Abstract
BACKGROUND AND AIMS: The estimated prevalence of metabolic dysfunction-associated steatotic liver disease (MASLD) in US adults is 31.3%. However, little is known about MASLD diagnosis rates in clinical practice. We used data from electronic health records and assessed the annual MASLD prevalence. METHODS: We conducted a repeated cross-sectional analysis of a retrospective dynamic cohort study among 2,040,688-2,852,060 Kaiser Permanente Northern California adult members without other chronic liver diseases from 2009 to 2018. Annual MASLD prevalence was identified based on International Classification of Diseases, Ninth or Tenth Revision, Clinical Modification diagnosis codes, the application of natural language processing of all radiology imaging report text that included the liver, and the application of the Dallas Steatosis Index, a MASLD prediction algorithm. RESULTS: Between 2009 and 2018, the estimated MASLD prevalence ranged from 0.37% to 0.95% using diagnosis codes, 0.88%-1.37% using imaging, and 6.14%-11.27% using the Dallas Steatosis Index. CONCLUSION: Reliance on diagnosis codes or natural language processing on hepatic imaging, or the Dallas Steatosis Index, to identify MASLD patients within a large health-care system yields much lower than expected prevalence. These results underscore an underdiagnosis issue and highlight the need for improved diagnostic practices.