Global burden of acute lower respiratory infection associated with human metapneumovirus in children under 5 years in 2018: a systematic review and modelling study

Introduction

Acute lower respiratory infections (ALRIs) are one of the leading causes of morbidity and mortality in children globally, accounting for 10% of deaths in children younger than 5 years in 2017.¹ Human metapneumovirus, first identified in 2001, is an important virus that causes ALRIs in young children.^{2, 3, 4, 5} Previous evidence indicates that most children are infected with human metapneumovirus by age 5 years, with most severe infections occurring in infants, including symptomatic and asymptomatic infections.^{6, 7, 8, 9, 10} Pooled analyses have shown that human metapneumovirus is associated with 6·1–6·4% of hospital admissions due to ALRI among patients younger than 20 years worldwide.^{11, 12} Few studies have reported the incidence and mortality for human metapneumovirus-associated ALRIs, and studies have rarely reported disease burden in narrower age groups (eg, those younger than 6 months). No global or regional burden estimates are available for children under 5 years.

A live-attenuated recombinant human metapneumovirus vaccine has been assessed in a phase 1 clinical trial; however, study findings showed that the vaccine had insufficient immunogenicity in children aged 6–59 months.¹³ Antivirals against human metapneumovirus infections and several other types of vaccines have been investigated, but have not reached clinical trials.^{14, 15, 16}

We aimed to estimate the global and regional number of human metapneumovirus-associated ALRI cases, hospital admissions, and mortality by age strata in children under 5 years for 2018, using published and unpublished data on laboratory-confirmed (culture, immunofluorescence assay, or molecular test) human metapneumovirus morbidity and mortality. Since the presence of human metapneumovirus in the upper respiratory tract in children does not indicate that ALRI symptoms were directly caused by these viruses, we also estimated the global burden of ALRI that is attributable to human metapneumovirus by accounting for the causal attribution of human metapneumovirus. The estimates would provide evidence for the further development of targeted interventions and treatment.

Methods

Results

We identified 159 studies with data on human metapneumovirus-associated ALRI community incidence, hospital admission rates, the proportion of human metapneumovirus-positive cases among hospitalised ALRI cases, and hCFRs (figure); 119 studies were published and 40 were unpublished studies from the Respiratory Virus Global Epidemiology Network. Of the 159 studies included, seven (4%) were from low-income countries, 36 (23%) from lower-middle-income countries, 67 (42%) from upper-middle-income countries, and 49 (31%) from high-income countries. The characteristics of the included studies are summarised in the appendix (pp 6–7, 34–53).

11 studies reported the incidence of human metapneumovirus-associated ALRIs (ie, ALRI cases with laboratory confirmation of human metapneumovirus). Ten studies provided data for the 0–59-month age group after imputation, including six studies from settings with high child mortality; five studies were from lower-middle-income countries, two studies from upper-middle-income countries, and three studies from high-income countries (table 1). Four studies reported incidence before 2010. Between 1990 and 2015, the mean human metapneumovirus-associated ALRI incidence meta-estimate was 22·1 per 1000 children per year (95% CI 17·0–28·7) in children aged 0–59 months in settings with high child mortality, and 18·9 (11·2–31·9) in settings with low child mortality. The high incidence observed in settings with low child mortality was mainly driven by two Australian studies (done between 1996 and 1999 and between 2010 and 2014).^{37, 38} Using these meta-estimates, we estimated that there were 14·2 million human metapneumovirus-associated ALRI cases (UR 10·2 million to 20·1 million) globally in children younger than 5 years.

We identified 39 studies reporting human metapneumovirus-associated ALRI hospital admission rates, including 29 studies reporting data by three narrow age groups (0–5 months, 6–11 months, or 12–59 months; table 2). The hospital admission rate meta-estimate was more than four times higher in infants aged 0–5 months and 6–11 months (2·2–3·3 admissions per 1000 children per year) than children aged 12–59 months (0·3–0·6 admissions per 1000 children per year) when stratified by World Bank income levels and child mortality settings. In the analysis stratified by child mortality settings, we estimated that there were 643 000 human metapneumovirus-associated ALRI hospital admissions (UR 425 000–977 000) globally in children younger than 5 years in 2018, of which 374 000 (58%) occurred in infants.

14 studies included hospital admission rates for human metapneumovirus-associated ALRIs with hypoxaemia by three age groups, including five studies from settings with low child mortality (table 2). In the analysis stratified by child mortality settings, we estimated that, globally, 112 000 (UR 29 000–522 000) hospital admissions were for human metapneumovirus-associated ALRIs with hypoxaemia in children aged 0–59 months in 2018, accounting for 17% (ie, 112 000 of 643 000 cases) of the human metapneumovirus-associated ALRI hospital admissions.

73 studies reported hCFRs for human metapneumovirus-associated ALRI in children younger than 5 years, including 28 studies with data stratified by three narrow age bands (table 3). hCFRs were highest in infants aged 0–5 months from settings with high child mortality (3·3%, 95% CI 1·7–6·1) and from lower-middle-income countries (4·5%, 2·3–8·6). hCFRs were lower in children from high-income countries who were aged 6–11 months (0·6%, 0·1–3·9) and 12–59 months (0·5%, 0–7·0) than infants aged 0–5 months from high child mortality settings and from lower-middle-income countries, with the estimates in older children having notably wide confidence intervals. On the basis of these meta-estimates, we estimated that there were 7700 human metapneumovirus-associated ALRI in-hospital deaths (UR 2600–48 800) in children younger than 5 years in 2018, of which 5500 (71%) occurred in infants. About 64% of these deaths were in young infants aged 0–5 months (4900 deaths of 7700 total in-hospital deaths), and 88% (6800 deaths of 7700 total in-hospital deaths) occurred in countries with high child mortality.

Across 28 countries or regions with low child mortality, 22–94% of children with pneumonia received care from a health provider (appendix p 16). About 71% of the data were for the period 2010–14. On the basis of these data, we estimated the median inflation factor was 1·3 across regions or countries, and thus that there were about 1100 human metapneumovirus-associated ALRI deaths (UR 100–28 800; after rounding in-hospital deaths), in hospitals and in the community, among children younger than 5 years in settings with low child mortality in 2018. Of the eight sites with data on pneumonia mortality in high child mortality settings, five sites were in rural areas, and six sites were in African countries. Six studies reported data between 2010 and 2016, and the remaining two studies reported data from 2008 onwards. The inflation factor ranged from 1·5 to 3·5 across the eight sites, with a median value of 2·2 (IQR 1·7–2·7). Thus, we estimated that there were 14 900 human metapneumovirus-associated ALRI deaths (UR 5600–59 700; after rounding in-hospital deaths), in hospitals and in the community, in settings with high child mortality. We therefore estimate that there were 16 100 human metapneumovirus-associated ALRI deaths (5700–88 000; ie, ALRI deaths in children with laboratory-confirmed human metapneumovirus) among children globally in 2018.

By applying the attributable fraction of human metapneumovirus to the estimated human metapneumovirus-associated burden, we estimated that 11·1 million ALRI cases (UR 8·0 million to 15·7 million) and 502 000 ALRI hospital admissions (UR 332 000–762 000) could be attributed to human metapneumovirus in children under 5 years (table 4). We estimated that the ratio of in-hospital case fatalities of human metapneumovirus-attributable ALRIs to human metapneumovirus-associated cases was 0·9 (appendix p 18), and the attributable fraction for human metapneumovirus-associated ALRI deaths was 70% (appendix pp 17–18), indicating that 11 300 ALRI deaths (UR 4000–61 600) globally could be attributed to human metapneumovirus in 2018, including 10 400 deaths (3900–41 800) in settings with high child mortality (table 4).

We did several sensitivity analyses to estimate hospital admissions, in-hospital deaths, and overall human metapneumovirus-associated ALRI deaths, and of human metapneumovirus-attributable ALRI deaths (by country development strata and World Bank income level) using different approaches (appendix pp 8–13,16,19). For global human metapneumovirus-associated ALRI hospital admissions, the estimates in children younger than 5 years ranged from 626 000 to 650 000 in analyses by different stratification groups (appendix p 9); the proportion-based approach yielded a broader range (282 000–902 000; appendix p 10). The point estimate of global in-hospital deaths ranged from 7200 to 9100 in children younger than 5 years, and an estimated 79% of the global in-hospital deaths occurred in low-income and lower-middle-income countries (7200 of 9100 deaths) when stratified by World Bank income level (appendix p 11). We estimated 19 900 human metapneumovirus-associated ALRI deaths (UR 12 100–33 200) and 9900 human metapneumovirus-attributable ALRI deaths (UR 2600–39 300) in children aged 1–59 months in settings with high child mortality using a different approach (appendix pp 16, 19).

Discussion

To our knowledge, this is the first study to report global human metapneumovirus-associated ALRI burden estimates both among children younger than 5 years overall and stratified by narrow age groups (0–5 months, 6–11 months, and 12–59 months). When compared with our previously published all-cause ALRI burden estimates, human metapneumovirus is identified in 11% of ALRI cases, 4–13% of ALRI hospital admissions, and 2% of ALRI deaths among children under 5 years globally.^{1, 32, 33} About 8% of ALRI cases, 3–10% of ALRI hospital admissions, and 1% of ALRI mortality can be attributed to human metapneumovirus. The wide URs of these burden estimates reflect differences across studies, arising from variation in human metapneumovirus epidemiology between populations and methodological differences, and from paucity of data, especially mortality data.

Similar to respiratory syncytial virus and influenza, we found that the human metapneumovirus-associated ALRI hospital admission rate was much higher in infants than older children, with about 58% of hospital admissions and 71% of in-hospital deaths due to human metapneumovirus-associated ALRIs in children under 5 years occurring in the first year of life.^{18, 19} The high burden in infants could be due to the immaturity of infants' immune systems and decaying maternal antibodies during the first several months of life.^{39, 40} The consistently high human metapneumovirus-associated ALRI hospital admission rate among infants across different settings highlights the importance of developing safe and effective maternal human metapneumovirus vaccines, particularly for infants.

Our hCFR estimates show that young infants aged 0–5 months are at an increased risk of human metapneumovirus–associated ALRI mortality compared with infants older than 5 months. hCFR estimates for infants aged 0–5 months varied substantially across settings, possibly reflecting the differences in disease severity at admission and in the quality of hospital care. Increased disease severity at presentation could be associated with the high prevalence of certain underlying conditions and delays in care-seeking.^{3, 33} The case-fatality ratio (overall and in-hospital) and overall mortality due to ALRI among children under 5 years has substantially reduced in the past 15 years due to socioeconomic development, reduced prevalence of pneumonia risk factors, and increased use of interventions.^{33, 41} However, most studies spanning several years reported one to three deaths from human metapneumovirus-associated ALRIs, and we were unable to identify any changes in the hCFR of human metapneumovirus-associated ALRI over time. When stratifying the data using 2010 as a cutoff and using available age-stratified data from studies done before 2010 (eligible if part of a study was done before 2010 and data could not be stratified by year), we estimated that there were 9600 human metapneumovirus-associated ALRI in-hospital deaths (UR 2300–51 200) among children under 5 years in 2010 (appendix p 12).

For settings with high child mortality, the inflation factor was based on scarce data.¹⁹ The estimates could be biased due to limited generalisability and location-specific characteristics (ie, prevalence of human metapneumovirus in childhood ALRI deaths). However, our reported estimates of overall human metapneumovirus-associated ALRI deaths for settings with high child mortality are conservative and could increase by approximately 30% (19 900 deaths) using an alternative estimation approach (appendix p 16). For settings with low child mortality, the inflation factor and overall mortality estimate are likely to be underestimated because the definition of care-seeking is broader than the definition of in-hospital care: contact with primary care is included as care-seeking in surveys, but not included in the in-hospital mortality estimates in the present analysis. The US vital statistics data showed that about 40% of under-5 ALRI deaths (International Classification of Diseases, 10th revision diagnostic codes J09–22; U04) occurred in outpatient or emergency departments between 2010 and 2017.⁴² Additionally, this analysis was based on one further assumption that no differences exist between the case-fatality ratio for hospitalised and non-hospitalised pneumonia cases. The direction of bias associated with this assumption could be complicated by the two-way association between disease severity and care-seeking: children with more severe symptoms are more likely to receive hospital care; whereas, supportive care in hospitals can reduce the risk of death, and lack of appropriate care or delays in care can lead to rapid deterioration.^{43, 44, 45} Moreover, data on pneumonia care-seeking in high-income countries were not available in published reports; extrapolation from other countries with low child mortality to high-income countries might result in bias. The estimates of inflation factor could also be affected by accuracy of assessment and completeness of documentation for ALRI or pneumonia. For example, the diagnosis of (presumptive) pneumonia in the UNICEF dataset is based on caregivers' report, thus might be inaccurate and affect the estimates of inflation factor.

Our study had several other limitations. Little data were available for Europe and Latin America, which could have affected the generalisation of the estimates. Moreover, heterogeneity in the methodology (ie, variations in precise case definitions, proportion of eligible ALRI cases tested for human metapneumovirus) existed across studies for each outcome and is likely to have biased our estimates (appendix p 7). We adjusted for levels of testing when estimating hospital admission rates by assuming that the proportion of tested patients with ALRIs who were positive for human metapneumovirus was the same as the proportion in untested patients with ALRIs. However, we did not adjust for the underdetection of human metapneumovirus when estimating hCFRs. The hCFR of human metapneumovirus-associated ALRIs and in-hospital mortality might be underestimated, as suggested by the higher hCFR in patients tested than untested patients (appendix p 24). The incidence and hospital admission rates of human metapneumovirus-associated ALRI are unlikely to be affected by the accuracy of test methods since 92% of the studies used PCR to detect human metapneumovirus. In addition to potential biases, we summed yearly data to ensure precision in the age-stratified analysis; however, we did not account for variation across years (appendix p 20).

Our results show that infants younger than 1 year have disproportionately high risks of severe human metapneumovirus infections across settings. Infants aged 0–5 months in low-income and lower-middle-income countries have an increased risk of mortality from human metapneumovirus-associated ALRIs compared with those in upper-middle-income countries and high-income countries, accounting for nearly eight of every ten global deaths occurring in this age group. These estimates highlight the importance of interventions for infants. In low-income and lower-middle-income countries, continued efforts are needed to reduce human metapneumovirus-associated ALRI mortality among young children, especially infants, in hospitals and communities through improving access to care and case management in hospitals. Considering the paucity of data at the regional level, these estimates should be viewed as preliminary estimates. In the future, additional high quality data on childhood human metapneumovirus-associated ALRI cases, hospital admissions, and mortality, especially age-stratified data, would help refine the estimates and track changes in human metapneumovirus-associated ALRI morbidity and mortality over time.

Supplementary Material