We read with interest the recently published comprehensive metaanalysis by Magherman et al. on the association between low to moderate alcohol intake and progressive nonalcoholic fatty liver disease (NAFLD). 1 Non-heavy alcohol consumption was significantly associated with the progression of liver fibrosis, cirrhosis, and hepatocellular carcinoma in NAFLD, while conversely may reduce the occurrence of steatohepatitis. In addition, low alcohol intake showed a protective effect against all-cause mortality. These are in line with more recently published studies that were not included in this meta-analysis. 2,3 These interesting findings suggest that this nuanced relationship may be influenced by methodological limitations and thus lead to heterogeneity that could influence clinical decision-making regarding alcohol abstinence in NAFLD.An important source of heterogeneity is in how alcohol intake is measured. Most studies use various self-report questionnaires or clinical interviews to obtain data on alcohol intake, which could have contributed to part of the heterogeneity and may have introduced recall bias. Therefore, the use of readily available biomarkers that accurately reflect alcohol intake may represent a new approach towards measurement. Blomdahl et al. employed blood phosphatidylethanol (PEth), a biomarker with high sensitivity and specificity that is formed only in the presence of ethanol, to capture alcohol consumption in patients with NAFLD. PEth correlated well with commonly used questionnaires or interviews, and PEth ≥48 ng/mL conferred the highest risk of significant liver fibrosis progression (adjusted odds ratio: 5.9). 2 Another recent study suggested that ethyl glucuronide in hair and urine can be applied as a biomarker for harmful alcohol intake, which demonstrated promising accuracy. 4 Another aspect contributing to heterogeneity can be attributed to incomplete or missing adjustment for potential confounding factors. Metabolic comorbidity 5 and physical activity 6 in patients with NAFLD can have both independent and combined/counteracting effects on the metabolic consequences of alcohol intake. In addition, patients with NAFLD who were alcohol consumers may have different dietary intake compared to non-drinkers, 7 although another study revealed that the association of alcohol intake with the prevalence of NAFLD was independent of dietary patterns. 8 These lines of evidence suggest that sufficient adjustment for confounding factors in included patients will be critical to minimise heterogeneity.Finally, since most of the available studies are retrospective or cross-sectional, causal relationships are hard to draw in the absence of compelling clinical trials.