Gastroenterology · Medical Literature
[Correspondence] The spectrum of steatotic liver disease in the USA – Authors’ reply
Abstract
We thank Kartik Jaidka and colleagues and Hans Scherübl for their thoughtful comments on our recent Article.1
Clinical relevance
In a US population-based analysis using NHANES data spanning 1988–2023, does correcting for alcohol underreporting materially alter the estimated prevalence and premature mortality burden across the steatotic liver disease (SLD) spectrum — specifically for MASLD, MetALD, and ALD — and what are the principal risk factors for premature mortality within each SLD subtype? __SNAPSHOT__ {"studyDesign":"Authors' correspondence reply (Letter) in The Lancet Gastroenterology & Hepatology, responding to reader commentaries on the primary article: Younossi ZM, Krag A, Zelber-Sagi S, et al. 'Binge drinking, metabolic dysfunction, and the spectrum of steatotic liver disease in the USA: a cross-sectional and longitudinal analysis.' Lancet Gastroenterol Hepatol. 2026 Mar 23. The reply itself contains no new data.","setting":"USA; population-based analysis using all available National Health and Nutrition Examination Survey (NHANES) cycles, 1988–2023. Affiliated with the Global Center for Liver Outcomes & Policy Research (G-CLORP), Georgetown University School of Medicine, Washington, DC.","sampleSize":"Prevalence cohort: n=41,100 participants; mortality cohort: n=24,707 participants (derived from NHANES III [1988–94] and NHANES 1999–2014, linked to the National Death Index through December 31, 2019).","population":"US adults aged ≥20 years who completed both the interview and examination components of NHANES and had sufficient data to define SLD subtype and alcohol consumption by self-report. The original article used the 2023 multi-society Delphi consensus nomenclature (MASLD, MetALD, ALD) and applied a correction algorithm for alcohol underreporting.","intervention":"Correction for self-reported alcohol underreporting to reclassify SLD subtypes and re-estimate prevalence and mortality across the MASLD–MetALD–ALD spectrum; longitudinal assessment of SLD subtype-specific premature mortality risk factors (including type 2 diabetes and binge drinking pattern).","comparator":"Uncorrected self-reported alcohol intake (standard classification) vs. alcohol-underreporting-adjusted classification. Mortality compared against a reference group of non-SLD abstainers.","primaryOutcome":"Prevalence trends of MASLD, MetALD, and ALD over time (1988–2023) before and after alcohol underreporting correction; age-standardized premature mortality rates and adjusted hazard ratios by SLD subtype.","followUp":"Median follow-up 17.4 years (410,293 person-years) in the mortality cohort.","fundingSource":"Not reported in the correspondence reply. The original article disclosed that some authors declared affiliations with biotech, pharmaceutical, and/or device companies."}
Key findings
The following findings are attributed to the primary original article (Younossi et al., Lancet Gastroenterol Hepatol, March 2026), as the August 2026 correspondence reply (Paik and Younossi) contains no new primary data — only acknowledgement of reader comments. The correspondence reply's abstract consists solely of a statement thanking correspondents. From the original article: After applying alcohol underreporting correction, the prevalence of MetALD and ALD approximately doubled compared to unadjusted estimates. In the 2021–23 NHANES cycle, ALD prevalence was 4.59% (95% CI 3.94–5.25%) with correction vs. 1.65% (95% CI 1.19–2.11%) without; MetALD prevalence was 4.10% (95% CI 3.38–4.82%) with correction vs. approximately 2.14% (95% CI 1.69–2.59%) without. Among participants with SLD after correction, 84.58% had MASLD, 11.56% had MetALD, and 3.86% had ALD. MASLD prevalence increased from 12.69% (95% CI 11.38–13.99%) in 1988–90 to 28.16% (95% CI 26.07–30.25%) in 2021–23 (P<0.0001); MetALD increased from 1.62% to 4.10% and ALD from 2.28% to 4.59% over the same period (all P<0.0001). In the mortality cohort, age-standardized mortality rates were 7.86 per 1,000 person-years (95% CI 6.33–9.48) for MASLD, 8.74 per 1,000 person-years (95% CI 4.30–13.82) for MetALD, and 14.91 per 1,000 person-years (95% CI 7.63–23.81) for ALD, compared with 4.76 per 1,000 person-years for non-SLD abstainers. Adjusted hazard ratios for premature mortality were 1.39 (95% CI 1.10–1.76) for MASLD, 1.45 (95% CI 1.03–2.04) for MetALD, and 2.21 (95% CI 1.45–3.37) for ALD. Type 2 diabetes was a consistent independent mortality risk factor within MASLD subgroups. Binge drinking was identified as a leading mortality risk factor across SLD subtypes. No NNT or NNH is calculable from the reported data. No specific adverse event data were reported, as this was an observational epidemiologic study.
Practice implications
CRITICAL LIMITATION: The document submitted for summary is a brief correspondence reply (letter format, Lancet Gastroenterology & Hepatology, August 2026) whose entire abstract consists of a single sentence acknowledging two sets of reader commentaries on the authors' prior original article. No new data, analyses, or conclusions are presented in the correspondence reply itself. All quantitative findings cited in this summary are drawn from the referenced primary article (Younossi et al., March 2026), reconstructed from secondary reporting, not from a full-text review of the original paper. Clinicians should consult the primary article directly before making clinical or policy decisions. What the original article established: Alcohol underreporting significantly distorts the apparent epidemiology of SLD in the USA. When corrected, the burden of MetALD and ALD is approximately double the conventionally reported estimates, with important implications for mortality surveillance and public health prioritization. The finding that ALD historically appeared to affect only ~1% of the general population despite being the leading cause of cirrhosis and liver transplantation in the USA and Europe is a recognized paradox — this study provides population-level evidence that systematic underreporting of alcohol intake is a major driver of this discrepancy. What was not established: Causality cannot be inferred from this cross-sectional and observational longitudinal design. The correction algorithm for alcohol underreporting relied on population-level assumptions rather than direct biomarker measurement (e.g., phosphatidylethanol [PEth]), limiting its precision at the individual level. MASLD was diagnosed using different modalities across NHANES cycles, introducing heterogeneity in steatosis ascertainment. A substantial proportion of participants was excluded due to incomplete data, potentially introducing selection bias. Steatosis can regress in advanced liver disease, which may underestimate true SLD prevalence, particularly for ALD. Generalizability: Findings apply broadly to US adults as represented by NHANES, a nationally representative survey. Findings may not generalize to healthcare settings, non-US populations, or individuals not captured by NHANES sampling. Clinical importance: The approximately two-fold underestimation of MetALD and ALD is clinically significant and directly relevant to risk stratification, coding practices, and resource allocation in hepatology and primary care. The identification of type 2 diabetes and binge drinking as leading mortality risk factors reinforces the need for integrated metabolic and alcohol use assessment in all patients presenting with liver disease. __LIMITATIONS__ 1. Publication type: This summary is based on a correspondence reply whose abstract contains no primary data; findings are reconstructed from the underlying original article via secondary sources, not full-text review — a critical methodological limitation of this summary itself. 2. Observational design: Cross-sectional prevalence analyses preclude causal inference. Longitudinal mortality data are subject to residual confounding despite multivariable adjustment. 3. Alcohol ascertainment: Self-reported alcohol intake is subject to systematic recall bias and social desirability bias; the underreporting correction algorithm relies on population-level assumptions rather than individual-level biomarkers (e.g., PEth, carbohydrate-deficient transferrin), reducing precision for individual risk reclassification. 4. Heterogeneous steatosis ascertainment: MASLD was diagnosed using different available modalities across NHANES cycles (ultrasound, controlled attenuation parameter), introducing systematic measurement variability over time. 5. Incomplete data exclusions: A substantial proportion of participants was excluded for incomplete data, with potential for non-random selection bias. 6. Steatosis regression: Steatosis may be absent in advanced liver disease (burnt-out ALD/cirrhosis), likely causing underestimation of true SLD prevalence, particularly in the ALD subgroup. 7. Mortality follow-up truncation: The mortality analysis is restricted to NHANES through 2014 linked to the National Death Index through 2019; more contemporary SLD burden and treatment effects are not captured. 8. Industry affiliations: Some authors declared affiliations with pharmaceutical and biotech companies; potential for selective framing of results cannot be excluded. 9. Generalizability: NHANES captures the non-institutionalized US civilian population; results may not apply to hospitalized, incarcerated, or homeless individuals who carry disproportionate ALD burden. __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ Correcting for alcohol underreporting in US population-based data reveals that the true burden of MetALD and ALD is approximately double conventionally reported estimates, with ALD conferring a more than two-fold increased risk of premature mortality compared to non-SLD abstainers. Type 2 diabetes and binge drinking are leading mortality risk factors across SLD subtypes, underscoring the need for integrated metabolic and alcohol use assessment in all SLD patients. This correspondence reply itself adds no new data; clinicians should consult the primary article (Younossi et al., Lancet Gastroenterol Hepatol, March 2026) directly, and confirmatory studies incorporating objective alcohol biomarkers (e.g., PEth) are needed before individual-level reclassification algorithms can be applied in routine clinical care.