P-512. Assessing the Validity of ICD-9/10 Codes and Microbiologic data for Congenital CMV Diagnosis in an Integrated Health System

Abstract Background Universal screening at birth for cytomegalovirus (CMV) detects congenital CMV (cCMV) in 0.4% to 1% of infants. Accurate coding of cCMV is necessary to identify cCMV in large databases. In a large integrated healthcare delivery system, we aimed to evaluate the accuracy of ICD-9/10 codes and microbiologic data in identifying children with cCMV.Table 1:Utility of cCMV diagnosis codes in predicting confirmed or probable cCMVTable 2:Utility of CMV diagnosis codes in first 45 days of life in predicting confirmed or probable cCMV Methods Children born between 2011-2020 with a cCMV diagnosis (ICD-9 771.1, ICD-10 P35.1) in childhood, a CMV diagnosis (ICD-9 078.5, ICD-10 B25) within 45 days of life, or viral testing for CMV under age 2 years were reviewed for clinical, laboratory and radiographic evidence of cCMV infection. Virologic studies included CMV antibody, CMV polymerase chain reaction (PCR) and viral / CMV culture. Children were categorized as cCMV confirmed, probable, unlikely or ruled out.Table 3:Utility of either cCMV diagnosis codes and / or CMV diagnosis codes in first 45 days of life in predicting confirmed or probable cCMV Results In a birth cohort of 401,043, 69 children (0.02%) had confirmed (48) or probable (21) cCMV. Our cohort of 329 children included 80 with a cCMV diagnosis code, 62 with a CMV diagnosis code in the first 45 days of life and 226 with CMV viral testing in the first 2 years of life. Of the 80 children with a cCMV diagnosis code, 47 (59%) had confirmed or probable cCMV (Table 1). Of 22 children with cCMV without a cCMV diagnosis code, 8 had a CMV diagnosis code including 7 with confirmatory CMV PCR in the first 21 days of life and 1 with microcephaly and positive CMV PCR at age 42 days. The 14 children without either a cCMV or CMV diagnosis code had abnormal CMV testing and symptoms of cCMV including microcephaly, sensorineural hearing loss, abnormal brain imaging, and/or, before postnatal day 21, thrombocytopenia or elevated direct bilirubin. The sensitivity and specificity of cCMV diagnosis codes were 68% and 87%, respectively (Table 1). The sensitivity and specificity of CMV diagnosis codes in the first 45 days of life were 49% and 89%, respectively (Table 2). Using either cCMV and/or CMV diagnosis codes the sensitivity was 80% and specificity was 82% (Table 3). Conclusion cCMV diagnosis codes appear unreliable: missing 32% of children with cCMV and falsely identifying 13% of children without evidence of cCMV. To improve case ascertainment, cCMV and CMV diagnoses and symptoms recorded in large databases should be confirmed with virologic evidence. Disclosures Tara L. Greenhow, MD, Moderna Therapeutics, Inc.: Grant/Research Support Gail J. Demmler-Harrison, MD, Elsevier: Royalties|Merck: Grant/Research Support|Microgen: Advisor/Consultant|Microgen: Grant/Research Support|Moderna: Advisor/Consultant|Moderna: Grant/Research Support|WEB MD Medscape: Honoraria|WEB MD Medscape: travel|Wolters Kluwer UpToDate: Royalties Yinge Qian, PhD, Moderna Therapeutics, Inc.: Grant/Research Support John D. Diaz-Decaro, PhD, MS, ModernaTX, Inc: Stocks/Bonds (Private Company) Colin Kunzweiler, PhD, Moderna Therapeutics, Inc.: Current employee|Moderna Therapeutics, Inc.: Stocks/Bonds (Public Company) Lisa A. Croen, PhD, Moderna Therapeutics, Inc.: Grant/Research Support