Gastroenterology · Medical Literature
[Correspondence] The spectrum of steatotic liver disease in the USA
Abstract
We read the recent analysis of the spectrum of steatotic liver disease in the USA by Zobair Younossi and colleagues1 with great interest. The study addresses an important question about the impact of alcohol under-reporting; however, we believe that the central premise—correction of self-reported alcohol intake—requires more cautious interpretation, as it underpins all subsequent conclusions.
Clinical relevance
Does correcting self-reported alcohol intake for systematic underreporting materially alter estimated prevalence and premature mortality across the steatotic liver disease (SLD) spectrum—including MASLD, MetALD, and ALD—in the United States? __SNAPSHOT__ {"studyDesign":"Correspondence (letter to the editor) commenting on a published cross-sectional and longitudinal analysis (Younossi et al., Lancet Gastroenterol Hepatol, March 2026); no original data generated by the correspondence authors","setting":"Not applicable (commentary); index study used US National Health and Nutrition Examination Survey (NHANES) data from 1988–2023, a nationally representative survey of the US civilian, non-institutionalized population","sampleSize":"Not applicable (correspondence); index study prevalence analysis: n=41,100; mortality analysis: n=24,707","population":"Not applicable (correspondence); index study enrolled US adults aged ≥20 years with sufficient data to define SLD subtype and alcohol consumption via self-report","intervention":"Application of a validated sex-specific and drinking-frequency-specific correction factor to self-reported alcohol intake, calibrated to national per-capita ethanol consumption, to reclassify SLD subtype (MASLD, MetALD, ALD)","comparator":"Uncorrected self-reported alcohol intake (standard approach in NHANES-based SLD research)","primaryOutcome":"Change in prevalence estimates and premature mortality hazard ratios (HRs) across the SLD spectrum after alcohol underreporting correction; the correspondence specifically challenges the methodological validity of the correction premise itself","followUp":"Not reported for the correspondence; index study mortality follow-up linked to the National Death Index through December 31, 2019","fundingSource":"Not reported for the correspondence; index study authors declared multiple affiliations with pharmaceutical and biotech companies (industry-affiliated)"}
Key findings
This is a correspondence letter responding to the Younossi et al. index study; it does not report original quantitative findings. The correspondence authors raise interpretive concerns about the central premise of alcohol underreporting correction underpinning all conclusions of the index study. From the index study itself (Younossi et al., Lancet Gastroenterol Hepatol, March 2026): after applying validated correction factors for alcohol underreporting, the prevalence of both ALD and MetALD nearly doubled compared to uncorrected self-report estimates. Binge drinking (defined as ≥5 drinks on a single occasion at least once in the past year) was identified as the dominant driver of premature mortality in ALD (population attributable fraction: 92.9%) and contributed substantially in MetALD (21%), while type 2 diabetes was the principal driver in MASLD (up to 44.8%). The coexistence of binge drinking with type 2 diabetes or hypertension was associated with up to a 17-fold increase in premature mortality risk (index study, specific HR not extractable from available abstract). Cox proportional hazards models were used for mortality analyses; population attributable fractions were derived from adjusted HRs and exposure prevalence. Specific p-values, confidence intervals, and NNT/NNH are not reported in the available abstract of either the correspondence or the index study. Steatosis was defined variably across NHANES cycles using ultrasound, the US Fatty Liver Index, or vibration-controlled transient elastography, depending on data availability.
Practice implications
This correspondence raises a conceptually important methodological challenge: whether population-level, aggregate correction factors derived from validation studies can validly be applied to reclassify individual patients' SLD subtypes. Correction of self-reported alcohol intake in the index study relied on sex- and drinking-frequency-specific factors calibrated to per-capita ethanol consumption at the population level—a technique that may introduce systematic error when applied to individuals, as per-capita calibration does not account for within-category heterogeneity in drinking patterns, socioeconomic variation, or drinking occasion type. The correspondence authors argue that all downstream conclusions of the index study—including the dramatic near-doubling of ALD and MetALD prevalence and the striking mortality attributable fractions—rest on this correction premise and therefore require more cautious interpretation. The findings of the index study, if robust, would have significant implications for clinical practice, as large numbers of patients currently classified as MASLD may carry unrecognized alcohol-related disease with different prognostic implications. However, because only the abstract of the correspondence is available, the specific alternative analyses or data proposed by the correspondence authors are not reported here. Clinicians should be aware that the alcohol thresholds defining MetALD and ALD in current SLD nomenclature are themselves based on arbitrary cut-points, and that the degree to which underreporting correction changes individual-level risk is uncertain. Findings from the index study are generalizable only to the US civilian, non-institutionalized population; institutionalized individuals, those with advanced liver disease causing steatosis regression, and those with incomplete NHANES data were not represented. Practice should not change based on this correspondence alone, but the broader signal—that routine clinical reliance on self-reported alcohol intake likely underestimates alcohol-attributable liver disease burden—is consistent with existing literature. __LIMITATIONS__ 1. Only the abstract of the correspondence is available; the specific criticisms, proposed reanalyses, and quantitative arguments raised by Jaidka et al. cannot be fully evaluated. 2. The correspondence generates no original data; it is hypothesis-generating in nature and cannot independently confirm or refute the index study findings. 3. The index study's core limitation—reliance on population-aggregate correction factors applied at the individual level—may introduce non-differential misclassification that could either over- or underestimate true SLD subtype prevalence. 4. Steatosis ascertainment in the index study was heterogeneous across NHANES cycles (ultrasound, US Fatty Liver Index, or vibration-controlled transient elastography), introducing measurement variability. 5. A substantial proportion of NHANES participants were excluded due to incomplete data, potentially introducing selection bias. 6. Steatosis regression in advanced liver disease may lead to underascertainment of ALD in NHANES. 7. The mortality analyses are restricted to NHANES III (1988–94) and 1999–2014 cycles, limiting contemporaneous applicability. 8. Multiple index study authors disclosed affiliations with pharmaceutical and biotech companies, introducing potential conflicts of interest. 9. The correspondence itself involves one of the same senior authors (Ajay Duseja) who is also a co-author on the index study, which represents a notable context for interpretation. 10. Specific correction factors may not generalize across racial/ethnic subgroups or geographic regions within the USA. __BOTTOM_LINE_CATEGORY__ low_certainty __BOTTOM_LINE__ This correspondence raises a legitimate and clinically important methodological concern: that population-level alcohol underreporting correction factors, when applied to NHANES data, may not validly reclassify individual SLD subtypes and that all downstream prevalence and mortality conclusions of the Younossi et al. index study require cautious interpretation. The broader clinical message—that self-reported alcohol intake systematically underestimates true alcohol exposure and that binge drinking pattern is a critical and underweighted driver of liver-related mortality—remains consistent with the existing evidence base. Clinicians should actively supplement patient self-report with validated tools (e.g., AUDIT-C) and consider emerging biomarkers such as phosphatidylethanol (PEth) when subtype classification has therapeutic or prognostic implications.