Benefit–Risk Assessment of mRNA COVID-19 Vaccines in Children Aged 6 Months to 4 Years in the Omicron Era

Abstract

Background

There is no risk and benefit assessment of COVID-19 vaccination for children younger than 5 years using a single health outcomes scale. The objective of this study is to compare the expected risk and benefits of the mRNA primary series of COVID-19 vaccines for children aged 6 months to 4 years in the United States using a single health outcome scale in the Omicron era.

Methods

The expected benefits and risks of the primary two-dose series of mRNA COVID-19 vaccines for children aged 6 months to 4 years were stratified by sex, the presence of underlying medical conditions, the presence of infection-induced immunity, and the type of mRNA vaccine (BNT162b2 or mRNA-1273). A scoping literature review was conducted to identify the indicators in the decision tree model. The benefit–risk ratio was the outcome of interest.

Results

The benefit–risk ratios ranged from 200.4 in BNT162b2 for males aged 6–11 months with underlying medical conditions and without infection-induced immunity to 3.2 in mRNA-1273 for females aged 1–4 years without underlying medical conditions and with infection-induced immunity.

Conclusions

The expected benefit of receiving the primary series of mRNA vaccines outweighed the risk among children ages 6 months to 4 years regardless of sex, presence of underlying medical conditions, presence of infection-induced immunity, or type of mRNA vaccines. However, the continuous monitoring of the COVID-19 epidemiology as well as vaccine effectiveness and safety is important.

INTRODUCTION

Vaccinating against coronavirus disease 2019 (COVID-19) is recommended by the United States Centers for Disease Control and Prevention (CDC) for everyone 6 months of age and older [1]. However, because of the ever-changing epidemiology of COVID-19 and emerging data on vaccine effectiveness and safety, regularly re-evaluating the benefits and risks of COVID-19 vaccines is critical to ensure recommendations are based on the most up-to-date evidence [2].

The benefits and risks of COVID-19 vaccination in adults have been evaluated previously. For example, one study conducted benefit and risk assessments for primary and booster doses of COVID-19 vaccination for adolescent and adult populations [3, 4]. In their analysis for the messenger ribonucleic acid (mRNA) COVID-19 vaccines, the numbers of COVID-19 cases, hospitalizations, and deaths as well as the number of myocarditis cases were estimated for those aged 12 or older. However, the analyses used the Vaccine Adverse Events Reporting System (VAERS) as their vaccine safety data source; VAERS is a passive surveillance reporting system, subject to underreporting, misreporting, and overreporting, and is only intended for signal detection, not for calculating the risk of adverse events following immunization (AEFI). Another study evaluated the risks and benefits of mRNA COVID-19 vaccines for adolescents aged 16–29 years using the vaccine safety datalink (VSD), an active surveillance database [5]. However, both studies used data on COVID-19 disease burden from the delta variant period and did not use a single health outcome scale to compare the benefits and the risks. Comparing benefits and risks in a single health outcome scale provides a more straightforward quantitative comparison than presenting benefit outcomes and risk outcomes in separate scales, allowing the results to be interpreted intuitively [6].

Our own recent study compared the risks and benefits of mRNA COVID-19 vaccines in a single health outcome scale, quality-adjusted life years (QALY), for those aged 5 years or older, using data on COVID-19 disease burden from the omicron variant period and data on vaccine safety from active surveillance [7]. We found the benefits of mRNA COVID-19 vaccines to outweigh the risks, irrespective of age, sex, and underlying medical condition.

Two mRNA COVID-19 vaccines—BNT162b2 (Comirnaty^®), manufactured by Pfizer-BioNTech and mRNA-1273 (Spikevax^®), manufactured by Moderna—have been authorized for use in children as young as 6 months [8]. However, vaccine coverage among young children is very low. As of June 2023, just 10.0% of children aged 6 months to 4 years received at least one dose of COVID-19 vaccines in the United States [9]. Safety concerns and low perceived benefits of vaccinating young children are associated with parental hesitancy to vaccinate their children [10, 11]. The objective of this study is to compare the expected risk and benefits of the mRNA primary series of COVID-19 vaccines for children aged 6 months to 4 years in the United States, using QALY.

METHODS

The expected benefits and risks of the primary two-dose series of mRNA COVID-19 vaccines for children aged 6 months to 4 years from an individual perspective were stratified by sex, the presence of underlying medical conditions, the presence of infection-induced immunity, and the type of mRNA vaccine (BNT162b2 or mRNA-1273). Although a three-dose schedule of BNT162b2 as the primary series was recommended in this age group, vaccine effectiveness of BNT162b2 with a detailed timeframe (i.e. vaccine effectiveness by each month) was only available for two doses in recent US data [12, 13]. Therefore, we evaluated the benefits and risks of primary two-dose series for both BNT162b2 and mRNA-1273. The age group was divided into 6–11 months and 1–4 years. The Consolidated Health Economic Evaluation Reporting Standards statement was followed. The benefit–risk ratio and the benefit–risk difference were the outcomes of interest.

The model development and analyses were performed in Microsoft Excel 2016 (Redmond, WA, USA) and Stata 14.2 (StataCorp, College Station, TX, USA).

Overview of Model Structure and Indicator Identification Process

The overview of the outcome flow in the model is presented in Figure 1. Both unvaccinated and vaccinated children can be infected with SARS-CoV-2 with a reduced chance of infection in the vaccinated group compared with the unvaccinated group during the study comparison period. Conditions of infection are divided into asymptomatic cases, symptomatic non-hospitalized cases, hospitalized cases without ICU (intensive care unit) admission, and critical cases with ICU admission. While a proportion of infected cases lead to death, the rest of the infected population recover after the duration of each stage.

Figure 1.

Outcome flow. Both unvaccinated and vaccinated children can be infected with SARS-CoV-2 with a reduced chance of infection in the vaccinated group compared with the unvaccinated group. Conditions of infection are divided into asymptomatic cases (no disease burden assumed), symptomatic non-hospitalized cases, hospitalized cases without ICU admission and, critical cases with ICU admission. While a proportion of infected cases lead to death, the rest of the infected population recover after the duration of each stage. Abbreviations: COVID-19, coronavirus disease 2019; AEFI, adverse events following immunization; ICU, intensive care unit; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

We conducted the scoping literature review using PubMed to identify indicators in the model with search terms related with the name of each indicator, COVID-19 and children (Supplementary Tables 1–5). Given the most recent data may have only been reported in national reports without publication in the peer-reviewed journals, reports from CDC were also reviewed to evaluate US nationwide data. The initial search was conducted on June 22, 2023, followed by an additional search on November 10, 2023. When multiple studies were identified, the following steps were taken to select a study for each model indicator.

1. Data specifically focusing on children aged 6 months to 4 years were prioritized. If data for this specific age group were lacking, available data for the closest age range were used (e.g. children aged 5–11 years).

2. Studies directly providing each indicator value were prioritized over studies providing indirect data from which the indicator value would need to be estimated.

3. Studies with more detailed timeframe was prioritized (e.g. vaccine effectiveness data by each month were selected over that of data only reporting for the overall study period or for uncertain duration).

4. More recent data were prioritized.

5. Data from the US were selected over data from other countries.

6. Nationwide data, followed by multi-state data and state-wide data, were prioritized over data of a single center.

Quality assessments of studies informing the model indicators were conducted using the checklists of the Joanna Briggs Institute (Supplementary Table 6) [14].

Outcome

All benefits and risks were calculated as QALY change/100,000 vaccinees. The benefits and risks of vaccinating are presented as:

$${\displaystyle \begin{array}{l}{\displaystyle \begin{array}{llllll}& {\displaystyle \mathrm{V}\mathrm{a}\mathrm{c}\mathrm{c}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{b}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{f}\mathrm{i}\mathrm{t}=\mathrm{V}\mathrm{a}\mathrm{c}\mathrm{c}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{e}\mathrm{f}\mathrm{f}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{s}\mathrm{s}}& \times \mathrm{C}\mathrm{O}\mathrm{V}\mathrm{I}\mathrm{D}\text{-}19\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{e}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{b}\mathrm{u}\mathrm{r}\mathrm{d}\mathrm{e}\mathrm{n},& {\displaystyle \mathrm{V}\mathrm{a}\mathrm{c}\mathrm{c}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{s}\mathrm{k}=\mathrm{D}\mathrm{i}\mathrm{s}\mathrm{e}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{b}\mathrm{u}\mathrm{r}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{o}\mathrm{f}\mathrm{a}\mathrm{n}\mathrm{a}\mathrm{p}\mathrm{h}\mathrm{y}\mathrm{l}\mathrm{a}\mathrm{x}\mathrm{i}\mathrm{s}}& +\mathrm{m}\mathrm{y}\mathrm{o}\mathrm{c}\mathrm{a}\mathrm{r}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{s}+\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{o}\mathrm{g}\mathrm{e}\mathrm{n}\mathrm{i}\mathrm{c}\mathrm{i}\mathrm{t}\mathrm{y}& \mathrm{s}\mathrm{y}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{o}\mathrm{m}\mathrm{s}(\mathrm{g}\mathrm{r}\mathrm{a}\mathrm{d}\mathrm{e}3\mathrm{o}\mathrm{r}4).\end{array}}\end{array}}$$

The benefits and risks of vaccinating by completing the primary two-dose series versus no vaccine dose were calculated. The timeframe to consider vaccine benefits of two doses were for 5 months (from month 1 through month 5 since the first dose) [12], whereas that of vaccine risk were considered immediately after each dose.

Then, the benefit–risk ratio, defined as the ratio of benefits to risks in QALY change, was calculated as:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{B}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{f}\mathrm{i}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{s}\mathrm{k}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}=\mathrm{V}\mathrm{a}\mathrm{c}\mathrm{c}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{b}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{f}\mathrm{i}\mathrm{t}/\mathrm{v}\mathrm{a}\mathrm{c}\mathrm{c}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{s}\mathrm{k}.}\end{array}}$$

The benefit–risk ratio quantified the relative magnitude of benefits and risks. A benefit–risk ratio greater than 1 indicates that vaccine benefits are expected to outweigh vaccine risks.

The benefit–risk difference, defined as the difference in QALY change between benefits and risks, was calculated as:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{B}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{f}\mathrm{i}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{s}\mathrm{k}\mathrm{d}\mathrm{i}\mathrm{f}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}=\mathrm{V}\mathrm{a}\mathrm{c}\mathrm{c}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{b}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{f}\mathrm{i}\mathrm{t}-\mathrm{v}\mathrm{a}\mathrm{c}\mathrm{c}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{s}\mathrm{k}.}\end{array}}$$

The benefit–risk difference measured the absolute difference between benefits and risk. A benefit–risk ratio greater than 0 indicates that vaccine benefits are expected to outweigh vaccine risks.

Disease Burden Estimation

First, COVID-19 disease burden and AEFI were converted into QALY loss. QALY was estimated as

$${\displaystyle \begin{array}{lll}{\displaystyle \mathrm{Q}\mathrm{A}\mathrm{L}\mathrm{Y}=}& \mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}\mathrm{s}\mathrm{o}\mathrm{f}\mathrm{l}\mathrm{i}\mathrm{f}\mathrm{e}& \times \mathrm{h}\mathrm{e}\mathrm{a}\mathrm{l}\mathrm{t}\mathrm{h}\text{-}\mathrm{r}\mathrm{e}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d}\mathrm{q}\mathrm{u}\mathrm{a}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}\mathrm{o}\mathrm{f}\mathrm{l}\mathrm{i}\mathrm{f}\mathrm{e}\left(\mathrm{H}\mathrm{R}\mathrm{Q}\mathrm{L}\right)\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e},\end{array}}$$

where the HRQL score was a value with 0 being death and 1 being perfect health. We adopted a 3% discount rate per year in the base case to calculate QALY [15, 16].

The disease burden of COVID-19 in QALY loss was calculated as

$${\displaystyle \begin{array}{l}{\displaystyle \begin{array}{lllll}& {\displaystyle \mathrm{C}\mathrm{O}\mathrm{V}\mathrm{I}\mathrm{D}\text{-}19\mathrm{d}\mathrm{i}\mathrm{s}\mathrm{e}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{b}\mathrm{u}\mathrm{r}\mathrm{d}\mathrm{e}\mathrm{n}\left(\mathrm{Q}\mathrm{A}\mathrm{L}\mathrm{Y}\mathrm{l}\mathrm{o}\mathrm{s}\mathrm{s}\right)}& {\displaystyle =\mathrm{S}\mathrm{y}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{o}\mathrm{m}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{c}\mathrm{n}\mathrm{o}\mathrm{n}\text{-}\mathrm{h}\mathrm{o}\mathrm{s}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{i}\mathrm{z}\mathrm{e}\mathrm{d}\mathrm{i}\mathrm{n}\mathrm{f}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}{\displaystyle }}& {\displaystyle +\mathrm{h}\mathrm{o}\mathrm{s}\mathrm{p}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{i}\mathrm{z}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{w}\mathrm{i}\mathrm{t}\mathrm{h}\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{c}\mathrm{a}\mathrm{r}\mathrm{e}\mathrm{u}\mathrm{n}\mathrm{i}\mathrm{t}\left(\mathrm{I}\mathrm{C}\mathrm{U}\right)\mathrm{a}\mathrm{d}\mathrm{m}\mathrm{i}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}{\displaystyle }}& {\displaystyle +\mathrm{I}\mathrm{C}\mathrm{U}\mathrm{a}\mathrm{d}\mathrm{m}\mathrm{i}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}+\mathrm{d}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{h}.}\end{array}}\end{array}}$$

In this study, the disease burden of long COVID in this age group was not considered because the pediatric long COVID study with the largest sample size to date did not show a reduced HRQL in young children with SARS-CoV-2 infection compared with those without SARS-CoV-2 infection [17]. The burden of multisystem inflammatory syndrome was not considered given its negligible incidence among US children in the past year [18]. No disease burden was assumed in cases with asymptomatic SARS-CoV-2 infection.

The reported incidence of COVID cases and COVID-19 deaths in the United States were obtained from August 2022 to March 2023, the duration that the nationwide COVID-19 case and death data among children aged 6 months to 4 years stratified by the vaccination status were reported [19]. The incidence of COVID-19 hospitalization and ICU admission in the same period were obtained through COVID-NET [20]. In the model, we assumed that 53.8% of hospitalized cases with SARS-CoV-2 infection were admitted due to COVID-19 as a primary reason for admission [21]. Underestimation was considered in the study, and the underestimation ratios were also obtained from US studies [22].

The effectiveness of prior infection against reinfection and against severe disease was reported in children younger than 5 years [12]. The seroprevalence of infection-induced immunity stratified by age group was obtained from the nationwide surveillance [23, 24].

The HRQL score and the life expectancy of the general US population were obtained from the national data [25–27]. The baseline HRQL scores in pediatric population with and without underlying medical conditions were obtained from proxy reports by parents [25]. The reductions in HRQL scores for symptomatic non-hospitalized cases, hospitalized cases, and cases with ICU admission were estimated by the difference between the HRQL score in the non-infected population versus that in those with SARS or febrile respiratory infection [28, 29].

Vaccine Effectiveness

Based on recent US data, monthly vaccine effectiveness of primary two-dose series of monovalent vaccines against any infection was reported among children ages 6 months to 4 years for 5 months (month 1 to month 5 since the first dose) stratified by mRNA vaccine type (BNT162b2 and mRNA-1273) and by the presence of prior infection [12]. While vaccine effectiveness against symptomatic infection is potentially higher than that against any infection, in our model we assumed the vaccine effectiveness against symptomatic infection was the same as against any infection. The vaccine effectiveness against severe disease was also assumed to be same as against infection given the lack of data in this specific age group as well as data showing comparable vaccine effectiveness against infection and severe disease among children aged 5–11 years [12].

Although effectiveness data for bivalent vaccines or 2023–2024 updated vaccines used as a primary series are yet to be published at the time this study was conducted, we explored the additional impact of bivalent compared with monovalent booster shots from the difference in vaccine effectiveness of bivalent and monovalent booster doses in children aged 5–11 years [12]. Because of this uncertainty regarding the vaccine effectiveness, a sensitivity analysis was conducted.

Vaccine Risk

The vaccine adverse reactions attributable to COVID-19 vaccination (vaccine risks) were divided into anaphylaxis, myocarditis, and reactogenicity symptoms. We did not include any passive surveillance data to estimate the incidence or risk of adverse reactions but instead restricted our data to those coming from active surveillance systems or large healthcare databases.

Given the lack of pediatric data with a large sample size regarding the incidence of anaphylaxis, the incidence data in adults were used in the model [30]. In the model, all cases with anaphylaxis were assumed to recover [31].

At the time the study was conducted, there was no active surveillance data with a large sample size on the incidence of myocarditis following COVID-19 vaccination or the risk of myocarditis attributable to the COVID-19 vaccination among ages 6 months to 4 years. Instead, we conservatively used the VSD data among children 5–11 years old for the incidence of myocarditis following COVID-19 vaccination [32]. In the model, we assumed that the length of hospital stay for myocarditis cases following COVID-19 was 2 days, 18.7% of which required ICU admission [33]. We assumed the reduction of HRQL score by 0.07 (the difference in scores of visual analog scale between those with ongoing myocarditis vs the general population) in cases with myocarditis following immunization for 143 days after the onset of myocarditis [34]. Because the overall HRQL score in cases with myocarditis following immunization at 143 days of median interval after the onset was similar to that in the general population, we did not consider the life-long impact in HRQL among cases with myocarditis [32, 33]. No death was assumed in cases with myocarditis [33].

Regarding reactogenicity symptoms, an adult study showed that symptoms with grade 1 or 2 were not associated with reduced HRQL score [35]. Therefore, the risk of developing grade 3 or 4 symptoms was obtained from randomized controlled trials [36–39]. The attributable risk was calculated as the rate of AEFI in the vaccination group subtracted by the rate of AEFI in the placebo group.

Sensitivity Analysis

Vaccine effectiveness, incidence of COVID-19, and COVID-19 severity (e.g. COVID-19 death) are subject to vary by new variants and other factors. In addition, the HRQL reduction data by acute COVID-19-related outcomes and adverse reactions (e.g. anaphylaxis and myocarditis) were limited at the time the study was conducted, which indicated the need to explore the impact of uncertainties on the reduction of HRQL scores by these conditions through the sensitivity analysis. Therefore, vaccine effectiveness and incidence of COVID-19, COVID-19 death, reduction in HRQL scores in COVID-19 outcomes, and reduction in HRQL scores in AEFI outcomes (i.e. anaphylaxis, acute myocarditis, and reactogenicity symptoms grade 3 or 4 attributable to the vaccination) were included in the sensitivity analyses.

First, a one-way sensitivity analysis was conducted using the above indicators. In addition, the probabilistic sensitivity analysis was implemented with 1,000 iterations using the same indicators evaluated in the one-way sensitivity analysis. Log-normal distribution was used for vaccine effectiveness, and beta-PERT distribution based on most likely, minimal, and maximal values was applied for other indicators.

Ethics

A review of human subjects research was exempted because the study only used publicly available data.

RESULTS

The estimated monthly average disease burden of COVID-19 ranged from 2.8 among females aged 1–4 years with infection-induced immunity and without underlying medical conditions to 30.4 in males aged 6–11 months without infection-induced immunity and with underlying medical conditions (Supplementary Table 7). The estimated rates of AEFI for the primary series of BNT162b2 were 0.52 and 0.48 for males and females, and for mRNA-1273 were 2.45 and 2.41 for males and females, respectively (Supplementary Table 7).

The benefit–risk ratios ranged from 200.4 in BNT162b2 for males aged 6–11 months with underlying medical conditions and without infection-induced immunity to 3.2 in mRNA-1273 for females aged 1–4 years without underlying medical conditions and with infection-induced immunity. The benefit–risk ratios were larger in children aged 6–11 months compared with those in children aged 1–4 months, in males compared with females, in children with underlying medical conditions compared with children without underlying medical conditions, and in children without infection-induced immunity compared with children with infection-induced immunity (Table 1).

Table 1.

Benefit–Risk Ratios and Benefit–Risk Differences Stratified by Age Group, Sex, Presence of Underlying Medical Condition, and Presence of Infection-Induced Immunity

	Male With Underlying Medical Conditions	Male Without Underlying Medical Conditions	Female With Underlying Medical Conditions	Female Without Underlying Medical Conditions
Benefit–risk ratios
Two-dose BNT162b2
Aged 6–11 months with no infection-induced immunity	200.4	122.1	194.1	118.5
Aged 6–11 months with infection-induced immunity	75.0	45.6	72.6	44.3
Aged 1–4 years with no infection-induced immunity	72.1	48.3	70.6	47.6
Aged 1–4 years with infection-induced immunity	26.8	17.9	26.2	17.6
Two-dose mRNA-1273
Aged 6–11 months with no infection-induced immunity	39.3	24.0	35.7	21.8
Aged 6–11 months with infection-induced immunity	14.6	8.9	13.2	8.1
Aged 1–4 years with no infection-induced immunity	14.1	9.5	13.0	8.8
Aged 1–4 years with infection-induced immunity	5.2	3.5	4.8	3.2
Benefit–risk differences (QALY change/100,000 vaccinated children)
Two-dose BNT162b2
Aged 6–11 months with no infection-induced immunity	104.1	63.3	93.0	56.6
Aged 6–11 months with infection-induced immunity	38.6	23.3	34.5	20.8
Aged 1–4 years with no infection-induced immunity	37.1	24.7	33.5	22.5
Aged 1–4 years with infection-induced immunity	13.5	8.8	12.1	8.0
Two-dose mRNA-1273
Aged 6–11 months with no infection-induced immunity	93.9	56.3	83.6	50.1
Aged 6–11 months with infection-induced immunity	33.2	19.2	29.4	17.0
Aged 1–4 years with no infection-induced immunity	32.2	20.8	28.9	18.7
Aged 1–4 years with infection-induced immunity	10.3	6.0	9.1	5.3

Abbreviations: AEFI, adverse events following immunization; QALY, quality-adjusted life year.

The benefit–risk differences ranged from 104.1 QALY/100,000 vaccinated children in BNT162b2 for males aged 6–11 months with underlying medical conditions and without infection-induced immunity to 5.3 QALY/100,000 vaccinated children in mRNA-1273 for females aged 1–4 years without underlying medical conditions and with infection-induced immunity.

In children ages 6 months to 4 years, our model estimated that benefit–risk ratios were greater than 1 or benefit–risk differences were more than 0, irrespective of sex, presence of underlying medical conditions, presence of infection-induced immunity, and type of mRNA vaccines in the base case scenario. The one-way sensitivity analysis of benefit–risk ratio for receiving primary dose series of mRNA vaccines is presented in Supplementary Table 8. In all scenarios in the one-way sensitivity analysis, the benefit–risk ratio ranged from 1.0 in mRNA-1273 for females aged 1 to 4 years without underlying medical condition with infection-induced immunity in low vaccine effectiveness scenario to 2,721.5 in BNT162b2 for females aged 6 to 11 months with underlying medical condition without infection-induced immunity in low reduction of AEFI HRQL score scenario. Among females aged 1–4 years without underlying medical conditions and with infection-induced immunity, the group with the lowest benefit–risk ratio, the benefit–risk ratio could be less than 1 in the following scenarios.

1. if vaccine effectiveness was below 3.5% for BNT162b2 and 17.4% and mRNA-1273,

2. if COVID-19 incidence was less than 5.5% for BNT162b2 and 31.2% for mRNA-1273 of the estimated COVID-19 incidence between August 2022 and March 2023, and

3. if COVID-19 death was less than 0.7% of the estimated death incidence between August 2022 and March 2023 for mRNA-1273.

The results of the probabilistic sensitivity analysis are presented in Supplementary Figure 1. The probabilities of benefit–risk ratio larger than 1 were >99% for all groups except for children with infection-induced immunity for mRNA-1273. For mRNA-1273, among children with infection-induced immunity, the probabilities of benefit–risk ratio larger than 1 in those aged 1–4 years were 98.1% in males with comorbidity, 96.5% in males without comorbidity, 97.7% in females with comorbidity, and 95.6% in females without comorbidity.

DISCUSSION

This study found that the benefits of receiving the primary series of the mRNA COVID-19 vaccine outweighed the risks among children ages 6 months to 4 years regardless of sex, presence of underlying medical conditions, the presence of infection-induced immunity, and type of mRNA vaccine in the base case analysis. However, if the situational changes (e.g. the significant reduction in the COVID-19 incidence or vaccine effectiveness) occur, the benefit of the vaccination was reduced. The disease burden of COVID-19 was larger in children aged 6–11 months compared to those aged 1–4 years.

Currently, the 2023–2024 updated COVID-19 vaccines are used for primary series, and the monovalent or bivalent vaccines are no longer used [40]. Of note, the data regarding the assumed bivalent vaccine effectiveness as primary dose series need to be interpreted with caution. While continuous monitoring of benefits and risks is still important, the available data to date and our study results suggest that the benefits of the currently available mRNA COVID-19 vaccines likely outweigh the risks among children ages 6 months to 4 years.

This study did not consider herd immunity or other secondary impacts, including parental productivity loss. The consideration of these factors would have resulted in larger benefits than those estimated. However, the evaluation of benefits and risks solely from the perspective of each individual’s health is important to many for making their vaccine decisions.

The study focused on the United States. Each country has different data regarding the epidemiology of COVID-19. Thus, our methodology of evaluating the benefits and risks of vaccinating using a single health outcome scale should be considered globally.

There are a few limitations in our model. First, data regarding health utility or HRQL score specifically for young children are scarce. Therefore, we had to use data from older children or adults for some disease outcomes. In general, the relative lack of pediatric HRQL data as well as the difficulty measuring HRQL scores (e.g. the need for proxy-reported score) for young children limits the quantitative evaluation of multiple outcomes in QALY scale for this specific age group. While more research to evaluate health utility in young children is needed, this does not negate our conclusions. Even if we assumed no reduction in HRQL scores by COVID-19-related outcomes except for death, the benefit of receiving COVID-19 vaccines still outweighs the risk in young children. Secondly, given the lack of data regarding the incidence of myocarditis after COVID-19 vaccination in children 6 months to 4 years old, we used the incidence among 5–11 year olds in our model. Because the incidence of myocarditis in 5–11 year olds is much less than that of older age groups (i.e. 12–17 years old), we expect the incidence of myocarditis in children 6 months to 4 years to likely be even less than that in 5–11 year olds [29]. With the evidence of smaller incidence of myocarditis after COVID-19 vaccination in children 6 months to 4 years old than those of older age groups, the benefit–risk ratios in our study results would have been higher (i.e. more favoring the vaccination).

In conclusion, this study showed that the expected benefits of receiving the primary series of mRNA vaccines outweighed the risks among children ages 6 months to 4 years regardless of sex, presence of underlying medical conditions, the presence of infection-induced immunity, or type of mRNA vaccine. These results support the current recommendation that both children with underlying medical conditions and otherwise healthy children receive mRNA COVID-19 vaccines. Continued monitoring of data on vaccine effectiveness, vaccine safety, and COVID-19 disease epidemiology is needed for regularly updated risk/benefit analyses.

Note

Data availability statement. The authors confirm that the data supporting the findings of this study are available within the article or its supplementary materials.

Financial support. No funding was secured for this study.

Potential conflicts of interest. Dr Salmon has received unrelated research grants from Merck, Vaccination Confidence Fund, which is jointly funded by Facebook and Merck and serves on advisory boards for Merck, Janssen, and Moderna. Dr Dudley has received unrelated research grants from Merck and Vaccination Confidence Fund, which is jointly funded by Facebook and Merck. All other authors report no conflicts.