Do Current Methamphetamine Users Drive Differently When Blood Levels Do Not Predict Their Simulator Performance? A 30-Driver Study

Answer: In a study of 30 adults, 15 current methamphetamine users drove a fixed-base simulator differently from 15 comparison drivers. The current-user group spent more of the course above the posted speed limit, showed more lateral lane-position variability, and accepted a shorter gap when turning across approaching traffic. Those three comparisons were statistically significant. Yet within the methamphetamine-user group, measured blood methamphetamine concentrations did not significantly correlate with lane-position variability, average speed, or distance traveled above the speed limit. The evidence therefore concerns group differences in simulated driving, not a validated blood-concentration threshold or proof that methamphetamine caused a particular motorist’s actions.

The findings are reported by David Bosanquet, Hamish G. MacDougall, Stephen J. Rogers, Graham A. Starmer, Rebecca McKetin, Alexander Blaszczynski, and Iain S. McGregor in the original primary investigation, Driving on ice: impaired driving skills in current methamphetamine users, published in Psychopharmacology (2013), 225:161–172. The original peer-reviewed study (DOI: 10.1007/s00213-012-2805-y) is a naturalistic comparison; it is not a randomized drug-administration experiment.

What is the central forensic answer?

Three questions must be kept separate. First, did people selected because of current methamphetamine-use history differ from matched nonusers on the simulator? For several outcomes, yes. Second, did the investigators chemically measure methamphetamine in a specimen obtained from those participants? Yes, but blood was obtained from 14 of the 15 current users, not all 30 participants, and the report leaves important fraction-processing details unresolved. Third, did a larger contemporaneous methamphetamine result statistically explain the different driving measures within users? No statistically significant association was established for the three driving measures reported in that analysis.

A positive group comparison is neither a demonstration that acute methamphetamine caused the group’s behavior nor proof that the same behaviors would occur at a particular drug concentration. Conversely, nonsignificant correlations in 14 specimens do not demonstrate the absence of acute drug effects. The small number of participants, restricted concentration distribution, differences in habitual use, and measured and unmeasured covariates limit both inferences.

Who were the independent participants?

Researchers initially screened 216 prospective recruits. Thirty-nine entered simulator testing and nine were excluded: five because of simulator-related illness, two because of inadequate knowledge of Australian road rules, and two because detailed interviews revealed substantial other illicit-drug histories that were incompatible with enrollment criteria. The analyzed sample contained 30 independent human beings: 15 current methamphetamine users and 15 comparison drivers. Repeated segments of a driving course, multiple vehicle encounters, questionnaire items, and several drug measurements do not increase the biological sample size beyond these 30 adults.

The methamphetamine group had to report using the drug at least weekly in the preceding three months. The comparison group lacked a history of chronic or binge drug use. Both groups had to be adults, drive at least weekly, have at least three years of driving experience, and satisfy cognitive and language eligibility requirements. Mean age was 36.53 years among users and 37.27 years among controls. Years of driving were similar, 18.73 versus 18.67, as were reported driving days per week, 5.80 versus 5.87. The groups were not randomly assigned to become methamphetamine users or nonusers.

Did investigators administer methamphetamine or control the last dose?

No experimental dose was given. Participants were recruited while following their own patterns of use. Six users reported their last methamphetamine use within six hours, six reported six to 24 hours, and three reported more than 24 hours before testing. These categories were self-reported histories, not a validated chemical clock or an investigator-administered dose schedule. The researchers did not standardize route, dose, product purity, isomer composition, tolerance, sleep after a binge, or the precise interval between dosing and simulator testing.

The report discusses methamphetamine and amphetamine as distinct detected analytes. It does not show a chiral laboratory method capable of separately quantifying d-methamphetamine and l-methamphetamine. A person who reports methamphetamine use is not thereby proved to have received a controlled laboratory dose of a particular stereoisomer. It would also be incorrect to replace the methamphetamine exposure here with a trial of oral dextroamphetamine, mixed amphetamine salts, or lisdexamfetamine.

What did the simulator actually reproduce?

The investigators used a fixed-base STISIM M400 simulator built around a vehicle body with steering wheel, pedals, horn, dashboard, and three projection screens spanning approximately 135 degrees. Participants first practiced handling the simulator for roughly ten minutes. Each then completed one approximately 9.5-kilometer route through simulated urban, rural, and semirural settings with varying speed-limit signs, intersections, pedestrians, slower cars, oncoming traffic, and a speed camera. It was not an actual public-road drive or a collision-reconstruction experiment.

The instruction combined two potentially competing incentives: obey road rules while reaching a friend quickly, because that friend would leave if the participant was late. That structure matters. It permitted voluntary speed choice, instead of fixing everyone at a constant study speed. Some previous driving simulations emphasize lane control at prescribed speed; here, the research design explicitly allowed risk-taking through discretionary speeding and passing behavior.

How was speeding quantified?

The most direct speeding endpoint was the percentage of the course distance driven above the applicable posted speed limit, called OSPD. The methamphetamine group averaged 16.95% (standard deviation 9.08), whereas the comparison group averaged 9.53% (standard deviation 5.65). The source reported p = .012. This amounts to a descriptive 7.42-percentage-point greater fraction of route length above the limit in the user group. It is not a statement that all users sped, nor that any participant exceeded a specific speed limit by a particular number of kilometers per hour for that fraction of the route.

Mean speed was also higher in the user group, 66.48 versus 62.58 km/h, but that between-group comparison was not conventionally significant in the original results (p = .052). The difference between a significant fraction of distance exceeding posted limits and a nonsignificant average speed illustrates why endpoint definitions matter. A driver might exceed a limit more often on limited sections without having a reliably higher mean speed over the entire route.

What did the lane-weaving measurement show?

The standard deviation of lateral position (SDLP) quantified variation in simulated within-lane position. Mean SDLP was 1.66 meters (SD 0.29) for current users and 1.45 meters (SD 0.24) for controls, with a reported p = .024. Greater SDLP in this setting indicates less stable lateral lane positioning on the simulated course. It does not by itself identify why that variability occurred or equate to a legally established level of individual driving impairment.

The authors noted that absolute SDLP values were comparatively small and attributed that possibility to the relatively short, complex, and effortful route. Simulator geometry, roadway width, visual projection, and instructed speed choice also influence how lateral variation should be compared across studies. Substituting a number derived from one platform into a different vehicle, road, or jurisdiction without validation would be unsound.

What happened when drivers turned across oncoming traffic?

A deliberately structured right-hand-turn decision created a gap-acceptance measure: how much distance separated the participant’s car from an oncoming vehicle when making the turn. The user group averaged 92.48 meters (SD 25.73) compared with 108.55 meters (SD 12.78) among controls. The published between-group result was p = .039. The descriptive gap difference was 16.07 meters shorter in the user group.

This outcome captures an operational driving choice under the particular simulator traffic sequence. It is not a direct count of actual collisions, it is not a calibrated time-to-collision threshold, and it is not proof that an individual user accepted an unsafe road gap. Interpretation also depends on the oncoming vehicle’s programmed speed and context. The paper does not supply an individualized legal threshold for gap acceptance.

Original evidence table: significant and nonsignificant driving outcomes

These numbers belong to a study-sized sample, not to an individual’s evidentiary record. The original table displays a statistic column labelled “t or χ²,” although several paired values resemble an analysis-of-variance F statistic; the exact label and the underlying calculation should be verified against the original analytical records before using those statistic labels in sworn testimony. The article retains the published group means and p-values without silently correcting the source.

Which observed outcomes did not differ significantly?

Several measures that might sound like direct expressions of unsafe driving failed to show significant group differences: the number of simulated crashes, the number of centerline crossings, and the number of individual speed exceedance events. The mean-speed comparison came close to but did not satisfy the conventional 0.05 criterion. Thus the original result is not “every measure of driving got worse.” It is a selective pattern across chosen measures.

The nulls should not be rephrased as proof that crashes, lane crossings, or average speed were truly identical across populations. A study with only 15 people per group has limited power to distinguish small or heterogeneous effects. Counts in one designed scenario also may not translate to real annual crash rates. Both negative and positive results need to remain attached to the study’s particular sample and simulator.

How were biological samples collected?

At the laboratory, an oral-fluid sample was tested with an Oratech III qualitative lateral-flow device before driving, and expired breath was assessed with an Alcolizer instrument. The oral-fluid screen had separate nominal targets and reporting cutoffs; notably, the device’s methamphetamine/MDMA channel did not distinguish those two compounds. Such a screening result should not be described as an independently confirmed exact methamphetamine identity or quantitative concentration.

After the simulator, 10 mL of blood was obtained by venipuncture into a red, silicon-coated plastic BD tube containing a clot activator, and stored at −80 °C. ELISA immunoassays screened for amphetamine-class compounds, benzodiazepines, opiates, cannabinoids, and cocaine. Positive blood screening results received gas chromatography–mass spectrometry (GC/MS) quantitative follow-up. One user could not provide blood because of damaged veins, so 14 of 15 current users supplied blood specimens, even though all 15 contributed driving data.

Matrix caution: the paper calls the later quantitative results “blood drug levels,” but the red-top clot-activator tube description does not independently document whether the final instrument aliquot was unseparated whole blood, serum obtained after clotting, or another prepared fraction. No centrifugation or aliquot-preparation validation detail in the primary report licenses a categorical courtroom claim that every reported value is a validated legal venous whole-blood concentration. Conversely, one must not simply relabel them “serum” without documentation. Collection container, processing, analytical extraction, and reporting matrix must be confirmed separately.

What did GC/MS identify in the users’ blood specimens?

Among the 14 users with a blood sample, 11 tested positive for methamphetamine and ten for amphetamine; the paper also reports seven participants positive for methamphetamine or amphetamine only. Source Table 5 gives methamphetamine mean 0.09 mg/L, median 0.06 mg/L, and range from below 0.03 to 0.21 mg/L. Amphetamine mean was 0.028 mg/L, median 0.03 mg/L, and range below 0.003 to 0.04 mg/L. These are specimen assay findings in a selected cohort, not controlled dose-response values and not individual legal thresholds.

Source Table 5’s percentage column uses the enrolled-user denominator of 15 for some drug counts, even though only 14 had blood collected; 11 of 15 is 73.3% whereas 11 of 14 would be a different proportion. A forensic summary should state the raw denominators explicitly rather than presenting the source’s displayed percentages as blood-collection-specific prevalence without qualification.

Other compounds appeared as well. Three participants had detected THC, five had THC-COOH, and three had MDMA; MDA and benzodiazepine findings also occurred. A screen’s positivity, a mass-spectrometric drug identity, a measured drug concentration, and an inference about the behavioral contribution of that drug are separate analytical and inferential steps. The source does not establish which substance, if any, caused the behavior of each participant.

Did contemporaneous methamphetamine concentrations predict simulator outcomes?

The investigators calculated correlations in the methamphetamine-user group with available blood data. None of the three reported correlations was statistically significant: methamphetamine concentration and SDLP, r = 0.154; concentration and mean speed, r = 0.034; and concentration and the percentage of course distance spent above the speed limit, r = −0.079. The paper reported p > .05 in each case. These values are near zero in this dataset, but they do not create an independently validated equivalence statement or prove the absence of any dose-related effect across concentrations, routes, or users.

The correlation analysis had only 14 available specimens and occurred within one selected user cohort. It could not use the 15 controls to estimate a methamphetamine concentration–response function because controls did not have comparably positive drug results. The timing of sampling, heterogeneous time since reported use, undisclosed dose, possible other drugs, and chronic-exposure characteristics are all potential determinants of scatter. In particular, a user-group contrast and a within-user biochemical correlation test ask different statistical questions.

Original evidence table: chemical results and drug–driving nulls

Neither the methamphetamine GC/MS results nor these three correlations validate a universal impairment concentration. A numerical cutoff in an oral-fluid screening device is also not a biological threshold of unsafe driving. Only a new study with appropriate calibration, specified matrix, controlled or carefully modeled exposure, and independently validated behavioral outcomes could address those questions.

Why might use history and blood level tell different stories?

A self-reported history of recurring drug consumption may reflect many exposures and associated characteristics, not merely the concentration present during the testing hour. The researchers observed greater scores on a Barratt impulsiveness scale in current users: 75.4 versus 55.73, with p < .001. Five users, versus none of the controls, met the MINI interview criteria for antisocial personality disorder; eight users, versus one control, met criteria for alcohol abuse. Court-recorded prior traffic conviction was reported by 11 users compared with four controls. These differences indicate potential confounding, not a proven causal explanation for the simulator outcomes.

The current-user group also reported substantially more lifetime and recent use of multiple substances. Other drugs appeared in some blood assays, and a few controls acknowledged remote methamphetamine exposure despite having no use in the preceding 12 months. These observations caution against treating the cohort label as a pure pharmacological contrast between the effects of methamphetamine and complete lifelong nonexposure. Nor does the study’s failure to find a difference in subjective Karolinska sleepiness scores or brief MMSE scores rule out more subtle fatigue, attention, or executive-control differences.

What does observational design mean for a causation opinion?

The central causal limitation is assignment: no one was randomly allocated to start or stop using methamphetamine. Cohort membership correlated with other features including prior drug history, personality measures, alcohol-use problems, and legal driving history. Therefore, the observed group difference could reflect the measured drug, withdrawal, chronic-exposure associations, other drugs, sleep variation, stable dispositions, or several factors together. The original investigators expressly acknowledged that their small design could not determine precise causes of riskier simulated driving.

While actual GC/MS findings are analytically meaningful, they occurred after the simulator and do not turn this into a standardized timed pharmacokinetic study. One cannot infer from a population mean or blood median precisely when a person consumed methamphetamine or what blood concentration existed at a different moment. A retrospective time-to-last-use reconstruction would require validated kinetics, a precisely identified specimen fraction, sampling times, verified administration information, individual covariates, and a justified uncertainty model.

Does a significant simulator result establish actual unsafe driving?

No. The simulated course has practical components—traffic controls, overtaking, turning, speeding, and lane keeping—but a fixed-base driving simulation does not reproduce every characteristic of real traffic, including physical acceleration forces, real collision consequences, navigation burden, or the deterrent effects of actual law enforcement. A laboratory p-value represents a group comparison under the test design. Whether a person broke a traffic rule, endangered others, or caused a specific crash requires independent event evidence.

The contrast between a statistically significant OSPD difference and a nonsignificant count of separate speed exceedance events is also instructive. Those endpoints emphasize duration or route extent versus event frequency and are not interchangeable. Likewise, three versus one simulator crashes is not an observed annual real-world crash-risk ratio. A defensible expert report should specify the measured endpoint without converting it into a qualitatively different conclusion.

How should an expert review specimen and assay integrity?

Start with the original collection documentation: time relative to simulator exposure, collector identity, chain of custody, tube additive, storage interval, freezing, thawing, serum separation if any, and extraction of a specimen fraction. Check the GC/MS method validation for actual matrix, methamphetamine and amphetamine identification criteria, calibrators, quality-control acceptance, detection limits, quantitation limits, and interference rules. Distinguish source-described values from validated legal evidentiary concentrations that may use a different tube or fraction.

Separate oral-fluid preliminary screening from laboratory confirmation. The authors note that their initial test cannot differentiate methamphetamine from MDMA on one assay channel. A positive initial band does not resolve isomer, confirm pharmacologically active stereochemistry, or quantify a drug in blood. For a foundational overview of specimen-specific interpretation, the published biological matrices guide explains why blood, plasma, serum, oral fluid, urine, and hair must not be treated as equivalent. The published confirmatory drug testing guide also describes why a result remains tied to its validated method and reporting purpose.

What can this study contribute to an actual forensic matter?

It is useful evidence that one small naturalistic cohort of current methamphetamine users exhibited selected simulator differences and that contemporaneous source-described blood methamphetamine values did not significantly track three of those performance endpoints. It is also useful as a caution against a simple “more nanograms equals worse driving” narrative. Its greatest strength is simultaneous collection of behavioral, drug-use-history, and specimen information; its greatest limitations include cohort selection, small biological n, uncontrolled dosing, possible drug mixtures, matrix-processing uncertainty, and simulator generalizability.

A case-level evaluation must still examine actual observed driving, contemporaneous signs, medical and medication history, specimen matrix and timed collection, instrument validation and uncertainty, and competing explanations. The original study does not make a drug-positive person automatically unsafe, and it does not make a person below some invented methamphetamine number automatically safe. Those categorical statements exceed the original empirical record.

Frequently asked scientific questions

1. Did the researchers administer a measured dose of methamphetamine before the simulated drive?

No. The groups were recruited by use history, and current users continued their ordinary patterns. No assigned route, dose, product purity, stereoisomer, or dosing time was experimentally controlled. The article measures current users, not a blinded crossover challenge with a known dose.

2. How many actual human participants had quantitative blood toxicology?

Thirty drivers supplied simulator results: 15 current methamphetamine users and 15 controls. Blood collection succeeded for 14 current users; one could not provide a specimen. Eleven of those 14 were methamphetamine-positive by the blood-testing workflow. The 9.5-kilometer route and multiple simulator events did not produce additional independent people.

3. Did users show more speeding and lane weaving?

Yes, for the specific endpoints. The group comparison for percentage of distance above posted limits had p = .012 and lateral-position variability had p = .024. Mean speed p = .052 and centerline crossings were not significant, so the broader statement that every aspect of simulated driving deteriorated would be inaccurate.

4. Did a higher methamphetamine concentration predict worse simulated driving within the users?

Not significantly for the three reported correlations: r = .154 for SDLP, r = .034 for average speed, and r = −.079 for distance above the limit, all reported p > .05. These nonsignificant results in 14 specimens do not establish a universal absence of pharmacological driving effects.

5. Were those methamphetamine results definitely confirmed legal whole-blood concentrations?

The paper describes postdrive venipuncture into a red clot-activator tube and later GC/MS findings it calls “blood.” It does not sufficiently establish in the published Methods whether the assayed fraction was unseparated whole blood, serum, or another prepared matrix. Do not relabel the result without laboratory records specifying tube processing and method validation.

6. Can the oral-fluid screening cutoff identify methamphetamine isomer or prove impairment?

No. The on-site assay is qualitative, one channel can respond to methamphetamine and MDMA, and the paper does not validate d-/l-methamphetamine differentiation through a chiral assay. A screen reporting cutoff describes detection conditions, not a clinical or statutory impairment threshold or a reliable last-use clock.

7. Could this study alone prove methamphetamine caused an individual’s collision?

No. The experiment included no real on-road crash, no randomized methamphetamine exposure, no individual reconstruction, and no validated blood-level rule for case causation. Its conclusions address group simulator behavior and limitations of concentration correlations. A particular incident requires evidence concerning the specific driver and events.

Bottom line: separate the driver, the simulation, and the specimen

Bosanquet and colleagues observed significant differences in simulated speeding by route distance, lateral position variation, and turn-gap acceptance among 15 current methamphetamine users versus 15 comparison drivers. They also reported genuine nulls for several other driving endpoints and for within-user correlations between measured methamphetamine values and three simulator measures. None of those findings can be replaced by a universal impairment cutoff, exact timing estimate, or a legal conclusion about a particular crash. Read the complete original primary investigation (DOI: 10.1007/s00213-012-2805-y) together with the tested matrix and the event-specific record before advancing any case-level opinion.