FormalPara Key Summary Points

Why carry out this study

COVID-19 booster vaccination programs have been successful in reducing hospitalization and deaths associated with COVID-19 during the Delta and Omicron periods.

Many modelling analyses do not consider the full complexity of COVID-19 transmission dynamics, such as booster vaccinations, vaccine waning and hybrid immunity.

This study used the most recently available data to assess the impact of the autumn 2022 and spring 2023 booster vaccinations on public health outcomes including hospitalizations, NHS occupancy (a key determinant of bed capacity), patient productivity, and mortality across several scenarios.

What was learned from this study

The model estimated that the autumn 2022 and spring 2023 vaccination campaign caused a 14% and 16% reduction in general ward and ICU admissions respectively, when compared with a no booster scenario. Similar reductions were observed in general ward and ICU bed days (14% and 16% reduction respectively). Patient productivity was partially restored with the number of paid and unpaid productive days lost reduced by 3% and 7%. Overall mortality decreased by 15%.

Further benefits in health and productivity outcomes were observed in additional scenarios with increased eligibility and uptake, with greatest clinical benefit predicted when coverage was highest in both autumn and spring booster campaigns with a 64% further reduction in hospitalizations and approximately twice as many ICU admissions prevented.

Expanding the eligibility for the COVID-19 booster vaccine programs has the potential to substantially reduce mortality, reduce general ward and ICU bed days and further alleviate the healthcare burden due to COVID-19.

Introduction

Current evidence suggests that the rapid development and implementation of the COVID-19 vaccination program in the UK in December 2020 achieved a substantial reduction in the number of clinically significant cases [1], the number of serious complications such as long COVID [2], hospitalizations associated with the disease [3], and ultimately COVID-19 mortality rates [3]. Primary vaccine courses were followed by booster campaigns against the Delta variant in 2021/2022 [4]. The Omicron variant emerged in late 2021, and was associated with higher rates of transmission, but less severe outcomes [5]. However, vaccines were less effective against the Omicron variant and vaccine effectiveness (VE) was observed to wane more rapidly [6, 7]. The VE against symptomatic disease from the Omicron variant was reported to drop to 8.8% after 25 weeks post-immunization whereas the VE against the Delta variant at the same timepoint was reported at 43.5% [7].

An updated bivalent vaccine booster program was initiated in autumn 2022 [8,9,10], and continued in spring 2023 [11] targeting vulnerable individuals at risk of developing severe COVID-19 or those in contact with individuals at risk of developing severe COVID-19, such as frontline health or social care workers. The emergence of the Omicron lineage of SARS-CoV-2 was hallmarked by an increase in infection rates, partly due to the variant’s ability to evade pre-existing immunity mechanisms [12, 13]. As the disease changes to a more endemic state, analyses accounting for the role of hybrid immunity from previous COVID-19 infections and vaccinations, vaccine waning and differences in vaccine efficacy against new variants are needed to help inform public health strategies for controlling SARS-CoV-2.

The aim of this study was to quantify the public health impact of the autumn 2022 and spring 2023 booster vaccination program in the UK over a 48-week time horizon, from September 2022 to August 2023, and to explore the impact of alternative strategies with extended eligibility and/or improved uptake. Exploring the impact of expanding the eligibility criteria or increasing booster vaccine uptake rates can help inform the design of future booster campaigns. This study can provide valuable insight into the benefits of vaccination programs in both the current eligible population and broader eligibility populations.

Methods

Overview

A previously described age- and risk-structured deterministic compartmental Susceptible-Exposed-Infectious-Recovered (SEIR)-type model [14, 15] was adapted and extended to evaluate the public health impact of six strategies for the autumn 2022/spring 2023 booster vaccination campaign in the UK over a 48-week time horizon as presented in Table 1.

Table 1 Summary of vaccination strategies considered
Full size table

Model Overview

The SEIR model of SARS-CoV-2 transmission [14] was expanded using R® [16] to include general hospital ward and intensive care unit (ICU) compartments, and COVID-19 deaths within the dynamic transmission system (Fig. 1).

Fig. 1
figure 1

Model schematic. A SEIRD compartmental model was developed where subscript k = vaccination status; a = stratification by age. Parameters in blue indicate fixed/calculated parameters and those in red show calibrated parameters. S: susceptible; Vk=0,1,2,3,4: susceptible, vaccinated; Ea,k: exposed; Iaa,k: infected asymptomatic; Isa,k: infected symptomatic; Ta,k: treatment; Ha,k: hospital general ward; ICa,k: intensive care unit; Ra,k: recovered; Da: covid death. Hybrid immunity (HI1–4) replaces Ra,k in Vk=0

Full size image

The UK population was stratified into age groups in line with contemporaneous vaccine recommendations from the Joint Committee on Vaccination and Immunisation (JCVI; Supplementary Table 12) [9], as well as two currently ineligible pediatric age compartments; ≤ 0.5 years and 0.5–4 years. Each age group was further stratified into high- and low-risk cohorts with clinical risk identified by the UK Health and Security Agency (UKHSA) and defined as those at heightened risk of experiencing severe illness—a considerable 24.4% of the UK population [17]. The population was also stratified by vaccination class to include unvaccinated, protected by vaccination with one dose, two doses or a booster dose and waned from booster protection. Booster doses and waned booster populations were further compartmentalized into pre-autumn 2022 boosters and autumn 2022/spring 2023 boosters.

A hybrid immunity (HI) compartment was included for each vaccination strata, containing vaccinated individuals who have recovered from prior infection, or previously infected individuals who have since been vaccinated. Notably, it was assumed that the combined effect of infection- and vaccine-derived immunity for individuals in the HI compartment was greater against infection, but was the same against hospitalization as protection from vaccine-induced immunity alone [18]. The starting population in the HI compartment for Vk=1–Vk=3 (Fig. 1) was 10.5%. The proportion of hybrid immunity per age group is provided in Supplementary Table 14. The waning immunity rate from HI was assumed to be the same as for individuals with only infection- or vaccine-induced immunity.

Model Inputs

Baseline inter-age contact rates in the population were calculated from contact matrices reported in the POLYMOD study [19]. Baseline age-specific probabilities of hospitalization from infection, ICU admission from infection and death from severe infection requiring hospitalization (Supplementary Table 5) were sourced from an England-based analysis conducted at the beginning of the pandemic [20], and assumed these were representative of the UK. Age-stratified severity scaling factors were calibrated to account for changes in severity between clinical metrics from the beginning of the pandemic and those relevant to the study period during Omicron dominance (Supplementary Table 6). VE against infection, symptomatic disease, hospitalization, ICU admission and death for each vaccination status and age group are taken from UKHSA data (Supplementary Table 1), assuming 100% Omicron variant distribution.

The model assumed 12-week cycles for vaccination campaigns. In the model, the population in the compartments other than symptomatic infected, inpatient, and death were eligible for booster vaccinations following the criteria of each campaign. For factual strategies, weekly vaccination uptake rates were estimated from observed uptake rates from UKHSA data (Supplementary Table 2). Uptake rates for high-risk individuals, including immunosuppressed patients, outside of the eligible age groups were derived from the Department for Health and Social Care (DHSC) autumn booster impact assessment [21]. Age-stratified uptake rates were obtained from the UKHSA reports (Supplementary Table 2). Assumptions and sources for counterfactual strategies are presented in Supplementary Table 3.

The number of long COVID cases was estimated using two conditional probabilities: the probability of experiencing persistent symptoms 4–12 weeks after the initial onset of infection, termed ‘ongoing COVID symptomatic’ and the probability of experiencing symptoms 12 weeks or more after the initial onset of infection, termed ‘post COVID syndrome’ (Supplementary Table 7). In both cases, the conditional probabilities were stratified by age (< 18 or ≥ 18) and by the level of care received, either receiving outpatient care, receiving inpatient care in the general ward or receiving inpatient care in the ICU [22].

The input parameters were estimated under uncertainty and the sensitivity was analysed via a deterministic sensitivity analysis (DSA). The model outputs are subjected thus to parameter, methodological, and structural uncertainty.

Model Analysis and Outputs

The public health impact of various alternative vaccination strategies was quantified on the basis of the changes in the following outputs:

  • Number of infections, calculated on the basis of the relative incidence of infection to both asymptomatic and symptomatic compartments.

  • General ward and ICU admissions, calculated from the compartmental occupancy of hospitalisation (H) and ICU (IC).

  • Corresponding general ward and ICU bed days, using the average number of days of hospital and ICU stay per admission.

  • Long COVID cases, which were calculated on the basis of the number of symptomatic infections, inpatient cases, and probabilities of long COVID obtained from literature. Double counting was mitigated by accounting for outpatient and inpatient cases separately.

  • The associated number of productive days lost as a result of hospitalization, for those in outpatient care, and as a result of fatal COVID infection. The hospitalization length of stay was assumed to directly correlate with reduced patient productivity. Patient productivity loss was conservatively calculated on the basis of ONS estimates [23, 24] and published estimates [25, 26] of paid and unpaid productive time. Additionally, ONS estimates were used to calculate premature deaths associated with COVID infection [27].

  • The number needed to vaccinate (NNV) to avoid one hospitalization (NNVh), one ICU admission (NNVi), and one death (NNVd) were also estimated by dividing the number of additional vaccinations required by the predicted number of averted hospitalizations or deaths over the prediction period.

The impact of each vaccination strategy was considered through the incremental difference of these outcomes compared to the counterfactual no booster strategy.

To explore whether the booster campaign could be optimized, several different strategies were considered, as presented in Table 1. These strategies considered the benefit of the current campaign compared to a no booster strategy, or a strategy where only one booster is administered per year. They then expand on the current strategy to consider the impact of increasing the current eligibility criteria and how increasing the uptake of the booster would affect hospitalizations, deaths, and wider societal benefits, such as reduced productivity loss. Improved uptake was defined by the maximum observed uptake rate among all in eligible age groups in the corresponding campaign. By modelling these strategies with a broader population or greater uptake, it is hoped that policy makers can identify areas in the current strategy that could be modified to ensure that future booster campaigns are as beneficial as possible.

Sensitivity Analysis

Sensitivity analyses were carried out to determine how the range of key parameters within the model impact key outcomes. Uncertainty around the clinical parameters such as relative infectiousness of symptomatic versus asymptomatic cases, duration of immunity (infection, vaccine, and hybrid), duration of vaccine protection, risk ratios, and length of stay for hospitalization and ICU were tested via DSA (Supplementary Table 8). Parameter ranges for sensitivity analysis were sourced from the literature (Supplementary Table 8).

Calibration

The model was calibrated over a 36-week period from January to August 2022 against three calibration targets: PCR-positivity estimates [28], hospital admissions [29], and COVID-19 mortality [30]. A simultaneous, non-trivial parallel flow approach was taken to estimate PCR positivity from modelled infections, per existing literature [31]. A full description of this approach is outlined in the Supplementary Material (Parallel flow). All three calibration targets were scaled, using the L-BFGS-B method in optim function in R® [32], and equally weighted to minimize the negative log-likelihood function or equivalently, maximize the log-likelihood of recovering input data. Further description of model calibration and results are presented in the Supplementary Material (Model calibration).

Statement of Ethics Compliance

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Results

Base Case

The impact of vaccination during the autumn 2022/spring 2023 booster period, hereafter referred to as autumn/spring, was modelled for several health and societal outcomes. Outcomes were compared against a counterfactual scenario where no booster was administered during autumn/spring, hereafter referred to as ‘no booster.’ When compared to a counterfactual no booster scenario, the model estimated that hospitalizations, categorized by general ward and ICU admissions, decreased by 17,785 (14% reduction; NNVh 1357), and 1136 (16% reduction; NNVi 21,247; Tables 2 and 4), respectively. General ward and ICU bed days (i.e. occupancy) were decreased by 237,253 (14%) and 16,184 (16%), respectively, and the number of paid and unpaid productive days lost decreased by 3.3 million (3%) and 0.5 million (7%), respectively (Table 3). An estimated 1463 (15%; NNVd 16,492) deaths were averted in the autumn/spring booster strategy compared with no booster strategy.

Table 2 Estimated clinical outcomes averted across all scenarios
Full size table
Table 3 Estimated productivity outcomes averted across all scenarios
Full size table

Given the considerable benefit of the autumn/spring booster on clinical outcomes, several strategies were considered where the eligibility and uptake of the vaccine were increased to determine how changing these factors could further enhance the benefit of the booster. The absolute results for each scenario are presented in Supplementary Tables 15 and 16.

Mortality

Those at highest risk of the most severe outcomes, including death, were vaccinated under the current eligibility criteria. When counterfactual scenarios were modelled whereby the eligibility of the booster campaign was expanded to include all adults and children over the age of 12, or vaccine uptake was increased to the highest uptake observed in any eligible age group during the actual autumn 2022/spring 2023 campaign (Supplementary Table 2) additional benefits in deaths averted were observed. The strategy that provided the greatest protection from mortality was expanding eligibility criteria of autumn and spring with improved uptake levels. In this strategy, the booster campaign was estimated to avoid 1877 (19%; NNVd 36,261) deaths compared to the no booster strategy (Tables 2 and 4).

Table 4 Estimated numbers needed to vaccinate (NNV) across all scenarios
Full size table

Figure 2 demonstrates additional insights into the patterns of COVID-19 deaths associated with the different vaccination strategies. All booster strategies reduced the number of deaths compared with a no booster strategy. Notably, the smallest reduction in deaths averted occurred where the booster was given in autumn 2022 only with 990 deaths averted (10% reduction; NNVd 20,047; Tables 2 and 4), compared with a no booster strategy, demonstrating that a biannual campaign is required for prevention of deaths due to vaccine waning.

Fig. 2
figure 2

Number of COVID-19 deaths averted against time for the whole population for each strategy. Strategy I: factual autumn 22/spring 23 booster campaign. Strategy II: autumn 22 booster only. Strategy III: autumn 2022/spring 2023 with improved uptake. Strategy IV: expanded autumn and expanded spring. Strategy V: expanded autumn and expanded spring with improved uptake. Strategy II had increased infections and thus deaths within the first 24 weeks of the autumn booster, due to a higher number of individuals in the susceptible and waned booster compartments

Full size image

Hospitalizations

When compared with the no booster scenario, increasing uptake rates for the autumn and spring campaigns with eligibility remaining the same led to an estimated reduction in general ward and ICU admissions by 18,013 (15%; NNVh 1428) and 1322 (18%; NNVi 20,877; Tables 2 and 4), respectively. The number of general ward admissions averted increased to 23,894 (20%; NNVh 1960; Fig. 3) with ICU admissions averted increased to 1840 (25%; NNVi 27,423; Fig. 4) when eligibility was expanded for both the autumn and spring campaigns, and uptake rates remaining the same. The greatest benefit was observed when eligibility was expanded and uptake rates increased in both autumn and spring campaigns to prevent 27,371 (23%; NNVh 2299) general ward and 2236 (31%; NNVi 30,441) ICU admissions compared with the no booster campaign. When considering hospital bed days, expanding the eligibility criteria and increasing uptake of both the autumn and spring campaigns had the greatest impact, reducing time spent in the general ward by 390,030 days (23%) and reducing time spent in the ICU by 31,867 days (31%).

Fig. 3
figure 3

Number of general ward admissions averted against time for the whole population for each strategy. Strategy I: factual autumn 22/spring 23 booster campaign. Strategy II: autumn 22 booster only. Strategy III: autumn 2022/spring 2023 with improved uptake. Strategy IV: expanded autumn and expanded spring. Strategy V: expanded autumn and expanded spring with improved uptake

Full size image
Fig. 4
figure 4

Number of ICU admissions averted against time for the whole population for each strategy. Strategy I: factual autumn 22/spring 23 booster campaign. Strategy II: autumn 22 booster only. Strategy III: autumn 2022/spring 2023 with improved uptake. Strategy IV: expanded autumn and expanded spring. Strategy V: expanded autumn and expanded spring with improved uptake

Full size image

Conversely, when comparing the no booster strategy to the autumn only strategy, reductions in general ward admissions of 11,501 (10%; NNVh 1598) and ICU admissions of 919 (13%; NNVi 21,596) were observed (Tables 2 and 4), substantially less than the factual observed campaign of autumn 2022/spring 2023, demonstrating that a biannual campaign is required for a reduction in clinical outcomes.

Long COVID

The autumn/spring booster campaigns were estimated to prevent 22,293 (2%) long COVID cases (Table 2). However, improved uptake levels in autumn and spring had the potential to prevent 28,838 (3%) cases compared to no booster strategy, which increased to 103,701 (11%) when the eligibility was expanded in the autumn and the spring, but uptake remained at the observed rates. The greatest reduction in long COVID cases was observed when the eligibility of both the autumn and spring strategies was expanded to include all adults and children over the age of 12, and uptake rates were increased, which was estimated to prevent 152,243 (16%) long COVID cases compared to a no booster strategy.

Productivity

When the wider impact of vaccination on individual productivity was considered, increasing the eligibility and/or increasing the vaccine uptake rates of the COVID-19 booster had the potential to substantially reduce productivity loss. This was observed in the strategy considering increased uptake rates for the autumn and spring booster, where productivity loss was estimated to decrease by over 4.9 million (5%) days averted (Table 3). This benefit was increased when eligibility was increased in the autumn and spring booster campaigns, which averted 13.7 (13%) million productive days lost compared to a no booster strategy. The greatest decrease in productivity loss was observed in the strategy considering expanded eligibility for both the autumn and spring boosters with improved uptake which averted 19.6 million (18%) productive days being lost. Similar trends were observed in productivity loss averted associated with deaths due to COVID infection. Expanding eligibility criteria in autumn and spring boosters would have resulted in 4.2 million (20%) productive days lost averted when compared with a no booster strategy. A further breakdown of averted paid and unpaid productive days lost is described in Table 3. This reduction in productivity lost could have substantial impact on the economic and societal burden of COVID-19.

Deterministic Sensitivity Analysis

One-way sensitivity of model outputs to certain parameters were selected for DSA, as shown in Fig. 5. The relative infectiousness of symptomatic and asymptomatic cases had the greatest influence across all clinical outcomes, followed by the mean duration of recovery for symptomatic and asymptomatic infection and risk ratios for hospitalization, ICU, and deaths. When the relative infectiousness of symptomatic versus asymptomatic cases was varied by 2.1–7.2 times the base case value, the number of averted hospitalizations ranged from 2.2 in the upper limit and 0.6 in the lower limit. Similar results were observed for averted ICU admissions and deaths which were increased by 2.1-fold and reduced by 0.6-fold of the base case results (Fig. 5).

Fig. 5
figure 5

Deterministic sensitivity analysis demonstrating prevented a hospitalizations, b ICU admissions, and c deaths

Full size image

Discussion

Given the evolving landscape of COVID-19 in the UK and globally, where new variants will continue to emerge and the profile of population immunity will shift over time, it remains crucial to account for the role of hybrid immunity and consider the benefit of adapting future vaccination campaigns. Since the emergence of SARS-CoV-2, studies have used mathematical modelling to investigate population dynamics and the effectiveness of public health interventions to control the COVID-19 pandemic [33,34,35,36,37]. While the effect of vaccination on COVID-19 transmission has been modelled in the UK before [14, 15], this study builds on the dynamics of the disease and vaccination strategy by additionally considering hybrid immunity to COVID-19. A growing body of evidence strongly suggests that hybrid population immunity profiles, acquired from both prior infection and vaccination, are potentially more protective against surges in COVID-19 infections and healthcare demand [38,39,40,41].

This study provides novel insights into the impact of the autumn 2022 booster campaign in the UK, explicitly accounting for the role of hybrid immunity in the population. Beyond the impact on averting infections and hospital admissions, this study quantified the impact of boosters in averting long COVID and patient productivity—two outcomes that are considered to have serious clinical and societal implications [42, 43]. The model assessed the impact of vaccination across several health outcomes for six vaccination strategies, compared to a counterfactual no booster campaign. Under conservative assumptions, the most beneficial results across all health outcomes were observed when coverage was expanded to all adults and children aged 12 and over, during both the autumn 2022 and spring 2023 campaigns with improved vaccine uptake rates observed during the campaigns. Enhanced uptake may be achieved by public health information campaigns driven by government and public health authorities [44].

Previous studies indicate booster vaccination has been consistently advantageous, with an estimated 100,000 hospitalizations and 23,000 deaths due to COVID-19 averted in England between October 2021 and December 2022 when compared with a counterfactual no booster scenario [45]. This study looked at a different time horizon, focusing on the current landscape of the COVID-19 pandemic dominated by the Omicron lineage of subvariants. Published evidence has shown that initial doses of the vaccine provided high levels of protection from severe disease, which is now diminished via waning and immune escape of the Omicron variant. Administration of a booster vaccine developed against the Omicron variant versus the ancestral vaccine designed for previous strains has shown that booster vaccines targeting specific and current circulating variants result in substantially more reductions in the number of severe outcomes associated with disease, demonstrating the long-term applicability of COVID booster vaccines for public health [46].

Current JCVI eligibility criteria recommend booster vaccinations for those at highest risk of developing severe COVID-19 or those with a higher risk of contracting and transmitting SARS-CoV-2 through occupational exposure, such as healthcare workers [47]. Expanding vaccine eligibility criteria to all adults and children over the age of 12 may indirectly protect those of advanced age and at risk of severe complications from COVID-19, which aligns with previous studies [48, 49]. Therefore, our study supports a growing argument that expanding the eligibility for COVID-19 booster vaccination would provide direct and indirect protection against serious COVID-19-related outcomes in the most vulnerable members of society, as well as the wider population [50].

COVID-19 is associated with reduced productivity [51, 52], which has widespread effects such as workplace absenteeism or inability to care for others. A recent study demonstrated that vaccinated individuals with COVID-19 had lower absenteeism rates compared to unvaccinated individuals (45.6% versus 65.0%) [51]. By increasing eligibility criteria to include standard risk adults for the autumn and spring booster campaigns, this study predicts a substantial reduction in productivity loss associated with COVID-19, which could result in significant economic benefits.

Further benefits of the booster campaign on long COVID have been modelled here. Long COVID has a considerable detriment to productivity and has a significant societal burden with 3.6% of individuals in the UK self-reporting as experiencing long COVID. Over half (57.0%) of those individuals reported that the condition negatively affected their well-being, and one-third reported that long COVID impacted their work [53]. Long COVID has a substantial clinical burden with an increased cost of £23.4 million attributed to primary care consultations per year in the UK [54]. Self-reported long COVID has been described to affect people across all age groups, with women most susceptible to the condition [53, 55]. Further, there is a 25% lower risk of long COVID in those aged > 70 years, and 6.0% lower risk in those aged 30–39 compared with those aged 18–30, indicating that long COVID may disproportionately affect younger generations [56]. In a study of 672 individuals with long COVID, defined as experiencing symptoms 12 months post-infection, a reduction in work ability scores was reported compared with those without long COVID symptoms [57]. Here, expansion of the eligibility criteria to include the wider population and increasing uptake rates for the 2022 autumn and 2023 spring booster campaign was estimated to substantially offset long COVID cases and could further restore productivity. Further analyses are required to examine this full productivity gain and the wider societal benefits associated with booster vaccination.

A reduction in hospital bed days by the administration of the autumn and spring booster demonstrates the potential benefit that vaccination may have on healthcare resources and bed capacity. Even small changes in occupancy translate to large differences in real-world hospital settings, particularly in the post-pandemic era. The NHS has experienced overwhelming pressures since the COVID-19 pandemic [58] and measures to alleviate such pressures, particularly during a winter season, could have substantial benefits. The potential to save 18,921 admissions over winter may have enhanced value, given the significant bed capacity limits of NHS during this time [59,60,61,62,63]. Vaccination reduces severe consequences of COVID-19 that result in hospitalization and mortality [64], and so, vaccination strategies that directly and indirectly protect those at risk may have substantial benefits for the NHS.

Our NNV estimates are aligned with those estimated by the UKHSA for the autumn 2023 COVID-19 booster campaign [65]. Our analysis shows that increasing the eligibility of the COVID-19 booster campaign has the potential to reduce the NNV to avoid one symptomatic case from 64 to 34, demonstrating that the impact of the booster campaign could be improved by widening the current eligibility criteria.

The model presented here accounts for real world dynamics and complexities of disease transmission. As with any modelling analysis, there are some limitations. Firstly, it should be noted that the model does not account for the various sublineages of the Omicron strain and variants that may influence transmission and hospitalizations. This is reflected using the transmission parameters obtained during the calibration period for predictions. As the model was calibrated to data from the 9 months prior to the prediction period, projections over time may be impacted as fewer people have severe reactions to COVID-19 owing to the combined effect of vaccine and prior infection, and a higher number of people would have been infected with different subvariants of Omicron with different risk profiles and waned protection over time, which may not be reflected in the calibrated parameters. The observed hospital admissions (until March 2023) and deaths (until July 2023) were 133,623 and 13,144 respectively [66, 67]. Using the point estimates of parameters and conservative assumptions for VE, the model underpredicts both hospitalizations (109,460) and deaths (8344). Further, any disparities in disease transmission due to differences in the microenvironment, such as hospital or community care, are not explicitly modelled.

There is no specific vaccine considered in the model, nor are the differences between vaccines investigated between outcomes. Productivity gains due to prevention of long COVID by vaccination have not been accounted for in this model; therefore, the societal benefits of the booster campaign are underestimated. We have not modelled adverse events because of the rarity of severe adverse events after vaccination [68, 69]. Adverse events after COVID-19 vaccination are mostly mild and their risk is largely outweighed by the benefit of vaccination in reducing the risk of severe disease [68, 69]. Guidance from the NHS supports seasonal vaccination to protect against severe COVID-19 [70] with the Medicines and Healthcare products Regulatory Agency (MHRA) continually reviewing suspected adverse events associated with vaccination via the Yellow Card reporting scheme. Reviews following the autumn 2022 booster campaign did not acknowledge any new safety concerns associated with vaccination [71]. A full description of model limitations is provided in the Supplementary Material.

Conclusion

This study estimates the impact of the autumn 2022/spring 2023 booster campaigns on COVID-19 morbidity and mortality and demonstrates the considerable public health benefit the booster campaigns had on healthcare and societal outcomes by reducing pressures on NHS bed occupancy and restoring patient productivity. Additional hypothetical vaccination strategies such as expanding the eligibility criteria and/or promoting an increase in vaccine uptake would considerably boost protection against severe outcomes, resulting in substantial social, economic and public health benefits.