Human metapneumovirus (HMPV), a virus with genetic, epidemiological, and clinical similarities to respiratory syncytial virus (RSV), was first identified in 2001 among children with acute respiratory infections (ARI) in the Netherlands.1 The prevalence of HMPV infection among ARI cases has been described in North America,2 South America,3 Asia,4 the Middle East,5 Africa,6 and Oceania.7 The first report of HMPV infections in Central America was of a pediatric case series (n = 9) in Costa Rica during 2008.8 A study among both outpatients (n = 1573) and inpatients (n = 183) with influenza-like illness in El Salvador, Honduras and Nicaragua from August 2006 to April 2009 detected only three HMPV infections using immunofluorescence.9 Schlaudecker et al.10 used a multiplex polymerase chain reaction test and found 28 (8%) cases positive for HMPV out of 345 ARI cases in children presenting to outpatient clinics in Honduras from February 2010 to June 2011.

Evidence from previous studies has been inconsistent regarding whether ARI cases that are HMPV-positive tend to be more or less severe than HMPV-negative cases. Studies have shown greater likelihood of a clinical pneumonia diagnosis and need for mechanical ventilation associated with HMPV compared with HMPV-negative cases,11,12 but another study showed that children with HMPV were less likely to require admission to an intensive care unit (ICU) than those with RSV or influenza A.2 Other studies comparing HMPV and RSV infections did not find any evidence of a difference in the probability of requiring mechanical ventilation, having longer hospital stay,12,13 or being admitted to the hospital or ICU.14 However, as these studies did not adjust for age, it is unclear to what extent these comparisons are biased by differences in the age distributions of the children infected with each virus.15

In this manuscript, we present estimates from Guatemala of population-based incidence rates of hospitalized cases of ARI that have HMPV infection (HMPV-ARI), and we use comparisons with cases RSV-positive and those negative for both HMPV and RSV (HMPV-RSV-negative) to highlight important features of the epidemiology of HMPV infection, particularly, seasonality, age distribution and disease severity.

We identified 7289 eligible ARI cases, 6597 (91%) were enrolled, and 6288 (95%) of those enrolled had a specimen tested. Of these, rRT-PCR assays detected HMPV in 596 (9%) and RSV in 1485 (24%) cases; 43 (0·7%) HMPV-RSV co-infections were identified. Of the HMPV-positive cases, 87% occurred in children aged <5 years, similar to proportion of the RSV-positive cases (91%), whereas only about half of the HMPV-RSV-negative cases occurred in this age group (Table1).

Of the 2295 HMPV-RSV-negative cases in children aged <5 years, we detected the following viruses: 180 (8%) influenza A, 43 (2%) influenza B, 99 (5%) parainfluenza virus type 1, 28 (1%) parainfluenza virus type 2, 251 (12%) parainfluenza virus type 3, and 293 (14%) adenovirus. Of the 1953 HMPV-RSV-negative cases in persons aged ≥5 years, we detected the following viruses: 153 (8%) influenza A, 30 (2%) influenza B, 28 (1%) parainfluenza virus 1, 10 (1%) parainfluenza virus 2, 62 (3%) parainfluenza virus 3, and 118 (6%) adenovirus. At least one other virus was also detected in 127 (24%) of the HMPV-positive cases in children aged <5 years and in 13 (17%) of the HMPV-positive cases among persons aged ≥5 years. At least one other virus was also detected in 271 (20%) of the RSV-positive cases in children aged <5 years and in 33 (26%) of the RSV-positive cases among persons aged ≥5 years.

Among children aged <5 years with ARI, the median age was older for those HMPV-positive (1 year) compared with those RSV-positive (6 months). There was no major difference in the respiratory signs and symptoms, presence of fever, or radiological diagnoses between the three etiological groups presented in Table1. Among children aged <5 years, hospital course (intensive care, mechanical ventilation, length of stay) and mortality proportions suggest that severity is lowest for HMPV-positive, intermediate for RSV-positive, and greatest for HMPV-RSV-negative. Among those aged ≥5 years, there was no consistent association between these indicators of severity and etiological group.

In Santa Rosa during the years 2008, 2009, 2010, 2011, and 2012, we detected HMPV in 42 (14%), 5 (1%), 75 (15%), 16 (3%), and 43 (10%) of ARI cases and RSV in 33 (11%), 242 (38%), 81 (17%), 161 (33%), and 9 (2%) of ARI cases, respectively. In Quetzaltenango during the years 2009, 2010, 2011, and 2012, we detected HMPV infection in 21 (3%), 111 (13%), 31 (7%), and 37 (8%) of ARI cases and RSV in 223 (34%), 195 (22%), 118 (26%), and 69 (15%) of ARI cases, respectively. In Guatemala City during 2010, we detected HMPV infection in 195 (16%) of ARI cases and RSV in 330 (27%). RSV infections mostly occurred from August to October, during the second half of the rainy season, with the exception of 2012, when there were few RSV cases and most occurred during November–December. In contrast, HMPV peaked at different times each year (Figure1).

Among children aged <5 years, statistically significant nonlinear relationships with age were found for both HMPV [degrees of freedom (d.f.) = 5·2, P-value <0·001) and RSV (d.f. = 2·1, P-value <0·001). The probability of HMPV was low during the first month of life and increased with age, peaking at 20% around age 16 months. The predicted probability of RSV was approximately 41% during the first month of life, after which it declined rapidly, becoming similar to the probability of HMPV by around age 2 years (Figure2). Among persons aged ≥5 years, the probability of HMPV declined from 8% at age 5 years to around 2·5% at age 20 years (d.f. = 3·6, P-value = 0·005). In contrast, there was no statistically significant evidence of an association between age and the probability of RSV in this age group (P-value = 0.402).

The crude annual incidence rates of hospitalized HMPV-ARI (Table3) were 56 cases per 100 000 children aged <5 years (95% CI: 49, 64), 1·4 cases per 100 000 children aged 5–17 years (95% CI: 0·8, 2·4), and 1·5 cases per 100 000 persons aged ≥18 years (95% CI: 0·9, 2·2). The estimated annual incidence rates adjusted for the proportion of eligible cases tested for HMPV and for proportions of the populations seeking health care at the surveillance hospitals were 102 cases per 100 000 children aged <5 years (95% CI: 75, 178), 2·6 cases per 100 000 children aged 5–17 years (95% CI: 1·2, 5·0), and 2·6 cases per 100 000 persons aged ≥18 years (95% CI: 1·5, 4·9).

In unadjusted logistic regression models (Table3), among children aged <5 years, having HMPV infection compared with RSV infection was associated with significantly lower odds of being admitted to the ICU, needing mechanical ventilation and being hospitalized ≥1 week. After adjusting these models for age, sex and surveillance site, there was no statistically significant association. Among children aged <5 years, after adjustment for potential confounders, we found that HMPV-positive cases versus HMPV-RSV-negative cases were associated with lower odds of ICU admission (P-value = 0·002), mechanical ventilation (P-value<0·001), hospital stay >1 week (P-value = 0·011), and death (P-value = 0·026). Among persons aged ≥5 years, there was no evidence of a difference in severity associated with etiological group, and we did not find any evidence of effect modification by age (Table3).

To our knowledge, this is the first report from Central America of HMPV infection in the context of systematic hospital-based surveillance for ARI, and these data from three sites in Guatemala covering a span of 5 years demonstrate that HMPV is a substantial cause of ARI hospitalization in Guatemala. Among children aged <5 years, we found population incidence rates of HMPV-positive ARI similar to a previous study among US children.24 In contrast, the incidence rate we found for adults was many times lower than that reported in Tennessee, USA.25 However, lower access to and utilization of healthcare services may result in underestimation of the burden in Guatemala, despite our attempt to correct for healthcare seeking behavior. The proportion of ARI cases HMPV-positive among children aged <5 years in our study (13%) was higher than other previous studies from USA (New York and Tennessee), Italy, France and Korea,26–29 but similar to studies in Alaska and Brazil.12,30 The proportion of hospitalized ARI cases that were HMPV-positive among adults ≥50 years (3%) was similar to that reported in TN, USA (4·5%).25

There was no clear seasonality of HMPV infection that we observed. In contrast, it has been reported that peak incidence of HMPV coincides with the latter half of RSV seasonal peaks in both the USA and Brazil.31,32 The broad distribution of HMPV cases over time, in contrast to the sharper peak of RSV, is, however, consistent with previous studies in both temperate and tropical environments.31–33 The wide variation we found between years in the numbers and proportions of ARI cases attributable to HMPV infection is also consistent with findings from prospective ARI surveillance in other countries.31,32

Our study is consistent with previous findings that HMPV and RSV are major causes of ARI during early childhood and fairly important during later childhood and adulthood, and that HMPV-ARI is less frequent than RSV-ARI.34 Hall et al. 35found that incidence of RSV hospitalization in the United States was highest for infants 1 month of age, which is similar to our findings for proportion of ARI cases RSV-positive. Our detailed analysis of the association between age and the probability of detecting each virus among ARI cases, however, provides new information about the relative burden of disease attributable to each virus at specific ages. For instance, during the neonatal period, HMPV was detected in only about 3% of ARI cases, whereas RSV was detected in 41%. Most population-based studies that have reported the proportions of ARI positive for HMPV and RSV have not provided this level of detail because they have grouped children aged <6 months or <2 years.27,36,37 The decline in RSV and increase in HMPV over the first year of life suggest that the timing of vaccination would have important implications for the relative burden of disease that would be prevented by potential future HMPV and RSV vaccines. The small number of HMPV cases in the first months of life suggests that an effective early childhood vaccine could prevent the majority of its disease burden, in contrast to RSV, which may require maternal immunization to prevent the high burden during the first few months of life.

Among children aged <5 years, after adjusting for potential confounders, ARI cases with HMPV were no less severe than those with RSV but were less severe than those negative for both viruses, which included 1512 (66%) negative for the eight viruses tested (data not shown). Among persons aged ≥5 years, there was no evidence of a difference in severity when comparing HMPV-positive to either RSV-positive or HMPV-RSV-negative cases, which included 1575 (81%) negative for the eight viruses tested. Our crude results are consistent with evidence that HMPV hospitalizations are less severe and of shorter duration compared with RSV hospitalizations.32,38 However, it is also informative to know, as the adjusted models indicate, that children aged <5 years of similar age and same sex admitted to the same hospital would be expected to have similar hospital course and outcomes, regardless of whether the infection is HMPV or RSV, but that HMPV-positive cases can be expected to be less severe than HMPV-RSV-negative cases.

Our study has a few important limitations. Our initial ARI case definition is likely to have missed a proportion of illnesses caused by HMPV and RSV, particularly among young infants and adults, as they are often afebrile and lack significant WBC abnormalities.39–41 However, this underestimation of burden and potential selection bias was no longer a limitation for children aged <5 years after we added the WHO pneumonia case definition in February 2011. Furthermore, the small proportion (9%) of cases after February 2011 in children aged <5 years that lacked one of these signs of infection suggests this source of bias was not substantial for this age group. Interpretation of the comparisons between HMPV and HMPV-RSV-negative cases is limited by the incomplete information on the etiologies of this heterogeneous reference group, among which we detected a viral infection in only 27% of the cases, as well as the diverse clinical and epidemiological features of these infections. Our crude population incidence rates of HMPV-ARI hospitalization are known to substantially underestimate the burden of severe disease in the community that should have led to hospitalization. On the other hand, the adjustment of incidence rates for healthcare seeking behavior was based on self-reported illness and healthcare seeking practices, which introduces substantial uncertainty due to small sample size, as reflected in the wider confidence intervals, and is also subject to reporting bias.

Our study suggests that HMPV is important though less frequent than RSV as a cause of ARI hospitalizations and mortality in Guatemala. These findings highlight the importance of considering HMPV testing within ARI surveillance platforms in Central America. Additionally, interventions to prevent ARI due to HMPV and evaluations of their impact should take into account the potentially unpredictable timing of incidence peaks, rarity of cases during the first few months of life, and lesser severity compared with HMPV-RSV-negative cases among young children.