Menu
Menu
Menu
Menu
Background: Influenza virus infection is an important cause of under-five mortality. Maternal vaccination protects children younger than 3 months of age from influenza infection. However, it is unknown to what extent paediatric influenza-related mortality may be prevented by a maternal vaccine since global age-stratified mortality data are lacking. Methods: We invited clinicians and researchers to share clinical and demographic characteristics from children younger than 5 years who died with laboratory-confirmed influenza infection between January 1, 1995 and March 31, 2020. We evaluated the potential impact of maternal vaccination by estimating the number of children younger than 3 months with in-hospital influenza-related death using published global mortality estimates. Findings: We included 314 children from 31 countries. Comorbidities were present in 166 (53%) children and 41 (13%) children were born prematurely. Median age at death was 8·6 (IQR 4·5–16·6), 11·5 (IQR 4·3–24·0), and 15·5 (IQR 7·4–27·0) months for children from low- and lower-middle-income countries (LMICs), upper-middle-income countries (UMICs), and high-income countries (HICs), respectively. The proportion of children younger than 3 months at time of death was 17% in LMICs, 12% in UMICs, and 7% in HICs. We estimated that 3339 annual influenza-related in-hospital deaths occur in the first 3 months of life globally. Interpretation: In our study, less than 20% of children is younger than 3 months at time of influenza-related death. Although maternal influenza vaccination may impact maternal and infant influenza disease burden, additional immunisation strategies are needed to prevent global influenza-related childhood mortality. The missing data, global coverage, and data quality in this study should be taken into consideration for further interpretation of the results. Funding: Bill & Melinda Gates Foundation.
The FLU GOLD study was initiated in October 2017. We invited our existing global respiratory syncytial virus (RSV) GOLD network [15], consisting of individual investigators, research groups, and clinicians, to share individual-level data of children aged 0–59 months who had died with laboratory-confirmed influenza infection between January 1, 1995 and March 31, 2020. YNL, NIM, FSV, and LJB had full access to all the data in the study. We excluded community deaths due to limited available data (n = 8) and children with influenza-related mortality after stem cell transplantation. Additionally, we searched the literature using PubMed for “influenza” combined with “death”, “deaths”, “died”, “mortality”, “fatality”, or “case fatality ratio (CFR)” and “pediatric”, “pediatrics”, “child”, or “children” and invited authors to share additional (unpublished) cases. Collaborators were invited to share data between October 13, 2017, and March 31, 2020 through a link to a questionnaire (Supplementary Material) in Research Online, an electronic data capture platform [16]. We collected demographic and clinical characteristics and compared these between children from different income groups. Countries of origin were categorised as LMIC (LIC and LMIC combined), upper-middle-income countries (UMIC), and HIC according to the World Bank classifications for 2020 [17]. We compared age distribution at time of death for the 3 income groups and between the following 3 patient populations: children with comorbidities, healthy term children, and healthy preterm children (without comorbidities). A comorbidity was defined as at least one underlying disease, such as congenital heart disease, chronic lung disease or a genetic disorder. Prematurity was defined as gestational age less than 37 weeks. If data for comorbidities or prematurity were not reported, we assumed that the children were healthy term. We calculated weight-for-age z-scores as previously described [15]. We determined the proportion of children who died within the influenza virus epidemic season by comparing age at death and seasonality within the country of origin as estimated by a recent systematic analysis on global patterns of monthly influenza virus activity [18]. We compared the proportion of in-hospital deaths under 6 months of age in our study to the proportions from published studies used for the recent global influenza burden study from the Respiratory Virus Global Epidemiology Network [1]. We calculated age at influenza infection by subtracting the number of days between onset of influenza-related symptoms and influenza-related death from age at influenza-related death. We then determined the proportion of children under 3 months of age at time of influenza infection. We differentiated between children with community-acquired and hospital-acquired influenza infection. In case the setting where influenza had been acquired was not provided, and if there were no strong indications of nosocomial infection based on timeframe and clinical disease course, we assumed the infection had been community-acquired. We assumed that deaths occurred within the hospital if data on location of death were missing. To evaluate the minimum expected impact of maternal vaccination on influenza-related deaths assuming 100% vaccine efficacy and complete vaccination coverage, we multiplied the proportion of children younger than 3 months at time of community-acquired in-hospital death by the estimated total number of global influenza-associated ALRI in-hospital deaths under 5 years of age for each World Bank income group [1]. We compared demographic and clinical characteristics between different income groups, excluding cases with missing data for comorbidities or prematurity. We performed subgroup analyses and compared characteristics for children with hospital-acquired and community-acquired influenza-related death. Furthermore, we differentiated between seasonal and pandemic influenza-related deaths by excluding children with influenza A(H1N1)pdm09 who died within the timeframe of the WHO-declared pandemic (June 2009 – August 2010) from the analyses. Lastly, we analysed to what extent our results were sensitive to the contribution of a large number of cases from Ecuador, United Kingdom, Kenya, Turkey and South Africa (n = 145) by excluding these countries from our analyses. Since de-identified secondary patient data were used in the FLU GOLD study, parental informed consent was waived by the institutional research board of the University Medical Centre Utrecht. Ethical approval was obtained for individual collaborating institutes when required. We report descriptive statistics for all variables. Continuous variables are presented as median with interquartile ranges (IQR). Categorical variables are presented as frequencies and proportions. We used the χ2-test or the Fisher’s exact test to determine statistical significance between groups for categorical parameters. We report conservative exact p values instead of asymptotic p values because of the small sample size. The Mann-Whitney U test was used for all continuous parameters. We applied the Bonferroni correction for multiple testing between World Bank income groups. All statistical analyses were performed with SPSS (version 21·0; IBM Corp, Armonk, NY). The funder had no role in study design, data collection, data analysis, data interpretation, writing of the report, or the decision to submit the paper for publication. YNL, NIM, FSV, and LJB had full access to all the data in the study and NIM had final responsibility for the decision to submit for publication.
