PCR-based Detection of Cytomegalovirus in Neonates and Children, with Emphasis on Congenital and Neonatal-Screening Cohorts: A Systematic Review and Meta-Analysis

Background:Cytomegalovirus (CMV) is one of the commoner infections of childhood and the leading non-genetic cause of sensorineural hearing loss. Diagnosis in children is not straightforward. Serology is hard to interpret in early life and viral culture is slow and often insensitive, and neither method suits testing at scale. CMV detection using PCR is easier, but sample type, format, and age affect accuracy. We reviewed how well PCR-based CMV testing performs in neonates and children, and how much the available studies differ from one another. Methods:The Cochrane Library, Web of Science, Scopus and PubMed were searched for English-language reports published between January 2000 and February 2025. Studies were eligible if they reported PCR-based CMV detection in saliva, urine, dried blood spots, whole blood, plasma, serum or amniotic fluid; accuracy was then compared across these specimen types. Quality was assessed with QUADAS-2. Pooled sensitivity, specificity and false-positive rate came from random-effects and bivariate Reitsma models, and further analyses were run by specimen type, age group and test format. Results:We identified 52 studies examining PCR-based CMV detection from neonates to 18 years of age. We analyzed diagnostic accuracy, specimen type performance, prevalence, PCR techniques, geographical distribution, and risk of bias using QUADAS-2. Seven 2×2 tables drawn from five publications, representing four independent research programmes, yielded a pooled sensitivity of 0.898 (95% CI: 0.489-0.988) and a pooled specificity of 0.998 (95% CI: 0.975-1.000), with an area under the summary ROC curve of 0.991. Results for saliva were the most consistent (sensitivity 0.974-1.000). Dried blood spots varied much more, from 0.283 under universal newborn screening to 0.953 where testing was directed at high-risk neonates. Between-study heterogeneity was substantial overall (I2 94.2% for sensitivity and 98.7% for specificity) but was concentrated in dried blood spots and universal screening; saliva-based and targeted testing subgroups were homogeneous (I2 0%). For pooled CMV positivity, heterogeneity reached 99.5%. Patient selection and interpretation of the index test were the weakest QUADAS-2 domains. Conclusion:Real-time PCR on saliva or urine appears accurate enough to be useful in neonatal CMV screening. The pooled estimates should still be read cautiously, given how much the studies differed and the lower sensitivity recorded for dried blood spots.