Fiorentino, D., Evans,S. W., & Page T. E. (2020) The usefulness of SFSTs in detecting drugs other than alcohol. (Final report). Van Nuys, CA: DF Consulting
The objective of the study was to determine whether the three standardized field sobriety tests (SFSTs, horizontal gaze nystagmus or HGN, walk and turn or WAT, and one leg stand or OLS), alone and in combination, can detect impairment caused by drugs other than alcohol. Data were collected from arrestees selected at random while they awaited processing at a jail.
The data collection site was the Genesee County Jail (also known as “Flint City Lockup”) in Flint, Michigan. Data collection started on 10/30/2018 and ended on 5/6/2019. Data collection days and hours depended on the availability of the site coordinator and the availability of the testers. In general, however, most data were obtained between 8:00 pm and 5:00 am.
Law enforcement officers trained in the administration and scoring of the SFSTs collected data for the study. Only officers trained in enforcement of driving under the influence (DUI) using the SFSTs were recruited for the study.
The three sobriety tests (SFSTs) were administered and scored in accordance with the National Highway Traffic Safety Administration (NHTSA) specifications. Only HGN, WAT, and OLS were administered.
A preliminary breath testing instrument was used to measure the participants’ breath alcohol concentration (BrAC). Calibration was verified with a dry gas simulator at least once a month.
A one-step, multi-drug urine drug screen was used to detect one or more of the following concentrations of drugs (expressed in nanograms of drug per milliliter of urine):
· Amphetamine (AMP), 300 ng/ml
· Barbiturates (BAR), 300 ng/ml
· Benzodiazepines (BZO), 300 ng/ml
· Buprenorphine (BUP), 10 ng/ml
· Cocaine (COC), 300 ng/ml
· Ecstasy (MDMA), 500 ng/ml
· Marijuana (THC), 50 ng/ml
· Methadone (MTD), 300 ng/ml
· Methamphetamine (mAMP), 1000 ng/ml
· Morphine (MOP), 300 ng/ml
· Oxycodone, (OXY) 100 ng/ml
· Phencyclidine (PCP), 25 ng/ml.
The array of drugs screened in urine also allowed examination of five of the seven Drug Recognition Expert (DRE) categories. They were central nervous system (CNS) depressants (alcohol, barbiturates, and benzodiazepines), CNS stimulants (cocaine, amphetamine, and methamphetamine), hallucinogens (MDMA), narcotic analgesics (morphine, methadone, buprenorphine, and oxycodone), and cannabis (marijuana or THC). Inhalants were not included in the array and no dissociative anesthetics (PCP) were detected.
In the event that a urine screen was positive for THC, a saliva screen test was also administered. The primary reason was to detect the active metabolite of THC, but four more drugs were tested as well. After data collection, it was determined that the saliva test was not accurate, and therefore saliva test data were not included in the analyses.
A total of 624 arrestees were approached and asked to participate in the study. Of those, 527 agreed to participate. Urine drug tests results were available from 524 participants, HGN results from 527 participants, WAT results from 524 participants, and OLS from 525 participants.
Four hundred and fifty-nine subjects (87.1%) had zero BrAC (expressed in grams of absolute alcohol per 210 liters of air, g/210 L), and 68 subjects (12.9%) had any positive BrAC. Of those, 31 subjects (5.8%) had BrACs between .001 and .079, and 37 subjects (7.0%) had BrACs of .08 and above.
The number of positive urine drug tests was very high. Three subjects tested positive for seven drugs, four subjects tested positive for six drugs, 11 subjects tested positive for five drugs, 25 subjects tested positive for four drugs, 37 subjects tested positive for three drugs, 131 subjects tested positive for two drugs, 219 subjects tested positive for one drug, and 94 subjects tested negative for all drugs in urine. Of those 94, 10 produced a BrAC .08 and above.
In general, correlations between variables were small to medium size. A large correlation (.63) was observed between HGN and CNS depressants. In the behavioral sciences, correlations of .10 are considered small, correlations of .30 are considered medium, and correlations of .50 are considered large (Cohen, 1992).
Each test was examined alone and in combination with the other standardized field sobriety tests. Note that officers in the field make decisions based on the totality of the circumstances, not on the results of a single test.
The negative/positive criterion for HGN, WAT, and OLS were four, two, and two clues, respectively. Receiver operating characteristic (ROC, see Appendix F for description) analyses were conducted to determine the optimal cut-point value for the combined three tests (positive values ranging from zero to three). The optimal cut-point, criterion, was determined to be two positive tests.
Each test and combination of tests was examined in its ability to detect impairment from individual drugs, or DRE categories of drugs, or any one or more drugs.
HGN alone had an overall accuracy above chance for all drugs and categories of drugs except marijuana (THC) and any combination of one or more drugs. Note that drugs that are not expected to exhibit HGN, like cocaine, may be found in cases where other drugs causing HGN are present.
WAT alone had an overall accuracy above chance for marijuana, CNS stimulants, and any one or more drugs.
OLS alone had an overall accuracy above chance for cocaine, marijuana, CNS depressants, CNS stimulants, and any one or more drugs.
SFSTs with two as criterion allowed correct classification (no drug vs. any one or more drugs) in 2/3 of the participants. In case of two or more positive SFSTs, there was a .94 probability that the participant had one or more drugs in their system. In case of fewer than two positive SFSTs, there was a .30 probability that the participant did not have one or more drugs in their system. The false positive rate (the proportion of positive SFSTs in participants with negative drug tests) was .23 and the false negative rate (the proportion of negative SFSTs in participants with positive drug tests) was .36. Thus, in cases when the SFSTs wrongly classified participants, false negatives were more common than false positives, an important characteristic at roadside, where an incorrect arrest is considered more harmful than an incorrect release.
Although the study had limitations, including a very high prevalence of drugs, and especially a faulty saliva test, it did provide evidence that the SFSTs can be used to detect impairment caused by drugs other than alcohol.