EVIDENCE LIBRARY
Does the Most Alcohol-Sensitive Laboratory Test Best Predict Driving Impairment?
A 24-person alcohol-calibration study found particularly large effects on balance, but sensitivity to alcohol and prediction of driving performance are different scientific questions.
No. A test can respond strongly to alcohol without being established as the best predictor of driving impairment. In a controlled study of 24 volunteers, body sway showed larger standardized alcohol effects than several cognitive tests. The experiment did not measure actual driving, so it could compare alcohol sensitivity but could not establish which test best predicted a driver’s performance on the road.
That distinction matters when laboratory, clinical or research findings are offered in a DUI case. Detecting a drug effect, measuring a skill relevant to driving and predicting an individual’s driving are related tasks, but they are not interchangeable.
What the calibration study measured
Jongen and colleagues tested 12 men and 12 women, with a mean age of approximately 23 years, under placebo and three alcohol conditions. The randomized, double-blind design used four conditions across two test days, separated by a washout. Lower-dose sessions preceded higher-dose sessions on each day; the tests also followed a fixed order.
The nominal alcohol targets were 0.2, 0.5 and 0.8 g/L, equivalent to 0.02, 0.05 and 0.08 g/dL. The investigators monitored alcohol using a breath instrument, rather than laboratory blood analysis. Breath-derived values changed during testing, and additional drinks were used to restore the targets. These were controlled exposure conditions, not perfectly constant measured blood concentrations.
The battery included critical tracking, divided attention, psychomotor vigilance, digit-symbol substitution, concept shifting, digit span, an attention-network task and balance on a force platform. The tasks sampled different functions and lasted from a few minutes to approximately 25 minutes.
Balance was sensitive, but the comparison has a specific meaning
At the 0.5 g/L target, standardized balance effects were approximately 0.77 with eyes open and 0.62 with eyes closed. At 0.8 g/L, they were approximately 1.50 and 0.96. These were comparatively large changes within this experiment.
Other measures also responded. Divided-attention reaction time and psychomotor vigilance were sensitive at the middle and higher targets. For the inverse-reaction-time vigilance measure, reported effect sizes were approximately 0.29 and 0.70; divided-attention reaction-time effects were approximately 0.39 and 0.65.
A standardized effect describes the size of a group difference relative to variability. It does not state the percentage of driving ability lost, identify a universal impairment cutoff or give the probability that a particular driver would crash.
Why sensitivity does not establish predictive validity
The experiment asked which tests detected alcohol effects under its conditions. It did not compare each participant’s test change with that participant’s on-road driving change. A direct prediction claim would require that additional relationship to be studied, with appropriate validation in the population and setting where the test would be used.
A balance task can therefore be an efficient experimental indicator of alcohol exposure effects without automatically becoming the strongest measure of lane control, hazard response or divided attention in traffic. The site’s alcohol simulator study examining driving errors illustrates why driving endpoints themselves must be specified: different outcomes can produce different statistical findings.
Low-dose findings need particular care
The paper reported a divided-attention reaction-time difference at the lowest target with nominal significance below 0.05. Its methods also described a stricter Bonferroni threshold of 0.0167 for relevant comparisons. The reported lowest-dose result does not clearly satisfy that stricter threshold. It should not be presented as an unqualified, correction-resistant demonstration of a universal 0.02 g/dL impairment boundary.
Some other outcomes were nonmonotonic or showed no consistent alcohol effect. Critical tracking, for example, did not yield a simple progressively increasing effect across all conditions. Digit span and concept shifting did not show consistent overall effects. A null result on one measure does not establish preserved ability across the battery, as discussed in the six-domain alcohol study.
Applying the findings to DUI evidence
The strongest supported conclusion is methodological: alcohol-calibration studies can help choose sensitive experimental measures, but their effect-size rankings do not by themselves rank tests for individual driving prediction.
The sample was small and young, testing involved repeated tasks, and fixed task order and the placement of higher doses later in each day complicate interpretation. Actual breath-derived exposure varied over time. The study did not establish legal standards, diagnose an individual driver’s condition or validate a roadside decision rule.
A case-specific opinion should identify the actual task, the measured endpoint, the specimen or breath measurement, timing, and the evidence connecting that task to the driving question. The label “alcohol-sensitive” alone does not supply that connection.
Primary source
Jongen S, Vuurman E, Ramaekers JG, Vermeeren A. Alcohol calibration of tests measuring skills related to car driving. Psychopharmacology. 2014;231:2435–2447. https://doi.org/10.1007/s00213-013-3408-y.