Identification of congenital CMV cases in administrative databases and implications for monitoring prevalence, healthcare utilization, and costs

Objective To critically review researchers' use of diagnosis codes to identify congenital cytomegalovirus (cCMV) infection or disease in healthcare administrative databases. Understanding the limitations of cCMV ascertainment in those databases can inform cCMV surveillance and health services research. Methods We identified published studies that used diagnosis codes for cCMV or CMV in hospital discharge or health insurance claims and encounters records for infants to assess prevalence, use of services, or healthcare costs. We reviewed estimates of prevalence and of charges, costs, or expenditures associated with cCMV diagnosis codes. Results Five studies assessed hospitalizations with cCMV diagnosis codes recorded in hospital discharge databases, from the United States (n = 3), Australia (n = 1), and the United Kingdom (n = 1). Six other studies analyzed claims or encounters data from the United States (n = 5) or Japan (n = 1) to identify infants with cCMV codes. Prevalence estimates of recognized cCMV ranged from 0.6 to 3.8 per 10,000 infants. Economic analyses reported a wide range of per-hospitalization or per-infant cost estimates, which lacked standardization or comparability. Conclusions The administrative prevalence of cCMV cases reported in published analyses of administrative data from North America, Western Europe, Japan, and Australia (0.6-3.8 per 10,000 infants) is an order of magnitude lower than the estimates of the true birth prevalence of 3-7 per 1,000 newborns based on universal newborn screening pilot studies conducted in the same regions. Nonetheless, in the absence of systematic surveillance for cCMV, administrative data might be useful for assessing trends in testing and clinical diagnosis. To the extent that cCMV cases recorded in administrative databases are not representative of the full spectrum of cCMV infection or disease, per-child cost estimates generated from those data may not be generalizable. On the other hand, claims data may be useful for estimating patterns of healthcare use and expenditures associated with combinations of diagnoses for cCMV and known complications of cCMV.