In the United States, an estimated 28,000 (0.7%) infants are born with congenital cytomegalovirus (CMV) infection annually, 3,600 (12.7%) of whom are estimated to be symptomatic at birth [6]. Symptomatic congenital CMV disease is characterized by petechiae, purpura, thrombocytopenia, hepatosplenomegaly, direct hyperbilirubinemia, microcephaly, intracranial calcifications, chorioretinitis, and hearing impairment [4, 10]. To confirm a diagnosis of congenital CMV infection, CMV must be detected in urine, saliva or blood within 2–3 weeks of birth [19, 27]. In a study of dried blood samples collected for newborn screening, the prevalence of congenital CMV infection was significantly higher among preterm and low birth weight infants [13]. As such, the prevalence of congenital CMV infection among infants admitted to neonatal intensive care units (NICUs) is likely higher than that of the overall population [22, 26, 28].

A subset of the NICU population, infants born <32 weeks gestational age or with a birth weight <1500g, is at risk for symptomatic CMV disease [7] from infection acquired peri- or postnatally through exposure to infected maternal genital secretions, breast milk, or blood transfusion [32]. The risk of CMV transmission by blood transfusion can be reduced by the use of CMV-seronegative or leukocyte-reduced blood components [1, 18, 35], which have become the standard of care for premature infants in the United States [1, 11, 31]; most CMV infections in neonates are likely transmitted by breast milk [7, 17, 34]. The clinical characteristics associated with postnatally-acquired CMV infection include hepatopathy, thrombocytopenia, neutropenia, petechiae, respiratory distress syndrome, and sepsis-like syndrome [7, 34].

Limited data are available on the burden of CMV disease in infancy and the extent to which it is severe enough to require hospitalization in NICUs. This study describes demographic and clinical characteristics of infants with reported CMV infection in California NICUs during 2005–2010 and estimate the proportion of expected live births with symptomatic congenital or acquired CMV disease identified in California NICUs during the study period.

During 2005–2010, 156 of 91,435 infants in CPQCC-member hospital NICUs were reported as having CMV infection; 87 of 33,588 VLBW infants and 69 of 57,847 infants >1500g. This corresponds to an overall prevalence of 1.7 infants reported with CMV infection per 1000 live births requiring NICU care; 2.6 per 1000 among VLBW infants and 1.2 per 1000 among infants >1500g. The prevalence of CMV infection increased from 1.1 to 2.5 per 1000 live births during 2005–2010 with marked increases among VLBW infants (1.5 to 4.6 per 1000 live births) (Figure 1). The prevalence of CMV infection was higher among infants with younger gestational age (<32 weeks: 2.4 per 1000, 32–36 weeks: 1.4 per 1000, and ≥37 weeks: 1.2 per 1000), and among infants with lower birth weight (<1500g: 2.6 per 1000, 1501g–2500g: 1.8 per 1000, and >2500g: 0.8 per 1000) (P<0.01).

The majority of infants reported with CMV infection were born to mothers 20–29 years of age (44%) (Table 1). The prevalence of CMV infection was highest among infants born to mothers <20 years of age, 1.8-fold that of infants born to mothers 30–39 years of age. The highest prevalence of CMV infection was observed among infants born to Asian/Pacific Islander mothers (3.4 per 1000 live births), followed by those born to Hispanics (1.8 per 1000 live births) and non-Hispanic blacks (1.7 per 1000 live births). Among VLBW infants, the prevalence of CMV infection among infants born to Asian/Pacific Islander mothers was nearly three-fold that of infants born to non-Hispanic white mothers (PR=2.9; 95%CI-1.5–5.5). Among infants >1500g, the prevalence of CMV infection among infants born to Hispanic mothers was nearly two-fold that of infants born to non-Hispanic white mothers (PR=1.9; 95%CI=1.1–3.5).

Overall, IUGR was diagnosed in 25 (17%) infants reported with CMV infection (Table 2), and was more common than in infants without CMV infection (Table 2). Among infants >1500g, 15 (23%) with CMV infection had anomalies detected prior to birth (brain anomalies, cardiac defects, or gastrointestinal malformations) two times as many as those without CMV infection (P<0.05). The proportion of infants with Apgar scores ≤3 did not differ by CMV status, even after stratification by birth weight. VLBW infants with CMV infection were more likely than those without CMV infection to have respiratory distress syndrome (OR=2.5; 95%CI=1.3–4.7) (P<0.01). Although only two infants reported with CMV infection had microcephaly listed as a congenital anomaly, 19 (14%) infants with CMV infection were determined to have head circumference ≤3^rd percentile, and 42 (31%) had microcephaly (head circumference ≤ 10^th percentile). Microcephaly was more common among infants with CMV infection (P<0.05).

Among infants reported with CMV infection, 99% of VLBW infants and 81% of infants >1500g were hospitalized in a NICU beginning on in the 1^st day of life. Fourteen (9%) infants reported with CMV infection died, including 8 (9%) of the VLBW infants and 6 (9%) of the infants >1500g (Table 2), at a median of 38 (range: 3–99) and 9 (range: 3–107) days of life, respectively. Among infants >1500g, the frequency of death was significantly higher among those with CMV infection. Among surviving infants with CMV infection, the median length of hospital stay was 99 (interquartile range: 72–125) days for VLBW infants and 48 (18–76) days for infants >1500g, which was significantly longer than for those without CMV infection, 60 (42–86) days for VLBW infants and 14 (8–27) days for infants >1500g (P<0.01).

Of a total of 3 272 565 live births in California during 2005–2010, an estimated 3834 would have symptomatic CMV disease - 2909 live births with symptomatic congenital CMV disease, and 925 VLBW live births with symptomatic acquired CMV disease (Table 3). The total number of live births with symptomatic CMV disease identified in California NICUs extrapolated from CPQCC data (n=179) represented 5% of the expected number of live births with symptomatic CMV disease in the newborn period during 2005–2010; this proportion increased from 3% in 2005 to 7% in 2010.

Limited data are available on the burden of disease caused by congenital or acquired CMV infection among infants in the NICU setting [22, 26, 28]. While congenital CMV infection can result in severe disease [4, 10, 15], the proportion of infants with disease severe enough to require NICU care is unknown. The overall reported prevalence of CMV infection was 1.7 per 1000 live births requiring NICU care in California, higher than the estimated prevalence of symptomatic congenital CMV disease in the general population (0.9 per 1000 live births) [6]. However, the proportion of infants identified with CMV infection in NICUs represents only a fraction of the total expected burden from symptomatic CMV disease, likely resulting from milder symptomatic cases that do not require NICU care. Nonetheless, CMV infection was associated with prolonged NICU stays, higher mortality rates, and neurological abnormalities.

Data on the prevalence of congenital or acquired CMV infection among NICU infants based on screening studies are limited. A study in Canada [33] found a prevalence of congenital CMV infection among VLBW infants of 1.5%. Another study in Spain [2] diagnosed 2.3% of VLBW infants with congenital CMV infection and 10.2% with peri- or postnatally acquired CMV infection. In our study, the true prevalence of CMV disease in NICUs was likely underestimated because even clinically apparent CMV infections may not be suspected or tested for in a timely manner, laboratory techniques have variable performance depending on the specimen type, and not all infections diagnosed may have been reported. Typical practice for considering and diagnosing CMV infection among NICU infants is currently unknown; the increasing prevalence of CMV infection in recent years of CPQCC data could reflect changing practices in CMV testing, diagnosis, and/or reporting, and merits further investigation. Some of the infants reported in our study may have had asymptomatic CMV infection. Our major limitation was the inability to classify reported infants as having congenital or acquired CMV infection. Because acquired CMV infection might occur in up to approximately 10% of VLBW infants [2, 17], screening VLBW infants for CMV infection at birth would allow for distinguishing between congenital and acquired CMV infection, which have different prognoses for long-term sequelae and might simplify follow-up of these infants.

In our study, the prevalence of CMV infection increased with younger gestational age and lower birth weight. In addition, CMV infection was associated with IUGR both in VLBW infants and infants >1500g. The mechanisms by which CMV infection may be associated with prematurity and IUGR are not completely understood. Altered placental cytokine expression is associated with IUGR, prematurity, and early membrane rupture in uninfected pregnancies [3]. Recent studies suggest that CMV infection could result in proinflammatory changes with important consequences for placental development and function, virus transmission, and fetal viability [9, 30]. Understanding immune mechanisms at the level of the placenta associated with CMV transmission is an active area of research with the potential to aid the development of vaccines and immunotherapies [29].

Microcephaly is a condition frequently reported among infants with congenital CMV disease [4, 10], but it was reported as a congenital anomaly for only 2 infants with CMV infection in our study. When the criteria of head circumference ≤10^th percentile using appropriate population-based growth charts was applied, 31% of infants with CMV infection were classified as having microcephaly. Microcephaly was more common among infants with reported CMV infection compared to those without CMV infection both among VLBW infants and infants >1500g. While nearly all VLBW infants were routinely screened for brain abnormalities, among infants >1500g, it was more likely among those reported with CMV infection than those without, suggesting that this group more likely has true congenital CMV infection for which evaluation for brain abnormalities might be a routine practice. The role of neuroimaging is likely to evolve as more experience is gained with use of antiviral treatments for infants with congenital CMV-related neurologic abnormalities.

In populations with high maternal CMV seroprevalence, the birth prevalence of congenital CMV infection tends to be higher as does the acquisition of CMV through breastfeeding [5, 12]. A study using dried blood spots from a representative sample of newborns in California found the highest maternal IgG CMV seroprevalence in Hispanics (89%), followed by Asians (87%), blacks (77%) and whites (57%) [13]. In our study, the prevalence of CMV infection was highest among infants born to Asian/Pacific Islander mothers (3.4 per 1000), followed by Hispanics (1.8 per 1000), non-Hispanic blacks (1.7 per 1000), and non-Hispanic whites (1.2 per 1000). Information on country of origin and other maternal sociodemographic variables were not available which could have helped to better understand the higher prevalence of CMV infection observed among infants born to Asian/Pacific Islander mothers.

Population-based studies are needed to better understand the spectrum of symptomatic congenital and acquired CMV disease as well as the sub-groups of infants in whom disease is most prevalent. Understanding the proportion of cases of congenital and acquired CMV infection that are severe and require NICU care is important for assessing the burden and costs of CMV disease in infancy. In the absence of universal newborn screening most cases of congenital CMV infection, both asymptomatic and symptomatic, are likely to be missed. Newborn screening for CMV has the potential to identify infants at risk of hearing loss and cognitive delays although more data are needed on the interventions these infants would need and the benefit that prompt identification through early intervention could provide. More complete ascertainment of infants with symptomatic congenital CMV disease may facilitate appropriate selection of infants with central nervous system disease who would benefit from treatment with antivirals [14, 16, 23, 24]. In addition, data on burden of acquired CMV disease in VLBW and premature infants are important for guiding use of breast milk in this population [17].