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

I. Introduction


Numerous studies have reported the prevalence of drug use (including alcohol) by impaired drivers, injured drivers, and fatally injured auto drivers. A representative sample of these studies is shown in Table 1. Early studies of fatally injured driving populations, such as that reported by Garriott et al. (1977) provided an understanding of the prevalence of alcohol and other drugs in fatally injured drivers in the U.S. However, substance use in transportation safety is not a problem unique to the US, as demonstrated by Robinson (1979) in Ireland and Cimbura et al. (1982) in Canada. The early studies by Garriott and Robinson provide a historical prospective, but their findings have limited applicability today because comprehensive testing was not performed, drug use patterns have changed and the analytical methods available to these authors lacked the sensitivity to screen and reliably quantify cannabinoids in biological fluids.

The study by Reeves et al. (1979) provided some of the first insight into the extent of marijuana use by motor vehicle operators. Reeves et al., showed that 16% of a selected sample of arrested drivers had cannabinoids in their urine. These data were consistent with those reported by Cimbura who also tested for cannabinoids in blood and urine. The fatally injured driver study by Terhune et al. (1992) further documented the extent of marijuana use and also indicated that cocaine use by drivers might be of growing concern.

Willette and Walsh (1998) pointed out that the full impact of drugs on traffic safety was unknown in the early 1980s and unfortunately this remains true today. However, there are some data that have emerged over the last decade that provide insight regarding the overall extent of the problem. Williams et al. (1985) reported a "high risk" sample of 440 young male auto drivers killed in California traffic crashes. This study showed that 70% of blood specimens collected from these drivers contained alcohol, 37% contained cannabinoids, and 11% contained cocaine. A report by Mason and McBay (1984, not shown in Table 1) also addressed the question of cannabinoid use by drivers. This study of 600 driver fatalities in North Carolina demonstrated that over 79% had detectable blood alcohol concentrations and 7.8% showed evidence of cannabinoids.

Table 1

  1977 1979 1982 1982 1979 1985 1988 1989 1992 1993
  Fatals Impaired Fatals Injured Impaired Fatals Volunteers Volunteers Fatals Fatals
  Auto Auto Auto Auto Auto Auto Truck Truck Auto Truck
TOTAL CASES 127 425 401 506 1792 440 300 500 1882 168
DRUGS                 18%  
ETHANOL 61% 83% 57% 25%   70% 1% NA 52% 13%
DRUGS 18% 27% 26% 22%   >40%   9% 18%  
DRUGS OR ETHANOL 70% 86% 69% 38%   81% 29% 10%   33%
THC NA NA 9% URINE 9% 16% 37% 16% 5% 7% 13%
      3% BLOOD              
COCAINE/BZE <1% NA <1% 2%   11% 2% 1% 5% 8%
DIAZEPAM/NOR 10% 19% 3% 7%   4% 4%   2%  
OTHER BENZOS NA 2% <1% 2%         1%  
AMPHETAMINE   NA       <1% AMP+METH 2% <1% <1% 4%
PHENTERMINE   NA   & RELATED 1%       <1%   <1%
PPA   NA       3% COMBINED     <1%
EPHEDRINE   NA       2% 12% 2.2%**   7%**
PCP   NA       4%        
BARBITURATES 3% 4% <1% 3%   2% 1%   2%  
CODEINE <1% <1% 2% <1%         <1%  
ANTIHISTAMINES 2% <1% 2% NA         <1%  
METHAQUALONE 2% <1% <1% <1%            
PROPOXYPHENE 2% NA <1% <1%            
OPIATES           2%     <1%  
2 OR MORE DRGS 5% 1% 5% 11%   43%       14%

* Center for Human Toxicology Participated in Study
**Ephedrine and Pseudoephedrine

Soderstrum et al. (1988) found that, of 1,023 patients admitted to The Maryland (Baltimore) Shock and Trauma Unit, 34.7% had very recently used marijuana (i.e., greater than 3 ng/mL tetrahydrocannabinol in serum) and 33% had BACs greater that 100 mg/dL (0.10%). Lindenbaum et al., (1989) at the Albert Einstein Trauma Center (Philadelphia) found 54% of admissions tested positive for cocaine, 37% for cannabis, 35% for alcohol, 10% for barbiturates, and 7% for benzodiazepines.

The recent comprehensive study by Terhune showed that these drugs remain among the most commonly detected by fatally injured drivers (Terhune et al., 1992). In this study, samples were collected from 1,882 fatally injured drivers from seven states with each sample tested for more than 40 drugs.

Table 1 also shows data from three studies of drug use by truck drivers. The Lund et al., (1988) study of 300 paid volunteer drivers randomly selected at an interstate weigh station demonstrated that 29% were positive for drugs and less than 1% were positive for ethanol. The table also shows a British Columbia study (Campbell, 1989) where truck drivers were also randomly selected at weigh stations. In this study, 2% of the drivers admitted to recent alcohol use and 9.6% tested positive for drugs. Crouch and his associates tested fatally injured drivers of large trucks and found that 1/3 contained psychoactive drugs or alcohol (Crouch et al., 1993). In this study, alcohol and cannabinoids were detected in 13% of the drivers, cocaine in 8%, and sympathomimetic amines (e.g., amphetamines, methamphetamines, and similar over the counter preparations) in 11.3% of the drivers.

Although alcohol, cannabinoids, and cocaine remain the most prevalent drugs detected in drugs-and-driving studies, Table 1 shows that other drugs were also detected and, therefore, should not be discounted. Amphetamine, methamphetamine and related sympathomimetic amines were frequently detected in the more recent studies (Williams, 1985; Lund, 1988; Terhune, 1992; Crouch, 1993). Some drugs, such as barbiturates (southern California), and PCP show regional or local areas of prevalence. Benzodiazepines (e.g., valium, xanax, etc.) have not been frequently detected in recent studies; however, they are among the most prescribed class of drugs in the US, can impair driving ability, and were detected in approximately 5% of the drivers in the Terhune fatal injury study. The same study indicated that 2% of the drivers were taking barbiturates.

Comparatively little data have been published on the prevalence of drugs in drivers detained by the police for erratic driving. Compton and Anderson summarized the studies in this area in a NHTSA staff report (Compton & Anderson, 1985). These authors reported that the prevalence of drugs in arrested drivers with BAC concentrations below 0.10% was between 14% and 50%. The most frequently encountered drugs in order of prevalence were marijuana, tranquilizers and sedatives, hallucinogens (PCP), cocaine, amphetamines, and opiates. In the Virginia data reported by Compton and Anderson, blood from 788 drivers was tested and 16% of the samples contained one or more drugs. A six-year California study showed annual prevalence rates of 30% to 50% for drugs in arrested drivers. A second California study showed that between 14.4% and 23% of the blood samples collected from impaired drivers contained marijuana.

As advancements in drug testing technology have developed, these new technologies have been applied to testing samples collected from impaired, injured, and fatally injured drivers. Recently, an expedient urinalysis drug screening technology was introduced. These commercially available "on-site drug screening devices" are immunoassay based, require no sophisticated instrumentation, and do not require a permanent laboratory or extensively trained personnel. Several devices are currently described in the literature and are discussed below. These devices have been advocated for use in clinical settings, the criminal justice system, nuclear power generating plants, offshore oil drill platforms, commercial trucking, and highway safety. Several studies have been performed to assess the sensitivity and selectivity of on-site testing devices (Fitzgerald, 1994; Wu, 1994; Koch, 1994; Baker, 1991; Armbruester, 1992; Ferrara, 1994; Jenkins, 1995; Towt, 1995). Unfortunately, these studies have been performed primarily in laboratories using trained laboratory professionals and have compared a single on-site drug test device to laboratory based immunoassay tests such as Enzyme Mediated Immunoassay Technique (EIA), Radioimmunoassay (RIA), and Flouresences Polarization Immunoassay (FPIA) and/or with Gas Chromatography/Mass Spectrometry (GC/MS) results.

The devices have several potential uses in support of NHTSA's efforts to improve highway safety. For example, on-site test device results could be used to corroborate the field sobriety assessments of drivers detained and evaluated by Drug Recognition Technicians who are trained using NHTSA-approved Drug Evaluation and Classification (DEC) training procedures. Results could augment the DEC evidence presented in legal proceedings of drivers charged with driving under the influence. Further, on-site tests could be used at the roadside to assist officers in deciding whether to arrest suspected drug-impaired auto or truck drivers. To evaluate the devices' potential for use in traffic safety, NHTSA designed and funded a study to have three analytical laboratories evaluate three on-site urine drug-testing devices. Specifically, the study was designed to be the first multi-site evaluation comparing on-site test results to urinalysis drug test results using EIA immunoassay testing and with GC/MS confirmation. Each laboratory operated independently in sample selection, analysis, data interpretation, and reporting. That study showed consistent results between the laboratories (Crouch, 1997).

Brookoff et al. (1994) used on-site testing devices in a study that found a 58% prevalence rate for drugs in subjects arrested for reckless driving (who were not found to be impaired by alcohol). The Brookoff team found that 33% of their sample tested positive for marijuana, 13% for cocaine, or 12% for both. (Because of sampling flaws in the study, these drug test rates should not be interpreted as drug prevalence rates for reckless drivers.) Interestingly, the on-site device (Microline) used by Brookoff and his colleagues generated a significant false positive rate for marijuana when compared to GC/MS results.

In a recent study by Walsh, Buchan, and their associates that is very similar to the study reported here, four on-site drug screening devices were evaluated in a law enforcement setting (Walsh et al., 1997; Buchan et al., 1997). As in the study reported here, the results of the on-site devices were compared to laboratory GC/MS tests (n=305). Prevalence rates for cannabinoids, cocaine/metabolites, and opiates were 15.5%, 13.2% and 0.7%, respectively. The four on-site devices used in that study " Triage®, Abu-Sign®, OnTrak®, and TesTcup® " were rated on ease of handling, time to conduct the test, specimen handling, reagent mixing, and readability of results by the three university-based evaluators. The four on-site devices were also assessed on their sensitivity and specificity (compared to GC/MS) and their cost. Abu-Sign® and OnTrak® were clearly superior to the other two tests, although the cost of OnTrak® was a fraction of the cost of the other three. All four tests displayed high (96% or higher) levels of specificity (Buchan et al., 1997). Walsh and his associates concluded that routine use of these devices was feasible and that the devices could be integrated into police operations. However, it should be noted that in their study, although law-enforcement personnel collected the urine specimens, laboratory personnel conducted the on-site test – not officers. Therefore, although the Walsh study provides valuable information on the accuracy of on-site drug screening devices, it did not assess how well police-administered on-site tests performed.

Table of Contents -- Forward