Field Test of On-Site
Drug Detection Devices
Final Report -- October 2000

III. Methods

In September 1997, ISA Associates, Inc. (ISA) and the University of Utah's Center for Human Toxicology (CHT) received a contract to conduct a field test of five on-site drug testing devices. The major tasks of the study were: 1) identify the pool of available on-site devices; 2) select five on-site drug testing devices for testing; 3) identify two study sites, each of which could provide a sample size of 400 arrested drivers; 4) train police officers and field technicians in specimen collection and device testing; 5) obtain subjective assessments of the on-site devices from the law enforcement officials and on-site research analyst; 6) conduct confirmation testing using GC/MS on all positive samples and 5% of the negative samples; and 7) submit reports to NHTSA throughout the project.

Identification of Devices

Two of the senior study personnel, Mr. Crouch and a project consultant, Dr. Michael Walsh, had performed research recently using on-site testing devices (Buchan, 1997; Crouch, 1997; 1998). This experience provided substantial knowledge of the scientific literature about the use and effectiveness of on-site devices as well as the manufacturer's literature on the devices. We supplemented this knowledge and our collective reprint files by performing a computer search of the scientific literature for references to on-site testing devices. These searches were performed both at CHT, through a computer interface with the library facilities at the University of Utah, and at ISA, where an extensive search was made on the Internet. The search was limited to the last ten years when these devices become commercially available and began to be evaluated scientifically. These computer searches were supplemented with abstracts from recent meetings of the American Academy of Forensic Sciences, the American Association of Clinical Chemistry, the Society of Forensic Toxicologists, and similar scientific meetings. We also performed extensive searches to identify on-site device manufacturers. Once manufacturers were identified, they were contacted and asked to provide information about their products and referrals to additional manufacturers.

Data about each of the devices was compiled from three major sources. We contacted the FDA to obtain a list of approved devices and contacted manufacturers to provide product information. The information requested included availability of devices, cost, volume reduction(s) in cost, testing procedures, drugs/metabolites tested, antibody target analytes, storage requirements, and any additional test materials that are needed. Manufacturers were also asked if their devices tested for the presence of a single drug or for multiple drugs simultaneously. The scientific literature and the experience of ISA staff, Mr. Crouch, and Dr. Walsh were used to assess the accuracy and reliability of the devices based on the following criteria: 1) manufacturer criteria; 2) procedural criteria; 3) practical considerations and experience with the product; 4) scientific credibility of the device; and 5) cost. These criteria are described below.

• Manufacturer Criteria. The manufacturer had to be a stable business entity, responsive, and able to provide the product expeditiously. Communications with the manufacturers were used to assess their willingness to cooperate and provide product information, their technical expertise, and their willingness to support users in the field.

• Procedural Criteria. The manufacturer's recommended procedures were reviewed for simplicity, analysis time, number of reagents, number of steps, stability of the test results, and applicability for use by non-technical analysts. Whenever possible, sample devices were obtained from the manufacturers and the tests were performed at CHT. Experience has shown that one difficulty encountered with on-site devices is determining whether a test result is positive or negative. The ambiguity of the test results was a major criterion. The selection criteria needed to accommodate the field evaluation requirement of this project. Here the experience of Dr. Walsh was invaluable since he had already performed a field assessment of four devices (Buchan, 1997). Assessments of the accuracy and reliability of the devices were based on previous laboratory assessments.

• Practical Considerations. Practical considerations included storage requirements, shelf life, special requirements for disposal, and the need for additional materials. The investigators relied primarily on their collective experience with the devices (Buchan, 1997; Crouch, 1997; 1998) to comment on the practical considerations of the devices.

• Scientific Credibility. Information gathered from the scientific literature review, manufacturers, and product use references, was used to assess the accuracy and reliability of the devices. Numerical ratings of specific elements within each of these criteria were conducted. These elements and ratings included:

The ratings provided a numerical score that was used to group the devices for possible selection. Devices were recommended that, at a minimum, tested for the drugs indicated in the RFP (referred to above as NIDA-5). These drugs included amphetamines (amphetamine and methamphetamine), cocaine/metabolites (BZE), opiates, marijuana metabolite (THC-COOH), and phencyclidine (PCP). With the exception of PCP, these drugs were among the most common drugs detected in previous drugs and driving studies (Table 1). Additional drug classes were omitted based on a joint decision between NHTSA, ISA, and CHT. Originally, approximately 30 devices were identified. Sixteen of those devices had sufficient information to allow the research team to evaluate and rate the device. The research team was unable to obtain sufficient information the manufacturers to evaluate the remaining devices. The sixteen devices that were evaluated and rated by the research team are listed in Table 2.

Table 2
Identified On-Site Devices

Selection of On-Site Devices

ISA and CHT recommended and NHTSA selected the following on-site devices for the evaluation:

AccuSign®, Princeton Biomedical Corporation, Princeton, NJ

OnTrak TesTcup-5®, Roche Diagnostic Systems, Inc, Somerville, NJ

OnTrak TesTstik®, Roche Diagnostic Systems, Inc, Somerville, NJ

Rapid Drug Screen®, American Bio Medica Corp, Ancramdale, NY

Triage®, Biosite Diagnostics, San Diego, CA

The cutoff values reported by the manufacturers for the on-site devices were consistent with the 1992/1993 federal recommendations and are shown below (DOT, 1992; DHHS, 1993).

The method of analysis for each of the devices is detailed in Appendix A: Research and Data Collection Guidelines. The following is a summary of the method used by each device.

The AccuSign® drug tests were performed by placing three drops of urine from the specimen into the sample well and allowing the reaction to proceed. After approximately two-to-five minutes, color appeared at the "Control Line" indicating that the analysis results could be read. Results were read within ten minutes.

The OnTrak TesTcup-5® is an integrated urine collection and drug-testing device. No urine or reagent handling was required after the urine was collected. To initiate an analysis, the lid of the cup was turned to the "TEST" position and the cup was tilted forward for approximately ten seconds then returned to its upright position. After about five-to-ten minutes, blue bands were visible in the "Test Valid" window and the drug test results could be read.

Each OnTrak TesTstik® is designed to detect only one class of illicit drug. This is in contrast to the other devices that tested for a battery of five drugs/metabolites simultaneously. To perform the TesTstik® analysis, the protective sample pad cover was retracted. The exposed sample pad of the TesTstik® was immersed in the urine for approximately ten seconds. The test was allowed to proceed until a distinct blue band formed in the "TEST VALID" window. The drug test result then was read. Five drug class specific TesTstiks® were needed for a complete drug screen.

To perform an analysis using Rapid Drug Screen®, the testing cup was first filled with urine. Then the Rapid Drug Screen® card was inserted through the safety seal tape covering a slit in the cap of the testing cup. The Rapid Drug Screen® was allowed to contact the urine and the urine to migrate up the testing card. Negative results were read after approximately three minutes and positive results were evident in eight-to-ten minutes.

To initiate a Triage® test for drugs of abuse, the operator used a pipette provided by the manufacturer to transfer a portion of urine sample (40) to a reaction cup on the device. The reaction mixture then was allowed to incubate for ten minutes at room temperature. Again using a pipette, the reaction mixture was transferred from the reaction cup to the detection area of the device. The mixture was allowed to soak completely though the absorbent detection area. Three drops of the wash solution then were added to the detection area and this was allowed to soak through the absorbent area. Results were read anytime within five minutes. This was the only device in which a color development indicated a positive test.

Site Selection

The study required the participation of two police departments with sufficient DUI arrests to obtain four hundred urine samples each during the data collection period (approximately eight months). In addition, the sites needed to have strong Drug Recognition Evaluation (DRE) programs. NHTSA staff identified ten states that had sufficiently mature DRE programs to meet the needs of the field test. These states were New York, Maryland, Minnesota, Wisconsin, Texas, Colorado, Arizona, California, Hawaii, and Oregon. With assistance from State DRE Coordinators and NHTSA Regional Offices, the following sites were identified:

Colorado: Denver

New York: Nassau County

Arizona: Phoenix, Mesa

Texas: Houston, San Antonio

Wisconsin: Milwaukee, Madison

Oregon: Multnomah County

ISA staff contacted each site and obtained detailed information about the following selection criteria: 1) characteristics of the law enforcement agency and jurisdiction (number of officers, number of DRE officers, number of traffic division officers, square mileage and population size of police district, etc.); 2) DRE program and procedures; 3) local DUI procedures; 4) availability of a central testing facility for DUI processing; 5) willingness of the department to allow officers or research staff to obtain a urine sample from DUI suspects; 6) legal barriers to obtaining consent to obtain a urine sample in DUI arrests; 7) types of drugs in the driving populations; and 8) willingness of the department to participate as a study site.

A number of the sites were eliminated early in the site selection process either because there was no central testing facility within the law enforcement jurisdiction or because the implied consent laws in DUI cases made it difficult to obtain urine samples. After careful consideration of all the selection criteria, Nassau County, NY and Houston, TX were selected as the field test sites.

Study Design

The study design, including training and site preparations, sample size, specimen handling procedures, data collection procedures, and confirmation procedures are described below.

Training and site preparations. ISA staff met with representatives from the Houston and Nassau County Police Departments as well as representatives from the local District Attorney's Office to discuss the study and data collection procedures. Informed consent documents were reviewed and approved by both police departments. The informed consent documents clearly indicated that the samples were to be collected anonymously and that the results of the tests could not be used against the participants in any legal proceeding.

In both Nassau County and Houston, an on-site research analyst was hired to lead the data collection. Dr. Hersch and Mr. Crouch supervised the data collection. The research analysts and the police central testing/traffic division personnel participated in a three-hour training session conducted by Mr. Crouch and Dr. Hersch. During the training session, videotapes, prepared by each of the manufacturers, were presented. Mr. Crouch then demonstrated how to perform each test with urine samples known to be drug-free and known to be cocaine positive. Finally, participants received hands-on experience with each of the devices. The manufacturers were invited by the research staff to visit each site and provide on-site training for the officers and research analysts. The research analysts received additional training on field-test procedures; specimen handling, labeling, and shipping; and record keeping.

Sample size. In each site, 400 urine samples were collected and tested. Sample sizes of 400 per site were proposed to allow for the collection of a sufficient number of specimens containing relatively low-prevalence drugs in the driving population; i.e., opiates, PCP and amphetamine/methamphetamine. If the study were designed to test the effectiveness of the on-site devices for marijuana and cocaine only, a smaller sample size would have been sufficient. Based on previous studies, we expected drug positive rates for DUI arrestees of approximately 10% to 15%, yielding 40 to 60 drug positive participants per site. However, previous studies indicated that users of opiates and amphetamines/methamphetamines occur in only 1% to 2% of the sample. It was believed that the sample size of 400 per site (or 800 total) would provide 8 to 16 specimens on which to test the reliability of the devices for these drugs. However, as will be discussed later, the drug positive rate for this study was much higher than anticipated and provided a larger sample on which to evaluate the devices.

Specimen handling procedures. CHT supplied urine collection kits. Each kit contained the following:

• 60 mL inert plastic collection bottles containing fluid tight, leak proof, self-sealing specimen bags.
• Each bag contained a liquid absorbing packet and specimen labels

ISA trained collectors at each site in the proper collection, handling, and shipment of urine specimens. All specimens were treated as biohazards during collection, handling, and shipment. All urine collection supplies such as latex gloves, indelible markers, manifests, labels, disposable laboratory coats, and safety glasses were provided to the data collection team as part of this contract (See Appendix A: Research and Data Collection Guidelines).

Law enforcement officials and analysts were given specific instructions to follow manufacturers' protocols explicitly and not to deviate from their procedures. The research team provided the research analysts a test result sheet to record the identification of the on-site device, sample identification, date and time of analysis, test results, his/her initials, and any additional comments. A brief rating form was also included for recording the police officers' subjective assessment of each device (see the dimensions listed under "data collection procedures" below).

CHT provided each site with shipping boxes; pre-printed shipping labels; and "bill recipient," overnight courier shipment forms. The research analysts shipped the specimens to CHT via overnight courier periodically. Upon receipt, CHT assigned the specimens a unique identification number reserved for specimens from this project. All specimens were stored frozen. For testing, CHT thawed the specimens and removed the liquid, usually 1 mL volume, for each confirmation test.

Data collection procedures. Data collection in Nassau County began in November 1998 and was completed in November 1999. In Houston, the data collection began October 1998 and was completed in July 1999. Data collection occurred on Friday and Saturday nights from 10:00 pm to 6:00 am. All individuals arrested for a suspected DUI offense and processed through the central testing facilities at each site were eligible to participate in the project. Written, informed consent was obtained from each participant. Participants were informed that their sample would be collected anonymously and used to assess the effectiveness of on-site drug testing devices. Research and police staff emphasized that the results of the tests would not be used against participants in any legal proceedings.

In Houston, seventy-seven percent (77%) of the 975 arrestees eligible to participate were approached to participate. The remaining 23% were unable to provide consent due to advanced inebriation or language barriers. Of the 751 arrestees asked to participate, 53% agreed and provided urine samples. The participation rate in Nassau County was higher. Of the arrestees asked to participate, approximately 79% agreed. The higher participation rate in Nassau County was due primarily to the fact that the Central Testing (police) staff approached a select sample of arrestees - those believed to be more likely to provide consent (typically those that complied with the DUI processing requirements and those that were not belligerent).

Each urine sample received a complete drug test with each of the five devices (so that officers had equal experience with each device). The sequence in which the devices were used was rotated. For example, Sample 1 was tested first with Triage®, then TesTcup®, AccuSign®, Rapid Drug Screen®, and TesTstik® in that order. Sample 2 was tested first with TesTcup®, then AccuSign®, Rapid Drug Screen®, TesTstik®, and finally Triage®. A police officer always conducted the testing with the first device in the sequence and the research analyst conducted the drug testing with the remaining four devices. After the officer completed the testing and recorded the results, s/he completed a rating form assessing the devices on five elements (ease of use, time needed to conduct the test, need for specimen handling, readability of results, usefulness of the test for routine DUI evaluation). Each element was rated on a five-point scale with "5" being the most positive response. If the urine sample tested positive for any of the five drugs on any of the five devices, the sample was sent to the CHT for GC/MS confirmation. In addition, as indicated earlier, 5% of the samples that tested negative on all five devices were also sent for GC/MS confirmation.

Confirmation testing procedures. CHT performed MS confirmations for all drugs presumptively identified as positive by any of the devices. In addition, 5% of the samples that tested negative (drug free) were randomly selected for confirmation testing. For each confirmation test, GC/MS confirmation cutoff sensitivities at lower concentrations than those of the federal standards (DOT, 1992; DHHS, 1993) were established. A summary of confirmation methods and testing limits used in this study can be found in Appendix B: Confirmation Testing Procedures.

Back -- Table of Contents -- Forward