EVIDENCE LIBRARY
Can Falling AST or ALT Rule Out Alcohol Exposure? What a 0.5% Beverage Trial Shows
In a 42-person crossover study, AST and ALT fell during low-dose alcohol exposure, but between-period changes were not significant. Learn the forensic limits.
Short answer: No. A decrease in serum aspartate aminotransferase (AST) or alanine aminotransferase (ALT) does not establish that someone abstained from ethanol. In a small randomized, double-blind crossover study, adults consumed a beverage containing 0.5% alcohol twice daily for four weeks; average AST and ALT values fell within that exposure period. Crucially, the changes between the alcohol and placebo periods were not statistically significant at the conventional 0.05 threshold. The study supplies an instructive example of why a laboratory trend cannot establish an individual’s exposure history, but it does not validate alcohol as a liver treatment or supply a forensic drinking-history test.
The primary evidence is Oshima, Shiiya and Kato (2022), “Effects of Regular Low-Level Alcohol Consumption in Healthy Individuals: A Randomized, Double-Blind, Placebo-Controlled Crossover Study,” Healthcare 10(5):882. Every numerical result below refers to that study, not to a general rule for all drinking patterns.
What exactly did the researchers give participants?
Forty-two healthy Japanese adults (15 men and 27 women), ages 40–69, completed a randomized, double-blind, placebo-controlled crossover trial. Participants had relatively low baseline scores on a general well-being questionnaire. People drinking alcohol more than once weekly, receiving medication treatment, or reporting alcohol dependence were among those excluded. This was not a study of patients with alcoholic hepatitis, advanced liver disease, or people investigated for impaired driving.
Each participant was assigned an order: four weeks of a study beverage containing 0.5% alcohol by volume or an otherwise matched water-based placebo, a four-week washout, and then four weeks of the other beverage. Both beverages came in 465 mL cans, with two cans consumed each day, one in the morning and one in the evening. The investigators described the ethanol quantity as approximately four grams daily. The exposure was intentional even though the percentage by volume was low. A beverage described as low-alcohol or even colloquially “non-alcoholic” is not necessarily ethanol-free.
All 42 completed the crossover schedule. Blood-chemistry analyses involved 41 participants because one participant had difficulty with venous sampling. The study’s primary laboratory observations concerned blood biochemistry, not serial concentrations of ethanol.
What was actually collected and analyzed?
Venous specimens were obtained at the beginning and end of each four-week intervention period. The tested variables included serum AST, ALT, lactate dehydrogenase (LDH), gamma-glutamyl transferase (GGT), alkaline phosphatase, bilirubin, other chemistry measurements and selected endocrine measurements. Well-being and sleep questionnaires were also administered. The researchers did not publish measured blood ethanol concentrations or an alcohol-elimination curve for each participant. Their discussion of hypothetical low blood-alcohol levels is not equivalent to a measured BAC result.
What happened to AST, ALT, LDH and GGT?
Table 2 in the original paper permits two separate statistical questions: did a value change from the beginning to the end of a single period, and was that change different between the two beverage conditions? Those questions must not be interchanged. The means below are reported in the study’s stated units; values are rounded as in the original table. The intervention-period biochemistry sample was 41.
| Analyte | 0.5% ethanol period: start → four weeks | Within-ethanol-period p | Placebo period: start → four weeks | p for difference in change between periods |
|---|---|---|---|---|
| AST | 22 → 20 (study table labels IU) | 0.034 | 21 → 21 | 0.059 |
| ALT | 22 → 18 U/L | 0.033 | 19 → 19 U/L | 0.064 |
| LDH | 173 → 169 U/L | 0.013 | 173 → 172 U/L | 0.340 |
| GGT | 22 → 22 U/L | 0.453 | 21 → 20 U/L | 0.718 |
The inferential distinction matters: AST, ALT and LDH showed statistically significant within-period declines during the low-alcohol condition. But the authors’ direct comparisons of change between beverage conditions did not cross p < 0.05 for any of these three markers. AST p = 0.059 and ALT p = 0.064 were described as trends, not definitive evidence of a between-condition effect. The LDH difference was especially unpersuasive at p = 0.340. A reader who quotes only p = 0.034 or p = 0.033 omits the principal comparison needed to support an alcohol-specific treatment effect.
The group averages also do not describe every individual’s trajectory. The means, standard deviations and plotted individual points show variation. A change of a few units does not constitute a calibrated ethanol-dose measure; the observation should not be turned into a personal prediction about alcohol exposure.
Is an AST or ALT measurement an alcohol test?
No. In this protocol, AST and ALT were measured as endogenous enzymes in serum. Their laboratory units express enzyme activity, not ethanol mass per volume. A result of “ALT 18 U/L” does not answer whether ethanol was detected in blood, when drinking occurred, what concentration was reached, or whether any cognitive or motor function changed. The presence of alcohol in a study beverage was established by the controlled intervention; it was not inferred from a positive ethanol assay in the participants.
Conversely, a direct ethanol measurement in blood or breath, if available, addresses a different analytical target. Even a properly obtained ethanol result is not interchangeable with an inference about liver injury, functional impairment or exact drinking history. For more on the difference between an analyte and its interpretation, see the site’s pharmacokinetics evidence guide and timed clinical laboratory interpretation guide.
What about GGT, bilirubin and glucose?
The authors reported no statistically significant before-versus-after changes during either condition for GGT, alkaline phosphatase, total bilirubin and blood glucose, among several other chemistry measurements. That is not proof that those analytes never respond to alcohol, nor that any given person remained biologically unaffected. It reflects results under this trial’s particular dose, selection criteria, sampling schedule and duration.
Does a decrease in liver enzymes mean that alcohol improved liver health?
Not on these data. A statistically significant difference within one arm can arise while the more relevant between-condition contrast remains unconvincing. This study did not measure histological liver injury, hepatic fibrosis, long-term progression, clinical liver outcomes, or mortality. It also did not establish a mechanism for the enzyme pattern. The authors expressly acknowledged that oxidative-stress measures were not collected, despite discussing possible oxidative-stress pathways.
A four-week observation with a low ethanol dose cannot establish that starting alcohol consumption treats a liver condition. Nor can it cancel established concerns about alcohol-related harms at different doses and durations. The participants were screened healthy adults with limited usual drinking, not a representative sample of all clinical or forensic populations. Accordingly, the measured enzyme pattern is evidence about these laboratory values during this intervention, not about net health benefits, disease prevention, or therapeutic advice.
Why are p = 0.059 and p = 0.064 not “proof of no effect”?
Failure to cross an arbitrary threshold does not demonstrate equivalence. The differences could be real but too uncertain to distinguish clearly with the available sample; they could also arise from variation rather than a beverage-specific effect. The study was exploratory and did not perform a prospective sample-size power calculation. Its reported between-condition statistics, variability and confidence limitations prevent either extreme claim: “ethanol definitely improved liver function” or “ethanol definitely had no effect.” The accurate result is that a statistically compelling between-condition difference was not demonstrated for AST or ALT.
What did the well-being and sleep questionnaires show?
The study measured the General Well-Being Schedule (GWBS) and Pittsburgh Sleep Quality Index (PSQI) before and after each intervention. In all 42 participants, the reported GWBS change did not differ significantly between the beverage periods (p = 0.206), and PSQI change was not distinguishable (p = 1.000). A stratified analysis of 30 participants with poorer baseline well-being reported a GWBS change of 16.0 versus 11.5 on the study’s median summary scale (p = 0.041). This smaller subgroup finding should not be promoted as a general effect on mental health, alertness, competence or driving performance.
Similarly, an endocrine subgroup comparison was described as a non-significant trend for free testosterone (p = 0.051), not a demonstrated hormonal treatment benefit. The study’s scores capture responses to particular questionnaires and laboratory assays. They are not field-sobriety testing, driving-simulator outcomes, reaction-time tests, or validated measures of an individual’s intoxication.
Which questions are unsupported in a forensic or clinical case?
| Proposed inference | What this study measured | Why the proposed inference is unsupported |
|---|---|---|
| “AST fell, so the person did not drink.” | Group-mean serum AST before and after known low-dose beverage exposure. | Some observed means fell despite administered ethanol; AST is not a specific ethanol assay. |
| “ALT is normal, so the BAC must have been zero.” | Serum ALT activity, not measured venous ethanol in the participant series. | No calibrated ALT-to-BAC conversion, individual BAC observations or diagnostic threshold. |
| “The beverage improved liver function.” | Short-term enzyme concentrations in a selected healthy sample. | Between-condition contrasts were not significant; clinical liver function or disease outcome was not established. |
| “No enzyme rise means no impairment.” | Chemistry panels and subjective questionnaires. | No driving, balance, reaction-time or case-specific functional impairment testing. |
| “The observed association explains an injury.” | A controlled intervention’s group measurements. | No case injury, exposure timing, competing medical causes or causation adjudication. |
An appropriate evaluation of a disputed alcohol-history opinion first specifies the exact claim. Is the question whether ethanol was in a tested specimen, whether a four-week biomarker trend occurred, whether tissue injury existed, or whether a person’s actions were impaired at a particular moment? The trial measured some of these quantities and not others. Mixing them creates a false impression that an ordinary chemistry panel can resolve an alcohol-exposure dispute.
What are the study’s most important limitations?
The research team described an exploratory trial; no prospective power calculation set the enrollment target. Only 42 adults completed it, and one did not contribute complete blood chemistry samples. Eligibility restricted participants to healthy middle-aged or older Japanese volunteers with comparatively low well-being scores and without more-than-weekly usual alcohol consumption. These factors materially restrict extrapolation to heavy drinkers, patients with liver disease, younger adults, people taking drugs, or a deceased individual’s laboratory specimens.
The study lasted four weeks per treatment condition with a four-week washout. It lacked measured blood ethanol pharmacokinetic curves, breath-test series or tissue biopsy evidence of liver benefit. Multiple laboratory variables and subgroup comparisons raise concerns about interpreting isolated p-values without cautious replication. All three authors disclosed employment by Asahi Quality & Innovations, Ltd.; they reported no external research funding. Sponsorship or employment does not by itself falsify results, but it is relevant context when weighing a health-benefit interpretation of an alcohol-containing beverage.
Above all, the published values are group statistics. They do not provide diagnostic sensitivity, specificity, a likelihood ratio, a validated individual classification rule, or a medically justified action threshold for alcohol exposure. The absence of such validation is a substantive limit, not a missing detail that a commentator can fill in from a single mean.
Frequently asked questions
Can a person consume alcohol and still have lower AST or ALT later?
Yes. In this low-dose intervention, the study reported falling mean AST and ALT values over four weeks. That does not predict whether either marker will rise, fall or stay constant after a different person’s consumption, nor whether that person’s liver disease has changed.
Were the reported AST and ALT changes statistically significant?
They were significant when each value was compared to its own baseline during the 0.5% alcohol period (AST p = 0.034; ALT p = 0.033). The more relevant change-versus-placebo contrasts were not significant at p < 0.05 (AST p = 0.059; ALT p = 0.064). Those are different statistical questions.
Did the researchers measure a blood alcohol concentration?
Not as a serial exposure endpoint in the published trial. Venous blood was collected for biochemical and endocrine testing. A predicted blood-alcohol concentration discussed as part of the rationale must not be labeled a participant’s measured BAC.
Does “0.5% alcohol” mean ethanol-free?
No. The intervention beverage explicitly contained 0.5% alcohol by volume and was administered twice daily. The study described its total daily ethanol quantity as approximately four grams. A regulatory or retail label may vary across jurisdictions and cannot replace the actual composition.
Did the trial prove alcohol improved liver health?
No. It recorded a short-term enzyme pattern without statistically decisive between-beverage differences for AST or ALT, tissue injury assessment, or long-term clinical outcomes. An isolated laboratory trend is not proof of clinical benefit or a reason to use ethanol therapeutically.
Can this study determine whether someone was impaired while driving?
No. There was no validated motor, vigilance, field-sobriety or on-road performance outcome and no forensic impairment threshold. Biochemical findings and ability to perform a task are distinct empirical questions.
What should an expert ask when AST or ALT is offered as evidence of drinking?
Identify the exact assay and units; obtain the time series, laboratory intervals, history, medications and relevant medical context; distinguish direct ethanol analysis from nonspecific enzymes; and ask whether a validated model exists for the particular inference. This trial does not supply an individual ethanol-exposure rule.