Influenza Vaccine Effectiveness Among Children: 2011–2020

In a pooled analysis of 9 influenza seasons, vaccine effectiveness across all flu types/subtypes was 46%.

Abstract

Background and Objectives

Infants and children are at increased risk of severe influenza virus infection and its complications. Influenza vaccine effectiveness (VE) varies by age, influenza season, and influenza virus type/subtype. This study’s objective was to examine the effectiveness of inactivated influenza vaccine against outpatient influenza illness in the pediatric population over 9 influenza seasons after the 2009 A(H1N1) pandemic.

Methods

During the 2011–2012 through the 2019–2020 influenza seasons at outpatient clinics at 5 sites of the US Influenza Vaccine Effectiveness Network, children aged 6 months to 17 years with an acute respiratory illness were tested for influenza using real-time, reverse-transcriptase polymerase chain reaction. Vaccine effectiveness was estimated using a test-negative design.

Results

Among 24 148 enrolled children, 28% overall tested positive for influenza, 3017 tested positive for influenza A(H3N2), 1459 for influenza A(H1N1)pdm09, and 2178 for influenza B. Among all enrollees, 39% overall were vaccinated, with 29% of influenza cases and 43% of influenza-negative controls vaccinated. Across all influenza seasons, the pooled VE for any influenza was 46% (95% confidence interval, 43–50). Overall and by type/subtype, VE against influenza illness was highest among children in the 6- to 59-month age group compared with older pediatric age groups. VE was lowest for influenza A(H3N2) virus infection.

Conclusions

Analysis of multiple seasons suggested substantial benefit against outpatient illness. Investigation of host-specific or virus-related mechanisms that may result in differences by age and virus type/subtype may help further efforts to promote increased vaccination coverage and other influenza-related preventative measures.

What’s Known on This Subject:

Studies have reported vaccine effectiveness for single influenza seasons; however, few have described vaccine effectiveness across multiple seasons. Influenza vaccine effectiveness has varied across seasons, depending on the predominating influenza virus type/subtype and across age groups within individual influenza seasons.

What This Study Adds:

In a pooled analysis, effectiveness of inactivated influenza vaccine was estimated for 9 influenza seasons. Patterns of pooled vaccine effectiveness across 9 seasons by age group showed that the 6- to 59-month-old group had the highest vaccine effectiveness across all influenza types/subtypes.

Influenza epidemics occur worldwide and have been responsible for as many as 41 million annual illnesses, 710 000 hospitalizations, and 52 000 deaths between 2010 and 2020 in the United States.¹ Within each of the influenza seasons from 2010 to 2011 through 2018 to 2019, the adjusted influenza-associated hospitalization incidence rates for children <18 years of age were as low as 10 per 100 000 persons and as high as 375 per 100 000 persons.² Among all children, the youngest ones had the highest rates of hospitalization and deaths; however, older children had a higher risk of severe outcomes, if hospitalized, during that same period.² Infants and children are particularly vulnerable to higher rates of infection and complications compared with adults, in part because of waning protection from maternal antibodies, still developing immune systems, fewer prior infections, and higher transmission rates.^3–9 Moreover, children may play an important role in seasonal influenza transmission and epidemic spread, with varying intensity depending on the predominant influenza virus type/subtype and season.^10,11 Preventing influenza virus infection in children can have both direct and indirect benefits to themselves and others.

The Advisory Committee on Immunization Practices (ACIP) recommended influenza vaccination for all children aged 6–23 months in 2004, and by 2008, recommended influenza vaccination for all children and adolescents aged 6 months through 18 years.^12,13 The ACIP recommends 2 doses of the influenza vaccine at least 1 month apart for children aged 6 months to 8 years during the first season they are vaccinated or for those who have not previously received 2 doses.¹⁴ Otherwise, a single, annual dose is recommended for all children.¹⁴ Once these vaccines are approved, effectiveness studies, rather than efficacy studies alone, provide important insight into how the vaccines perform under the myriad of conditions that a person may be receiving a vaccine under. Unlike vaccine efficacy studies—which provide the initial information as to whether the vaccines are safe and effective under optimal conditions—vaccine effectiveness (VE) studies enable us to better assess how the vaccine performs in more diverse settings and groups of people, which is what this study seeks to do. Therefore, vaccine effectiveness studies’ results such as those from this one may not indicate as high levels of protection as do results from vaccine efficacy studies.

VE is a measure of protection against influenza. In this study, VE measures the reduction in the likelihood of influenza illness among patients presenting with acute respiratory illness symptoms. VE against symptomatic influenza virus infection and serious illness varies by age, influenza season, and influenza virus type and subtype.^15–17 Further information on the effectiveness of influenza vaccination in pediatric age groups can help communicate the benefit of vaccination to parents and caregivers. The objective of this study was to examine influenza VE in the pediatric population against outpatient influenza illness by virus type/subtype during 9 seasons after the H1N1 influenza pandemic of 2009 and before the COVID-19 pandemic of 2020. Analysis of multiple influenza seasons yielded more precise VE estimates across pediatric age groups for each influenza virus type/subtype.

Methods

Study Population

The study population consisted of children and adolescents enrolled in the US Influenza Vaccine Effectiveness Network (US Flu VE Network) over 9 influenza seasons from 2011 to 2012 through 2019 to 2020. The US Flu VE Network is a long-standing research network that annually estimates effectiveness of licensed US vaccines.¹⁸ During the influenza seasons in this analysis, the US Flu VE Network included 5 sites. At health facilities associated with study sites in 5 states (Michigan, Pennsylvania, Texas, Washington, and Wisconsin), outpatients aged ≥6 months with acute respiratory illness were enrolled after 2 consecutive weeks of increasing local influenza activity at the onset of the influenza season.¹⁸ This analysis included participants aged 6 months to 17 years. Few individuals were enrolled more than once in a season. When they were, each episode >14 days from the previous episode was included. Each site’s institutional review board approved study procedures, forms, and consent documents. The US Centers for Disease Control and Prevention relied on the approval determination made at each site’s institutional review board.

Influenza Testing

Participants were tested for influenza virus infection using real-time, reverse-transcriptase polymerase chain reaction assays developed by the US Centers for Disease Control and Prevention. Specimens were tested for influenza virus type and subtype. Details of the testing and vaccine status assessment methodology of the US Flu VE Network have been published previously.^19–21

Enrollment Interview and Electronic Health Records Review

Enrollment interviews provided information on patient demographic characteristics, illness onset date, symptoms, and general and current health status in addition to immunization status. Participants themselves or a parent identified a child’s sex as male or female, race as white, Black or African American, Asian, Native Hawaiian or other Pacific Islander, American Indian or Alaska native, mixed race, or other race, and ethnicity as Hispanic or non-Hispanic. Chronic medical conditions were identified in electronic health records via documentation of International Classification of Diseases 9th Edition or 10th Edition, Clinical Modification diagnostic codes corresponding to high-risk conditions designated by the ACIP.^22–24 The codes included the following diseases and can be found in the Supplementary Table of Shang et al (2018): asthma, blood disorders, cerebrovascular diseases, chronic obstructive pulmonary diseases and other lung diseases, diabetes, heart diseases, immunosuppressive conditions, kidney diseases, liver diseases, and neurologic conditions.²⁵

Vaccination Status

Vaccination status was obtained from electronic health records or immunization information systems. Children younger than 9 years of age who had received 2 doses of influenza vaccine in the current influenza season or 1 current season dose after receipt of 2 doses in 1 or more previous seasons were classified as vaccinated for the current season.¹⁴ Children aged ≥9 years who had received 1 dose of current season influenza vaccine were considered vaccinated. Any children vaccinated 0 to 13 days before illness onset were excluded. Children and adolescents who received live attenuated influenza vaccine (n = 1248) were excluded from this analysis because of reduced observed effectiveness of live attenuated influenza vaccine and low uptake of the vaccine during this period.^26,27

Statistical Analysis

We pooled data from 9 seasons to compare characteristics of laboratory-confirmed influenza cases and test-negative influenza controls by virus type and subtype and age group (6–59 months, 5–8 years, and 9–17 years). The ratio (odds) of patients with laboratory-confirmed influenza virus versus a negative test for influenza virus was compared among vaccinated versus unvaccinated children. VE was calculated using a test-negative case-control design, as (1 – odds ratio) × 100%, where odds ratio represents the odds of vaccination among patients with laboratory-confirmed influenza virus infection (test-positive cases) divided by the odds of vaccination among all influenza-negative patients (test-negative controls).^28,29 Logistic regression models were adjusted for study site, influenza season of enrollment, subjective general health status (excellent, very good, good, or fair/poor), interval from illness onset to network enrollment (0–2 days, 3–4 days, or 5–7 days), month of illness onset, and the presence of 0 versus ≥1 chronic medical condition associated with increased risk for complications because of influenza virus infection during at least 1 medical encounter in the 12 months before enrollment.²⁵ Patient self-identified sex and race/ethnicity categories were removed from multivariate models. We estimated pooled VE for the 9-year period by influenza virus type/subtype and age group as well as pooled VE against A(H3N2) viruses by antigenic similarity to vaccine reference viruses. Pooling VE against H3N2 by age group was a secondary analysis based on predominance of H3N2 in multiple seasons. Analyses were conducted using SAS statistical software, version 9.4 (SAS Institute Inc).

Results

In total, 24 148 children and adolescents aged 6 months to 17 years were included in the study; 6767 (28%) tested positive for influenza virus infection (Table 1). The breakdown for positive cases for influenza A(H3N2), A(H1N1)pdm09, and influenza B viruses was 3017, 1459, and 2178, respectively. There were 41 people who had influenza A and B coinfections. Among those who tested positive for influenza B virus, ∼51% tested positive for influenza B/Victoria lineage, ∼48% for influenza B/Yamagata lineage, and ∼1% for an undetermined influenza B lineage. The season with the greatest number of pediatric cases and highest proportion of positive cases was 2018–2019. The youngest age group had the lowest proportion of participants test positive for any influenza virus (Table 1). The oldest age group had the highest percentages of people testing positive for influenza A(H3N2) and influenza B compared with the other 2 age groups.

TABLE 1.

Comparison of Characteristics of Pediatric Cases and Controls, US Flu VE Network 2011–2012 Through 2019–2020

		Influenza Positive (Cases)		Influenza Negative (Controls)
Influenza type; A Subtype; B Lineage	Total Children	No. Cases a	% (Column Total)	No. Controls a	% (Column Total)
All influenza A and B	24 148	6767	28.0 b	17 381	72.0 b
Influenza season
2011–2012	2023	283	4.2	1740	10.0
2012–2013	2321	909	13.4	1412	8.1
2013–2014	1659	254	3.8	1405	8.1
2014–2015	3381	771	11.4	2610	15.0
2015–2016	2428	390	5.8	2038	11.7
2016–2017	2602	748	11.1	1854	10.6
2017–2018	3016	1113	16.4	1903	11.0
2018–2019	3608	1215	18.0	2393	13.8
2019–2020	3110	1084	16.0	2026	11.7
Age
6–59 mo	9553	1647	24.3	7906	45.5
5–8 y	5707	2113	31.2	3594	20.7
9–17 y	8888	3007	44.4	5881	33.8
Sex a
Female	11 586	3224	47.7	8362	48.1
Male	12 558	3540	52.3	9018	51.9
Race/ethnicity c
Non-Hispanic white	15 285	4271	63.2	11 014	63.5
Non-Hispanic Black or African American	2831	918	13.6	1913	11.0
Hispanic	3238	827	12.2	2411	13.9
Non-Hispanic other race	2737	737	10.9	2000	11.5
General health state d
Excellent	12 730	3853	57.0	8877	51.1
Very good	7453	2027	30.0	5426	31.2
Good	3272	746	11.0	2526	14.5
Fair/poor	679	136	2.0	543	3.1
Presence of ≥1 chronic medical condition(s)
Yes	5955	1470	21.7	4485	25.8
No	18 193	5297	78.3	12 896	74.2
Type of vaccine received e	9468	1992		7476
Trivalent	3027	480	24.2	2547	34.4
Quadrivalent	6366	1502	75.8	4864	65.6
Site
Michigan	4719	1443	21.3	3276	18.8
Pennsylvania	3614	973	14.4	2641	15.2
Texas	5289	1274	18.8	4015	23.1
Washington	3650	836	12.4	2814	16.2
Wisconsin	6876	2241	33.1	4635	26.7
Influenza A f	—	4548	67.2	—	—
A(H3N2)	—	3017	67.4	—	—
A(H1N1)pdm09	—	1459	32.6	—	—
Influenza B	—	2178	32.2	—	—
Victoria lineage	—	1115	51.2	—	—
Yamagata lineage	—	1034	47.5	—	—
Undetermined lineage	—	29	1.3	—	—
Influenza A and B coinfection	—	41	0.6	—	—

^a3 with unknown sex for influenza A and B cases and 1 with unknown sex for controls.

^bRow total.

^c14 cases with unknown race and 43 controls with unknown race.

^d5 cases and 9 controls refused to answer health state question.

^e10 cases with unknown vaccine type and 65 controls with unknown vaccine type.

^f72 with influenza A that was unable to be subtyped.

—, not applicable.

Overall, 9472 (39%) participants were fully vaccinated. Among vaccinated cases, 480 (24%) received a trivalent vaccine and 1502 (76%) received a quadrivalent vaccine. Among all children aged 6 to 59 months, 5 to 8 years, and 9 to 17 years, 49%, 38%, and 30% were vaccinated, respectively. Overall, 29% of cases compared with 43% of controls were fully vaccinated. Vaccination coverage among influenza-negative patients varied by age group, with the highest percentage (52%) among those aged 6 to 59 months (Fig 1).

FIGURE 1.

Effectiveness of pediatric influenza vaccination against laboratory-confirmed influenza, stratified by age group, US Flu VE Network, influenza seasons 2011 - 2012 through 2019 - 2020. Vaccine effectiveness was calculated as (1 – adjusted odds ratio) × 100, in which the odds ratio is the odds of pediatric vaccination among children with influenza versus controls. Will not add to total for influenza A(H3N2) and A(H1N1)pdm09 because some influenza A virus specimens were not able to be subtyped.

Overall, pooled VE against any influenza virus infection was 46% (95% confidence interval [CI], 43–50) (Fig 1). VE varied by virus type/subtype: VE was lowest against A(H3N2)-associated illness (33% [95% CI, 27–39]), whereas VE estimates were similar for influenza B (54% [95% CI, 49–59]) and influenza A(H1N1)pdm09 (57% [95% CI, 51–62]). There was no statistically significant difference for the VEs between participants aged 5–8 years and 9–17 years. Against A(H3N2), A(H1N1)pdm09, and influenza B, VE for children aged 6–59 months was 47% (95% CI, 37–55), 63% (95% CI, 54–70), and 63% (95% CI, 53–71), respectively. Overall VE was substantially higher for children aged 6 to 59 months compared with VE for children aged 5 to 8 and 9 to 17 years. Similarly, children aged 6 to 59 months had substantially higher VEs for influenza A subtypes and influenza B than did children aged 5 to 8 or 9 to 17 years.

In addition to observing pooled overall influenza A(H3N2) VE for all age groups, influenza A(H3N2) VE was also confirmed for a few influenza seasons. Influenza A(H3N2) viruses predominated during 5 of the 9 included influenza seasons: 2011–2012, 2012–2013, 2014–2015, 2016–2017, and 2017–2018.^{16,21,30–33} In 2014–2015, although vaccination offered little to no protection for the older pediatric age groups, VE for the youngest age group was 52% (95% CI, 29–68) (Fig 2). For this influenza season, VE was not observed for children aged 5 to 8 or 9 to 17 years. In other A(H3N2)-predominant seasons, patterns of VE by pediatric age group varied (ie, VE within each A(H3N2)-predominant season was not consistently highest for the youngest age group). Among the 4 matched A(H3N2)-predominant seasons, VE was highest for the youngest age group during 3 of the seasons (2011–2012, 2012–2013, and 2017–2018) and highest for those aged 5 to 8 years during one (2016–2017).

FIGURE 2.

Effectiveness of pediatric influenza vaccination against influenza A(H3N2), stratified by pediatric age group, antigenic match between vaccine and circulating A(H3N2) virus, and season. Seasons depicted here are those in which influenza A(H3N2) circulated predominantly.

Discussion

Over 9 influenza seasons, inactivated influenza vaccine reduced the risk of medically attended laboratory-confirmed influenza virus infection by 46% (95% CI, 43–50) in our study population of children aged 6 months to 17 years. VE was highest against influenza A(H1N1)pdm09 virus (57% [95% CI, 51–62]), followed by influenza B virus (54% [95% CI, 49–59]), and lowest for influenza A H3N2 virus (33% [95% CI, 27–39]). VE varied by age group across all virus types/subtypes and seasons. For all types/subtypes, pooled VE estimates were highest among children aged 6 to 59 months, whereas VE varied among older children and adolescents aged 5 to 17 years across seasons and influenza virus types/subtypes. Age-specific patterns may vary by season and circulating virus, and the heterogeneity we observed is consistent with other reports.^34,35 In the 2014–2015 influenza season, predominant A(H3N2) viruses were antigenically distinct from the vaccine component.³⁶ We found that vaccine-derived protection against the influenza A(H3N2) virus was effective in 4 of 5 influenza A(H3N2)-predominant seasons.

It is important to bear in mind the difference between vaccine efficacy and VE when interpreting our results. VE measures vaccine protection under ideal conditions, from easy vaccine storage to healthy participants.³⁷ This VE study refers to the protection the vaccine confers under suboptimal circumstances, such as when people have underlying medical conditions or the circulating influenza type/subtype has changed from the previous year. Our findings in pediatric age groups are consistent with previous studies in older ages showing differences in VE by virus type and subtype. A meta-analysis reporting pooled VE of influenza A subtypes H3N2 and (H1N1)pdm09 and influenza B found that VE was higher against influenza A(H1N1)pdm09 and influenza B compared with influenza A(H3N2).³⁸ Similarly, in our analysis among pediatric age groups, VE for influenza A(H1N1)pdm09 and VE for influenza B were similar to each other and higher than VE for influenza A(H3N2). Influenza viruses undergo continuous antigenic evolution, which may result in decreased VE. Antigenic change occurs 5 times faster with influenza A(H3N2) than it does with influenza B.³⁹ Furthermore, influenza A(H3N2) antigenic change happens 17 times faster than antigenic change for influenza A(H1N1)pdm09.³⁹ Antigenic changes to the viruses themselves in recent years may have affected the VE observed during our study period. However, despite these alterations, pooled VE across the 9 influenza seasons shows that the influenza vaccine provided substantial protection against any influenza virus infection across all pediatric ages.

In this study, we focus on the prevention of medically attended acute respiratory illness resulting from laboratory-confirmed influenza in the outpatient setting, which can generalize to impeding the development of complications and more severe illness. Our results show a variation in VE but also an overall benefit across all age groups of children. Individual seasons do not fit the overall pattern of the youngest children consistently deriving the most protection. Inconsistencies in VE from year to year result from the differing virus types that dominate across years (eg, which virus type circulated the most) and antigenic matches across seasons. When we pool data, however, the advantage of influenza vaccines is seen across multiple seasons.

Studies comparing influenza VE among young children compared with older children and adults have reported inconsistent patterns. Hu et al showed that young US children aged 6 months to 4 years had higher pooled VE across any influenza virus type than US children from 5 years to 17 years of age for influenza seasons 2016 to 2017 through 2019 to 2020.³⁴ Stein et al found that VE estimates were not consistently higher in younger children (ages 0.5–4 years of age) compared with older children (5–17 years of age) in European countries during the 2016 to 2017 and 2017 to 2018 influenza seasons.³⁵ Differences in vaccination history and influenza vaccine products, as well as differences in circulating influenza viruses, may contribute to discordant VE among children in the United States and Europe, among other factors. One strength of the US Flu VE Network is estimation of VE over several seasons with a consistent network throughout the period and similar populations of participations.

Factors associated with age-related differences in VE are not well understood. In our study, the youngest age group had the highest vaccination coverage. The relative decrease in influenza-associated mortality and health care use after increased influenza vaccination coverage in Ontario, Canada, may provide support for the association between high vaccine coverage and public health benefits.⁴⁰ It is also possible that the youngest age group may be the most likely to live in households in which members make an extra effort to be vaccinated. This may be a residual approach to protecting young children because studies and campaigns emphasize this strategy of vaccinating pregnant and postpartum mothers of infants because it has been found to reduce influenza-like illness and health care–seeking behavior in and for infants in such households.^41–44 Another possible reason for higher VE among young children may be a more broadly reactive immune response rather than a more focused one as the immune system matures and gets boosted after multiple new exposures to influenza viruses.^45,46 Children aged 6 to 59 months are likely to have had fewer influenza exposures than older children have had.⁴ Social interactions and mixing may also increase frequency of exposure to influenza viruses among school-aged children, and that higher force of infection could put increased pressure on the protection conferred by vaccines.^4,47 VE also varies by age and first or early influenza virus infections (ie, immune imprinting).⁴⁶ Understanding the roles that host-specific, behavioral, and viral factors play in VE may help improve vaccines for pediatric age groups.

Our study has several limitations. Despite comparing VE across multiple seasons, sample sizes were small for A(H1N1)pdm09 and influenza B, which limited examination of VE against antigenically distinct A(H1N1)pdm09 viruses and against B virus lineages. Our study population may not be generalizable to all children in the United States because vaccination histories may differ between those who sought clinical care for acute respiratory illness and the general population of US children.⁴⁸ Additionally, children with underlying conditions are at increased risk for severe influenza virus infection and influenza-associated complications and may have different vaccine coverage from those without underlying conditions. Misclassification of vaccination status may have occurred because of incomplete immunization records. For example, children may have been vaccinated in a state other than the one they were enrolled in for this study. It is also possible that pharmacies may not have reported vaccinations to the immunization registry. However, these do not occur often, and registries for children are usually complete.

Overall, vaccines provided substantial protection against influenza illness among pediatric age groups, with the highest vaccine coverage and vaccine protection among those younger than 5 years of age. Understanding reasons for the higher VE estimates in young children might inform efforts to design improved influenza vaccines. The benefits of current influenza vaccination can be increased by improving vaccine coverage, which was less than 50% among control group children who sought care for acute respiratory illness. Continued monitoring of VE in children and maximizing benefits of current vaccines by increasing vaccine coverage are all needed to decrease influenza-related illness among children.

Glossary

ACIP

Advisory Committee on Immunization Practice

CI

confidence interval

US Flu VE Network

United States Influenza Vaccine Effectiveness Network

VE

vaccine effectiveness