EVIDENCE LIBRARY
Can PEth Verify a 14-Day Drinking History? A Two-Year Clinical Study
A two-year AUD-treatment study compared PEth with 14-day drinking diaries. Learn what a 0.22 µmol/L cutoff, sensitivity and imperfect agreement can prove.
A phosphatidylethanol (PEth) blood result can provide independent evidence consistent with recent alcohol consumption, but it cannot verify the exact number of drinks in a person’s diary. In a 2024 analysis of 181 people receiving treatment for alcohol use disorder, the association between PEth 16:0/18:1 and a 14-day drinking history strengthened over two years: the reported Spearman correlation was 0.42 at baseline and 0.69 at 104 weeks. The relationship was meaningful, but far from an individual-level equation converting a PEth result into a specific dose or schedule of drinking.
The investigators also tested whether a blood PEth threshold could distinguish patients who reported at least 10 Swedish standard drinks per week. The pooled receiver-operating-characteristic (ROC) analysis selected 0.22 µmol/L as a useful threshold in this particular treatment cohort. That threshold was derived against retrospectively reported consumption—not independently observed drinking—and is not a universal proof-of-abstinence, impairment, or legal-drinking cutoff. The original primary research is Hammarberg and colleagues, Alcohol and Alcoholism (2024), DOI 10.1093/alcalc/agae070.
What does a PEth blood test measure—and what does it not measure?
PEth refers to a group of abnormal phospholipids associated with ethanol exposure. This study quantified the specific homologue PEth 16:0/18:1 in blood using a mass-spectrometric laboratory method. It did not measure a contemporaneous blood ethanol concentration as a blood-alcohol-concentration (BAC) test would. A person may have no reportable ethanol at the time of collection while a PEth result still reflects prior drinking; conversely, a below-threshold PEth result does not prove lifetime abstinence or exclude a particular isolated drinking event.
The authors reported PEth in micromoles per liter (µmol/L). Under the reporting convention applied in their analysis, values below 0.05 µmol/L were classified as negative and entered as zero for statistical purposes. A value assigned zero in an analytical dataset is not the same as a demonstration that no PEth molecules were present, nor is it definitive proof that no alcohol had been consumed.
PEth is therefore relevant to alcohol-consumption history, not a direct measurement of alcohol’s immediate effects on the brain. It does not by itself establish acute intoxication, motor impairment, decision-making capacity, a person’s precise recent BAC, the number of drinks on any particular day, or why an incident occurred.
Who was studied, and why does the sampling frame matter?
The PEth analysis used information from a randomized clinical trial conducted at three specialized addiction outpatient clinics in Stockholm. The larger trial enrolled 250 adults diagnosed with alcohol use disorder who wanted to reduce drinking rather than necessarily become abstinent. Participants had been allocated to behavioral self-control training or motivational enhancement therapy. For this secondary biomarker analysis, the investigators combined treatment arms, because comparing those therapies was not the question being tested.
Only 181 patients had usable matched PEth and drinking-history information for the baseline analysis. Their median age was 53 years (range 28–70); 87 were men and 94 were women. Median baseline self-reported intake was 21.2 standard drinks per week, and median baseline PEth was 0.4 µmol/L. The sample comprised treatment-seeking people with alcohol use disorder, not a random cross-section of drivers, criminal defendants, employees, or the general population.
Blood and interview assessments were scheduled at baseline and 12, 26, 52, and 104 weeks. Complete matched observations diminished over time: 100 participants contributed matched data at 104 weeks, and a separate longitudinal comparison considered 77 participants who had complete data at both baseline and the final follow-up. Late blood collection, missing questionnaires, and attrition affected analyzable samples. Those are material qualifications when evaluating the apparent changes in association over time.
How was the 14-day drinking history obtained?
The reference history came from the Timeline Follow-Back (TLFB) interview method, which retrospectively records alcohol use by day. Although participants were asked about a longer period, this analysis matched PEth against the 14 days immediately before the blood draw. A Swedish standard drink in this study represented 12 grams of ethanol.
Matching the blood collection date to the interview window improves temporal relevance. But TLFB remains a recollection-based self-report, not a camera, continuous transdermal monitor, witnessed dosing record, or chemically verified count of every beverage. Reporting could be affected by memory, estimating serving sizes, motivation, shame, perceived expectations, or improved practice at keeping a drinking record. The investigators could not verify every stated drink independently.
This is particularly important for litigation. Calling TLFB the “reference” in an ROC analysis identifies the study’s chosen classification standard; it does not establish that every classification represents objective ground truth.
Did PEth agree more closely with drinking histories after two years?
Yes, on the study’s group-level rank correlation measure. The baseline Spearman coefficient was ρ = 0.42; by 104 weeks it was ρ = 0.69, which the authors described as a statistically significant increase. An examination limited to the 77 participants with data at both endpoints showed a similar overall pattern, arguing against attrition as the only simple explanation.
Nevertheless, neither coefficient establishes perfect agreement. A Spearman correlation describes the extent to which higher reported intake tends to accompany higher PEth concentrations across people. It is not the percentage of accurate histories, the percentage of drinks recovered, a probability of deception, or a rule predicting an individual’s exact consumption from a single measurement. The trial did not independently confirm whether enhanced diary accuracy, changes in actual consumption, differential retention, or other changes produced the stronger relationship.
| Study quantity | Reported finding | Permitted interpretation |
|---|---|---|
| Original treatment trial | 250 enrolled | Parent randomized trial; not the sample size for every biomarker analysis |
| Matched PEth/TLFB observations at baseline | 181 patients | Analyzable baseline biomarker-history data |
| Matched observations at 104 weeks | 100 patients | Smaller late-follow-up sample; missingness matters |
| Complete baseline-plus-week-104 pairs | 77 patients | Subset used to investigate retention effects |
| PEth versus past-14-day TLFB, baseline | Spearman ρ = 0.42 | Moderate group-level rank association, not exact intake reconstruction |
| PEth versus past-14-day TLFB, week 104 | Spearman ρ = 0.69 | Stronger association in a reduced follow-up sample, not proof of perfect reporting |
| PEth versus other biomarkers | ρ = 0.54 with CDT; ρ = 0.24 with GGT | Correlations with distinct clinical markers, not forensic equivalents of PEth |
What does the proposed PEth threshold of 0.22 µmol/L mean?
The researchers evaluated whether PEth could identify the study’s self-report-defined category of at least 10 Swedish standard drinks each week, averaged across the 14 days before collection. With 12 grams of ethanol per standard drink, that operational threshold corresponds to about 120 grams of ethanol per week. It is a particular Swedish clinical-guideline category; it is not a criminal BAC threshold, a universal international definition of hazardous drinking, or a statement that drinking below that rate cannot cause harm.
The authors used ROC curves and Youden’s J statistic to identify a threshold balancing sensitivity and specificity. Their combined time-point analysis selected approximately 0.22 µmol/L. The original Figure 5 labels the pooled operating point around 81% sensitivity and 66% specificity. The separate week-104 curve selected approximately 0.21 µmol/L with 88% sensitivity and 71% specificity. These are not one interchangeable set of estimates: the week-104 operating characteristics belong to the follow-up curve, while 0.22 µmol/L characterizes the pooled analysis.
Those percentages compare the test to reported intake. Even if a PEth assay were analytically excellent, a disagreement between PEth and TLFB could reflect inaccurate recall, differences in biomarker formation or elimination, sampling timing, drinking pattern, or another unmeasured factor. It would be scientifically unwarranted to call every positive result “proof the patient lied” or every negative result “proof the patient did not drink.”
| Analysis set | PEth cutoff | Sensitivity versus TLFB | Specificity versus TLFB |
|---|---|---|---|
| All study time points pooled | Approximately 0.22 µmol/L | Approximately 0.81 | Approximately 0.66 |
| Week-104 follow-up only | Approximately 0.21 µmol/L | Approximately 0.88 | Approximately 0.71 |
These are rounded values labeled on the original research figure, not independent replication or cross-validation. “Positive” means meeting the specified self-reported drinking category, not proving impairment or an objectively observed dose.
Is 0.22 µmol/L a legal or universal PEth cutoff?
No. Selection of a threshold depends on the definition of the outcome, population, collection timing, and the balance between falsely classifying a reporting participant and missing someone whose reported intake meets the chosen category. The article discusses a different Swedish clinical convention of 0.30 µmol/L as a higher threshold suggesting elevated consumption. It also notes other studies recommending lower values in other populations with differently defined drinking levels.
A case-specific analysis must not quietly substitute this single cohort’s optimized statistical threshold for a legal requirement or another laboratory’s validated interpretive policy. A clinician’s longitudinal treatment-monitoring question is not necessarily the same as a professional-licensing, child-custody, workplace, criminal, or postmortem interpretation question. The consequences of misclassification also differ.
Can PEth reveal exactly when someone drank, or whether they were impaired?
No. PEth measurements integrate biological responses over time; formation and elimination differ among individuals. The study could not distinguish in a rigorous way whether the same total weekly drinking came from regular daily consumption or one or two binge episodes. The timing of a drinking episode relative to sampling may change the observed value. A 14-day statistical association cannot be inverted into a verified chronology of a specific person’s drinking.
Nor does the study report directly measured blood ethanol concentration at the time of the alleged event, breath alcohol during a driving episode, field-sobriety performance, controlled psychomotor tasks, a collision outcome, or medical causation. A PEth value should not be relabeled “BAC,” “current intoxication,” “degree of impairment,” or “proof of conduct.” The relevant specimen, targeted PEth homologue, measurement units, collection date, reporting limit, documented prior values, and supporting history must remain explicit.
What preanalytical and clinical uncertainties should be checked?
The study recognized that PEth can be affected by post-collection formation in blood containing ethanol. It did not experimentally quantify that artifact for each individual participant. Interpretation therefore requires the laboratory’s collection, transport, storage, and handling details, particularly when a sample may have contained ethanol when collected. The study also lacked body-mass-index data, lacked a continuously observed drinking record, and could not isolate the effects of binge-versus-daily patterns.
The available study data reflect a selected treatment-seeking cohort, people with alcohol use disorder, and repeated contact with care. Transfer to people without that condition or to an adversarial evidentiary setting is an extrapolation requiring separate validation. Even clinically meaningful changes in PEth are better evaluated as part of a documented series of measurements than as a stand-alone arithmetic reconstruction of beverages.
For information about specimen timing and handling, see this site’s specimen-collection guide. For the separate analytical question of measuring ethanol itself, see blood-alcohol testing by headspace GC-FID. For why blood levels and modeled exposure histories require pharmacokinetic context, see pharmacokinetics in forensic toxicology.
How should an expert explain this study in deposition or trial?
A defensible description is narrow: the researchers found that blood PEth 16:0/18:1 was associated with self-reported consumption during the preceding 14 days in adults receiving treatment for alcohol use disorder, with stronger rank correlations at later follow-up. An ROC-selected threshold near 0.22 µmol/L classified a specified Swedish self-report drinking category with imperfect sensitivity and specificity. This evidence supports PEth’s usefulness as a clinical adjunct to drinking histories. It does not establish that a measured value maps to a particular number of drinks, that a patient intentionally misreported consumption, or that any person was impaired during a legally relevant episode.
Before drawing individualized conclusions, ask whether the method measured PEth or ethanol; the reported matrix and units; the collection-to-analysis sequence; whether other blood ethanol was present at collection; laboratory detection and quantitation limits; the person’s longitudinal biomarker pattern; and the precise proposition the opinion is intended to establish. The statistical association in this primary study is not a substitute for these records.
Frequently asked questions
Does a positive PEth test prove that someone drank during the preceding 14 days?
It provides evidence consistent with prior ethanol exposure, but it does not independently date every drinking episode. This study compared PEth with a 14-day self-reported history; it did not continuously witness or chemically document each beverage.
Does a negative PEth test prove abstinence?
No. This study classified results below 0.05 µmol/L as negative and assigned them zero for statistical analysis. A negative report must be interpreted against the assay’s reporting threshold, sampling time, and individual variation; it is not proof that no alcohol was consumed.
Was 0.22 µmol/L validated as a BAC or impairment threshold?
No. The value was selected to classify a self-reported average of at least 10 Swedish drinks per week in a particular treatment cohort. It is not a contemporaneous ethanol concentration or a psychomotor performance cutoff.
Does a PEth concentration tell exactly how many drinks a person consumed?
No. The paper documented considerable variation between individuals and could not reliably distinguish different drinking patterns producing similar weekly totals. A Spearman rank correlation cannot be converted into a trustworthy individual drink count.
Why did the PEth and drinking-history correlation rise from 0.42 to 0.69?
The association strengthened during follow-up. More accurate diary keeping, changes in drinking, participant retention, and other factors may contribute. The design did not independently verify every dose or prove a single cause of the change.
Are the 81%/66% and 88%/71% sensitivity/specificity pairs contradictory?
No. The original Figure 5 presents different ROC analyses: approximately 81%/66% at the pooled 0.22 µmol/L threshold and approximately 88%/71% at the week-104 threshold near 0.21 µmol/L. They should not be merged into one allegedly universal operating point.
Can an expert use this study to conclude that a defendant was impaired?
Not by itself. It contains no individual driving performance, offense-time ethanol measurement, or accident-causation validation. Its results concern group-level agreement between a biomarker and recalled drinking in a specific clinical setting.
Primary research and interpretation boundary
Hammarberg A, Ingesson Hammarberg S, Redegren Cuellar S, Guterstam J. Phosphatidylethanol as an outcome measure in treatment aimed at controlled drinking. Alcohol and Alcoholism. 2024;59(6):agae070. https://doi.org/10.1093/alcalc/agae070. This discussion interprets that original 181-patient biomarker analysis. It is general scientific education, not a substitute for method validation, a complete laboratory record, clinical evaluation, or case-specific forensic analysis.