Performance Evaluation of Four On-Site Drug-Testing Devices for Detection of Drugs of Abuse in Urine

Transcription

1 Journal of Analylical Toxicology, Vol. 24, October 2000 Performance Evaluation of Four On-Site Drug-Testing Devices for Detection of Drugs of Abuse in Urine Michelle R. Peace 1, Lisa D. TarnaP, and Alphonse Poklis 1,* 1Department of Pathology, Medical College of Virginia Campus at Virginia Commonwealth University, Richmond, Virginia and 2Scientific Testing Laboratories, Inc., Richmond, Virginia Abstract On-site drug tests are becoming increasingly more popular because of their easy test protocols and instantaneous results. This study evaluates the performance of four on-site drug testing devices that use competitive binding immunoassays to qualitatively determine the presence of drugs in urine: Triage Panel for Drugs of Abuse plus TCA, QuickScreen TM Pro-Multi Drug Screening Tests, Syva Rapid Test d.a.u. TM 5 and d.a.u. TM 2, and Rapid Drug Screen TM. All devices simultaneously determine the presence of the following drugs of abuse: amphetamine (AMP), benzoylecgonine (BE), 11-nor-9-carboxy-A%tetrahydrocannabinol (THCA), opiates (OPI), and phencyclidine (PCP). Triage and Rapid Drug Screen also simultaneously test for benzodiazepines (BZB) and barbiturates (BRB), whereas QuickScreen and Rapid Test require separate devices for the BZB and BRB analyses. Urine specimens (222) containing drug concentrations around or above cutoff values were screened by ONLINE or EMIT II immunoassays. Of these, 199 yielded positive gas chromatography-mass spectrometry results with at least 17 positive specimens in each drug class. Specimens with the target drugs added at 16.7% above and below the cutoff, 33.3% above and below the cutoff, and 66.7% above the cutoff were also used to evaluate the test devices. Sensitivity and specificity calculations demonstrated that Triage performed most predictably in the donor urine specimens and the drug-added specimens. In addition, it required the least amount of test volume and was the only device in which the appearance of a colored line indicated a positive result. Therefore, of the devices studied, Triage was the most dependable and reproducible on-site drug-screening device. screen for illicit drug usage. Primarily, drug screening eliminates the specimens/individuals who tested negative for certain drugs such that the more complex and time-consuming confirmation process is not overburdened. However, in the emergency room, screening also serves to identify drugs so that a patient will receive the appropriate treatment. Drug rehabilitation programs and justice departments use on-site screening devices to immediately verify compliance, and employers randomly screen employees to curtail absenteeism and workplace accidents. Therefore, on-site urine drug-screening devices need to be reliable and rapid. They must also be cost effective and simple to use because testing may be performed outside the laboratory by individuals with little training. On-site drug-screening devices attempt to fulfill these requirements. The purpose of this study was to evaluate the ability of four on-site drug-screening devices to detect the presence of amphetamine (AMP), benzoylecgonine (BE), 11-nor-9-carboxy- A9-tetrahydrocannabinol (THCA), opiates (OPI), phencyclidine (PCP), benzodiazepines (BNZ), and barbiturates (BRB) in urine: Triage Panel for Drugs of Abuse plus TCA, QuickScreen Pro-Multi Drug Screening Tests, Syva Rapid Test d.a.u. 5 and d.a.u. 2, and Rapid Drug Screen. All devices simultaneously determined the presence of five drugs of abuse: AMP, BE, THCA, OPI, and PCP. Triage and Rapid Drug Screen also simultaneously detected the presence of BNZ and BRB. QuickScreen and Rapid Test required separate test cards for BNZs and BRBs. Triage can also determine the presence of tricyclic antidepressants (TCA). However, TCAs were not analyzed in this study (2). All screening results were confirmed by gas chromatography in tandem with mass spectrometry (GC-MS). Introduction The Department of Health and Human Services reports that 11% of the population over the age of 12 (-24 million people) voluntarily admitted to using illicit drugs in 1996, with 6.4% of the population having used an illicit drug within one month (1). Because of such prevalence, emergency rooms, rehabilitation facilities, probation officers, drug counselors, and employers * Address all correspondence to Alphonse Poklis, Ph.D, Box , Depadment of Pathology, Medical College of Virginia Campus at Virginia Commonwealth University, Richmond, VA 23298~)165, apoklis@vcu.edu Materials and Methods Materials Screening assays. Triage Panel for Drugs of Abuse plus TCA from Biosite Diagnostics (San Diego, CA), QuickScreen Pro-Multi Drug Screening Tests from Phamatech (San Diego, CA), Syva Rapid Test d.a.u. 5 and d.a.u. 2 from Behring Diagnostics, Inc. (Cupertino, CA), and Rapid Drug Screen from American Bio Medica (San Diego, CA) were acquired. All devices utilized a competitive binding immunoassay to detect the presence of drugs in urine. Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission. 589

2 Triage Panel for Drugs of Abuse plus TCA devices simultaneously determined the presence of AMP, BE, THCA, OPI, PCP, BNZ, and BRB. Drug present in donor urine specimens competed with gold conjugated drug for antibody in a reaction well. This reaction mixture was transferred to the detection membrane. Antibody immobilized in the detection membrane bound the free gold conjugated drug from the reaction mixture and produced a colored line indicating a positive result. A negative specimen was indicated by the lack of a line because the gold conjugated drug was not available to bind with the antibody immobilized in the membrane. The device was operating correctly if the control negative region showed no line and the control positive region produced a line. QuickScreen Pro-Multi Drug Screening Tests, Syva Rapid Test d.a.u. 5, and Rapid Drug Screen simultaneously determined the presence of AMP, BE, THCA, OP], and PCP. QuickScreen and Rapid Test required separate devices to test for BNZ and BRB. Rapid Drug Screen tested for BNZ and BRB on the same device. These devices utilized an antibody-dye conjugate and an immobilized drug conjugate. Dye-conjugated antibody bound with the drug-protein conjugate immobilized in the detection membrane and created a colored line, indicating a negative test. Drug in donor urine bound the dye-conjugated antibody and prohibited it from interacting with the drug-protein conjugate immobilized in the detection area. Therefore, the absence of a line indicated a positive test. The device was operating correctly if the control region produced a line. Specimens. All urine specimens except those containing either BRBs or BNZs were obtained from specimens submitted for drug analysis to Scientific Testing Laboratories, Inc. in Richmond, Virginia. Urine specimens containing BRBs and BNZs were obtained from the clinical toxicology laboratory of the Medical College of Virginia in Richmond, Virginia. Specimens chosen for evaluation contained at least one drug with a concentration near or above the screening cut-off concentration. Each drug class contained at least 17 specimens confirmed posi- tive by GC-MS. Reference material and test specimens, d-amp, BE, PCP, THCA, codeine, alprazolam, and pentobarbital were purchased from Radian Corp. (Austin, TX). Six specimens were prepared by adding these drugs to drug free urine to produce concentrations at the cutoff, 16.7% above and below the cutoff, 33.3% above and below the cutoff, and 66.7% above the screening cutoff values of each drug. Each drug-added specimen contained all seven drugs. Lyphocheck Urine Toxicology Screen Control with GC-MS confirmed concentrations 20-25% above cutoff concentrations of AMP, BE, THCA, free morphine, PCP, secobarbital, and oxazepam was also purchased from BioRad (Anaheim, CA) and tested. Methods Confirmation testing. Specimens received at Scientific Testing Laboratories, Inc. were first screened by Roche's ONLINE KIMS monoclonal immunoassay. Specimens received at the c]inical toxicology laboratory at Medical College of Drug narne Barbiturates Virginia were screened by EMIT li monoclonal immunoassay. All positive specimens were subsequently confirmed using the Hewlett-Packard model 6890 temperature-programmed GC in tandem with the Hewlett-Packard model 5973 mass selective detector. On-site device protocols. The Triage package included the pipet and wash solution needed to perform the analyses. The reaction well was checked for three reaction beads (purple, white, and yellow). These beads corresponded to the labeled drug, antibodies, and buffer, respectively (3). Using the pipet, 140 ~tl of urine specimen was pipetted into the reaction well and incubated for 10 rain at room temperature. Using the same pipet fitted with a new tip, the reaction mixture was transferred to the detection membrane and allowed to soak through completely. As soon as the mixture soaked through, three drops of wash solution were added to the detection membrane. After the wash solution absorbed into the membrane, results were read within 5 min (4). The Rapid Drug Screen package included a cup in which each analysis was performed. The urine specimen was poured into the cup to the designated level, and the test device was inserted through a slit in the cap until it touched the bottom of the cup. The device incubated approximately 5 rain (within the 3- to 10- rain window designated in the package insert), and the results were read (5). The Syva Rapid Test package included a dropper to perform each analysis. The device was placed on the table, and three drops of urine specimen were placed into the reaction well. The device was incubated approximately 5 rnin (within the 3- to 10-rain window designated in the package insert), and the results were read (6). The protocol for QuickScreen deviated from the manufacturer's instructions. The device was intended to be placed vertically into a cup containing the urine specimen. However, no cup was supplied, so approximately 0.75 ml of urine specimen was Table 1. On-Site Drug-Screening Device Cutoff Concentrations for Drugs/Drug Metabolite PCP Cocaine Amphetamine Marijuana/THC Opiates Benzodiazepines * Morphine-3-glucuronide. Kits did not specify target analytes. Device Target analyte Cutoff value (ng/ml) All PCP 25 All Benzoylecgonine 300 All d-amphetan~ine 1000 All 11 -nor-a'~-thccooh 50 Triage Morphine 300 Rapid Drug Screen Morphine/M3G* 300 Rapid Test Morphine/M3G* 300 QuickScreen Morphine/M3G" 300 Triage Barbiturates* 300 Rapid Drug Screen Barbiturates t 300 Rapid Test Secobarbital 300 QuickScreen Secobarbital 200 Triage Benzodiazepines ~ 300 Rapid Drug Screen Benzodiazepines ~ 300 Rapid Test Oxazepam 200 QuickScreen Oxazepam

3 placed in the sample wells, and the test was performed horizontally. The device was allowed to incubate for 10 min, and the results were read (7). On-site device screening. On-site device testing was performed such that the analyst knew the drug class, but the drug concentration was unknown. Drug-added urine specimens, drug-free urine specimens, and the urine control specimens were labeled with a special code such that the identity of the specimen was unknown to the analyst during testing. All positive and negative results from each assay were reported regardless of the specific drug class being tested. An exception to this rule was that for specimens known to contain BNZs and BRBs only the results of BNZ and BRB were recorded. All urine specimens had been either frozen or refrigerated and were brought to room temperature prior to analysis. Device protocols were followed as previously described, and results were recorded in the following manner: if the testing device tested for AMP, BE, THCA, OPI, and PCP simultaneously, all results were recorded for each device. Because Triage could also simultaneously test for BNZ and BRB, the results for these analytes were also recorded for all urine specimens. BNZ and BRB results were recorded for QuickScreen, Rapid Test, and Rapid Drug Screen when the urine specimens contained these analytes. Each device detected the target drugs/drug metabolites at the screening cutoff concentrations as listed in Table I. A negative result was recorded if a line appeared next to a drug on the QuickScreen, Rapid Test, and Rapid Drug Screen devices. A positive result was recorded in the absence of a line next to a drug on these same devices. For the Triage devices, if a line appeared next to a drug, a positive result was recorded, and a negative result was recorded in the absence of a line. If any device responded incorrectly in its control region, that specific test was considered invalid and the results for that device were discarded. All results from the screening devices were compared to the quantitative results previously determined by GC-MS. False positives and false-negative results were tallied. A false positive (FP) was the indication by the device that the urine drug concentration was equal to or above the screening cutoff concentration when it had been determined by GC-MS to be below the screening cutoff concentration. A false negative (FN) was the indication by the device that the urine drug concentration was below the screening cutoff concentration when it had been determined by GC-MS to be above the screening cutoff concentration. A true positive (TP) was the correct indication by a device that the urine drug concentration was equal to or above the screening cutoff concentration. A true negative (TN) was the correct indication that the urine drug concentration was below the cutoff concentration. Calculations The analytical efficiency of each on-site device was determined by calculating the sensitivity and specificity when testing the donor urine specimens. Sensitivity, the device's ability to determine the presence of a substance, was calculated using equation 1. (TP) Sensitivity = Eq. 1 (TP+ FN) Specificity, the device's ability to determine the absence of a substance was calculated using equation 2. Specificity = (TN) Eq. 2 (TN + FP) Efficiency, the device's ability to correctly determine the presence or absence of a substance was calculated using equation 3. Efficiency = (TN + TP) Eq. 3 (TN + TP + FN + FP) Table II. Performance Evaluation on Control Samples Triage Rapid Drug Screen Rapid Test Quick Screen Blank Urine Benzoylecgonine (0) 100% (-)* 100% (-) 100% (-) 100% (-) Phencyclidine (0) 100% (-) 100% (-) 100% (-) 100% (-) Amphetamine (0) 100% (-) 100% (-) 100% (-) 100% (-) 11 -nor-a%thccooh (0) 100% (-) 100% (-) 100% (-) 100% (-) Free morphine (0) 100% (-) 100% (-) 100% (-) 100% (-) Secobarbital (0) 100% (-) 100% (-) 100% (-) 100% (-) Oxazepam (0) 100% (-) 100% (-) 100% (-) 100% (-) Screen Control Benzoylecgonine (545 nglml) 100% (+)t 70% (+) 100% (+) 100% (+) Phencyclidine (32 nglml) 100% (+) 40% (+) 100% (+) 60% (+) Amphetamine (1298 nglml) 100% (+) 0% (+) 100% (+) 0% (+) 11-nor-t~%THCCOOH (71 ng/ml) 100% (+) 100% (+) 100% (+) 90% (+) Free morphine (401 ng/ml) 100% (+) 70% (+) 100% (+) 100% (+) Secobarbital (364 ng/ml) 70% (+) 80% (+) 90% (+) 100% (+) Oxazepam (380 nglml) 80% (+) 80% (+) 100% (+) 100% (+) * (-), quantity of tests were negative. ~" (+), quantity of tests were positive. Results and Discussion As seen in Table I, the on-site test devices were manufactured with specific antibodies toward PCP, BE, d-amp, THCA, and morphine and its glucuronide metabolite. Despite the designed specificity for a drug, each device had a degree of cross reactivity to other substances. However, in this study, potentially cross-reactive species in the donor urines were not identified. Single specific BRB and BNZ antibodies were not identified for these two drug classes in Triage and Rapid Drug Screen devices, but the devices were designed to generally identify several drugs in each class. Rapid Test and QuickScreen specifically targeted secobarbital and oxazepam. Therefore, a caveat concerning the FP finding when testing for BNZ must be noted. Confirmation analysis was not performed to detect all possible BNZ drugs and/or possible BNZ metabolites. Given the number of BNZ metabolites that may cross-react with the 591

4 Journal of Analytical Toxicology, Vol, 24, October 2000 on-site devices, it is possible that some unidentified BNZs were present in the donor urines. Table II summarizes the results of the performance evaluation of Triage, Rapid Drug Screen, Rapid Test, and QuickScreen testing drug-free urine specimens and the Lyphocheck Urine Toxicology Screen Control. The manufacturer confirmed the concentration of AMP, BE, THCA, free morphine, PCP, secobarbital, and oxazepam in the control specimen via GC-MS to be % above the cutoff concentrations. Ten devices from each manufacturer were tested using the drug free urine and the positive control specimens. All devices were able to verify the absence of drug in a specimen. However, each device had FN results for at least one drug class when testing the positive control specimen, indicating difficulty in identifying positive specimens near the cutoff concentration. Performance of the on-site screening devices in donor urine specimens was evaluated. Given that donor specimens may have contained multiple test drugs, a total of 1364 drug challenges (negative or positive) were analyzed with the Triage. Each of the other devices simultaneously tested for five drugs in each specimen resulting in a total of 996 drug challenges (refer to Methods section for data recording protocol). Figure 1 shows the overall test results for AMP, BE, THCA, OPI, PCP, BRB, and BNZ for the Triage device and only AMP, BE, THCA, OPI, and PeP for Rapid Drug Screen, Rapid Test, and QuickScreen. The percentages of false-positive results were 3.9% for Triage, 3.8% for Rapid Drug Screen, 5.8% for Rapid Test, and 6.2% for QuickScreen. False-negative results for Triage, Rapid Drug Screen, Rapid Test, and QuickScreen were 0.5%, 2.9%, 1.3%, and 2.4%, respectively. No control failures were noted for the Triage device over the course of the study. The calculated sensitivity and specificity of each on-site urine test device in donor urine specimens and drug-added specimens are presented in Table [[I. AMP, BE, THCA, OPI, and PCP were evaluated using 184 donor urines for each device except for Triage, which was used to test 222 urine specimens because of its simultaneous analysis of BNZ and BRB. Because all 38 urine specimens were confirmed positive for BRBs and BNZs and were not evaluated simultaneously with the other drugs, sensitivity and specificity values were not calculated for Rapid Drug Screen, QuickScreen, and Rapid Test devices. Overall performance was calculated by tabulating the total numbers oftp, TN, FP, and FN for Rapid Drug Screen, QuickScreen, and Rapid Test using the data from AMP, BE, THCA, OPI, and PCP only. Overall performance values for Triage were calculated with data from all seven analytes. As shown in Tables III and IV, Triage had the highest sensitivity, specificity, and efficiency in determining the presence or absence of drugs in donor urines and drug-added specimens. For the drug-added specimens, 60 devices each of Triage, Rapid Drug Screen, and Rapid Test were tested using the six drug-added urines as previously described. Only 36 QuickScreen devices were evaluated using these six drug-added specimens. Overall sensitivity, specificity, and efficiency for each device were calculated by tabulating the total number of TP, TN, FP, and FN from all seven drug classes. The Rapid Test data are notable in that sensitivity in the donor urine specimens was 6a.3. However, it was zero in the drug-added specimens. These disparate data could possibly arise from some discrepant devices used to test the drug-added specimens that were antibody specific for methamphetamine. Although this was not indicated on the package insert accompanying the devices, amphetamine-selective devices were only available outside the United States. As a result, methamphetaminespecific devices were mistakenly shipped to us. A recent study by Table III. Sensitivity and Specificity of Each On-Site Urine Test Device in Donor Urine Specimens and Drug-Added Specimens Triage Rapid Drug Screen Rapid Test QuickScreen Sensitivity Specificity Sensitivity Specificity Sensitivity Specificity Sensitivity Specificity Donor Specimens Phencyclidine , Amphetamine , ] Benzoylecgonine 1( , ], ,7 Opiates 1 (] (] O nor-A~THC-COOH , Barbiturates NA* NA NA NA NA NA Benzodiazepines N A N A N A N A N A N A Overall , Drug-Added Specimens Phencyclidine Amphetamine Benzoylecgonine Opiates O0 11-nor-t~gTHC-COOH t00 10(] ,7 Barbiturates Benzodiazepines Overall ,2 88, * NA is designated in the barbiturate and benzodiazepine categories because calculations could not be made be(ause of the lack of negative samples tested. 592

5 Taylor et al. (8) evaluated several immunoassays, including Rapid Drug Screen. Using drug-added specimens with 25% above and 25% below the cutoff concentrations of methamphetamine, BE, THCA, morphine, and PCP, they also demonstrated a moderate rate of false positives and false negatives in most drug classes for Rapid Drug Screen. Analysis of their data shows 76% sensitivity and 46% specificity for Rapid Drug Screen. Performance around the screening cutoff concentration was analyzed for each device (Figure 2). After plotting the percentage of positive results against the concentrations in drug-added spec- 10 g imens, the slope of each line was analyzed for linearity. Ideally, concentrations below the cutoff should have 0% positive results, concentrations at and above the cutoff should have 100% positive results. Buechler et al. (9) referred to this type of analysis as threshold analysis. Rapid Drug Screen showed poor discrimination around the cutoff concentration with its high rate of falsepositive and false-negative results in each drug class. Triage performed most accurately with nearly 100% positive results in specimens at and above the cutoff and nearly 0% positive results below the cutoff concentration. Buechler et al. (9) tested the discrimination around the cutoff concentration for THCA, PCP, and BE for 3Yiage with similar results. They attributed the marginally higher number of false-negative results in their BNZ evaluations to the antibody used by Triage having a greater affinity for the drug metabolite conjugate than for the drug metabolite. However, since that study, the antibodies used in the device have been modified to detect the free drug and/or metabolite. Overall, they found Triage demonstrated > 96% sensitivity and specificity in all drug classes (9). Wu et al. (10) evaluated the Triage system in emergency toxicology testing for drugs of abuse, and con Triage Rapid drug screen Rapid test Quick screen On-site testing devices 9 False positives 9 False negatives D Invalid Table IV. Overall Efficiency Rates for Each On-Site Testing Device in the Donor Urine Specimens and the Drug-Added Specimens Donor urine efficiency Drug-added efficiency Figure 1. Results from total number of drugs in donor urines evaluated for each test product. Total number of drugs tested on Triage devices was Syva, QuickScreen, and Rapid Drug Screen each determined the presence of 996 drugs. (FP = false positive, FN = false negative). Triage Rapid Drug Screen Rapid Test QuickScreen >= Triage Rapid Drug Screen r 30 ~ I g =.,-~ 30 J~. ' -33~0% ' -le.~0% " Cutoff ' 16.70% ' 33.00% 66.70% ~ % % Cutoff 1@70% 33.00% 0(5.70% Rapid Test 10o ~ = = -- 8O '~ 30 "5 Y o A A A 9,L Jt Q ~ ~ Cutoff 10.70% 33.OO'~ ~70~ t 8O & O*" m~ I1..~., QuickScreen *~.~K)% -Ie.70% ~off t8.70% 33,00% 66.?0% Figure 2. Percentage of positive results in drug-added samples for each test device illustrating linearity around the cutoff concentration, n = 60 for each device. ~, BE; m, PCP; A, AMP; x, THCA; o, OPI; e, BRB; and +, BNZ. 593

6 cluded that Triage performed equivalently to accepted highvolume instrument-based methods such as EMIT. However, Poklis and Edinboro (11) cautioned that Triage may not detect all BNZs for three reasons: (1) manufacturers only target a small percentage of BNZs in their on-site devices, (2) a vast number of BNZ metabolites cross-react with on-site devices, and (3) a discrepancy between antibodies in the test kits and their calibrators exists. They concluded that if enhanced BNZ detection is required, the clinic should combine Triage with another immunoassay targeted to other BNZs (12). Triage was the most complex assay requiring about 12 total minutes for testing. Rapid Test, QuickScreen, and Rapid Drug Screen were simpler and required a total test time of 6-10 min. Rapid Drug Screen, QuickScreen, and Rapid Test had two potential problems that could ultimately invalidate the test results: (1) incomplete wicking in that the urine may not saturate the detection areas and (2) the misapplication of urine onto the devices such that the device was submerged too far into the specimen and urine mistakenly entered the detection membrane without passing through the specimen placement wells. Triage alleviated these potential problems. More importantly, assessment of test results was most straightforward on the Triage devices because any formation of a line indicated a positive specimen. For the other screening devices, the appearance of a colored line indicated a negative result. This was counter-intuitive and confusing. Lines barely appearing in QuickScreen, Rapid Test, and Rapid Drug Screen devices could have been indicative of a specimen that may have drug concentrations just below the screening cutoff. However, the intensity of this colored line was not a quantitative indicator and a barely visible line was considered negative. Studies have also shown that "self-learning" during Triage evaluations was effective (13). Conclusions The analytical performance as defined by sensitivity, selectivity, and efficiency of four on-site drug testing devices was evaluated: Triage Panel for Drugs of Abuse plus TCA, QuickScreen Pro-Multi Drug Screening Tests, Syva Rapid Test d.a.u. 5 and d.a.u. 2, and Rapid Drug Screen. Overall, the Triage device demonstrated superior sensitivity, selectivity, and linearity around the cutoff concentration. Triage was also the only on-site testing device in which the appearance of a colored line indicated a positive result, making the assessment of a positive drug test more intuitive. With the increased popularity of on-site drug testing, great care must be exercised in their application. Despite the advances in technology to produce more reliable results, on-site testing devices remain only a screening tool. Therefore, employers, officers, doc- tors, and counselors using these tools must regard the results of any immunoassay with caution. To be assured of analyte identification, immunoassay findings must be confirmed by GC-MS. Acknowledgments Triage Panel for Drugs of Abuse plus TCA devices were supplied by Biosite Diagnostics (San Diego, CA). References I. United States Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, National Household Survey On Drug Abuse Main Findings Rockville, MD, April A. Poklis, L.E. Edinboro, J.S. Lee, and C.R. Crooks. Evaluation of a colloidal metal immunoassay device for the detection of tricyclic antidepressants in urine. Clin. ToxicoL 35:77-82 (1997). 3. A. Poklis and C.L. O'Neal. Potential for false-positive results by the TRIAGE TM panel of drugs-of-abuse immunoassay. J. Anal. ToxicoL 20: (1996). 4. Triage Directions for Use. Biosite Diagnostics, Inc., San Diego, CA, Rapid Drug Screen TM Information and Test Procedures. American Bio Meclica Corp., Hudson, NY. 6. Rapid Test Rapid Test package insert. Behring Diagnostics, Cupertino, CA, QuickScreen TM Test Instructions. Phamatech, San Diego, CA, E.H. Taylor, E.H. Oertli, and J.W. Wolfgang. Accuracy of five on-site immunoassay drugs of abuse testing devices. J. Anal. Toxicol. 23: (1999). 9. K.F. Buechler, S. Moi, B. Noar, D. McGrath, J. Villela, M. Clancy, A. Shenhav, A. Co]leymore, G. Valkirs, T. Lee, J.F. Bruni, M. Walsh, R. Hoffman, F. Ahmuty, M. Nowakowski, J. Buechler, M. Mitchell, D. Boyd, N. Stiso, and R. Anderson. Simultaneous detection of seven drugs of abuse by the Triage panel of drugs of abuse. Clin Chem. 38: (1992). 10. A.H.B. Wu, S.S. Wong, K.G. Johnson, J. Callies, D.X. Shu, W.E. Dunn, and S.H.Y. Wong. Evaluation of the Triage system for emergency drugs-of-abuse testing in urine. J. Anal. Toxicol. 17: (1993). 11. A. Poklis and L.E. Edinboro. Discordant benzodiazepine results by immunoassay. J. Anal. Toxicol. 18: (1994). 12. L.E. Edinboro and A. Poklis. Detection of benzodiazepines and tribenzazolams by TRIAGE: confirmation by solid-phase extraction utilizing SPEC 3ML 3 MP3 microcolumns and GC-MS. J. Anal. ToxicoL 18: (1994). 13. R. de la Torre, A. Domingo-Salvany, R. Badia, G. Gonzalez, D. McFarlane, L. San, and M. Torrens. Clinical evaluation of the Triage analytical device for drugs-of-abuse testing. Clin. Chem. 42: (1996). Manuscript received March 1,2000; revision received June 2,