EVIDENCE LIBRARY
Can a Saliva Alcohol Test Establish BAC? Detection Versus Quantification
In 89 nightclub patrons, saliva and breath agreed on alcohol detection but showed poor quantitative agreement. The study did not measure blood alcohol.
A saliva alcohol result should not automatically be treated as a blood alcohol concentration. In a study of 89 Norwegian nightclub patrons, oral-fluid and breath tests agreed on whether alcohol was detected, but their numerical concentrations showed poor agreement. Importantly, the researchers did not collect blood for a direct BAC comparison.
The study addresses a practical DUI interpretation problem: a method may identify alcohol presence under particular sampling conditions without accurately estimating how much alcohol is in another specimen.
What was actually collected and measured?
Gjerde and colleagues recruited a convenience sample of 95 nightclub patrons. Six oral-fluid samples were excluded because too little fluid had been collected, leaving 89 paired comparisons.
Before testing, participants drank a glass of water to rinse their mouths. Oral fluid was collected with a Quantisal pad, transferred into preservative buffer, stored and later analyzed using an enzymatic alcohol method. The investigators weighed samples and calculated the concentration in the original, undiluted oral fluid rather than reporting only the concentration of the diluted buffer mixture.
Breath testing used Lion Alcolmeter 500 instruments with fuel-cell sensors. The instruments displayed an estimated BAC in g/L using a programmed 2,000:1 blood-to-breath ratio. For their analysis, the researchers converted those displays back to breath concentrations in mg/L. A displayed BAC estimate is not a collected and analyzed blood specimen.
Agreement on detection was not agreement on concentration
The investigators classified oral fluid above 0.10 g/L and breath above 0.05 mg/L as positive. These were study cutoffs for different matrices and units; they were not established as equivalent legal thresholds.
Eighty-eight participants were positive in both specimens, and one was negative in both. Thus, no discordant positive-versus-negative results occurred in this sample.
That observation is narrower than a claim of perfect diagnostic performance in all drivers. With only one breath-negative participant, the study provides very little information about performance in a population containing many alcohol-negative people. Breath testing was also the comparator, not an independent blood measurement.
For the numerical comparison, the intraclass correlation coefficient was 0.40, with a 95% confidence interval of 0.21–0.56. The authors classified this as poor agreement. Many oral-fluid concentrations were lower than expected from breath, while others were higher. Perfect agreement on these positive-versus-negative classifications therefore coexisted with poor quantitative agreement.
Recent drinking and mouth rinsing matter
The study deliberately examined a difficult sampling situation involving recent beverage consumption. Drinking water can change the alcohol concentration in the mouth, and the resulting oral-fluid sample need not immediately reflect an equilibrated concentration.
The researchers discussed departures from the manufacturers’ waiting instructions. They also did not record the time and type of each participant’s last food or drink. Their results cannot determine the exact waiting period that would have corrected each individual’s sample, or demonstrate that a glass of water guarantees valid quantitative saliva or breath testing.
The appropriate inference is tied to the tested collection procedure. This study does not establish that every saliva alcohol method performs poorly under every controlled sampling protocol.
Other limitations affect a numerical BAC inference
The sample was small and not randomly selected. Participants were nightclub visitors, rather than a representative sample of drivers. Uncertainty in collected fluid volume, buffer volume, weighing, the alcohol assay and storage could also contribute to discrepancies.
The correction for buffer dilution was necessary, but it did not eliminate poor agreement with breath. A reported saliva concentration should therefore retain its specimen identity even when a laboratory has accurately measured the alcohol present in that sample.
What the findings mean in a DUI context
A case review should identify the actual specimen, collection method, timing, reported units, analytical cutoff and any conversion applied. Oral-fluid alcohol in g/L, breath alcohol in mg/L and an estimated blood-equivalent display are distinct results.
This experiment did not test driving, coordination or cognition. Its detection results do not establish impairment, and its concentration comparison cannot determine a particular driver’s BAC at an earlier time.
For a related matrix problem, see the site’s 100-volunteer study of blood–breath ratios. The three-case analysis of breath slope detectors addresses a separate question about instrument interpretation; it does not supply an error rate for this saliva collection method.
Primary source
Gjerde H, Bretteville-Jensen AL, Furuhaugen H. Poor correlation between alcohol concentration in oral fluid and breath in subjects consuming beverages immediately before testing. Biochemia Medica. 2022;32(1):010902. https://doi.org/10.11613/BM.2022.010902.