Model Description

Lifetime clinical outcomes and economic costs of IPD and all-cause NBP were projected using a probabilistic cohort model with a Markov-type process in a hypothetical population of Argentinean adults aged 18–99 years. Figure 1 depicts the model structure, which has previously been used in evaluation of PCV20 vaccine [26]. The model population is initially characterized based on age in 1-year increments and risk profile (i.e., low, moderate, and high risk). Persons are allowed to transition to from low-risk to moderate-risk or from moderate-risk to high-risk group, but not to a lower risk group, during the modeling horizon.

Fig. 1

Persons in the model population may receive PCV20 alone, PCV13 → PPSV23, PCV15 → PPSV23, or no vaccine at model entry, depending on the vaccination strategy and coverage. Clinical outcomes and economic costs are projected annually for the model population based on age, risk profile, disease/fatality rates, vaccination status, vaccine type, and time since vaccination. IPD includes bacteremia and meningitis, and NBP is stratified by care setting (inpatient vs. outpatient). Persons vaccinated at model entry may be at lower risk of future IPD and NBP; the magnitude of vaccine-associated risk reduction depends on clinical presentation (i.e., IPD or NBP), as well as the vaccine(s) received, proportion of disease that is vaccine-preventable, age, time since vaccination, and risk profile. Risk of death from IPD, NBP requiring inpatient care, and other causes (i.e., other than IPD and NBP) depends upon age and risk profile. Expected costs of medical treatment for IPD and NBP are generated based on event rates and unit costs in relation to the setting of care (i.e., inpatient vs. outpatient), age, and risk profile. Costs of vaccination—including the vaccine and its administration—are tallied in the year(s) in which the vaccination occurs (i.e., year 1 for single vaccine strategies, years 1 and 2 for sequential vaccination strategies). Clinical outcomes and economic costs are projected over the modeling horizon for the alternative vaccination strategies considered and include expected numbers of cases of IPD and NBP (inpatient and outpatient), number of deaths due to IPD and inpatient NBP, number of life-years (unadjusted and quality-adjusted), costs of medical treatment for IPD and NBP, and costs of vaccination.

Model Estimation

A summary of methods used in estimation of key model parameters is presented below; base case model inputs are included in Table 1.

Table 1 Base case model input valuesPopulation

The size of the Argentinean population was based on the 2019 Argentina Encuesta Permanente de Hogares (EPH) Survey [27]. Persons in each age group (i.e., 18–49, 50–64, 65–74, 85–99 years) were allocated into low-, moderate-, and high-risk subgroups based on risk distribution presented in two Spanish database studies [28, 29]. The low-risk group included immunocompetent persons without underlying medical conditions, the moderate-risk group included immunocompetent persons with ≥ 1 underlying medical conditions (e.g., chronic heart disease, chronic pulmonary disease, diabetes, alcoholism, chronic liver disease, smoking, obesity), and the high-risk group included immunocompromised persons (e.g., due to immunosuppressive therapy, HIV, cancer) [28, 29].

Rates of Disease

IPD incidence was estimated using incidence of hospitalized pneumonia from Lopardo et al. (which was based on prospective active surveillance) [30], and the proportion of hospitalized pneumonia attributable to pneumococcus and the proportion of pneumococcal pneumonia that is invasive from Said et al. [31]. Because Lopardo et al. reported results for all persons aged 65–99 years in a single age group, estimates of pneumonia incidence were allocated across ages 65–74, 75–84, and 85–99 years based on relative rates of incidence by age group from Buzzo et al. [32] and the Argentinean population distribution [27]. As Lopardo et al. captured all cases of pneumonia (i.e., including all-cause NBP, treated in the inpatient and outpatient settings), incidence of IPD was derived by multiplying incidence of all cases of pneumonia by the proportion of pneumonia requiring hospitalization (68%) [30], proportion of hospitalized pneumonia attributable to pneumococcus (27.3%) and proportion of pneumococcal pneumonia that is invasive (24.8%) based on Said et al. [31]. These estimates were further apportioned across bacteremia (95%) and meningitis (5%) based on Drijkoningen et al. [33]. Age-specific rates of bacteremia and meningitis were then allocated across risk groups based on US CDC 2017/2018 data provided to Pfizer Inc. [34] and underlying population risk distributions.

Age-specific rates of inpatient and outpatient NBP were based on values reported by Buzzo et al. [32]. Because Buzzo et al. reported rates only for persons aged ≥ 50 years, incidence of inpatient NBP for persons aged 18–49 years was estimated using relative rates for 18–49 years (vs. 50–64 years) from Pelton et al. [35]. Similarly, incidence of outpatient NBP for persons aged 18–49 years was estimated using relative rates for 18–49 years (vs. 50–64 years) from Nelson et al. [36]. These estimates were further apportioned across risk profile using assumed relative risk-based published sources [35, 37] and underlying population risk distributions.

Case-Fatality and Mortality Rates

Age-specific case-fatality rates (CFRs) for bacteremia were based on data from a longitudinal study conducted in a single center in Argentina [38]. To derive CFRs across age groups of interest, point estimates were anchored to selected ages within the available age groups and methods of linear interpolation/extrapolation were employed. CFRs were apportioned across risk groups based on relative risk of mortality for moderate and high risk [34] and the age- and risk-specific distributions of bacteremia and meningitis cases. The CFR for meningitis was assumed to be the same as that for bacteremia. CFR for inpatient NBP was based on Buzzo et al. [32]. Because Buzzo et al. reported results only among persons aged ≥ 50 years, relative risk of death for persons aged 18–49 years (vs. 50–64 years) based on Averin et al. [39] was used to estimate inpatient NBP CFR for those aged 18–49 years. Age-specific CFRs were then apportioned across risk groups based on relative risk of mortality for moderate and high risk (vs. low risk) [39] and the age-specific risk distributions of inpatient NBP cases. Persons affected by outpatient NBP were assumed to not have an elevated risk of mortality.

Age-specific risk of death due to other causes (i.e., not related to IPD or inpatient NBP) was based on 2019 general population mortality rates in Argentina reported by INDEC [40] and apportioned across risk groups based on assumed RRs for mortality and age-specific risk distributions. Age- and risk-specific mortality rates were downwardly adjusted to account for death due to IPD and inpatient NBP.

Vaccine Effectiveness

For PCV20, vaccine effectiveness (VE) against vaccine type (VT) disease was assumed to be the same as that for PCV13 based on published results of an evaluation of PCV20 safety, tolerability, and immunogenicity [41]. Lacking alternative data, VE-PCV15 against VT disease was also assumed to be the same as VE-PCV13. For low-/moderate-risk persons aged ≥ 65 years, initial (i.e., year 1 of modeling horizon) VE of PCVs against vaccine type (VT) IPD and VT-NBP was assumed to be the same for all persons based on data from the Community-Acquired Pneumonia Immunization Trial in Adults (CAPiTA) per-protocol population; for persons aged 50–64 years, initial VE was derived from post hoc analyses of CAPiTA; and for persons aged 18–49 years, initial VE was assumed to be the same as that for persons aged 50 years [5, 42]. For high-risk persons aged 18–99 years, initial VE against VT-IPD and VT-NBP was assumed to be equal to 80% of corresponding values for low-/moderate-risk persons [43, 44]. Initial VE-PCV was assumed to persist for the first 5 years of the modeling horizon [5, 45]. Beyond year 5 of the modeling horizon, VE-PCV was assumed to wane, as follows: 5% annual decline during years 6–10, 10% annual decline during years 11–15, and no efficacy from year 16 through the end of the modeling horizon [42].

Initial VE of PPSV23 against VT-IPD for low-, moderate-, and high-risk persons aged ≥ 18 years was estimated from published literature [46, 47]. VE-PPSV23 against VT-CAP was assumed to be zero based on various published sources and consistent with base case assumptions employed in several economic studies [6,7,8,9,10,11,12,13,14,15,16,17,18,19, 48,49,50,51,52]. Beyond year 1 of the modeling horizon, VE-PPSV23 against VT-IPD was assumed to wane, as follows: linear decline to 76.2% of initial VE by year 5, and linear decline to no efficacy by year 10 and through the end of the modeling horizon [46].

Serotype Coverage

The percentage of IPD and NBP due to vaccine serotypes in year 1 of the modeling horizon was based on age-specific data from the Surveillance Network System of Agents Responsible for Bacterial Pneumonia and Meningitis (SIREVA II) (Supplemental Tables S1 and S2) [53]. Serotype coverage for IPD and NBP was assumed to be invariant across risk groups.

As childhood vaccination programs for PCV15 and PCV20 are expected to be implemented in model years 2 and 3, respectively, herd effects for serotypes unique to PCV15 and PCV20 were assumed to begin in model years 3 and 4, respectively. We assumed no serotype replacement in the model and consequently re-based the proportion of disease due to vaccine serotypes annually (model years 3–10) to account for the reduction in overall disease.

Vaccine Coverage

Vaccine coverage estimates were based on internal Pfizer data and varied by age (18–64 years, 17%; ≥ 65 years, 65%) [54]. For persons receiving sequential strategies, all persons who received a PCV in year 1 were assumed to receive PPSV23 in year 2 (if alive).

Costs

Costs of medical care for persons aged 18–64 years and for those aged ≥ 65 years were based on a micro-costing study conducted by Soul Consulting (data on file) [54]. For both IPD and outpatient NBP, age-specific costs were apportioned across risk groups using relative costs of disease (i.e., for moderate- and high-risk vs. low risk)—for IPD and outpatient NBP, respectively—from Weycker et al. [37] and risk distribution among cases. Medical care costs for inpatient NBP among those in low- and high-risk groups were based on results from the study by Soul Consulting [54]. The average of medical care costs for low- and high-risk groups was used as the estimated medical care costs for moderate-risk persons with inpatient NBP.

Since PPSV23 is procured through PAHO’s revolving fund, the list price was used as the vaccination price [55]. The price of PCVs were markedly higher than the price of PPSV23 and were based on confidential Pfizer pricing. Vaccine administration cost varied across 18–64- and 65–99-year-old age groups and was assumed to be $6.45 and $7.58, respectively, based on an internal analysis conducted by Soul Consulting [54].

Utility and Disutilities

Age-specific health state utilities for the general population in Argentina were based on Szende et al. [56] where a self-report of population health was converted to utility values using a visual analogue scale value set specific to Argentina. Estimates were apportioned across risk groups using relative utilities (moderate and high risk vs. low risk) [57] and underlying population risk distribution [28, 29].

Disease-related disutilities were based on published literature. Disutility due to IPD (bacteremia and meningitis) and inpatient NBP was assumed to be 0.13 in the year of occurrence based on Mangen et al. [58]. For NBP requiring outpatient care only, disutility was assumed to be 0.004 based on were based on Melegaro et al. [59]. Disutility was assumed to be invariant by age or risk.

Analyses

Clinical outcomes and economic costs over the remaining years of life for moderate-risk and high-risk adults aged 18–64 years and all adults aged 65–99 years at model entry were evaluated for PCV20 versus PCV13 → PPSV23 and, alternatively, considering the use of PCV20 versus PCV15 → PPSV23. Analyses were conducted from the healthcare system perspective; future costs, life-years (LYs), and quality-adjusted LYs (QALYs) were discounted at 3% annually [60]. Incremental cost-effectiveness ratios (ICERs) were calculated based on differences in costs and QALYs for each comparison. All costs are expressed in January 2023 US dollars (1 USD = 183.25 ARS $).

Probabilistic sensitivity analyses (PSAs) were conducted (1000 replications) to account for uncertainty surrounding key model parameters. For each input parameter, the appropriate distribution was determined based on information available in the source material (Supplemental Table S3). Certain parameters such as population risk distribution, vaccine coverage and vaccine price were not varied in the PSA either because their impact on results was believed to be minimal, or due to limitations of the model structure that prevented varying these estimates. Lastly, price sensitivity analysis was conducted where PCV20 vaccine price was varied across a range of values and the corresponding ICERs were estimated.

Ethics Compliance

This study was conducted using publicly available data and previously conducted studies. It does not contain any new studies with human participants or animals performed by any of the authors, and hence was not subject to institutional review.