Validation of Administrative Health Data for Identifying Cytomegalovirus Infection After Solid Organ Transplantation

BACKGROUND: Cytomegalovirus (CMV) infection is among the most common infections after solid organ transplantation and leads to high rates of morbidity and mortality. With the proliferation of large, claims-based datasets, there is a need to assess the validity of ICD-10-CM codes in identifying CMV infection. METHODS: We conducted a retrospective cohort study of solid organ transplant recipients (SOTRs) from a regional health system who underwent first transplantation between January 1, 2016, and June 30, 2025, had at least 30 days of post-transplant follow-up, and had known donor/recipient CMV serostatus. We performed cross-sectional and time-to-event analyses comparing the test performance of B25 ICD-10-CM codes in identifying CMV infection, clinically significant CMV infection (csCMVi), CMV disease, and CMV DNA reaching various thresholds. RESULTS: Of 8169 SOTRs, 3084 (37.8%) had B25 coded during follow-up. In addition, 4287 (52.5%) developed CMV infection, 2356 (28.8%) csCMVi, 297 (3.6%) CMV disease, and 1669 (20.4%) CMV DNA ≥ 1000 IU/mL. B25 codes were most specific to any CMV infection (88.5%), most sensitive to CMV DNA ≥ 100 000 IU/mL (98.2%), and csCMVi had the highest Youden's Index (0.651; sensitivity 84.1%, specificity 81.0%). Multivariable Cox regression provided similar results but with variations in effect measure between ICD-10-CM codes and clinical CMV definitions. CONCLUSIONS: ICD-10-CM codes showed modest performance in identifying CMV infection, most closely resembling csCMVi. ICD-10-CM codes poorly reflect CMV severity, however, and may overestimate CMV infection by some clinical definitions. These limitations should be considered when using CMV diagnosis codes for claims-based studies of SOTRs.