The population considered in this analysis was adults aged ≥ 50 years old in five countries in Southeast Asia: Indonesia, Malaysia, Philippines, Thailand, and Vietnam. These five countries are among the most populated in the region, and generally share similar socioeconomic and sociodemographic ranges [7].

A static multicohort Markov model, developed in Microsoft Excel [15, 18,19,20,21,22], was adapted to assess the public health burden of HZ and public health impact of RZV vaccination in the general population aged ≥ 50 years old in Southeast Asia. The model followed individuals from the year of vaccination and over their estimated remaining lifetime, with a cycle length set at 1 year. Two strategies were compared in each country: (1) no RZV and (2) vaccination with RZV.

HZ cases, PHN cases, and other non-PHN complications (classified as ocular, neurological, cutaneous, or other non-pain-related complications) were generated from the model scenario of no RZV to reflect the public health burden of HZ on a population level. HZ cases avoided, PHN cases avoided, and other non-PHN complications avoided by RZV were generated from the model scenario of vaccinating with RZV to reflect the absolute public health impact of RZV and were divided by the total cases to show the relative public health impact of introducing RZV.

Further details on the model design are provided in previous publications [15, 18,19,20,21,22].

The base population included adults aged ≥ 50 years old, stratified according to five age cohorts (50–59 years, 60–64 years, 65–69 years, 70–79 years, and 80+ years) residing in Indonesia, Malaysia, Philippines, Thailand, and Vietnam (Table 1). National population sizes for each country were taken from population pyramids reported in 2022 [23], and the annual probability of all-cause mortality was obtained from the World Health Organization (Supplementary Table 2) [24].

Epidemiology inputs included HZ annual incidence, proportion of patients with PHN or other complications (classed as ocular, neurological, cutaneous, or other non-pain-related complications), recurrence rate of HZ and PHN, and case fatality rates of HZ.

Asian regional age-specific HZ incidence data were taken from a worldwide meta-regression study, which reported that geographic location, gender, and year of analysis had no statistically significant impact on HZ incidence [13]. Recurrence rates of HZ episodes were assumed to be the same as the first occurrence [25].

The proportions of HZ cases with PHN or non-PHN-related complications were selected from a pooled analysis of PHN risk among patients with HZ in North and South America and Asia (Canada, Brazil, Mexico, Argentina, Taiwan, South Korea, and Thailand) [26] and a literature review of HZ burden in the Asia–Pacific region (Australia, China, Hong Kong, India, Indonesia, Japan, Korea, Malaysia, New Zealand, Philippines, Singapore, Taiwan, Thailand, and Vietnam) [8], respectively (Table 1). Recurrent HZ cases were assumed to be associated with the same risk of developing PHN.

Case fatality rates were conservatively set at 0% as no credible sources were available and the impact of HZ on life expectancy was minor.

The assumed first-dose coverage for RSV was set at 30% [27], with a second-dose compliance rate of 70% [28]. A 2-month time window between two doses of RZV was assumed.

RZV vaccine efficacy rates against HZ and PHN were calculated on the basis of granular data taken from the ZOE-50 and ZOE-70 clinical studies, and the long-term follow-up (LTFU) study ZOE-LTFU (Table 1) [20, 29,30,31,32]. Vaccine efficacy for individuals receiving one dose of RZV was assumed to be 90.1% for individuals aged 50–69 and 69.5% for individuals aged ≥ 70 years, while efficacy in individuals receiving both doses was assumed to be 98.9% and 95.4%, respectively [20, 29,30,31,32]. Waning rates for individuals receiving both doses of RZV were age- and time-dependent: efficacy waned at 1.5% annually for individuals aged 50–69 years old, while efficacy waned at 2.3% for individuals aged ≥ 70 years old [29, 31].

For individuals receiving only one dose of RZV, waning rates were assumed to be the same as that of the live attenuated zoster vaccine (5.4% in the first 4 years post-vaccination and 5.1% in subsequent years) based on short-term and long-term persistence studies [33, 34]. This assumption was also used in a German public health impact study, where it was validated by an advisory board [20].

However, some adjustments for vaccine efficacy were made in the first 2 years post-vaccination. In the first year (first cycle of the model), an average vaccine efficacy of 2 months for individuals receiving a single dose of RZV, and 10 months for individuals receiving both doses were applied, as the time window between the two doses was assumed to be 2 months. In the second year, a half-cycle correction was applied to capture the full decline of efficacy from the completion of both doses until the end of the second year. In the subsequent years, age-dependent waning rates were used as described above.

The numbers of events (HZ cases, PHN cases, and non-PHN complications) over the remaining lifetime of adults in each age cohort and overall adults aged ≥ 50 years old were reported for each country and summed to generate an overall estimate. The absolute numbers of events avoided with RZV vaccination were reported for the same cohort for their remaining lifetime and were divided by total number of events to generate relative cases avoided.

The number of individuals needed to vaccinate (NNV) to avoid one case of PHN was calculated using the following equation:

$$}=1\div \left(\left(\frac}}\right) -\left(\frac}}\right)\right).$$

A one-way deterministic sensitivity analysis (OWSA) was performed to assess the robustness and uncertainty of base case inputs for each country setting included in the study (Supplementary Table 3). Each model input was adjusted ± 20% or 95% confidence interval from the base case value to evaluate the relative impact on the number of HZ cases avoided from vaccination. A probabilistic sensitivity analysis (PSA) was also carried out to determine the impact of full uncertainty in the model inputs and the impact on the result estimates. A total of 1000 Monte Carlo simulations were performed, sampling input values from available probability distributions. A correlation coefficient of 0.5 was assumed for all age-specific incidence parameters.

Scenario analyses were performed by altering model assumptions to explore the impact of vaccine coverage rates, second-dose compliance, and changing demographics or population size receiving the vaccine. One scenario applied demographics data from local statistics offices in Malaysia (applying projected population sizes in 2022 and natural mortality rates in 2018 from the Department of Statistics Malaysia) [35], Philippines (applying projected population sizes in 2022 and natural mortality rates in 2016 from the Philippine Statistics Authority) [36], and Thailand (applying projected population sizes in 2022 and natural mortality rates in 2012 from the National Statistic Office of Thailand) [37]. A second scenario analysis compared the best case scenario of first-dose coverage (80%) and second-dose compliance rates (95%) with worst case scenario of first-dose coverage (20%) and second-dose compliance (45%).

The model was based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.