We used prospective, population-based, laboratory-confirmed active surveillance to define the burden of HMPV in hospitalized young children. The contribution of HMPV to pediatric hospitalization for ARI or fever was substantial with 1.2 hospitalizations per 1000 children <5 years old per year (95% CI, 0.9–1.6) compared with 3.0 per 1000 (95% CI, 2.8–3.4) for RSV, 0.9 per 1000 (95% CI, 0.8–1.1) for influenza virus, and 0.5 per 1000 (95% CI, 0.4–0.6) for PIV3 [23
]. The rate of HMPV-associated hospitalization was highest in infants 0–5 months old at 4.9 per 1000 (95% CI, 2.9–7.2), compared to 16.9 per 1000 (95% CI, 15.3–18.5) for RSV, 4.5 per 1000 (95% CI, 3.4–5.5) for influenza virus, and 1.6 per 1000 (95% CI, 1.0–2.2) for PIV3 [23
]. These data suggest that HMPV causes hospitalizations for respiratory symptoms and fever in children <5 years old at rates similar to those of influenza and parainfluenzaviruses. Extrapolating from US census data, our data would project approximately 27,000 HMPV hospitalizations in children <5 years of age annually, underscoring the impact of HMPV on children. HMPV is rarely detected in asymptomatic young children [2
Although there have been reports of HMPV disease in persons with underlying medical conditions such as prematurity or asthma, few of these studies have been population-based and have enrolled this large number of children [8
]. One-third of the hospitalized children with HMPV in our study had high-risk conditions, suggesting that these children are at higher risk for hospitalization with HMPV infection. However, two-thirds of HMPV hospitalizations occurred in otherwise healthy children. We did not observe an association between HMPV hospitalization and a history of asthma, in contrast to the association between asthma and human rhinoviruses that has previously been reported in this cohort [24
]. Consistent with previous reports, HMPV was associated most frequently with the diagnosis of bronchiolitis [2
]. The presenting symptoms of HMPV in this cohort generally were not distinct from those children with other viruses, consistent with other reports [1
]. However, very few HMPV-infected children (5%) had fever alone without respiratory symptoms.
HMPV was most prominent at both study sites during the late winter and spring months with peak illness rates in March, April and May. Overall, HMPV accounted for up to 15% of all hospitalizations for children less than five years with ARI or fever with a late spring distribution and was more common than influenza or RSV during this time period [23
]. However, the lack of distinctive clinical symptoms and the overlap with other community respiratory viruses highlights the potential utility of rapid sensitive diagnostic tests such as RT-PCR to detect HMPV and other respiratory viruses.
We identified all four subgroups of HMPV during each year and at each study site, although the distribution of subgroups varied by year. The extent of antigenic variability between HMPV subgroups is not clear in animals or in humans [30
], but the presence of multiple HMPV strains in a single season could affect the design of vaccines or prophylactic antibodies if antigenic variability leads to partial immune escape. We did not find that specific subgroups were associated with more severe disease (data not shown); some studies have suggested this phenomenon [42
], while others have not [43
]. Correlation between infecting subtype and disease severity has been shown for RSV [44
]. A postulated mechanism for alternating circulation of different HMPV subgroups is population immune pressure [46
]. Long-term, prospective, population-based surveillance is needed to establish whether different subgroups of HMPV vary in virulence and whether circulation patterns are the result of immune pressure.
Despite the strength of our prospective population-based surveillance system, our study has limitations. First, surveillance was performed for only two years at only two geographically distant sites (representing two US regions, Northeast and South). Greater regional and temporal differences in viral circulation may have been evident if the study included more years or sites, since others have reported substantial year-to-year variability in HMPV prevalence [1
]. Second, institutional differences in medical practice might have contributed to variation in HMPV hospitalization rates, and although enrolled and non-enrolled children did not differ in demographic characteristics, unknown biases may have affected our HMPV burden estimates. Coinfections were present in a minority of children, but the clinical importance of these coinfections is not clear. HRV is identified frequently in asymptomatic individuals and thus especially difficult to interpret as a co-detected virus [48
]. A large number of children in the study did not have a virus identified by the testing used; this could be due to false negatives or the presence of other pathogens not tested for. Nearly all of the HMPV-infected children had ARI symptoms and not fever alone, but we enrolled children with fever alone in addition to children with ARI. Thus, since the population-based incidence was calculated using all hospitalized children including those with fever alone, we may have underestimated the true incidence of HMPV (and other viruses) in children with ARI. Finally, this study only assessed hospitalized children and the impact of HMPV on emergency department and outpatient visits is unknown.