Public health impact and economic value of booster vaccination with Pfizer-BioNTech COVID-19 vaccine, bivalent (Original and Omicron BA.4/BA.5) in the United States

Abstract

Objective

To assess the public health impact and economic value of booster vaccination with the Pfizer-BioNTech COVID-19 Vaccine, Bivalent in the United States.

Methods

A combined cohort Markov decision tree model estimated the cost-effectiveness and budget impact of booster vaccination compared to no booster vaccination in individuals aged ≥5 years. Analyses prospectively assessed three scenarios (base case, low, high) defined based upon the emergence (or not) of subvariants, using list prices. Age-stratified parameters were informed by literature. The cost-effectiveness analysis estimated cases, hospitalizations and deaths averted, Life Years (LYs) and Quality Adjusted Life Years (QALYs) gained, the incremental cost-effectiveness ratio (ICER), the net monetary benefit (NMB), and the Return on Investment (ROI). The budget impact analyses used the perspective of a hypothetical 1-million-member plan. Sensitivity analyses explored parameter uncertainty. Conservatively, indirect effects and broad societal benefits were not considered.

Results

The base case predicted that, compared to no booster vaccination, the Pfizer-BioNTech COVID-19 Vaccine, Bivalent could result in ∼3.7 million fewer symptomatic cases, 162 thousand fewer hospitalizations, 45 thousand fewer deaths, 373 thousand fewer discounted QALYs lost, and was cost-saving. Using a conservative value of $50,000 for 1 LY, every $1 invested yielded estimated $4.67 benefits. Unit costs, health outcomes and effectiveness had the greatest impact on results. At $50,000 per QALY gained, the booster generated a 34.2 billion NMB and probabilistic sensitivity analyses indicated a 92% chance of being cost-saving and 98% of being cost-effective. The bivalent was cost-saving or highly cost-effective in high and low scenarios. In a hypothetical 1-million-member health plan population, the vaccine was predicted to be a budget-efficient solution for payers.

Conclusions

Booster vaccination with the Pfizer-BioNTech COVID-19 Vaccine, Bivalent for the US population aged ≥5 years could generate notable public health impact and be cost-saving based on the findings of our base case analyses.

Keywords:

• Cost-effectiveness
• COVID-19
• bivalent vaccine booster
• Pfizer-BioNTech
• public health impact
• SARS-CoV-2
• United States

JEL CLASSIFICATION CODES:

• I10
• I1
• I
• I19

Introduction

The public health and societal impact of COVID-19 and the associated economic and social fallouts have been far-reaching^1,2. More than three years in the pandemic, the SARS-CoV-2 virus has infected over 750 million people worldwide and caused almost 7 million deaths². The US has borne a significant health^2,3 and economic⁴ burden with over 100 million infections and 1 million deaths^2,3.

The Omicron variant of SARS-CoV-2 emerged in late 2021 and caused a rapid increase in COVID-19 cases, including in the US and other countries with substantial COVID-19 vaccination coverage and infection-induced immunity levels^2,3,5. Boosting recommendations continue to be a critical component of the US Centers for Disease Control (CDC) public health response to reduce adverse outcomes from COVID-19^6,7. During September-October 2022, the FDA removed the emergency use authorization (EUA) for monovalent COVID-19 booster doses and amended the EUA of the Pfizer-BioNTech COVID-19 Vaccine to authorize the Pfizer-BioNTech COVID-19 Vaccine, Bivalent (Original and Omicron BA.4/BA.5), hereafter referred to as Pfizer-BioNTech COVID-19 Vaccine, Bivalent for use as a single booster dose in individuals aged ≥5 years of age and older at least two months following primary or booster vaccination with any authorized or approved monovalent COVID-19 vaccine^6,7. The CDC broadly recommended the updated bivalent COVID-19 boosters to potentially increase the immune response and restore protection following waning in the face of the circulating Omicron sub-lineages^6,8. The recommendations were based on an assessment of the data on immunogenicity and safety from clinical trials of mRNA bivalent vaccines^6,8,9.

COVID-19 vaccines have been provided for free by the US government during the course of the public health emergency¹⁰.

Health economic evaluations using pandemic price levels showed that, in the US, COVID-19 vaccines generated substantial public health impact and were cost saving^11–15. With high population immunity levels, uncertain trajectory of the pandemic and the transition to traditional commercial market settings, research is warranted to assess how the cost-effectiveness and affordability profile of COVID-19 vaccines may change in response to changes in the burden of disease and costs. The aim of this study was to estimate the public health impact and economic value of bivalent booster vaccination in the US under commercial settings, across several burden of disease scenarios. Based upon CDC estimates, as to mid of December 2022, the Pfizer-BioNTech COVID-19 Vaccine, Bivalent represented the highest share of the US population boosted with a bivalent, accounting for ∼65% of all the bivalent booster doses administered^3,5. The primary objective of this study was to assess the societal impact and cost-effectiveness of the Pfizer BioNTech COVID-19 Vaccine, Bivalent and the secondary objective was to estimate the budgetary impact of its adoption from the perspective of a hypothetical 1-million-member health plan.

Methods

Study design

A cost-effectiveness and budget impact analysis were conducted comparing vaccination with the COVID-19 bivalent booster with no booster vaccination, incorporating the effects and costs related to COVID-19 disease and vaccination. The decision analytic model used in this economic evaluation was previously described¹¹. The cost-effectiveness model structure builds on a combined cohort Markov and decision tree with transition probabilities based on COVID-19 literature (Figure 1(A,B)). The budget impact model (BIM) structure uses the decision tree portion of the cost-effectiveness model (Figure 1(B)). Model inputs were derived from published evidence and stratified by age when data was available.

Figure 1. Economic model structure combining (A) Markov cohort model and (B) decision tree component used in both the cost-effectiveness and the budget impact analyses. In the Markov model component, individuals transition weekly through health states related to COVID-19 protection and infection. If individuals transition into any infection state, they immediately enter (B) the decision tree component for sorting. Green circles indicate decision points, with probabilities informed by the level of protection against COVID-19 in the particular infected state from which an individual arrives. Red triangles indicate decision tree end points, from which participants re-enter the Markov model component. Abbreviations. ICU, intensive care unit; SBV, second booster vaccination.

The analysis was conducted in individuals aged 5 years of age and older, the target population eligible for a single bivalent booster dose (30 mcg for 12 years of age and older, 10 mcg for 5 through 11 years of age) administered at least two months after either completion of primary vaccination with any authorized or approved COVID-19 vaccine, or receipt of the most recent booster dose with any authorized or approved monovalent COVID-19 vaccine. The CDC recommends COVID-19 vaccination for this population regardless of history of symptomatic or asymptomatic infection^6,8,9. The intervention modeled was a single booster dose of the Pfizer-BioNTech COVID-19 Vaccine, Bivalent booster and was compared to no booster vaccination, based on CDC guidance^6,8,9.

The analytic time horizon was governed by vaccine-induced duration of protection assumptions and set to a maximum of one year, starting the last week of August 2022, when the bivalent booster received authorization¹⁶. This time horizon was chosen based on (1) existing evidence regarding vaccine-induced duration of protection, which has been shown to be short-lived against infection and more stable against severe outcomes, (2) the rapidly changing nature of the SARS-CoV-2 virus introducing uncertainty in projecting long-term model inputs, and (3) expectations from public health authorities to transition to an annual implementation of updated COVID-19 vaccines^16–18.

Given the uncertainty on the future trends of COVID-19, this analysis simulated three scenarios: (1) a base case assuming the emergence of sub-lineages with similar presentation as the ones prevalent during the second half of 2022 (e.g. BA.1, BA.4, BA.5, BQ.1, BQ.1.1)³, and similar levels of protection from natural and vaccine-induced immunity as reported in that period, (2) a high case, assuming the emergence of an immune-escape variant leading to increased levels of disease severity and faster immunity waning than the base case, and (3) a low case, assuming no emergence of new variants and a receding pandemic with attack rates trending toward a severe flu season.

The cost-effectiveness analyses were conducted from a societal perspective, in alignment to ACIP guidelines¹⁹ and the Second Panel on Cost-Effectiveness in Health and Medicine²⁰. The budget impact analyses used the base case inputs of the cost-effectiveness model and were conducted from a payer perspective, with indirect costs as sensitivity.

The model was programmed in Microsoft Excel 2021. Given the nature of the study, ethical approval was not required. This publication has been prepared in accordance with the 2022 Consolidated Health Economic Evaluation Reporting Standards guidelines (Supplemental Table 1)²¹.

Table 1. Main model parameters for the base case.

Category	Input description	5–11 years	12–17 years		18–29 years			30–49 years		50–64 years	65–74 years	75+ years	References
Infection inputs	Population with infection-induced immunity at the start (%)	44.2	45.6		36.5			36.5		28.8	19.1	19.1	^26
	Annual attack rate (%)	25.2	28.5		29.4			31.0		28.6	28.8	37.7	^23
	Duration of protection from infection-induced immunity	3 months											^25
Vaccine inputs	Booster vaccine coverage (%)	8.3	17.6		21.4			26.6		35.3	44.0	46.6	^3^,^5
	Vaccine-induced duration of protection against infection	6 months											^29^,^30^,^32–36
	Vaccine effectiveness (VE) (infection/symptomatic/hospitalization)	60% / 70% / 80%
Clinical inputs	Hospitalization rate among symptomatic patients (%)	0.1	0.3		1.1	1.3				2.5	4.4	9.0	^44
	Among hospitalized patients:
	ICU admission rate (%)	29.7	29.7		11.9	17.6				24.3	26.9	21.3	^47
	IMV rate in general ward (%)	0.9	0.9		1.4	3.1				5.9	7.8	7.4
	IMV rate in ICU (%)	27.6	27.6		39.1	48.8				60.5	63.6	53.2
	Probability of COVID-19 death among infected individuals:
	General ward without IMV (%)	0.0	0.1		0.0	0.1				1.1	4.3	12.9
	General ward with IMV (%)	0.0	0.0		15.1	23.1				40.2	53.5	66.3
	ICU without ventilation (%)	0.0	0.0		0.6	0.8				3.6	10.4	25.4
	ICU with ventilation (%)	9.1	9.1		20.0	32.3				48.2	60.4	69.7
	Outpatient care setting (%)	0.0	0.0		0.0	0.0				0.2	0.5	1.8	^48
	Asymptomatic patients (%)	0.0	0.0		0.0	0.0				0.0	0.0	0.0	Assumption
	Probability of long COVID among infected individuals:
	Asymptomatic patients (%)	4.2	20.8		20.8	20.8				20.8	26.9	26.9	^49^,^54
	Patients having received outpatient care (%)	4.2	20.8		20.8	20.8				20.8	26.9	26.9	^49^,^54
	Patients having received inpatient care (%)	10.2	49.2		49.2	49.2				49.2	63.6	63.6	^49^,^54
Vaccine costs	Vaccine administration cost per dose (USD)	28.4											^59
	Vaccine acquisition cost per dose for 10 mcg bivalent for individuals 5 to 11 years (USD)	77											^60
	Vaccine acquisition cost per dose for 30 mcg bivalent for individuals 12 years and older (USD)	115											^60
Direct costs in the acute phase	COVID test (USD)	43.6											^59
	Number of tests/case (outpatient/inpatient/ICU)	2 / 3/ 3											^57
	GP visit number/ cost (USD)	2 / 229											^13^,^57
	OTC pain medication (USD)	205											^57
	In-patient treatment (USD)
	General ward (without/with IMV)	14,325 / 41,769											^47
	ICU (without/with IMV)	25,688 / 78,245
Direct costs of sequelae	Long COVID treatment cost (USD)	5,969											^13^,^47^,^55
Indirect costs	Asymptomatic (USD)	0	1	18			27		24		7	2	^64–68
	Symptomatic outpatient (USD)	0	74	922			1,351		1,180		365	122
	General ward without IMV (USD)	0	119	1,484			2,175		1,900		588	196
	General ward with IMV (USD)	0	1,354	16,865			24,725		21,597		6,684	2,228
	ICU without IMV (USD)	0	145	1,806			2,648		2,313		716	239
	ICU with IMV (USD)	0	1,354	16,865			24,725		21,597		6,684	2,228
	Patients with Long COVID (USD)	0	666	8,294			12,160		10,621		3,287	1,096
QALY outcomes	Utility Decrement (symptoms, GW, GW IMV, ICU, ICU IMV)	−0.093, −0.3, −0.6, −0.5, −0.6, 0.034											^69–74
	Duration (days) (symptoms, GW, GW IMV, ICU, ICU IMV)	28, 4, 9, 8, 15, 365											^47^,^70

Note: Refer to the Supplemental Material for further details on the model inputs for the different scenarios.

Abbreviations. COVID-19, coronavirus disease 2019; ICU, intensive care unit; IMV, invasive mechanical ventilation; GP, general physician; OTC, over the counter; QALY, quality-adjusted life year; USD, United States dollar; VE, vaccine effectiveness; GW, general ward.

Model structure

The model was built using a combined cohort Markov (Figure 1(A)) and decision-tree structure (Figure 1(B)) that was previously described¹¹.

Briefly, the Markov model simulated the transition of the target population through mutually exclusive health states (Figure 1(A)) based on a Susceptible-Infected-Recovered (SIR) structure and Vaccinated states. The “infected state” of the Markov model was linked with a Decision tree (Figure 1(B)) for transition probabilities of symptomatic and asymptomatic infections, and sequelae such as hospitalization, receiving intensive care (ICU) and/or invasive mechanical ventilation (IMV), and long COVID.

In alignment with the bivalent booster authorized indication⁷, all cohorts entering the model were previously vaccinated with a primary series, with or without a monovalent booster regimen. The model assumed that cohorts could enter the model either with infection-induced immunity in the “Recovered” health state, or with vaccine-induced immunity in one of the “Vaccinated” health states (“Full primary vaccination” or “First booster vaccination” or “Second booster vaccination”), or in the “Susceptible” (i.e. no longer protected) state. The cohorts moved through health states in weekly cycles over a one-year time horizon. In each cycle, cohorts could transition to other health states according to age-dependent probabilities. Individuals in the “Recovered” state were assumed to be protected by infection-induced immunity, with their immunity assumed to wane at a rate of 1/duration of protection, at which point they transition to “Susceptible”. An age-dependent annual attack rate determined the proportion of those in the “Susceptible” state that would transition to the “Infected” state. The age-dependent attack rate was reduced by the measure of vaccine effectiveness to determine the proportion of those in the “Vaccinated” health state that transitioned to the “Infected” state. Individuals were eligible to receive the bivalent booster from any health state at model start; the uptake was defined based on vaccine coverage assumptions. Upon vaccination, vaccine effectiveness began to wane at a rate of 1/duration of protection, and cohorts transitioned to the “Susceptible” state with a probability governed by the attack rate and vaccine effectiveness against infection.

Upon infection, cohorts entered the decision tree component and moved through it according to outcome probabilities based on the health state from which they entered the tree. Outcome probabilities were informed by scenario-specific efficacies against symptomatic disease and hospitalization. Symptomatic cases were assumed to be managed in the outpatient or hospitalized setting, and incurred healthcare costs for clinical care including visits, testing, medication, and hospitalization treatment. Hospitalized patients remained in the general ward or were admitted to the intensive care unit (ICU) and could receive invasive mechanical ventilation (IMV) in either setting. Both symptomatic and asymptomatic individuals who survived infection could move further in the decision tree and experience long-term outcomes of COVID-19 (Long COVID) and incurred the costs of managing these symptoms. The outcome probabilities corresponding to the decision tree node “Survive” or “Die” determined whether individuals returned to the Markov model component in the “Recovered” or “Death” states, respectively. Death during the “Infection” state was attributed to COVID-19, whereas death during any other state was attributed to all other cause mortality.

Model parameters

A targeted literature review was conducted on PubMed and Google Scholar between August and September 2022 to explore the literature and collect suitable data fit-for-purpose to inform model estimates (Supplementary Appendix A). The review was complemented by weekly searches through January 2023, incorporating relevant new evidence as it became available. The main model inputs for the base case and scenario analyses are presented in Table 1 and further detailed in Supplemental Tables. Supplemental Table 2 summarizes the datasets used to inform model development.

Table 2. Projected COVID-19 health and economic outcomes in the absence of COVID-19 vaccination in individuals ≥5 years old in the US under three burden of disease scenarios over one year.

Outcomes	Low	Base case	High
Symptomatic cases, n	38,444,718	51,636,594	70,279,345
Deaths, n	91,287	381,647	1,128,515
Outpatient cases, n	38,099,528	50,164,125	66,742,405
Hospitalizations, n	345,191	1,472,469	3,536,940
General ward without IMV	272,603	1,076,952	2,388,050
General ward with IMV	13,044	70,437	198,524
ICU without IMV	32,902	142,078	306,638
ICU with IMV	26,641	183,003	643,728
QALY lost, discounted	1,408,054	3,726,701	10,007,493
LY lost, discounted	961,432	3,632,115	10,560,212
Direct costs, USD	$85,385,041,552	$163,591,156,731	$320,700,098,148
Vaccine-related	$0	$0	$0
Inpatient	$5,903,701,333	$36,338,181,341	$120,895,657,496
Outpatient	$23,850,304,226	$33,258,814,545	$46,986,653,116
Testing	$3,352,379,427	$4,566,910,645	$9,346,748,916
Long COVID	$52,278,656,566	$89,427,250,198	$143,471,038,620
Indirect costs, USD	$133,755,889,763	$199,979,502,224	$321,330,814,086
Total costs, USD	$219,140,931,315	$363,570,658,954	$642,030,912,234

Note. Vaccine-related costs include acquisition, administration and adverse events management.

Abbreviations. ICU, intensive care unit; IMV, invasive mechanical ventilation; LY, life year; QALY, quality-adjusted life year; USD, United States dollar.

Population

Age-stratified estimates of population size were adopted from the US Census Bureau population projections for 2020 (Supplemental Table 3)²². The population eligible for a bivalent booster was derived from the CDC reported percentages of the US population aged five years and older whom had completed primary vaccination approximately two months prior to the bivalent booster rollout⁵ including individuals whom received their first and/or second booster of the previously authorized monovalent vaccines, which are no longer authorized for booster doses⁸. US age-specific population estimates²² were used to adjust the age-stratified percentages to match model age groups (Supplemental Table 4).

Table 3. Results of the cost-effectiveness analysis of booster vaccination with the Pfizer-BioNTech COVID-19 vaccine, bivalent (Original and Omicron BA.4/BA.5) in individuals ≥5 years old in the US under three burden of disease scenarios over one year.

Outcomes	Low	Base case	High
Averted symptomatic cases, n	−3,551,543	−3,717,921	−3,780,114
Averted deaths, n	−13,214	−44,792	−99,044
Averted outpatient cases, n	−3,503,226	−3,555,582	−3,488,114
Averted hospitalizations, n	−48,317	−162,339	−292,000
General ward without IMV	−37,771	−117,439	−195,009
General ward with IMV	−1,977	−8,302	−17,322
ICU without IMV	−4,685	−15,780	−25,180
ICU with IMV	−3,883	−20,819	−54,490
QALY saved, discounted	−159,279	−373,236	−782,991
LY saved, discounted	−129,153	−404,635	−888,238
Direct cost savings, USD billions	$4,205,400,521	−$2,428,806,475	−$11,803,441,719
Vaccine-related	$11,353,818,257	$9,740,303,035	$8,134,993,843
Inpatient	−$838,267,489	−$4,063,361,251	−$10,112,847,012
Outpatient	−$2,193,019,707	−$2,357,351,040	−$2,455,632,536
Testing	−$309,694,540	−$331,280,673	−$507,170,116
Long COVID	−$3,860,210,202	−$5,440,148,746	−$6,882,373,190
Indirect cost savings, USD billions	−$10,569,238,666	−$13,107,055,487	−$17,575,200,366
Total cost savings, USD billions	−$6,363,838,145	−$15,535,861,961	−$29,378,642,085
ICER – payer perspective	$26,237	Dominant	Dominant
ICER – societal perspective	Dominant	Dominant	Dominant
ROI	2.13	4.67	10.07
Net monetary benefit	$14,354,190,603	$34,197,669,630	$68,528,177,908

Notes: The reference scenario is no COVID-19 booster vaccination with the Pfizer-BioNTech COVID-19 Vaccine, Bivalent. The payer perspective includes only direct medical costs. The societal perspective includes indirect costs as well as direct medical costs. Vaccine-related costs include acquisition, administration and adverse events management.

Abbreviations. ICU, intensive care unit; IMV, invasive mechanical ventilation; LY, life year; QALY, quality-adjusted life year; USD, United States dollar; ROI, return on investment.

Table 4. Base case results of the budget impact analysis of booster vaccination with the Pfizer-BioNTech COVID-19 vaccine, bivalent (Original and Omicron BA.4/BA.5) in the US for a 1-million-member plan over one year.

Cost (USD)	No Vaccination with the Pfizer-BioNTech COVID-19 Vaccine, Bivalent (Original and Omicron BA.4/BA.5)	Vaccination with the Pfizer-BioNTech COVID-19 Vaccine, Bivalent (Original and Omicron BA.4/BA.5)	Difference
A. Base case budget impact results for a 1-million-member plan in individuals ≥5 years old.
Vaccine-related costs	$0	$34,645,683	$34,645,683
Total direct costs	$603,148,823	$538,277,085	−$64,871,737
Inpatient	$134,178,121	$114,394,783	−$19,783,338
Outpatient	$122,713,285	$109,915,735	−$12,797,550
Testing	$17,009,025	$15,199,046	−$1,809,979
Long COVID	$329,248,392	$298,767,521	−$30,480,871
Net direct costs	$603,148,823	$572,922,769	−$30,226,054
Net cost PMPY – payer perspective	$603	$572	−$30
Sensitivity analyses with indirect costs	$1,182,619,244	$1,100,841,989	−$81,777,255
Net cost PMPY – societal perspective	$1,182	$1,100	−$82
B. Base case budget impact results for a 1-million-member plan in individuals ≥12 years old.
Vaccine-related costs	$0	$34,371,649	$34,371,649
Total direct costs	$597,053,743	$532,342,353	−$64,711,390
Inpatient	$134,178,121	$114,394,783	−$19,783,338
Outpatient	$119,126,652	$106,428,802	−$12,697,849
Testing	$16,536,104	$14,739,275	−$1,796,829
Long COVID	$327,212,866	$296,779,492	−$30,433,374
Net direct costs	$597,053,743	$566,714,001	−$30,339,742
Net cost PMPY – payer perspective	$597	$566	−$30
Sensitivity analyses with indirect costs	$1,176,524,164	$1,094,633,221	−$81,890,943
Net cost PMPY – societal perspective	$1,176	$1,094	−$82

Note: The payer perspective includes only direct medical costs. The societal perspective includes indirect costs as well as direct medical costs.

Abbreviation. PMPY, Per Member Per Year.

Infection inputs

The data from the CDC COVID Data Tracker^5,23 was used to inform the distribution of the starting population for the model, and served as the primary source for infection statistics.

Age-stratified point prevalence of infection-induced, vaccine-induced and hybrid immunity were not available in the literature at the start of the model time horizon. The proportions of the population entering the model as recovered (i.e. with infection-induced immunity) and vaccinated (i.e. with vaccine-induced immunity) were therefore defined based on a review of the timing of rollout of primary series, first and second monovalent boosters⁸, evidence on the duration of vaccine-induced and infection-induced immunity during Omicron and the latest CDC seroprevalence data^18,24–30. The CDC seroprevalence data indicating that ∼55% of the population had infection-induced immunity during Omicron peak was used for the low scenario²⁶. The low scenario also conservatively assumed that approximately half of the individuals whom received a second monovalent booster (authorized at the end of March 2022) retained vaccine-induced immunity⁵. As such, this scenario assumed ∼60% hybrid immunity in the population at model start, which was directionally consistent although more conservative than the preliminary analyses of unweighted NHANES data, which found that 42% of adults had prior infection during Omicron²⁴. The base case scenario assumed a level of infection-induced immunity similar to the one reported in seroprevalence studies prior to the emergence of Omicron²⁶, and assumed a ∼35% hybrid immunity in the population at model start. Finally, the high case used the upper range of the low and base case scenarios for the level of infection-induced immunity, and assumed that, due to the waning of vaccine-induced immunity observed during Omicron, especially against infection, none of the individuals who had received the second monovalent booster had residual vaccine-induced immunity at the start of the time horizon (Supplemental Table 4).

The remaining portion of the population entered the model as “Susceptible”. In the base case, the annual attack rates determining the percentage of the susceptible cohort that would become infected with SARS-CoV-2 were derived from CDC reported age-stratified weekly infection rates for the period immediately preceding the booster rollout²³, projected to an annual rate. US age-specific population estimates²² (Supplemental Table 3) were used to adjust CDC age-stratified rates to match model age groups. These estimates were adjusted for CDC under-reporting infection ratios, which were not age-stratified³¹. Attack rates from prior periods and flu were reviewed for a qualitative comparison of trends and to inform plausible estimates for the scenarios. The low scenario used early incidence data reported by the CDC for the time period in which the original strain was predominant³¹, and the high scenario was derived from data from the other two scenarios (Supplemental Table 5).

Table 5. Base case population-level outcomes in individuals ≥5 years old in additional sensitivity analysis using alternative values for main model inputs.

Variables	Symptomatic cases averted		Deaths averted		Hospitalizations averted		Direct care cost savings (USD billion)		Indirect cost savings (USD billion)
	N (% difference)		N (% difference)		N (% difference)		N (% difference)		N (% difference)
Base case	3,717,921	N/A	44,792	N/A	162,339	N/A	$2.4	N/A	$13.1	N/A
Duration of Infection-induced immunity
2 months	4,184,526	12.5%	48,606	8.5%	176,420	8.7%	$3.9	62.1%	$15.0	14.3%
4 months	3,316,049	−10.8%	41,577	−7.2%	150,303	−7.4%	$1.1	−53.3%	$11.5	−12.3%
Vaccine duration of protection against infection
7 months	4,225,178	13.6%	50,126	11.9%	182,395	12.4%	$4.1	67.8%	$15.0	14.2%
5 months	3,123,832	−16.0%	38,503	−14.0%	138,823	−14.5%	$0.5	−79.5%	$10.9	−16.6%
Vaccine acquisition cost for ≥12 years old
− 30% ($80.5 / dose)	3,717,921	N/A	44,792	N/A	162,339	0%	$4.8	96.4%	$13.1	0%
+ 30% ($149.5 / dose)	3,717,921	N/A	44,792	N/A	162,339	0%	$0.1	−96.4%	$13.1	0%
GP visit unit cost
− 30% ($160)	3,717,921	N/A	44,792	N/A	162,339	0%	$1.3	−48.5%	$13.1	0%
+ 30% ($298)	3,717,921	N/A	44,792	N/A	162,339	0%	$3.6	48.5%	$13.1	0%
Initial VE against infection and VE symptomatic disease
50%, 60%	2,932,279	−21.1%	40,966	−8.5%	162,162	−0.1%	$0.5	−79.9%	$10.6	−18.9%
60%, 60%	2,918,697	−21.5%	40,919	−8.6%	162,377	0.0%	$1.80	−25.0%	$12.10	−7.6%
70%, 80%	4,522,791	21.6%	48,722	8.8%	162,515	0.1%	$4.40	83.3%	$15.60	19.1%
Initial VE against hospitalization
70%	3,718,507	0.02%	40,182	−10.3%	133,729	−17.6%	$1.60	−33.3%	$12.50	−4.6%
90%	3,717,335	−0.02%	49,402	10.3%	190,949	17.6%	$3.10	29.2%	$13.6	3.8%
Long COVID care
Lower re-admission: 4%^11	3,717,921	N/A	44,792	N/A	162,339	0%	$2.0	−16.1%	$13.1	0%
Higher re-admission: 15%^11	3,717,921	N/A	44,792	N/A	162,339	0%	$3.5	43.0%	$13.1	0%
Care management costs
Vaccine Administration costs ($16.9 per dose)^56	3,717,921	N/A	44,792	N/A	162,339	0%	$3.3	32.5%	$13.1	0%
Vaccine Administration costs ($40 per dose)^56	3,717,921	N/A	44,792	N/A	162,339	0%	$1.6	−32.5%	$13.1	0%

Note: The reference scenario is no vaccination with the Pfizer-BioNTech COVID-19 Vaccine, Bivalent.

Abbreviations. GP, general physicians; VE, vaccine effectiveness.

Vaccine inputs

Vaccine-specific inputs such as coverage, vaccine effectiveness, and duration of protection, were integrated from various sources and based upon medical opinion.

The demographic trends of individuals aged ≥5 years receiving the first COVID-19 monovalent booster vaccination reported by the CDC⁵ were used to define age-dependent booster vaccination coverage inputs for the base case. US age-specific population estimates²² were used to adjust CDC age-stratified uptake rates to match model age groups (Table 2, Supplemental Table 6).

In the absence of vaccine efficacy data from randomized controlled trials (RCTs), vaccine effectiveness (VE) and duration of protection data was sourced from real-world studies^{29,30,32–36} assessing the effects of the Pfizer-BioNTech COVID-19 bivalent (or, in absence, the monovalent booster), against mild and severe SARS-CoV-2 infection during the predominance of Omicron sub-lineages. Initial relative VE of a first booster against monovalent primary series was used as the main basis for clinical data, as the primary series was the alternative regimen approved at the time of writing, and the established practice from a competing resource perspective, given EUA revocation for monovalent boosters.

Tartof²⁹ reported that, among immunocompetent US adults, the VE against BA4/BA5 of a third dose monovalent booster was in the range of 55% to 70% for outpatient and emergency department (ED) less than three months since the last dose, and >70% for hospitalization three to five months after the third dose. The same research group reported VE of a third monovalent booster against BA1/BA2 in the range of 75% to 80% for hospitalization and 60% to 75% for ED³⁰. Duration of protection against outpatient/ED waned within approximately six months and was more stable against severe disease. Surie³⁵, Link-Gelles³³, and Tenforde³⁶ reported early VE estimates of mRNA bivalent boosters compared with no vaccination in the US, compared with receipt of last monovalent dose, stratified by time of prior vaccination. Surie³⁵ reported VE against hospitalizations among US elderly of ∼80% (range: 60% to 90%) compared to receipt of ≥2 monovalent doses >6 months earlier. Link-Gelles³³ reported VE against symptomatic disease of ∼50% (range of 37–52%) compared to receipt of 2 to 4 prior monovalent doses. Tenforde³⁶ reported VE against ED, urgent care or hospitalization in the range of 45% to 60% compared to receipt of ≥2 monovalent doses >5 months earlier. Arbel³² reported 81% VE against hospitalization and a borderline statistically significant 86% VE against death among elderly who received a bivalent booster in Israel. Building on this body of evidence, the base case assumed that initial VE against infection (60%) was lower than against severe disease (80%)^{6,9,18,25,27,29–32,34–36}. The scenarios used alternative values assuming lower and higher vaccine protection and uptake in the high and low scenarios, respectively. The VE assumptions were in line with the ones used in prior models⁹.

There is currently no evidence regarding the effectiveness of bivalent boosters against transmission^9,11. The existing literature is limited and not conclusive or specific yet as to the patterns and the sustainability of such effects; the CDC acknowledges that the protection provided by the monovalent vaccines against transmission is lower than against severe disease and diminishes over time, especially against the Omicron sub-lineages²⁵. As such, indirect effects (herd immunity) were conservatively not considered, and the duration of protection against infection was assumed to be six months in the base case (scenarios: five to seven months), along with a duration of protection against severe disease of eight months (scenarios: seven to nine months), both tested in scenario analyses.

There is growing evidence that COVID-19 vaccination can attenuate symptoms, duration of illness and medical care seeking³⁷, as well as increase protection after prior infection^38–41. Conservatively, such effects were also not considered.

Health inputs

Health inputs are detailed in Table 2 and Supplemental Table 7.

Systematic literature reviews informed the probability of an individual with COVID-19 infection experiencing symptoms for the low scenario⁴² and the high scenario used data from a US study conducted during the Omicron surge⁴³. The base case used the midpoint of these two estimates. Studies assessing the effectiveness of Omicron infection-induced immunity reported time to re-infection as short as 60 to 90 days^25,27,28. Consistent with the CDC definition of re-infection²⁵, the cost-effectiveness model assumed that Omicron infection conferred a limited duration of protection of three months; this parameter was not used in the budget impact model (BIM) calculations as the BIM does not include Markov health sates; individuals are tracked as vaccinated or unvaccinated and enter the decision tree when infected. The probability of an individual with symptomatic infection being hospitalized were derived from the Coronavirus Disease 2019 Associated Hospitalization Surveillance Network (COVID-NET) age-stratified weekly hospitalization rates⁴⁴. The high scenario used rates reported in early 2022 during the Omicron peak. The low scenario used rates from the period immediately preceding the booster rollout, when rates were the lowest for all non-elderly age groups compared to the whole pandemic period, and comparable to the pre-Delta period for the elderly. The base case used the age-stratified mid-points. The weekly rates were projected to an annual rate, adjusted based on the distribution of the US population age groups²² and hospitalization under-reporting³¹. The hospitalization rates were qualitatively compared to those reported by other US datasets for relevant time periods, and were found to be plausible^45,46. A retrospective analysis of the US Premier Healthcare COVID-19 Hospital database⁴⁷ was used to inform several age-dependent inpatient inputs: the distribution of symptomatic cases by general ward or ICU and receipt of IMV, case fatality rates for hospitalization. Outpatient mortality inputs were based on a database of US academic centers⁴⁸. We assumed that individuals with asymptomatic infection did not die or require any inpatient or outpatient treatment but could experience long COVID complications¹¹. The probability of long COVID was differentiated by inpatient, outpatient, and asymptomatic cases, and was derived from published literature reporting prevalence of persistent COVID-19 symptoms lasting beyond the acute phase^49–53 as well as probabilities of readmission^{49,50,52–55}. Literature reporting prevalence of long COVID was found to be highly heterogeneous, especially for non-hospitalized individuals^{49,50,52–54}. For individuals ≥12 years of age, the scenarios assumed an average probability of long COVID among non-hospitalized individuals of 18% (low) to 27% (high), and, respectively, 43% to 64% among hospitalized individuals. This assumption was in line with meta-analyses and US-based studies reporting point prevalence of 30 to 60% among non-hospitalized individuals and over 50% for hospitalized individuals during the year following infection^49,50,52,53. More conservative assumptions (5–10%) were used for the probability of long COVID among children aged 5 to 11 years⁵⁴.

Healthcare resource use and direct cost inputs

Individuals experiencing COVID-19 infection could incur direct costs (Table 2 and Supplemental Table 8) in the acute phase for COVID-19 tests, disease management in inpatient (hospitalization) or outpatient (general physician [GP] visits, emergency room, and over-the-counter medication) setting. Acute phase hospitalization costs, derived from a large US hospital claims database⁴⁷, were divided by ward, ICU and IMV use. Individuals with long COVID could incur further direct costs for additional COVID-19 tests, GP visits, specialist visits, and re-hospitalization. Outpatient inputs were derived from large retrospective claims database studies^56,57. The inpatient and outpatient cost inputs used in the base case were pre-Omicron and pre-Delta and were conservative compared to FAIR Health national estimates⁵⁸.

Vaccine-related costs included administration costs⁵⁹ and vaccine cost per unit⁶⁰, both under non-pandemic commercial settings. No serious adverse events (AEs) related to the vaccine were reported for the bivalent booster in early clinical data^6,9; conservatively, the model used the AEs observed for the first monovalent booster^61,62. Hospitalization costs associated to AEs were sourced from HCUPnet (Supplemental Table 9)⁶³.

Indirect costs

Indirect costs included productivity loss due to acute illness, long COVID and premature death (Table 2 and Supplemental Table 10). Lost productivity costs due to illness were calculated based on the workforce participation rate⁶⁴, labor cost per week⁶⁵, and number of work days lost. The work days lost estimates were informed by the CDC isolation requirements⁶⁶, studies assessing length of stay during hospitalization treatment⁴⁷, long-term consequences after critical illness⁶⁷ and long COVID⁶⁸. Specifically for long COVID, a US study⁶⁸ indicated that about 1 in 4 (25.9%) long-haulers reported that their long COVID condition affected employment or work hours. While a majority remained employed, they experienced reduced work hours. Based on the above, the base case assumed that one fourth of the working time of a long hauler during the one-year time horizon (i.e. 90 days) is affected by long COVID. Productivity loss due to premature death was accounted for on a lifetime horizon and discounted at a rate of 3% per year in accordance with economic evaluation guidelines^19,20. Conservatively, work productivity loss for caregivers was not considered. All costs were expressed as 2022 US dollars.

Health utilities

Quality-adjusted life years (QALYs) lost due to acute illness and long COVID were estimated by accumulating utility decrements corresponding to the short and long-term symptomatic outcomes for the duration of illness^69–74. Further details are provided in Table 2, Supplemental Table 11 and Supplemental Methods. Long-term QALY loss associated with early death was included on a lifetime horizon and discounted at 3% annually^11,13. Health utilities parameters were not used in the BIM calculations as the BIM only include costs²¹.

Model analyses

The model predicted the health outcomes averted by vaccination (COVID-19 symptomatic cases, hospitalizations, and deaths), alongside the corresponding direct healthcare and productivity costs saved, and QALYs gained, compared with no booster vaccination.

The cost-effectiveness results were summarized using the incremental cost-effectiveness ratio (ICER) per QALY gained. The ICER was calculated by dividing differences in costs by differences in QALYs under a societal perspective and was estimated based on willingness to pay thresholds commonly used in the United States¹³. Additional summary metrics analyzed were the net monetary benefit (NMB) and the Return on Investment (ROI). The NMB was calculated using a threshold of $50,000 per QALY gained^13,14; it was estimated by multiplying the discounted QALY gained by the threshold and subtracting the net direct and indirect cost savings. The ROI was estimated with a method used to estimate the ROI for the COVID-19 vaccination program in New York⁷⁵ by summing the direct and indirect cost savings with the monetary value of discounted lifetime life years (LY) saved (valued at $50,000 per LY) and dividing by the total vaccine-related costs.

The budget impact analyses projected the health outcomes and costs of a “current scenario” of no booster vaccination and those associated to an “alternative scenario” with the introduction of the Pfizer BioNTech COVID-19 Vaccine. The incremental budget impact was calculated by subtracting the costs of the “alternative scenario” from the “current scenario”, under both a payer and societal perspective. For simplicity, results were presented for a hypothetical health plan of 1 million members.

The base case results were followed by those of the low and high scenarios. Such scenarios used extreme but plausible alternative values of model parameters, informed by (1) burden of disease data of time periods of, respectively, lower and higher severity and transmissibility of SARS-CoV-2 and (2) medical opinion. Of note, lower bounds were used for the low scenario, and the upper bounds were used for the high scenario for all parameters except those related to the proportion of individuals with infection-induced immunity at model start, and the duration of protection from infection-induced immunity; for both the parameters, their upper bound values were used in the low scenario.

Deterministic and probabilistic sensitivity analyses were conducted to explore further variation in the base case results. A deterministic sensitivity analysis was performed to assess the impact of individual parameter variability on both the ICER. Input parameters were varied ±20% from the base case value and presented in a tornado diagram. Additional sensitivity analyses were run to assess vaccine impact in the broadest population of individuals ≥12 years and using more extreme values for model inputs driving results.

In a Monte Carlo probabilistic sensitivity analysis (PSA) the ICER was calculated 1,000 times based on variable parameter input according to probability distributions for model costs, outcomes, and probabilities. Distributions applied to model parameters were beta (e.g. attack rate, vaccine coverage, probability of symptomatic disease, of ICU care, and of mechanical ventilation), gamma (e.g. costs and duration of illness), log-normal (e.g. probability of symptomatic disease, attack rates, AE rates), and normal (e.g. probability of hospitalization, for which the sample size in the source data was high enough to assume that the central limit theorem holds)⁷⁶. Published confidence intervals were used to parameterize the distributions when available; otherwise, conventional assumptions about the parameters’ standard error and standard deviations were used. Results of the PSA were presented in a cost-effectiveness plane and a cost-effectiveness acceptability curve.

Results

Burden of disease in the absence of COVID-19 booster vaccination

Table 2 shows the estimated burden of disease in the absence of COVID-19 vaccination in the base case, low and high scenarios. In the base case, the model predicted 52 million new symptomatic cases, 382 thousand deaths, 1.5 million hospitalizations, $164 billion in direct costs of COVID-19 health outcomes (e.g. hospitalizations, outpatient treatments), $200 billion in indirect costs, and $364 billion in total costs. Across the low and high scenarios, the model predicted 38 and 70 million new symptomatic cases, 91 thousand and 1.1 million deaths, and 345 thousand to 3.5 million hospitalizations in the absence of vaccination, respectively. The associated direct costs in the low and high scenarios were estimated to be $85 billion and $320 billion, respectively without vaccination. Indirect costs in the low and high scenarios were estimated to be $134 billion and $321 billion, respectively without vaccination. Total costs in the low and high scenarios were estimated to be $219 billion and $642 billion, respectively.

Base case

The base case results are presented in Table 3. At a population-level, the model predicted that the Pfizer-BioNTech COVID-19 Vaccine, Bivalent booster would contribute to avert over 3.7 million symptomatic cases, over 162 thousand hospitalizations, almost 45 thousand deaths related to COVID-19, resulting in an estimated $2.4 billion direct and $13.1 billion discounted indirect cost savings, as well as discounted gains of 373 thousand million QALYs. The vaccine-related costs were estimated to amount to $9.7 billion and to be offset by the cost savings in direct medical costs alone. The $15.5 billion net cost savings yielded a negative ICER; hence the base case model predictions indicated that, using the commercial list prices and a societal perspective, the Pfizer-BioNTech COVID-19 bivalent booster was a dominant strategy (i.e. more effective, less costly) compared to no booster vaccination. The bivalent booster strategy was predicted to generate a net monetary benefit of $34.2 billion at a cost-effectiveness threshold of $50k/QALY gained and $52.9 billion at a threshold of $100k/QALY gained.

The base case budget impact analyses for a hypothetical 1-million-member plan are shown in Table 4. In subjects ≥5 years old, vaccine-related costs of $29.6 million were offset by a total of $64.8 million estimated direct cost savings, resulting in $30.2 million net cost savings (4 A). Budget impact analysis in subjects ≥12 years old yielded slightly higher cost savings (4B).

Scenario analyses

The results of scenario analyses are presented in Table 3. The benefits of the vaccine increased in the high scenario, driven by higher burden of disease despite assuming lower vaccine effectiveness and duration of protection than the base case. The high scenario estimated that the Pfizer-BioNTech COVID-19 bivalent booster could prevent over 3.8 million symptomatic cases, over 292 thousand hospitalizations, over 99 thousand deaths, and remained cost-saving. Symptomatic cases averted were similar to the base case because attack rates were similar for younger ages. Using more conservative estimates of burden of disease and clinical effectiveness, the low scenario estimated that the Pfizer-BioNTech COVID-19 Vaccine, Bivalent could prevent over 3.5 million symptomatic cases, over 48 thousand hospitalizations, over 13 thousand deaths and was highly cost-effective with an ICER of $26,237 per QALY gained from the payer perspective and was cost-saving from the societal perspective.

Sensitivity analyses

The one-way sensitivity analysis suggested that the base case results were robust. In the worst case, the Pfizer-BioNTech COVID-19 Vaccine was cost saving, reducing total societal costs by $12.6 billion. The results were most sensitive to varying estimates of duration of protection of infection-induced immunity, acquisition cost for individuals ≥12 years old, GP visits for outpatient cases and Long COVID, probability of symptomatic infection, hospitalization rate, vaccine effectiveness, and the probability of Long COVID (Figure 2). Table 5 presents additional one-way sensitivity analyses with higher and/or different estimates than the deterministic sensitivity analyses on selected parameters expected to have the highest impact on direct costs. The base case results were relatively robust and showing cost savings when using a higher duration of infection-induced immunity (four months instead of three months), a lower vaccine duration of protection against infection (five months instead of six months), a 30% higher acquisition cost, a 30% higher GP visit unit cost, a lower probability of readmission (4% instead of 7%), and use of the highest administration fee per dose ($40 instead of $28.4). Results were robust as well based on the PSA. Figure 3 shows the cost-effectiveness plane with 92% of the simulations in the south-east quadrant (lower cost, higher efficacy versus no booster vaccination) and 98% of the simulations below the reference ICER threshold of $50,000 per QALY gained. As such, the cost-effectiveness acceptability curves (Figure 4) showed an average 98% probability of the Pfizer-BioNTech COVID-19 Vaccine, Bivalent being cost-effective at $50,000 per QALY gained.

Figure 2. Deterministic sensitivity analysis: Tornado diagram for incremental cost – societal perspective in individuals ≥5 years old. Color of the bar indicates whether the parameter is at its lower or upper bound. Over 400 parameters were tested and this figure includes only the 20 most impactful parameters.

Figure 3. Probabilistic sensitivity analysis: cost-effectiveness plane for societal perspective in individuals ≥5 years old.

Figure 4. Probabilistic sensitivity analysis: cost-effectiveness acceptability curve for societal perspective in individuals ≥5 years old.

The DSA and PSA results from a payer perspective are presented in Supplemental Figures 1, 2 and 3. 70% of the simulations were in the south-east quadrant (lower cost, higher efficacy versus no booster vaccination) and 96% of the simulations were below the reference ICER threshold of $50,000 per QALY gained.

Supplemental Figures 4–6 show the PSA results for individuals ≥12 years old from a societal perspective: 92% of the simulations were in the south-east quadrant and 99% of the simulations were below the reference ICER threshold of $50,000 per QALY gained. The PSA results for individuals ≥12 years old from a payer perspective are presented in Supplemental Figures 7–9: 70% of the simulations were in the south-east quadrant and 96% of the simulations were below the reference ICER threshold of $50,000 per QALY gained.

Discussion

This health economic analysis predicted that, under base case assumptions, booster vaccination with the Pfizer-BioNTech COVID-19 Vaccine, Bivalent in the US population aged ≥5 years would confer considerable health and economic benefits.

The vaccine effectiveness against Omicron infection of the original monovalent COVID-19 vaccines has been shown to wane, but protection against severe illness was more stable and long-lasting^{6,9,18,25,27,29–31,34}. Initial real-world effectiveness US data suggests that the bivalent booster, designed to protect against Omicron, enhances protection relative to no booster vaccination^33,35,36 while being well tolerated⁹. Our model estimations showed that, relative to no booster vaccination, the bivalent booster could avert over 3.7 million symptomatic cases, over 162 thousand hospitalizations, almost 45 thousand COVID-19-related premature deaths and over 373 thousand QALYs lost. These averted health outcomes were estimated to translate in up to $15.5 billion in net cost savings, considering a societal perspective incorporative of direct costs and productivity losses for the affected individuals.

The base case results were robust to parameter variations. In scenario and sensitivity analyses, the meaning of the results remained largely the same: the Pfizer-BioNTech COVID-19 Vaccine, Bivalent is either cost saving or highly cost-effective in the population aged ≥5 years. The budget impact analysis for a 1-million-member health plan showed that vaccine-related costs were offset by direct cost savings. Similar to prior models of COVID-19 vaccination^11,13,14, the results were most sensitive to inputs related to burden of disease and clinical effectiveness of the vaccine. The model was moderately sensitive to the list price for the population ≥12 years old, however, in sensitivity analysis in this population, a 20% (Figure 2) and a 30% (Table 5) higher price still led to direct cost savings. It was estimated that, at a threshold of $50,000 per QALY gained, the maximum price for the value of the bivalent booster in the population 12 years and older would be $619 from a societal perspective and $426 from a payer perspective. As such, approximately 70% to 80% of the economically justifiable value of a single booster dose was estimated to be distributed to society and payers.

Using a conservative estimate of $50,000 to value a life year (LY) lost, the model estimated savings of $4.67 for every $1 spent toward vaccination. This estimate is more conservative although directionally consistent with prior studies assessing the return on investment of COVID-19 vaccination. Valuing a LY at $100,000 and using a societal perspective, Sah et al.⁷⁵ estimated $10.19 savings for every $1 dollar invested in the New York COVID-19 Vaccination campaign⁷⁵. Using a similar assumption of $100,000 for a LY, the estimate of our model would increase to $6.73 savings⁷⁵.

There are limited published data on the public health impact and cost-effectiveness of bivalent booster vaccination, especially during Omicron predominance. Comparison with other published models is limited by different assumptions on model structures and parameters. In a pre-Omicron modeling study, Li et al.¹⁴ estimated that booster vaccination with a single booster of the monovalent among US adults aged ≥ 65 years is likely to be cost-effective from a payer perspective. When incorporating broader value elements related to socio-economic activity, Kirson et al.¹² estimated that COVID-19 vaccines have the potential to generate some of the largest cost savings in vaccine history. More recently, Pandey et al.¹⁵ estimated significant public health impact and direct cost savings from the fall bivalent booster vaccination. Our estimates are substantially lower, likely due to several major design and methodological differences between the studies. For example, Pandey et al.¹⁵ captured the impact of the overall COVID-19 vaccination program and used a dynamic transmission framework. A strength of the model included its COVID-19-specific inputs – many of which were age-stratified – from published literature and national surveillance data. Furthermore, the model used conventional model structures^19,20 and aligned to reporting standards²¹. While integrating over 600 parameters related to epidemiological, health, economic and humanistic data, the model remained relatively parsimonious and interpretable. The model inputs and assumptions have been documented in detail in the Supplemental Material for transparency and replicability. Outcomes were evaluated from a societal perspective, which provided a comprehensive estimate of quality of life and indirect cost impacts for those infected compared to existing models adopting a payer perspective. Scenario and sensitivity analyses affirmed the robustness of the model and gave context to the base case results. The study showed that the economic case for the Pfizer-BioNTech COVID-19 Vaccine, Bivalent, remains strong in potential endemic commercial settings. With the transition away from the emergency phase characterized by centralized financing and procurement, these findings could be applied to real-world situations to support evaluation processes for healthcare decision makers and healthcare systems. By presenting evidence of benefits, cost-effectiveness and affordability, the study could potentially inform health plans’ projections of the vaccine’s impact on their enrolled population and support financial planning.

The model adopted several assumptions and simplifications to limit model complexity. While parameter uncertainty was tested in sensitivity analyses, other forms of uncertainty (i.e. structural and methodological) were not explicitly tested. Due to limited usable data on the spread of SARS-CoV-2 and the effect of bivalent booster vaccination on transmission, our model simplified the epidemiology dynamics and time-dependent effects in a static structure. Moreover, the model did not assess the potential therapeutic effects of vaccination in reducing the severity of cases, the impact of long COVID, and broader value elements (e.g. spill-over effects, macroeconomic effects). A growing body of evidence recognizes that the COVID-19 pandemic showed that vaccines have benefits and externalities that extend beyond the categories traditionally assessed in health economic evaluations for Health Technology Assessments, and this contributes to a potential systematic under-evaluation of value⁷⁷. The incorporation of broader value elements (e.g. in the form of billions of macroeconomic cost savings, as done in prior models¹⁵) would have substantially increased our estimates. Hence, this parsimonious model framework may have generated conservative estimates. Further studies assessing implementation costs are warranted to improve the understanding of the full resource use associated with the COVID-19 vaccination program. As this analysis used the Pfizer-BioNTech COVID-19 Vaccine, Bivalent as the intervention, results are not generalizable to other vaccines currently approved and authorized in the US, or in other countries. Similarly, the present study did not explicitly capture the effect of different existing or future interventions such as COVID-19 treatments, which may have a synergistic effect in further reducing disease severity. Furthermore, we examined a hypothetical one-year time period in which the vaccine is procured under commercial settings for the entire period with highly conservative booster coverage rates from a prior booster campaign. This makes the vaccine acquisition estimates used in the model conservative when directly applied to Fall/Winter 2022/2023. Due to the absence of granular age-stratified VE data for the bivalent vaccine, VE estimates were assumed to be the same across the age groups, and no further stratification to risk groups was assessed. The findings of this study may not fully reflect current or future disease trends, vaccine clinical profile, healthcare resource use and costs. Future research should leverage the surveillance data that is accumulating during the fall of 2022 and winter of 2022/2023 to retrospectively estimate the impact of the booster campaign. Results are neither generalizable to indications nor to populations not specifically covered in this analysis. The pediatric group was characterized by the highest evidence uncertainty; despite their small impact on the results of this study, future analyses with more updated data are warranted in this group. No subgroup analysis was conducted by ethnicity or socioeconomic status. However, this is an important area of future research. As COVID-19 data continues to rapidly expand and evolve, future studies are warranted to assess the public health impact in additional settings as well as in the longer-term.

Conclusion

This analysis showed that the use of the Pfizer-BioNTech COVID-19 Vaccine, Bivalent booster has the potential to generate notable gains in health outcomes and cost savings among the US population aged ≥5 years old. It supports FDA and CDC recommendations for broad use of the vaccine and highlights the opportunity to continue widespread uptake to prevent COVID-19 related disease and generate substantial societal benefits.

Transparency

Declaration of funding

This study was sponsored by Pfizer Inc.

Declaration of financial/other relationships

MDF, ET, TLW, JY are employees of Pfizer and may hold stock or stock options of Pfizer. KM and BY are employees of Evidera, which received financial support from Pfizer in connection with the study and the development of this manuscript. Peer reviewers on this manuscript have received an honorarium from JME for their review work but have no other relevant financial relationships to disclose.

Author contributions

All named authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for authorship for this article. All authors contributed to study conception and design, data acquisition, analysis, and interpretation, drafting and revising of the manuscript.

Acknowledgements

The authors acknowledge Mary M Moran and Alejandro Cane (Pfizer, Inc. Employee) for specific contributions to this research project. Assistance with model conceptualization and development and input collection was provided by Ruth Chapman, Gyorgyi Feldmajer, and Cecilia Duan (Evidera). Medical writing was provided by Jacqueline Janowich Wasserott, and Jonathan Pitt (Evidera) and was funded by Pfizer, Inc.

Data availability statement

Aggregate data generated or analyzed during this study are available upon request.