Hangover severity cannot be used as a validated estimate of an individual’s alcohol elimination rate. A small exploratory analysis found an association between faster breath-alcohol disappearance and less severe hangover, but it combined two separate experiments involving only eight people. The laboratory alcohol challenge and the night that produced the hangover occurred about a year apart.
The 2020 study by Mackus and colleagues offers a hypothesis about alcohol metabolism and next-day symptoms. It does not supply an equation for calculating an earlier blood alcohol concentration from how sick someone feels.
Two studies supplied different pieces of the comparison
Five men and three women participated in both studies. They were young, healthy social drinkers; mean age was 20.4 years during the 2015 hangover study and 21.4 years during the 2016 laboratory challenge.
In the laboratory, participants ate a standardized meal and received an individually calculated oral ethanol dose mixed with orange juice to make a 250 mL beverage, consumed within five to ten minutes. The mean ethanol amount was 33.7 mL, with a standard deviation of 8.9 mL. This is ethanol volume, not grams per kilogram. Dosing targeted a reported breath alcohol concentration of 0.05%.
Breath measurements were taken every five minutes until two consecutive readings were zero. The investigators calculated an average disappearance rate by dividing peak breath concentration by elapsed time from peak until zero. The mean was 0.016 in the paper’s reported BrAC percentage units per hour, with a range of 0.012–0.019. This was not serial laboratory measurement of whole-blood ethanol, and the calculated average is not a direct assay of an enzyme’s activity.
The other study followed a normal night out. Researchers did not control beverage selection, drinking duration, activity or alcohol amount. Participants retrospectively reported a mean 11.8 drinks, with a standard deviation of 5.8. An adjusted Widmark calculation produced a mean estimated BAC of 0.20%, standard deviation 0.10%. Those were model estimates from reported drinking, not measured peaks from that evening.
The statistical signal depended on the analysis
Participants rated overall hangover from zero to ten at seven hourly assessments, from 10 a.m. through 4 p.m. Partial correlations between elimination rate and severity, controlling for estimated BAC, were negative at every timepoint. Faster laboratory disappearance therefore tended to accompany lower reported symptoms in this small dataset.
The original-sample p values in Table 2 ranged from 0.130 to 0.467; none was below 0.05. After 10,000 bootstrap resamples, bias-corrected confidence intervals excluded zero at 11 a.m., 1 p.m. and 2 p.m. They included zero at the other four assessments. The abstract and table identify those three times; the Results prose contains an inconsistent clock-time label, which should not be treated as an additional finding.
At 1 p.m., for example, the original partial correlation was −0.508, with p = 0.244. The reported bootstrap correlation was −0.472 with a 95% interval of −0.922 to −0.221. Reporting only “significant correlation” would conceal both the small sample and the dependence of significance on the resampling approach.
Bootstrap resampling repeatedly draws from the same eight participants. It can explore uncertainty under the selected method, but it does not turn eight people into 10,000 independent observations. Seven related timepoint comparisons also warrant caution when interpreting a selected subset of positive findings.
The year between studies matters
The analysis assumes that disappearance measured after the lower laboratory exposure meaningfully ranks participants’ elimination during a much heavier drinking occasion. That transfer was not directly measured. The authors discussed the possibility of different elimination kinetics at higher exposures and called for replication measuring both processes in one study.
Most recorded demographic characteristics did not differ significantly across the two studies, but that does not establish identical physiology, food intake, sleep, exposure or circumstances. A year-separated association is weaker evidence for an event-specific mechanism than simultaneous measurements would be.
Symptoms, elimination and impairment remain separate questions
The analysis did not test driving, establish a diagnostic cutoff for hangover, or validate an individual impairment classifier. It cannot establish that a person without symptoms eliminated alcohol unusually quickly, that severe symptoms imply residual ethanol, or that either person was safe to drive.
It also did not randomize a treatment intended to accelerate metabolism. An association between elimination and symptoms cannot by itself prove that a supplement or other intervention prevents hangover. The source reports several authors’ commercial research or consulting relationships; the study itself reported no external funding.
The useful forensic lesson is to preserve the chain of measurements: a breath-derived disappearance estimate from one occasion, retrospective drinking estimates from another, and subjective symptom scores the next day. None substitutes for a specimen collected at the disputed event. For a separate functional question, see our analysis of hangover and combined multitasking performance.
Primary scientific source
Mackus M, van de Loo AJAE, Garssen J, Kraneveld AD, Scholey A, Verster JC. The Association between Ethanol Elimination Rate and Hangover Severity. International Journal of Environmental Research and Public Health. 2020;17:4324. doi:10.3390/ijerph17124324.
Leave a Reply