8% (1000 ng/ml) making this compound effectively
invisible.
A small number of sensitivity errors for the reports of
analyte groups were attributable to errors with the chroma-
tographic confirmation techniques. On one occasion, a
laboratory failed to confirm and hence report the presence
by GC–MS of four analytes, secobarbital, diazepam, metha-
done and phencyclidine, despite the respective immunoassay
tests being reported positive.
The major source of error observed was thus of sensi-
tivity where immunoassays failed to achieve their specified
cut-off. The issue of variable cross-reactivity of group assays
to different compounds in the group is potentially relevant to
the barbiturate, benzodiazepine and opiate immunoassays
but is only represented in the current data by the failure to
detect oxycodone. If these data are omitted, the sensitivity
for the opiate group increases to 93%. The analytical per-
formance of laboratories participating was comparable to
that observed for some commercial immunoassays used in
urine drug testing [8]. The increased difficulty that might
result from the concentration of drugs being typically lower
in oral fluids than in urine appears not to be a significant
factor in poly drug identification. A lack of precision in
quantitative measurements for some analytes when mea-
sured in oral fluid is evident in the scale of the coefficients of
variation observed. A low level of unexplained errors can
affect mass-spectrometric techniques applied to either
matrix.
References
[1] Department of Health and Human Services. Proposed Revisions
to Mandatory Guidelines for Federal Workplace Drug Testing
Programs, Federal Register 69 (71) (2004) 19673–19732.
[2] United Kingdom Laboratory Guidelines for Legally Defensible
Workplace Drug Testing, Version 1.0, March 2001. http://
www.wdtforum.org.uk/pdfs/wdtgde�1.pdf.
[3] International Organization for Standardization, General require-
ments for the competence of testing and calibration laboratories,
ISO/IEC 17025, 1999.
[4] R.S. Niedbala, K.W. Kardos, D.F. Fritch, S. Kardos, T. Fries, J.
Waga, Detection of marijuana use by oral fluid and urine
analysis following single-dose administration of smoked and
oral marijuana, J. Anal. Toxicol. 25 (5) (2001) 289–303.
[5] E.J. Cone, L. Presley, M. Lehrer, W. Seiter, M. Smith, K.W.
Kardos, D. Fritch, S. Salamone, R.S Niedbala, Oral fluid testing
for drugs of abuse: positive prevalence rates by InterceptTM
immunoassay screening and GC–MS–MS confirmation and
suggested cutoff concentrations, J. Anal. Toxicol. 26 (8)
(2002) 541–546.
[6] E.J. Cone, Saliva testing for drugs of abuse, Ann. NYAcad. Sci.
694 (1993) 91–127.
[7] P.P. Mortimer, J.V. Parry, Detection of antibody to HIV in saliva:
a brief review, Clin. Diagn. Virol. 2 (1994) 231–243.
[8] J.F. Wilson, B.L. Smith, Evaluation of detection techniques and
laboratory proficiency in testing for drugs of abuse in urine: an
external quality assessment scheme using clinically realistic
urine samples, Ann. Clin. Biochem. 36 (5) (1999) 592–600.