EVIDENCE LIBRARY
Can Urine Sorbitol or an NMR Fingerprint Time Alcohol Use? A Crossover Study’s Limits
A 20-person crossover study found urine metabolite changes after alcohol. Learn why sorbitol and NMR fingerprints cannot prove BAC or drinking time.
Quick answer: An experimental urine nuclear magnetic resonance (NMR) profile can show time-dependent metabolic changes after alcohol ingestion, but urinary sorbitol, hypoxanthine, or a multimetabolite pattern cannot reliably establish when a particular person drank, what their blood alcohol concentration was, or whether they were impaired. In a controlled crossover study, Irwin and colleagues identified increased urinary sorbitol and other metabolites after an alcohol intervention. The researchers did not validate a forensic test, an individual drinking-time equation, or an impairment threshold.
The original primary research is Irwin C, van Reenen M, Mason S, et al., The 1H-NMR-based metabolite profile of acute alcohol consumption: A metabolomics intervention study, PLOS ONE (2018), doi:10.1371/journal.pone.0196850. The observations concern urine metabolites in a defined volunteer experiment, not a validated replacement for blood, breath, or case-specific forensic interpretation.
What was the study actually designed to answer?
The researchers investigated which urinary metabolic changes follow a single alcohol intervention and whether taking a nicotinamide adenine dinucleotide (NAD) supplement altered selected responses. This was a metabolic-mechanism and analytic-method investigation. Its key question was not whether a positive urinary sorbitol finding could convict or exonerate a driver.
The intervention enrolled 24 healthy male social drinkers between 20 and 24 years old. Three did not complete all intervention conditions, and one additional participant’s data were excluded following an identified outlier, leaving 20 participants for the principal analysis. A crossover design allowed comparisons within participants, but the final sample was small, young, exclusively male, and experimentally controlled. It cannot directly describe all patients, populations, or real-world drinking scenarios.
Over seven Saturdays, the protocol included four conditions: flavored-water vehicle alone, vodka with the vehicle, a 50-mg NAD-containing supplement with the vehicle, and the supplement plus vodka and vehicle. The alcohol intervention was reported as 1.5 grams of 43%-alcohol vodka per kilogram of body mass; critically, that wording denotes a quantity of the beverage, not a stated dose of 1.5 grams of pure ethanol per kilogram. The supplement, when applicable, was taken one hour before the intervention.
Participants arrived fasting and supplied urine at one hour before intake (time −1), immediately before intake (time 0), and at one, two, three and four hours after intake. Thus each complete condition involved six scheduled specimens per participant. The beverage vehicle contained sodium benzoate, relevant because its metabolism can contribute to urinary hippuric acid. The researchers considered that background rather than treating every NMR change as ethanol-specific.
What did the laboratory measure: urine alcohol, a biomarker, or blood BAC?
These are not interchangeable measurements. Proton NMR spectroscopy detected resonance patterns from compounds present in urine, including ethanol itself and several endogenous metabolites. The study used a spectral preprocessing strategy, normalizing data against a creatinine resonance and retaining 289 spectral bins after quality-control filtering. A bin is an analytical spectral feature: it is not automatically one distinct chemical, one person, or one validated test result.
Some compounds, such as sorbitol, required assignment checks because multiple metabolites overlap in the same NMR spectral region. The authors supported the sorbitol identification by comparing a representative post-intake sample with a pure sorbitol reference using two-dimensional NMR correlation spectroscopy (COSY). Separately, urinary uric acid was assayed with a clinical-chemistry analyzer because it was not detectable through the relevant NMR proton signal at physiological pH. Those measurements should not be merged into an undifferentiated claim that “NMR measured everything.”
There was no parallel validated forensic whole-blood ethanol time series, no breath-alcohol curve, and no determination of the tested participants’ legally relevant BAC at a prior event in this study. Nor did investigators test driving, divided attention, lane control, crash risk, or individual clinical impairment. A creatinine-normalized urine-metabolite profile is not a direct blood concentration or a psychomotor-performance measurement. For the broader distinction between specimen and interpretive claim, see the site’s published specimen collection guide.
Which alcohol-related urine signals changed in the experiment?
Comparison of the vehicle-alone and alcohol-plus-vehicle conditions showed experimental group separation after ingestion, especially at two hours. The researchers reported at least 58 of 289 spectral bins (approximately 20%) meeting their screening criteria at two hours, using p ≤ 0.05 and an absolute fold-change threshold of at least two. By four hours, 13 bins met the stated between-condition criteria. These are exploratory feature-screening counts, not 58 separately identified biomarkers or the sensitivity of a forensic panel. A significant group contrast is not equivalent to validated individual classification.
Seven identified urinary metabolites were reported increased at one hour relative to the pre-intervention time point: ethanol, hypoxanthine, indoxyl sulfate, lactic acid, methylamine, sorbitol, and trimethylamine-N-oxide (TMAO). Five—ethanol, hypoxanthine, lactic acid, sorbitol, and TMAO—remained significantly elevated at the measured time points through four hours within the alcohol condition. Uric acid followed a different pattern, decreasing significantly at two and three hours before returning toward baseline.
| Observation in the 20-person analysis | What was measured or computed | What it does not prove |
|---|---|---|
| At least 58/289 NMR spectral bins differed at two hours; 13 at four hours. | Screened, creatinine-normalized urine spectral features comparing the two intervention conditions. | Not 58 confirmed individual-specific alcohol biomarkers, a validated test panel, or a quantified detection window. |
| Urinary ethanol, sorbitol, hypoxanthine, lactic acid and TMAO remained increased through sampled hours 1–4. | Serial urine analyte profiles in a controlled within-subject experiment. | Not a blood ethanol curve, precise ingestion clock, or proof of legal or clinical impairment. |
| Hypoxanthine reached its observed peak at one hour for 18 participants and two hours for two. | Individual urinary-metabolite time courses under one controlled protocol. | Not an external validation of a reliable personal post-drinking interval. |
| Uric acid fell at hours 2 and 3 and approached baseline at hour 4. | A separate enzymatic clinical-chemistry urinary uric-acid assay. | Not an NMR uric-acid result, a direct hepatic redox measurement, or an impairment measure. |
Why did urinary sorbitol attract attention?
Sorbitol is a polyol and an organic osmolyte involved in cellular volume regulation. The authors considered its rapid urinary increase after the alcohol intervention a noteworthy metabolic finding, supported by NMR spectral assignment and reference-standard comparison. Most participants displayed sorbitol excretion within the first post-intake hour, while the smoothed group trajectory peaked near two hours. Individual patterns varied substantially.
Importantly, identifying sorbitol in urine does not demonstrate that ethanol uniquely generated it, identify the organ producing it, or directly measure hepatocyte swelling. The paper discusses plausible mechanisms involving altered NAD-dependent sorbitol degradation and cell-volume regulation, but the investigators did not directly measure individual hepatocyte volume changes or establish an exclusive source pathway. Sorbitol has metabolic roles outside alcohol exposure, including pathways affected by glucose metabolism. A forensic laboratory could not adopt it as an alcohol-specific clock based on this experiment alone.
The researchers’ explanation is best read as a biologically informed hypothesis attached to measured urinary data: it helps explain why an osmolyte might rise, without converting urine sorbitol into an established diagnostic marker of alcohol ingestion or impairment.
Does hypoxanthine provide a more exact drinking-time marker?
No. Hypoxanthine is a purine-metabolism intermediate, and its urine excretion increased after the intervention. The authors reported a strong relative increase (fold change about 8.19 with p ≤ 0.0001) and an observed one-hour peak in 18 of 20 participants, with the remaining two peaking at two hours. These results describe a particular controlled sample and its discrete collection schedule. A peak assigned to one of five observation points does not determine the moment of alcohol intake in an unobserved person.
Renal excretion, hydration, diet, exercise, health conditions, variation in purine metabolism, specimen completeness and timing could alter urine patterns in other settings. This paper did not validate a classification threshold, reference population, forensic measurement uncertainty, time-since-use prediction interval or external replication cohort. Even if a urine profile resembled the group pattern, it would not prove that the person drank exactly one or two hours earlier.
The investigators interpreted hypoxanthine and uric-acid trajectories using known NAD-dependent metabolism and competing renal-handling mechanisms. Those pathways are mechanistic explanations, not direct assays of hepatic NAD+/NADH concentration in the volunteers.
Why is the distinction between a group curve and a single person’s timing so important?
A research graph often displays a smoothed trajectory across participants, but an evidentiary question involves a particular specimen collected at a particular time. The authors highlighted individual differences: one participant had an especially early, prominent increase in several metabolites without being excluded as a statistical outlier. Four participants displayed early and marked lactic-acid responses. Even within a fairly homogeneous group, the curves did not follow an identical timetable.
Further, urinary excretion integrates processes occurring before collection. A specimen may reflect urine accumulated in the bladder, fluid intake, renal handling and the timing of voiding. Its concentration-time pattern does not track systemic blood ethanol instantaneously. The study’s measured four-hour observation window also cannot establish behavior at eight, twelve or twenty-four hours, and cannot support extrapolation to repeated drinking, delayed absorption, disease or postmortem specimens.
Individual back-calculation would require a separately validated model, defensible assumptions, relevant measurements and quantified uncertainty. To understand why a biological measurement is not automatically a retrospective concentration estimate, see the published pharmacokinetics guide.
Did the NAD supplement prevent the alcohol-related metabolic changes?
Not in a manner the authors considered a justification for its use as an alcohol countermeasure. They modeled selected hypoxanthine and sorbitol measurements across the four intervention conditions. For sorbitol, the material contrast was between interventions containing alcohol and those without it; the two alcohol conditions did not differ significantly from one another in the comparisons reported. Some hypoxanthine comparisons differed later after alcohol plus NAD, but not as a simple normalization: the supplemented condition showed even greater hypoxanthine excretion in some comparisons.
This was an investigation of urinary metabolic responses after a 50-mg NAD-containing supplement, not a clinical trial showing prevention of impaired driving, faster ethanol elimination, sobriety restoration or medically meaningful benefit. A metabolite that shifts in a statistical comparison is not evidence that a person has regained motor or cognitive capacity. The findings should not be generalized to other supplement products or dosages.
How should a forensic toxicologist read the NMR evidence?
The scientific chain has separate links. An NMR resonance assigned to urinary ethanol can support the presence of ethanol in that urine specimen if identification and quality control are adequate. A measured urinary sorbitol concentration supports that this endogenous compound was detected under the validated assay conditions. A multivariate pattern in 20 volunteers describes a group distribution. A proposed redox or osmoregulation mechanism explains a hypothesis. None alone supplies a calibrated estimate of whole-blood ethanol, a forensic time of ingestion, a behavioral impairment finding, or the cause of a disputed event.
| Question in a forensic matter | Evidence from the Irwin study | Additional evidence needed for case-level interpretation |
|---|---|---|
| Was ethanol present in the submitted urine? | Ethanol resonances were directly observed in urine after administration, under research NMR conditions. | Validated specimen-specific identification, controls, chain of custody, collection/storage history and reporting limits. |
| Was sorbitol increased because the individual drank alcohol? | A sorbitol increase was associated with assigned alcohol intervention in this crossover group. | Investigations of specificity, alternative medical/dietary causes, within-person baseline, confounding and assay reproducibility. |
| Exactly when was alcohol consumed? | Experimentally scheduled sampling revealed varying group and individual trajectories over hours 1–4. | Prospectively validated individual timing model, comparable conditions, prediction intervals and known specimen collection history. |
| What was the person’s blood or breath alcohol concentration? | Neither forensic whole-blood nor breath ethanol concentration was measured as an outcome series for model validation here. | Appropriate blood/breath method and a defensible measured or independently validated estimated value. |
| Was a driver impaired or did alcohol cause the incident? | No driving, psychomotor, medical impairment, collision or case-causation endpoint. | Independent behavioral, temporal, scene, clinical and analytical evidence evaluated with uncertainty. |
What are the primary study limitations?
Population and sampling: analysis of 20 healthy young men, selected from 24 recruited volunteers, limits generalizability by age, sex, health, medications, diet, tolerance and disease status. Sampling stopped four hours after consumption; a sparse fixed time grid cannot establish the precise moment of a biological peak between collections.
Intervention: a controlled substantial vodka dose was ingested with a flavored water vehicle. That beverage included sodium benzoate and other ingredients that can influence the urinary fingerprint. The vehicle was considered in the analysis, but the results are not a universal alcohol-only metabolite signature under every diet or beverage choice.
Analytic scope: untargeted high-dimensional NMR discovery and relative, creatinine-normalized spectral values are not equivalent to a routinely validated quantitative forensic assay of every named metabolite. The reported bin-screening threshold does not supply clinical specificity or a multiple-population false-positive rate. Uric acid came from a separate assay.
Mechanistic inference: increased urinary lactic acid, sorbitol and hypoxanthine are compatible with proposed biochemical responses, but the study did not directly measure liver-cell swelling, tissue redox ratios, cellular injury or each pathway’s contribution. Group statistical significance alone does not establish causal exclusivity.
Absent forensic validation: there was no blinded external test set for unknown subjects, standardized forensic cutoffs, case-level likelihood ratios, demonstrated measurement uncertainty for individual time attribution, BAC-conversion validation or impairment endpoint. These omissions are decisive when an attorney or expert asks whether the paper proves a disputed individual fact.
What practical questions should counsel ask about a purported urine-metabolomics alcohol opinion?
Request the complete test method, chromatograms or NMR spectra, annotations, identification criteria, calibration and QC, analyte units, creatinine normalization, specimen collection and voiding history, relevant metabolic or medical history, reference populations, decision thresholds, independent validation and the statistical model’s individual-level error. Ask specifically whether a claimed “alcohol-specific” biomarker has been tested against nonalcohol causes and whether the testimony relies on group averages rather than calibrated individual predictions.
If the opinion purports to infer past intoxication or an accident cause, distinguish those assertions from what was measured. A well-characterized urine analyte may still leave the time, dose, blood concentration, behavior and causal mechanism unresolved. The site’s published forensic toxicology evidence library organizes additional analytical and interpretive subjects; it is not a substitute for the original data and case record.
Frequently asked questions
Can an NMR urine test determine that alcohol was consumed two hours earlier?
Not on the basis of this study. Certain group profiles were prominent near two hours after a scheduled intervention, but individual patterns varied. No external, individual-level timing algorithm with validated prediction limits was established.
Is urinary sorbitol an alcohol-specific biomarker?
No alcohol-specific diagnostic validation was shown. The study found increased sorbitol after the alcohol intervention and investigated plausible metabolic explanations, but did not establish exclusivity or quantify performance against other causes of elevated urinary sorbitol.
Did researchers measure ethanol in urine or only its metabolites?
They detected urine ethanol itself by NMR as well as several metabolic compounds. Direct urine ethanol detection must still be distinguished from creatinine-normalized endogenous metabolites and from whole-blood or breath ethanol measurements.
Do 58 significant spectral bins mean 58 positive alcohol biomarkers?
No. The 58 figure concerned screened spectral bins meeting study-defined statistical and fold-change criteria at two hours. Multiple bins may relate to a single compound or overlapping resonances; it was not a validated 58-analyte forensic panel.
Does the hypoxanthine one-hour peak identify when a specific person drank?
No. Eighteen of the 20 participants had their recorded urinary peak at one hour in this protocol, while two peaked at two hours. It is neither a universal biological clock nor an independently validated personal timing rule.
Did the NAD supplement reverse intoxication or make driving safe?
Neither conclusion was tested or established. Selected metabolite differences were analyzed; the supplement did not convincingly reverse the alcohol-related urine signature, and no driving or behavioral sobriety endpoint was measured.
Can these urine results establish impairment or the cause of a collision?
No. Such opinions need separate time-specific analytical and behavioral evidence, a defensible inferential framework and explicit consideration of competing causes. This metabolomics intervention study cannot supply those missing elements on its own.
Primary scientific source
Irwin C, van Reenen M, Mason S, Mienie LJ, Wevers RA, Westerhuis JA, Reinecke CJ. The 1H-NMR-based metabolite profile of acute alcohol consumption: A metabolomics intervention study. PLOS ONE. 2018;13(5):e0196850. https://doi.org/10.1371/journal.pone.0196850. The article and its original figures, source tables, methods and supporting-material links provide the primary basis for the study-specific statements above.