Modeling the potential public health impact of different vaccination strategies with an omicron-adapted bivalent vaccine in Malaysia

ABSTRACT

Background

Coronavirus disease 2019 (COVID-19) case numbers have increased following the emergence of the Omicron variant. This study estimated the impact of introducing and increasing the coverage of an Omicron-adapted bivalent booster vaccine in Malaysia.

Research Design and Methods

A combined cohort Markov decision tree model was used to compare booster vaccination with an Omicron-adapted bivalent COVID-19 vaccine versus no booster vaccination in Malaysia. The model utilized age-specific data from January 2021 to March 2022 derived from published sources. The outcomes of interest included case numbers, hospitalizations, deaths, medical costs, and productivity losses. The population was stratified into high-risk and standard-risk subpopulations, and the study evaluated the benefits of increased coverage in different age and risk groups.

Results

Vaccinating only high-risk individuals and those aged ≥ 65 years was estimated to avert 274,313 cases, 33229 hospitalizations, 2,434 deaths, Malaysian ringgit (MYR) 576 million in direct medical costs, and MYR 579 million in indirect costs. Expanding vaccination coverage in the standard-risk population to 75% was estimated to avert more deaths (31%), hospitalizations (155%), infections (206%), direct costs (206%), and indirect costs (281%).

Conclusions

These findings support broader population Omicron-adapted bivalent booster vaccination in Malaysia with potential for significant health and economic gains.

KEYWORDS:

• Bivalent booster vaccine
• COVID-19
• Malaysia
• Omicron
• Omicron-adapted vaccine
• SARS-CoV-2

1. Introduction

Coronavirus disease 2019 (COVID-19), caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus, continues to be a public health threat worldwide. Globally, there were over 765 million confirmed cases and over 6.9 million deaths due to COVID-19 [1]. Between 2020 and 2021, COVID-19 was responsible for the loss of 336.8 million years of life worldwide [2]. As of 12 March 2023, there were over 5 million confirmed cases and over 35,000 deaths due to COVID-19 in Malaysia, whose total population was approximately 34 million in 2023 [1,3]. Malaysia has witnessed multiple spikes in COVID-19 cases, including two major outbreaks between July and October 2021 and between January and March 2022, followed by stability since April 2022 [1,3]. Notably, the number of reported cases likely reflects an underestimation of the true disease burden [4].

Vaccines are vital in the fight against COVID-19. After their development and regulatory approval, full vaccination with COVID-19 vaccines, including ancestral monovalent messenger ribonucleic acid (mRNA) vaccines, substantially reduced COVID-19 infections, hospitalizations, and deaths in Malaysia and worldwide [5]. Globally, as of 15 June 2023, over 13 billion vaccine doses, including all available platforms, were administered [1]. In Malaysia, over 72 million vaccine doses were administered, most of which were inactivated or mRNA vaccines [1]. Primary vaccination coverage among adults, adolescents, and children aged ≥5 years was 98.4%, 91.8%, and 43.5%, respectively, approaching the Malaysia Ministry of Health target of 50% coverage in the pediatric population aged 5–11 years [6]. Booster vaccination coverage among adults, adolescents, and children was 69.1%, 1.9%, and 0.2%, respectively [7]. Nevertheless, the emergence and spread of Omicron variants resulted in a surge of COVID-19 cases worldwide, including in countries with high vaccine coverage rates, such as the United States, Malaysia, Thailand, and Singapore [1,8].

The Omicron variant of SARS-CoV-2 emerged in Malaysia in December 2021. Its transmission quickly surpassed that of the Delta variant. The Omicron variant spreads more rapidly than the Delta variant, resulting in more rapid transmission in the community and higher incidence levels than previously observed during this pandemic [9]. By February 2022, 98.7% of all submissions to the Global Initiative on Sharing All Influenza Data (GISAID) from Malaysia were caused by the Omicron variant [10]. In response to the increase in the number of cases caused by the Omicron variant, the Malaysian government implemented the National COVID-19 Immunization Programme-Booster (PICK-B) [10]. PICK-B advocated for booster vaccination with BNT162b2 mRNA vaccines, monovalent ancestral strain mRNA vaccines, inactivated vaccines and viral vector ancestral monovalent vaccines first for high-risk individuals and frontline workers and then for the entire population. A 78.3% coverage rate was achieved by 31 March 2022 [10]. However, while these vaccines were effective against the Delta variant, their effectiveness against Omicron variants was significantly lower and waned faster [11–13].

Bivalent booster vaccines, including the Omicron-adapted BNT162b2 vaccine (Original and Omicron BA.4/BA.5; hereafter, Omicron-adapted vaccine), were developed to combat the waning efficacy of ancestral monovalent vaccines against Omicron variants. Bivalent vaccines provide longer-lasting and wider protection against circulating and previous variants [14]. In December 2022, the Malaysian Drug Control Authority granted conditional approval for the use of the Omicron-adapted vaccine [15] as booster doses for individuals aged ≥12 years.

The objectives of this modeling study were to estimate the public health and economic impact of the introduction of the Omicron-adapted bivalent vaccine in the high-risk and standard-risk populations and increasing vaccination coverage in Malaysia. The health outcomes explored included the number of COVID-19 cases, hospitalizations, outpatient cases, and death, while the economic outcomes included the number of doses administered, COVID-19-related direct medical costs, and productivity losses. Because this was not a cost-effectiveness analysis, quality-adjusted life years (QALYs) were not explored, and vaccine acquisition costs were not included.

2. Methods

2.1. Model structure and overview

The public health impact of booster vaccination with an Omicron-adapted bivalent COVID-19 versus no booster vaccination (2 doses of the primary COVID vaccine series) was assessed using a previously published decision analytic model developed using Microsoft Excel [16]. The model utilized a combined decision tree-Markov approach as previously described (Figure 1). The model tracked the health outcomes of individuals aged ≥6 months after receiving a single booster dose of the Omicron-adapted vaccine or receiving no booster vaccination over a hypothetical 1-year time horizon based on historical data. A 1-year time horizon was selected because existing evidence suggests that the duration of vaccine-induced protection is limited, and the SARS-CoV-2 virus continuously evolves, introducing uncertainty in estimating long-term model inputs.

Figure 1. Model structure. The Markov component (top) and decision tree component (bottom) of the model are shown. Abbreviations: ICU = intensive care unit; PASC = post-acute sequelae of COVID-19.

The Markov component of the model utilized a susceptible-infected-recovered (SIR) structure (Figure 1), which compartmentalized individuals in the population to estimate the numbers of individuals who are susceptible to disease, individuals who are infected, and individuals who have recovered. SIR models are commonly used in studies modeling infectious disease based on real-world data to predict disease transmission within communities and estimate the effects of interventions and there is a growing body of literature applying this modeling approach to COVID-19 [17–20]. The model simulated individuals’ transition through the health states based on their vaccination status. Individuals entered the model with either infection-induced immunity (recovered health state), vaccine-induced immunity (vaccinated health state), or no immunity (susceptible health state). The model did not include hybrid immunity, which leads to an underestimation of the benefits of vaccination.

At model entry, individuals in all health states could receive the Omicron-adapted booster, and the uptake rate was defined according to vaccine coverage assumptions. Depending on age-dependent probabilities, individuals could transition to another health state in each weekly cycle. The transition probabilities and model inputs were informed by the literature and stratified by age when possible. Infection-induced immunity in the recovered state and vaccine-induced immunity in the vaccinated state were assumed to wane at a rate of the reciprocal of the duration of protection, followed by a transition to the susceptible state. Individuals in the susceptible state transitioned to the infected state based on an age-dependent yearly attack rate. To calculate the proportion of individuals who transitioned from the vaccinated state to the infected state, the age-dependent yearly attack rate was decreased by the vaccine effectiveness measure. Following a transition to any infection state, individuals were stratified by the probabilities of symptomatic and asymptotic infections, hospitalization, admission to an intensive care unit (ICU), receiving invasive mechanical ventilation (IMV), developing post-COVID condition, and death.

The model was static but did incorporate a feature allowing the risk of infection to depend on the proportion of infected individuals. The model assumed that symptomatic patients were treated in an outpatient or inpatient setting, and hospitalized patients were assumed to be admitted to a regular ward or ICU and could receive IMV in either setting. Individuals surviving infection could experience long-term outcomes, such as post-COVID condition, regardless of whether they experienced symptoms in the infection state. Any death in the infected state was assumed to be caused by COVID-19, and death in the other states was considered all-other-caused mortality.

The model was validated following the International Society for Pharmacoeconomics and Outcomes Research and the Society for Medical Decision Making (ISPOR-SMDM) guidelines [21]. Specifically, internal validation was conducted by experts not involved in the development of the model. These experts evaluated all equations, inputs, and outputs for consistency and accuracy. Experts on the modeling team conducted face validity checks of all inputs and results. No cross-validation, external validation, or predictive validation were conducted.

2.2. Model outcomes

The model estimated the health and economic outcomes of booster vaccination with the Omicron-adapted vaccine from both the payer and societal perspectives. The health outcomes of interest included the number of COVID-19 cases (derived based on the attack rate) and the number of COVID-19-related hospitalizations (derived based on the hospitalization rate among infected patients), outpatient cases (derived based on the rate of cases not requiring hospitalization), and deaths (derived based on the mortality rate among hospitalized patients). The economic outcomes of interest included the number of booster vaccine doses administered, COVID-19 related costs (e.g. testing, adverse events [AEs], treatment, and post-COVID condition), and productivity losses. Treatment costs and productivity losses were specific to each setting (e.g. outpatient, general ward, and ICU). Inputs used to derive health and economic outcomes are presented in the model parameters section below. All health and economic outcomes were discounted by 3% according to the Pharmacoeconomic Guidelines for Malaysia by Ministry of Health Malaysia [22].

2.3. Model parameters

2.3.1. Population inputs

The population size estimates in Malaysia were informed by the Department of Statistics Malaysia [23]. The population was stratified into the following age groups: 6 months-4 years, 5–11 years, 12–17 years, 18–54 years, 55–64 years, 65–74 years, and ≥75 years. The population was further stratified into standard-risk and high-risk populations (Supplementary Table S1). Individuals in the standard-risk population had no comorbidities, while the high-risk population comprised individuals with at least one comorbidity, including chronic heart disease, liver or heart failure, asthma, immunodeficiency, etc. To avoid double counting, hypertension was excluded as a comorbid condition, and the prevalence of comorbid conditions was approximated by the most prevalent risk factor per age group as follows: cancer in the 5–17 years age group; smoking in the 18–54 and ≥75 years age groups; and cardiovascular disease in the 55–74 and ≥75 years age groups. This approach avoided double-counting, but likely leads to undercounting of the high-risk population. The inputs related to diabetes were informed by the International Diabetes Federation (IDF) Diabetes Atlas [24]. The inputs related to chronic obstructive pulmonary disease [25], chronic kidney disease [26], immunodeficiency [27] and cardiovascular disease [28] were informed by the literature. In the case of immunodeficiency, only data on primary immunodeficiency were available, which may be an underestimation of the full prevalence by omitting severe immunodeficiency. The inputs related to the smoking rate were informed by Macrotrends [29], and the inputs related to cancer were informed by the World Health Organization (WHO) [30].

2.3.2. Infection inputs

The model assumed that Omicron (BA.4 and BA.5 subvariants) was the only variant in circulation and was responsible for 100% of infections. The initial protection provided by the primary vaccination series (i.e. residual immunity conferred by two doses of an ancestral monovalent COVID-19 vaccine) was assumed to have mostly waned. Therefore, at the start of the simulation, the proportion in percentage of population vaccine-induced immunity was assumed to be 10% of the primary vaccination series coverage. The proportion of the population with infection-induced immunity at the start of the simulation was informed by case data from 1 October 2021 to 31 December 2021 reported by the Ministry of Health Malaysia [7]. The proportion of the population susceptible to COVID-19 was informed by the Ministry of Health Malaysia [7] and calculated using the following equation: 100% $-$ proportion of infection-induced immunity $-$ proportion of vaccine-induced immunity. The yearly attack rate in the susceptible population by age group was informed by the Ministry of Health Malaysia [7] (see Supplementary Table S2 for a detailed description of the infection inputs by age).

2.3.3. Vaccine inputs

It was assumed that 100% of the Malaysian population indicated for a COVID vaccine was eligible to receive booster vaccination with the Omicron-adapted bivalent vaccine (Ancestral and Omicron BA.4/BA.5). The proportion of the Malaysian population projected to receive booster vaccination by age was informed by coverage of the ancestral monovalent booster vaccine reported by the Ministry of Health Malaysia [7] (Supplementary Table S3).

Based on assumptions and initial limited real-world data [12,31–35], three vaccine efficacy profiles were explored. The profiles differed in the degree to which they could reduce infections (50%, 60%, and 70%), symptomatic infections (60%, 70%, and 80%), and severe disease (70%, 80%, and 90%). The duration of protection under each profile was 5 months, 6 months, and 7 months, respectively. Waning was modeled by moving a portion of the cohort from the vaccinated health state to the susceptible health state at a rate of the reciprocal of the duration of protection. The duration of protection was assumed to be the same across all age groups. Infection-induced immunity was assumed to last 3 months.

2.3.4. Health inputs

The probability of symptomatic infection was informed by Ng et al. [36]. The hospitalization rate among symptomatic patients, the critical care/ICU admission rate among hospitalized patients, the ventilation rate among hospitalized patients in regular wards, and the ventilation rate among patients in ICUs were informed by surveillance data from the Ministry of Health Malaysia [7] (Supplementary Table S5). Hospitalization data are available for the entire Malaysian population, but hospitalization data subdivided by age group in Malaysia are not available. Therefore, data regarding the age distribution of hospitalizations from the United States reported by the Centers for Disease Control and Prevention (CDC) for January 2022 and August 2022 [37] and data regarding the age distribution of COVID-19 cases in Malaysia [7] were utilized to estimate the probability of hospitalizations by age in the Malaysian population. The probability of death among inpatients and outpatients by age group were informed by the per capital inpatient and outpatient deaths from January to March 2022 reported by the Ministry of Health Malaysia [7] (Table S4). Due to limited data availability and the lack of published data differentiating mortality among inpatients based on the hospital ward and/or ventilation usage, we assumed that the probability of death did not differ between inpatients in regular wards and inpatients in ICUs or between inpatients with and without ventilation.

Severe COVID-19 is associated with preexisting comorbidities, particularly hypertension, diabetes, obesity, heart disease, and kidney disease. The relative risk of individuals in the high-risk population (i.e. individuals with comorbidities) developing complications was informed by published odds ratios/hazard ratios and risk probabilities of hospitalization (weighted average of 3.12) derived from Mattey-Mora et al. [38], severe COVID-19 requiring ICU admission (weighted average of 1.46) derived from Sim et al. [39], and mortality (weighted average of 3.99) derived from Surendra et al. [40].

Given the limited data available concerning post-COVID condition specifically in Malaysia, the probability of developing post-COVID condition was informed by data from the United States [41]. Similarly, due to the lack of data concerning the probability of developing post-COVID condition by age, the probability of developing post-COVID condition was assumed to be the same in all age groups and was further stratified by asymptomatic patients (36.9%), patients receiving outpatient care (36.9%), and patients receiving inpatient care (45.7%).

The adverse events (AEs) due to vaccination with the Omicron-adapted vaccine and its comparators considered in the model include myocarditis, pericarditis, myopericarditis, acute allergic reaction requiring hospitalization, disseminated intravascular coagulation, cerebral venous sinus thrombosis with thrombocytopenia, capillary leak syndrome, Guillain-Barre syndrome, pulmonary embolism, and acute myocardial infarction. The probabilities of all AEs were informed by Hause et al. [42], except for the probability of acute myocardial infarction, which was informed by Klein et al. [43] (Table S6). The AE rates were assumed to be the same in all age groups.

2.3.5. Cost inputs

The model adopted the Malaysian payer perspective and societal perspective and considered both direct and indirect costs. The model did not include vaccine costs, including vaccine acquisition and administration costs. The healthcare costs included testing costs, treatment costs, hospitalization costs, post-COVID condition costs, and AE-related costs. It was assumed that all patients, including asymptomatic patients, patients treated in the inpatient or outpatient setting, and patients in the ICU, received two COVID-19 tests (i.e. one test for diagnosis and one test to confirm treatment) at a cost of Malaysian ringgit (MYR) 112.32 per test (average of polymerase chain reaction [PCR] and rapid antigen tests) [44]. The unit cost of a general practitioner (GP) visit (MYR 83.99) was informed by the cost reported for Malaysia by the WHO [45], and patients were assumed to attend two GP visits. The cost of over-the-counter pain medication (MYR 5.96) was derived from the Ministry of Health Malaysia [46]. It was assumed that no asymptomatic patients visited a GP, and their treatment cost was assumed to be MYR 0.

Hospitalization costs were based on the median COVID-19 treatment cost (i.e. MYR 870) per day [47]. The same values were applied to protected and susceptible patients and patients treated with or without ventilation due to a lack of data. The length of stay in a regular ward with or without ventilation was assumed to be 11 days for a total cost of MYR 10,237, while the length of stay in the ICU was assumed to be 20 days for a total cost of MYR 18,613 [47].

The cost of managing post-COVID condition is included as a one-off cost applied at the time of infection; although the cost is incurred immediately, it represents the accumulation of costs accrued throughout the duration of post-COVID condition. Patients who experience post-COVID condition are assumed to attend four GP visits and one specialist visit and complete four COVID-19 tests. In total, 4% of patients are assumed to be hospitalized for post-COVID condition. The cost of a specialist visit (MYR 166) is assumed to be twice the cost of a GP consultation, and the cost of hospitalization for post-COVID condition (MYR 10,237) is assumed to be the same as the cost of inpatient admission. Costs related to vaccine-induced AEs were informed by data from the Healthcare Cost and Utilization Project (HCUP) [42] and converted from US dollars (Table S6).

The indirect costs considered in the model included productivity loss due to illness, post-COVID condition, or premature death. The productivity loss costs were based on the workforce participation rate and labor cost per week [47] (Table S7). The data source informing the costs and workforce participation rate covers individuals aged 15 to 64 years. Therefore, the workforce participation rate of those older than 64 years was extrapolated as follows: individuals aged ≥75 years were assumed to have a 0% workforce participation rate, while the workforce participation rate of those aged 65–74 years was extrapolated by assuming that these individuals had a rate similar to those aged 50–64 years. The working time lost among asymptomatic patients was based on the proportion of asymptomatic patients diagnosed with COVID-19 (9%), percentage of asymptomatic patients unable to work from home (44%) [48], and working time lost among asymptomatic patients unable to work from home [49]. Due to a lack of data from Malaysia, the working time lost among patients receiving outpatient care was derived from the isolation guidelines by the Centers for Disease Control and Prevention [50]. The working time lost among hospitalized patients admitted to a regular ward or ICU with or without ventilation by age was informed by Thant et al. [51] (Table S8). The indirect costs due to premature death were computed as lifetime market productivity based on the age group of the death, the expected life expectancy from life tables, workforce participation rate, and average wage rate.

2.4. Analysis

3 This study assessed the health and economic benefits of vaccination programs and increased coverage in the Malaysian population stratified into high-risk and standard-risk subpopulations. Specifically, the benefits of increasing coverage in different age groups and risk categories were explored. Furthermore, the impact of increasing vaccination coverage in standard-risk individuals was explored by estimating the outcomes with no vaccination among the standard-risk population and the incremental gains if coverage in the standard-risk population increased to 25%, 50%, and 75%. Coverage among the high-risk population and individuals aged ≥65 years was held constant at the base case assumption in the analysis. Additionally, alternative base cases were explored. Alternative burden of disease scenarios were explored in a sensitivity analysis by varying the age-specific attack rate, probability of hospitalization, and inpatient and outpatient death rates. The alternative rates were parameterized using data from alternative peaks of COVID-19 infections [7] (Table S9). We conducted a one-way, deterministic sensitivity analysis to further assess the impact of parameter uncertainty on results. This deterministic sensitivity analysis used the base-case strategy of high-risk population and individuals aged ≥55 years and varied over 350 parameters ± 20% of the base case.

3. Results

This analysis explored the health and economic benefits of vaccination programs targeting different age groups and risk categories. In the base case, only individuals aged ≥65 years and individuals considered high risk were eligible for vaccination, resulting in the administration of 5,469,261 additional booster vaccine doses. The incremental health gains included 274,313 cases, 33229 hospitalizations, and 2,434 deaths averted. Regarding the direct medical costs, the results were as follows: incremental cost savings of MYR 59 million in testing costs, MYR 347 million in inpatient treatment costs, MYR 347 million in outpatient treatment costs, and MYR 128 million in post-COVID condition treatment costs and an incremental cost increase of MYR 149,908 due to AE management, corresponding to incremental cost savings of MYR 576 million in direct medical costs. Regarding the indirect medical costs, the results were as follows: incremental cost savings of MYR 574 million in productivity loss from illness, MYR 2 million in productivity loss from death, and MYR 3 million in education loss, corresponding to incremental cost savings of MYR 579 million in indirect medical costs (Table 1).

Table 1. Base case results and expanding vaccination to other age and risk groups.

	Base Case: Age ≥65 Years and High Risk			Age >55 Years and High Risk			Age >55 Years, Age 6 Mo-5 Years, and High Risk
	No Vaccination	Vaccination	Incremental Impact	No Vaccination	Vaccination	Incremental Impact	No Vaccination	Vaccination	Incremental Impact
Vaccine doses administered	0	5,469,261	5,469,261	0	5,992,760	5,992,760	0	6,576,703	6,576,703
Deaths	13,387	10,953	2,434	13,387	10,795	2,593	13,475	10,875	2,600
Hospitalizations	347,084	313,855	33,229	347,084	310,486	36,598	375,536	336,527	39,009
Infections	5,031,118	4,756,805	274,313	5,031,118	4,733,676	297,442	5,320,001	5,001,249	378,751
Direct medical costs (million MYR)	8,387	7,811	576	8,387	7,757	630	8,953	8,280	672
Testing cost (million MYR)	1,268	1,209	59	1,268	1,204	64	1,340	1,272	69
AE management cost	0	149,908	−149,908	0	164,256	−164,256	0	180,262	−180,262
Inpatient treatment cost (million MYR)	3,603	3,256	347	3,603	3,220	382	3,895	3,488	407
Outpatient treatment cost (million MYR)	815	773	347	815	769	382	860	811	407
Post-COVID condition treatment cost (million MYR)	2,702	2,574	128	2,702	2,563	139	2,858	2,709	148
Indirect costs (million MYR)	11,584	11,006	579	11,584	10,957	628	11,585	10,957	629
Productivity loss due to illness (million MYR)	11,409	10,835	574	11,409	10,786	623	11,409	10,786	623
Productivity loss due to death (million MYR)	11	9	2	11	9	2	11	9	2
Education loss (million MYR)	164	161	3	164	161	3	164	161	3

*Base case coverage in the high-risk population: 0% (5–11 years), 1.5% (12–17 years), 58.2% (18–29 years), 69.3% (30–54 years), 73.6% (55–64 years), 72.6% (65–74 years), and 61.9% (≥75 years)

Abbreviations: AE = adverse event, MYR = Malaysian ringgit.

In the analysis, vaccination eligibility was expanded to individuals aged >55 years and those considered high risk, which increased the number of vaccine doses administered by 10% and averted more cases (27%), hospitalizations (10%), deaths (7%), and direct (9%) and indirect (8%) costs. The incremental health gains included 297,442 cases, 36598 hospitalizations, and 2,593 deaths averted. Regarding the direct medical costs, the results were as follows: incremental cost savings of MYR 64 million in testing costs, MYR 382 million in inpatient treatment costs, MYR 382 million in outpatient treatment costs, and MYR 139 million in post-COVID condition treatment costs and an incremental cost increase of MYR 164,256 due to AE management, corresponding to incremental cost savings of MYR 630 million in direct medical costs. Regarding the indirect medical costs, the results were as follows: incremental cost savings of MYR 623 million in productivity loss from illness, MYR 2 million in productivity loss from death, and MYR 3 million in education loss, corresponding to incremental cost savings of MYR 628 million in indirect medical costs (Table 1).

Finally, we compared individuals aged >65 years and those considered high risk with vaccination eligibility expanded to those aged >55 years, those aged >6 months to 5 years, and those considered high risk, resulting in an increase in the number of vaccine doses administered (10%) and averting more cases (27%), hospitalizations (8%), and direct (6%) costs, with no change in indirect costs and minimal change in deaths (0.2%). These results reflect the low hospitalization rate and near-zero mortality rate in the group aged >6 months to 5 years. Specifically, the incremental health gains included 378,751 cases, 39009 hospitalizations, and 2,600 deaths averted. Regarding the direct medical costs, the results were as follows: incremental cost savings of MYR 69 million in testing costs, MYR 407 million in inpatient treatment costs, MYR 407 million in outpatient treatment costs, and MYR 148 million in post-COVID condition treatment costs and an incremental cost increase of MYR 180,262 due to AE management, corresponding to incremental cost savings of MYR 672 million in direct medical costs. Regarding the indirect medical costs, the results were as follows: incremental cost savings of MYR 623 million in productivity loss from illness, MYR 2 million in productivity loss from death, and MYR 3 million in education loss, corresponding to incremental cost savings of MYR 629 million in indirect medical costs (Table 1).

Next, the impact of higher vaccination coverage rates in the standard-risk population aged 5 years to 65 years was explored. Expanding the coverage rate in the standard-risk population to 25%, 50%, and 75% greatly increased the number of vaccine doses administered compared to the base case by 98%, 195%, and 293%, respectively. Importantly, these additional doses averted significantly more COVID-19-related deaths (by 10%, 21%, and 31%), hospitalizations (by 52%, 103%, and 155%), infections (by 96%, 138%, and 206%), direct costs (by 69%, 138%, and 206%), and indirect costs (by 94%, 187%, and 281%) (Table 2). In further analyses, we explored the impact of increasing vaccination coverage in the standard-risk population aged 5–54 years while using alternative base cases, and similar results were observed (Tables 3 and 4).

Table 2. Scenario analysis results expanding vaccination in the standard-risk population compared to base case of Age ≥65 and High Risk.

	Base case: Age ≥ 65and High Risk*	Standard RiskCoverage: 25%	Standard RiskCoverage: 50%	Standard RiskCoverage: 75%
Vaccine doses administered	5,469,261	10,811,476	16,153,691	21,495,905
Deaths averted	2,434	2,684	2,935	3,185
Hospitalizations averted	33,229	50,385	67,541	84,698
Infections averted	274,313	537,856	801,399	1,064,942
Direct medical costs averted (million MYR)	576	972	1,368	1,765
Indirect costs averted (million MYR)	579	1,121	1,663	2,205

*Base case coverage: low-risk population: 0% (5–64 years), 72.6% (65–74 years), and 61.9% (≥75 years); high-risk population: 0% (5–11 years), 1.5% (12–17 years), 58.2% (18–29 years), 69.3% (30–54 years), 73.6% (55–64 years), 72.6% (65–74 years), and 61.9% (≥75 years)

Standard-risk population comprises individuals with no comorbidities.

Abbreviations: MYR = Malaysian ringgit.

Table 3. Scenario analysis results expanding vaccination in the standard-risk population compared to base case of Age ≥55 and High Risk.

	Base case: Age > 55and High Risk*	Standard RiskCoverage: 25%	Standard RiskCoverage: 50%	Standard RiskCoverage: 75%
Vaccine doses administered	5,992,760	11,157,203	16,321,646	21,486,089
Deaths averted	2,593	2,789	2,986	3,182
Hospitalizations averted	36,598	52,610	68,622	84,634
Infections averted	297,442	553,131	808,819	1,064,508
Direct medical costs averted (million MYR)	630	1,008	1,386	1,764
Indirect costs averted (million MYR)	628	2,161	3,065	3,968

*Base case coverage: low-risk population: 0% (5–64 years), 72.6% (65–74 years), and 61.9% (≥75 years); high-risk population: 0% (5–11 years), 1.5% (12–17 years), 58.2% (18–29 years), 69.3% (30–54 years), 73.6% (55–64 years), 72.6% (65–74 years), and 61.9% (≥75 years)

Standard-risk population comprises individuals aged 5–54 years with no comorbidities.

Abbreviations: MYR = Malaysian ringgit.

Table 4. Scenario analysis results expanding vaccination in the standard-risk population compared to base case of Age 6 mo-4 years or > 55 and High Risk.

	Base case: Age 6 mo-4 yearsor >55 years and High Risk*	Standard RiskCoverage: 25%	Standard RiskCoverage: 50%	Standard RiskCoverage: 75%
Vaccine doses administered	6,576,703	11,741,146	16,905,589	22,070,031
Deaths averted	2,600	2,796	2,993	3,190
Hospitalizations averted	39,009	55,021	71,033	87,045
Infections averted	378,751	575,440	830,128	1,085,817
Direct medical costs averted (million MYR)	672	1,050	1,428	1,806
Indirect costs averted (million MYR)	1,301	2,204	3,107	4,011

*Base case coverage: low-risk population: 25% (6 months-4 years), 0% (5–64 years), 72.6% (65–74 years), and 61.9% (≥75 years); high-risk population: 25% (6 months-4 years), 0% (5–11 years), 1.5% (12–17 years), 58.2% (18–29 years), 69.3% (30–54 years), 73.6% (55–64 years), 72.6% (65–74 years), and 61.9% (≥75 years)

Standard-risk population comprises individuals aged 5–54 years with no comorbidities.

Abbreviations: MYR = Malaysian ringgit.

In the sensitivity analysis, the burden of disease was varied by varying the attack rate, probability of hospitalization, and inpatient and outpatient death rates using data from alternative time periods: wave 1 (Alpha variant) and wave 2 (Delta variant). The Alpha variant was associated with a greater probability of hospitalization than the Omicron variant. However, the attack rate of the Omicron variant was much greater than that of the Alpha variant, possibly due to underreporting. The Delta variant was associated with a substantially higher probability of hospitalization and inpatient mortality than the Omicron variant, especially in individuals aged ≥55 years (Table S9).

In the scenario analysis using burden of disease estimates from wave 1 (Alpha variant) when analyzing the base case of the high-risk and age ≥ 65 population, booster vaccination was estimated to avert 436 deaths, 13270 hospitalizations, 30473 infections, MYR 163 million in total direct costs, and MYR 70 million in indirect costs (Table 5). Using the burden of disease estimates from wave 2 (Delta variant), booster vaccination was estimated to avert 9,740 deaths, 45497 hospitalizations, 195,947 infections, MYR 633 million in total direct costs and 523 million in indirect costs (Table 5). The base case results fall between these estimates for deaths, hospitalizations, and costs averted and are higher for infections averted. These results are largely driven by the low attack rates associated with the Alpha variant and the high probabilities of hospitalization and inpatient death associated with the Delta variant. Overall, the Omicron-related results are situated between those of the Alpha and Delta variants. These results more closely resemble the results obtained during Wave 2.

Table 5. Scenario analysis results varying the disease burden in the base-case analysis of the high-risk population.

	Wave 1 (Alpha Variant)	Wave 2 (Delta Variant)	Base Case: Wave 3 (Omicron Variant)
Vaccine doses administered	5,469,261	5,469,261	5,469,261
Deaths averted	436	9,740	1,446
Hospitalizations averted	13,270	45,497	21,799
Infections averted	30,473	195,947	274,371
Direct medical costs averted (million MYR)	163	633	458
Indirect costs averted (million MYR)	70	523	573

Wave 3 served as the base case.

Wave 1 (Alpha variant) was between January 2021 and March 2021, Wave 2 (Delta variant) was between July 2021 and September 2021, and Wave 3 (Omicron variant) was between January 2022 and March 2022.

High-risk population comprises individuals with at least one comorbidity and individuals aged ≥65 years.

Abbreviations: MYR = Malaysian ringgit.

Figure 2 presents a tornado diagram of the one-way deterministic sensitivity analysis using the base case of high-risk and age ≥ 55 populations. The 25 parameters that had the greatest impact on results were reported. The parameters whose variation most impacted results were productivity-related (wages and participation rate) and clinical (probability of symptomatic infection, attack rate, severity of hospitalization, etc). Even under the worst case scenario of the parameters that had the greatest impact, results still showed a total cost saving of more than one billion.

Figure 2. One-way deterministic sensitivity analysis. Tornado diagram showings the top 25 parameters with the greatest impact on the results.

4. Discussion

In this study, the Malaysian population was stratified into a standard-risk population (defined as the population with no comorbidities) and a high-risk population (defined as the population with comorbid conditions). The current analysis evaluated the health and economic impact of increasing the coverage of booster vaccination with the Omicron-adapted vaccine in the standard-risk population in Malaysia. The results revealed that while focusing on vaccination in the high-risk and age ≥ 65 populations has the potential to achieve a large public health and economic impact, expanding the recommendation to the population aged ≥55 years has the potential to have an even greater public health impact, averting more cases (27%), hospitalizations (10%), deaths (7%), direct costs (9%), and indirect costs (8%). Expanding to the population aged ≥55 years was projected to increase doses administered by 10%, which is greater than the estimated percentage increase in deaths and direct costs averted, but less than or equal to the estimated cases and hospitalizations averted. Additionally, expanding booster vaccination coverage in the standard-risk population aged 5–64 years has the potential to have an even larger impact. Specifically, expanding vaccination coverage to 75% in the standard-risk population aged 5–64 years has the potential to further increase the number of deaths averted by 31%, hospitalizations averted by 155%, infections averted by 288%, and direct and indirect costs averted by 206% and 281%, respectively. Achieving 75% vaccination coverage may be challenging, but these results demonstrate the potential gains from achieving such a high coverage rate.

An early study in Malaysia reported that between 1 February 2022 and 30 May 2022, during the Omicron BA.1./BA.2 wave, most cases (92%) were mild, with a case fatality rate of 0.7% [8]. In Malaysia, severe COVID-19 requiring intubation and/or IMV has been associated with older age (≥60 years), experiencing certain symptoms (i.e. fever, cough and anosmia), and underlying conditions (i.e. hypertension, chronic kidney disease, chronic pulmonary disease, and diabetes) [8,52,53]. The proportion of patients requiring intubation and/or IMV in Malaysia (2.2% [53]) is similar to that in China (2.3–3% [54,55]) but somewhat lower than that in Singapore (5.5%) and Thailand (7.1%) and substantially lower than that in the United States (20.2–22.3%) [52,56]. This highlights the importance of developing a model that is specific to the Malaysia context to capture these important differences in COVID severity and healthcare resource utilization.

Historically, in the context of supply concerns and inequities in vaccine access, booster vaccination campaigns tended to target high-risk and elderly individuals because these populations are at a higher risk of hospitalization and death [57]. It has been suggested that the booster vaccine is especially important for the elderly and high-risk populations [58]. While the elderly population has been historically defined as those aged ≥65 years [59], various definitions of the elderly have been adopted (i.e. the elderly is defined as those aged ≥60 years by the United Nations [60]). Therefore, in this study, we compared the base case to vaccination of only individuals aged ≥55 years or vaccination of only individuals aged 6 months to 5 years. Expanding vaccination to these age groups improved the outcomes, averting more cases (27%), hospitalizations (8%), deaths (0.2%), and direct (6%) costs, and the number of additional doses administered only increased by 10%; however, the best outcomes were observed when vaccination was expanded to standard-risk individuals aged 5–64 years or 5–55 years. Our results illustrate that increasing vaccination coverage among young individuals results in significant health and economic benefits.

The results of this study suggest that broader recommendations to vaccinate standard-risk and younger populations would lead to substantial reductions in cases, hospitalizations, medical costs and productivity losses. These data are particularly relevant for decision makers formulating policies related to the incorporation of booster vaccination in national vaccination programs. Such policy decisions should be evidence based and target populations of interest. While there may be upfront challenges to the healthcare system of implementing a broader vaccination strategy, the health and economic benefits that accrue over the year would be substantial.

In the sensitivity analysis, burden of disease inputs from earlier waves of COVID-19 were explored as each wave was associated with a different set of clinical probabilities. The results demonstrate the importance of the burden of disease on the estimated impact of vaccination. The results were sensitive to using these alternative inputs for the attack rate, probability of hospitalization, and probability of inpatient mortality. For example, vaccines were estimated to have prevented more cases when using burden of disease inputs from the Omicron period but were estimated to have prevented more hospitalizations and deaths using burden of disease inputs from the Delta period. These results demonstrate that the benefit of vaccination in the future depends on the future burden of disease.

Certain limitations of this analysis should be considered when interpreting the results. First, this study applied a static model, providing a simplified depiction of COVID-19 disease transmission and its consequences. In contrast, dynamic models are superior in estimating indirect effects. However, data concerning SARS-CoV-2 transmission and the impact of booster vaccination with a bivalent Omicron-adapted vaccine are limited, justifying the use of a static model to simplify the epidemiological dynamics. Furthermore, static models have been shown to be preferable for analyses of the economic benefits of vaccines [61]. Nevertheless, given the static nature of the model, the risk of infection was captured exogenously as an input parameter rather than modeled directly. Additionally, only the direct effects of vaccination were captured, the model did not consider herd immunity or population effects, and the model did not explicitly consider interactions between infected and uninfected individuals. Therefore, the model may have provided a conservative estimate of the indirect economic benefits of the Omicron-adapted bivalent booster vaccine. Second, some inputs were not available for Malaysia and were derived from data from the United States or were based on assumptions. For example, the AE-related costs were derived from United States data and converted to MYR from United States dollars, which could have resulted in an overestimation of the costs. The costs associated with post-COVID condition are difficult to estimate because patients are treated based on acute complaints rather than a diagnosis of post-COVID condition. Additionally, post-COVID condition is a diagnosis of exclusion [62]; therefore, clinically, a patient may attend more specialist visits and undergo more tests than assumed in the model to rule out other conditions, potentially resulting in an underestimation the related healthcare resource use and costs. Furthermore, data regarding the probability of death differentiating hospitalized patients in regular wards from those in ICUs are lacking; therefore, the same probability was applied regardless of the ward in which patients are treated, potentially leading to an underestimation of deaths in ICUs. Granular hospitalization data differentiated by age group in Malaysia are lacking, preventing a more in-depth analysis of the impact of COVID-19 booster vaccination on age-specific hospitalization rates. Third, due to a lack of precise data regarding the prevalence of comorbidities in Malaysia, hypertension was excluded to avoid double counting given the high co-occurrence of hypertension and other comorbidities. Therefore, individuals with hypertension without other comorbidities could have been included in the standard-risk population, which may have resulted in an underestimation of the high-risk prevalence. Additionally, the source used for immunodeficiency only included primary immunodeficiency and so omits severe immunodeficiency and underestimates the prevalence of immunodeficiency. Fourth, the model did not consider vaccine acquisition and administration costs but rather estimated the economic impact of booster vaccination in terms of healthcare resource use and the indirect costs associated with COVID-19 and/or its sequelae. This study estimated that the associated costs with booster vaccination would be lower than those without booster vaccination but did not assess the overall cost-effectiveness of implementing a vaccination program. Fifth, the results are sensitive to changes in vaccine efficacy, duration of protection, and disease burden. To explore the burden of disease, data from three waves of COVID-19 (from January 2021 to March 2022) were compared; however, data from earlier waves of COVID are not as reliable. Specific groups may respond with different vaccine efficacy (e.g. those with immunodeficiency), but no data are available to indicate such differences and whether the response is higher or lower in each population. Furthermore, vaccine efficacy was not age specific and was based on assumptions drawn from the bivalent vaccine. Future vaccines may be needed to address different strains. However, similar results may apply to future variant adapted vaccines depending on their observed efficacy and duration of protection compared to the assumptions in this modeling study. The assumptions applied in this study were necessary due to a lack of data, particularly granular, detailed data. The collection of more specific and accurate data in future studies could shed light on the impact of these limitations and the robustness of the results. Additionally, future analyses using more recent data of the Omicron variant are warranted. Despite these limitations, our model represents the first model specifically constructed for Malaysia. This is an important advancement because it is essential that the models used to evaluate policy choices are adapted to the context of a specific country. It is possible that this model can be refined in the future to incorporate new data and also adapted to other countries.

5. Conclusions

These findings support broader population Omicron-adapted bivalent booster vaccination in Malaysia with the potential for significant gains in public health and economic outcomes. These data can guide decision makers aiming to formulate the most effective COVID-19 vaccination policies for different age groups. Our results suggest that a strategy targeting booster vaccination in standard-risk individuals aged ≥55 years and those <5 years should be prioritized because the results of the strategy targeting this group showed the highest potential gains. Additionally, the results suggested that additional gain can potentially be achieved by efforts to increase uptake and a gradual implementation of booster vaccination among standard-risk individuals aged 5–54 years. It is important for policymakers to use data to guide policy; thus, future research should continue to improve the model parameters and retrospectively assess the impact of the chosen strategy.

Declaration of interest

KT, JS, MHK, JY, and CM are employees of Pfizer and may hold stock or stock options of Pfizer. BY is an employee of Evidera, which received financial support from Pfizer in connection with the study and the development of this manuscript. Medical writing and editorial support was provided by Dr. Ruth Sharf-Williams at Evidera and was funded by Pfizer.

Reviewer disclosures

Peer reviewers on this manuscript have received honoraria for their review work. Peer reviewers on this manuscript have no other relevant financial or other 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.

Acknowledgments

We would like to thank Dr. Sharlini T Surendran, Country Medical Director, Pfizer Malaysia for her support. Assistance with model conceptualization and development and input collection was provided by Josie Dodd, Solene De Boisvilliers, and Paul Lozowicki (Evidera). Medical writing was provided by Dr. Ruth Sharf-Williams (Evidera) and was funded by Pfizer, Inc.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/14760584.2023.2245465

Additional information

Funding

This study was funded by Pfizer.