Purpose <p>Despite the introduction of new eating disorder diagnoses in 2013 with the DSM-5, the characterization of population-level trends and sociodemographic patterns of these expanded eating disorder definitions remain limited. This study examined temporal changes in eating disorder diagnoses and sociodemographic characteristics among adolescents and young adults between the years 2015 and 2023 using the Epic Cosmos database, an aggregate of electronic health records across various healthcare systems.</p> Methods <p>This is a retrospective cohort study using Epic Cosmos data, identifying individuals aged 9&#xa0;years–25&#xa0;years with an ICD-10 code consistent with an eating disorder diagnosis. Adjusted multivariable regression models assessed temporal changes and associations between sociodemographic characteristics and specific eating disorder diagnoses.</p> Results <p>Among 195,015 individuals, the most common eating disorder diagnoses during the study period were unspecified eating disorder (31%), other specified eating disorder (20%), and anorexia nervosa (20%). Avoidant restrictive food intake disorder had the fastest annual increase followed by binge eating disorder. Significant differences in diagnostic trends were observed by sex, race, ethnicity, and preferred language, with notable increases in avoidant restrictive food intake disorder among females and in anorexia nervosa among Hispanic/Latino and non-English-speaking individuals.</p> Conclusion <p>The diagnosis patterns of eating disorders are changing, with increases in avoidant restrictive food intake disorder and binge eating disorder as well as changes based on sociodemographic characteristics. These findings highlight the need for improved and continued surveillance of the “newer” eating disorder diagnoses and warrant further investigation for treatment capacity and long-term outcomes.</p> Level of evidence <p>Level V, based on descriptive study.</p>