Assessing the Performance of CCCv3 in Epic Cosmos: A Foundation for Future Innovation

OBJECTIVE: Children with medical complexity (CMC) are a heterogeneous population with chronic multisystem conditions, high utilization, and poor outcomes. While large administrative datasets (eg, Pediatric Health Information [PHIS], Kids' Inpatient Database [KID], Medicaid claims) have characterized these populations, they often lack clinical detail. We evaluated the performance of the Pediatric Complex Chronic Condition System version 3 (CCCv3) in Epic Cosmos (Epic Systems Corp), a national deidentified electronic health record dataset with potential to expand CMC research. METHODS: We conducted a retrospective cohort study of pediatric hospitalizations (patients ≤18 years old) in Cosmos between 2015 and 2025. CMC were identified using CCCv3 applied to admission, discharge, billing, and problem list diagnoses. We described complex chronic condition (CCC) distribution, hospitalization burden, and demographics. Associations between CCC count and outcomes were measured using negative binomial regression for length of stay (LOS) and modified Poisson regression for in-hospital mortality. RESULTS: Among 4 524 010 admissions, 43.3% of patients had at least 1 CCC, accounting for 55.8% of hospitalizations and 69.0% of hospital days. Patients with 4 or more CCCs represented 4.4% of the cohort but accounted for 14.3% of admissions and 27.2% of hospital days. Each additional CCC was associated with a 21% increase in LOS (incidence risk ratio, 1.21; P < .001). Mortality risk increased stepwise with each additional CCC. Compared with administrative datasets, Cosmos identified a higher absolute prevalence of CCCs but preserved relative trends in utilization and outcomes. CONCLUSIONS: CCCv3 applied in Cosmos stratifies patient complexity in patterns consistent with prior administrative studies, despite higher absolute prevalence. Recognizing this classification system can be used in Cosmos adds a clinically rich platform available for advancing pediatric research and clinical decision support.