Abstract
BACKGROUND AND AIMS: Despite causing significant morbidity and mortality, nonalcoholic fatty liver disease (NAFLD) is underdiagnosed. Clinical indices developed to identify hepatic steatosis are often used by providers but their potential for use at the population level remains unexplored. We assessed clinical phenotypes for their ability to identify potential patients with NAFLD and nonalcoholic steatohepatitis (NASH) in the electronic health record. METHODS: We conducted a single-center retrospective cohort study of adult patients from January 1, 2016, to December 31, 2022. We developed 4 phenotypes: clinical NAFLD (C-NAFLD), clinical NASH (C-NASH), NAFLD with diagnosis (D-NAFLD) and NASH with diagnosis (D-NASH) and compared characteristics across them to identify differences between patients with and without International Classification of Diseases diagnoses. RESULTS: Each of the (C) phenotypes identified a cohort of patients who had clinical evidence suggestive of disease without a documented diagnosis. Black patients were overrepresented in the (C) relative to (D) groups (C-NAFLD 24.3% vs D-NAFLD 21.2%; C-NASH 28.5% vs D-NASH 14.0%). Patients with D-NASH were more likely to be prescribed medications that may be effective in treating NAFLD-NASH spectrum disease, ie, glucagon-like peptide 1 receptor agonists (C-NASH 5.0% vs D-NASH 16.7%, P < .001). Fewer patients with D-NASH had cardiovascular (C-NASH 58.0% vs D-NASH 46.3%, P < .001) and heart failure (C-NASH 33.9% vs D-NASH 24.8%, P < .001) hospitalizations than those with C-NASH. CONCLUSION: Noninvasive clinical indices may improve identification of patients with or at risk for NAFLD-NASH at the population level. Systematic differences between populations with and without International Classification of Diseases diagnoses of NAFLD-spectrum disease suggest disparities in the application of screening and diagnostic procedures.