1.1. Resurgence of cholera

Multiple, converging factors have led to a resurgence of cholera: extreme weather events caused by climate change, humanitarian crises (e.g., forced displacement, armed conflicts, earthquakes), poorly planned urbanization, and inadequate access to basic infrastructure and services [1–3]. The situation is further exacerbated by chronic underfunding, limiting progress on cholera control efforts [4]. The number of countries reporting cholera cases has grown over time, from 35 in 2021 to 45 in 2023 [5]. From 2022 to 2023, there was a 71% increase in deaths and 13% increase in cases, largely in Africa, pointing to inadequate access to both prevention and treatment. Inadequate surveillance, non-specific clinical presentations, and reluctance by authorities to acknowledge cholera means the real burden is likely much higher [6]. In 2015, an estimated 1.3 billion people were at risk of cholera in endemic countries [7]. Like for other neglected diseases, the greatest burden is in countries with the weakest health systems and among the most vulnerable populations [8]. Socioeconomic losses associated with cholera can be severe, making vaccines especially cost-effective in regions with high incidence [9].

1.2. Role of oral cholera vaccines in global disease control

Oral cholera vaccines (OCV) play an important role in multisectoral interventions for the management of cholera. They target the O1 serogroup of the V. cholerae bacteria, which causes the majority of cases globally, helping protect populations by reducing local transmission and contributing to herd immunity [10,11]. OCV use is recommended across two settings [12]: i) reactive use for outbreak response and humanitarian crises, and ii) preventive use for priority areas in endemic countries. A global OCV stockpile was established in 2013, with priority given to reactive use. The stockpile is continuously replenished to maintain a pre-defined target emergency reserve (S^*). Available doses beyond the S^* can be used to meet demand in preventive settings. When supply is scarce, reliance on short-term, emergency vaccination delays long-term preventive efforts.

Vaccination with a safe and effective vaccine, although not a replacement for large-scale infrastructure investments in water, sanitation, and hygiene (WASH), bridges short-term outbreak response and long-term disease prevention. Therefore, the use of OCV helps achieve objectives outlined in a global strategy on cholera launched in 2017 by the Global Task Force on Cholera Control (GTFCC) [13]: 1) 90% reduction in outbreak deaths (relative to a 2017 baseline), and 2) elimination in up to 20 endemic countries by 2030. To achieve these objectives, all endemic countries are expected to identify Priority Areas for Multisectoral Interventions (PAMIs) and implement national cholera control plans (NCPs). Since 2023, Gavi, the Vaccine Alliance (GAVI) provides funding and technical assistance to implement multi-year phased preventive vaccination programs (pOCV), while continuing financial support for emergency deployment of doses from the global stockpile managed by the International Coordinating Group on Vaccine Provision (ICG) [14].

1.3. Key challenges

Achieving and sustaining a healthy market is not trivial, especially for vaccines with uncertain demand. The ability for OCV supply to meet demand has been increasingly difficult in recent years, with OCV market health categorized as ‘unacceptable’ based on GAVI’s Healthy Markets Framework [15]. This can partly be explained by the following demand dynamics: i) growing evidence on the cost-effectiveness of OCV and feasibility of implementing large campaigns [16], ii) unanticipated surge in outbreaks leading to a sustained spike in requests for OCV use in reactive settings [17], and iii) growing interest in pOCV following expansion of GAVI funding. Additionally, numerous factors contribute to supply scarcity. The time needed for technology transfer, process scale-up, batch production, and lot release is long, leading to delays in closing the supply gap. Furthermore, since a majority of doses are delivered to reactive settings through UNICEF public procurement, low predictability of demand combined with thin profit margins have led to limited attractiveness for suppliers to enter and stay in the market [18]. As a result, there is an insufficient number and geographic diversity of suppliers (Fig A in S1 Appendix) despite sustained product innovation over the past few decades [19].

Current market dynamics accentuate challenges faced by decision-makers across the OCV value chain. The surge in outbreaks means doses are prioritized for reactive settings, with erratic orders and little visibility for suppliers to plan production in the long-term. Due to recurring shortages [20], preventive vaccination has been suspended since 2022, while WHO has recommended 1-dose reactive vaccination rather than the typical two-dose regimen [21]. Although dose sparing, single-dose vaccination has shown to confer reduced protective effect [22]. Furthermore, the current focus on emergency response makes it difficult for funders to anticipate future programmatic needs and for countries to implement NCPs, especially when faced with competing health priorities and limited resources. Recognizing these challenges, GAVI has published a roadmap to align mid- and long-term market-shaping objectives across partners and define necessary interventions to inform procurement strategies and decisions. Specifically, large demand is expected once pOCV resumes, estimated between 85–220 million doses annually on average over 10 years [15]. However, it is unclear how quickly new suppliers will enter the market and what impact procurement decisions will have on future market dynamics. In the meantime, the full programmatic potential of OCV remains out of reach, delaying cholera control efforts.

1.4. Research aims & contributions

There has been extensive modeling on the epidemiology of cholera, prediction of outbreaks, and impact of control measures, including OCV [23]. Models have been proposed to compare the impact of a one-dose versus two-dose OCV vaccine regimen on disease transmission during an outbreak [24,25]. Given the importance of high vaccine coverage, one study explores the effect of population mobility and waning vaccine efficacy on vaccine-derived herd immunity [26]. Others look at the cost-effectiveness of different vaccination strategies, considering target age groups, geographic areas, and frequency of re-vaccination [27–29]. Most often, models focus on one (or few) geographic areas and a single outbreak event. To inform global OCV demand, some studies forecast the population at-risk of cholera across individual countries and regions [2,7,30]. No studies integrate OCV supply and demand at a global level over long periods.

This paper has several aims. First, compare the impact of different OCV supply and demand assumptions, including programmatic design, on market dynamics. Although this study focuses on OCV, it also considers synergistic actions (e.g., WASH investments) aimed at reducing the size of populations at risk of cholera. This is particularly relevant given the role multisectoral interventions have in modulating long-term OCV demand. Second, test the impact of market shaping strategies, defined by policies and interventions, on OCV market health. Here, market health is defined as the ability to align supply and demand both in quantity and time. This is critical to continuously meet programmatic needs and accelerate cholera control. Therefore, the study considers select key performance indicators (KPIs) such as doses requested and produced over time. Third, present a tool that can be used by practitioners to design, test, and compare scenarios, as well as researchers to determine avenues for future research.

These aims are accomplished by building a quantitative system dynamics (SD) model to simulate OCV market dynamics between 2013 and 2035. The model captures key interdependencies (e.g., feedback dynamics) between OCV supply and demand to better anticipate the individual and combined impact of interventions. It also considers the capacities and incentives of stakeholders in the OCV market. Additionally, a user interface allowed for rapid and interactive exploration, validation, and testing of scenarios together with stakeholders involved in market shaping activities. It displays key model and data assumptions, market shaping strategies, and KPIs of interest to stakeholders. By enhancing visibility of OCV supply and demand, the model promotes shared understanding of tradeoffs across strategies aimed at improving market health.

2.1. Study design & setting

SD modeling is particularly well-suited to provide strategic-level insights into complex problems characterized by non-linear behavior. It succeeds in capturing interactions between physical and institutional structures of a system, including relevant decision-making processes [31]. Since SD models capture the underlying structural drivers of observed behavior, they can effectively anticipate the long-term and unintended consequences of decisions, especially in complex settings [32]. The ability for SD to capture evolving feedback dynamics, often with long and variable delays, despite limited data made it more suitable than alternative modeling approaches such as econometric, agent-based, discrete-event, and optimization methods. Rather than testing interventions at a single point in time, SD models endogenously capture how policies alter system structures over time, often leading to unintended consequences and path-dependent outcomes. The complex interactions between multiple, concurrent interventions over time provide insight into system-level policy effects. The mechanisms inherent to SD models make this approach uniquely positioned to analyze future market dynamics and the dominant challenges addressed in this work, including understanding the tradeoffs between market shaping strategies under uncertainty. A participatory approach, through extensive stakeholder engagement, was used to design, develop, and validate the model. Model simulations are used to test the impact of deviations from baseline model assumptions and parameter values, based on user-defined scenarios.

The model captures historic (2013–2024) and future (2025–2035) dynamics of OCV supply and demand, providing visibility into both short- and medium-term market dynamics. This study considers the role of OCV, as well as complementary multisectoral interventions, in accelerating objectives outlined in GTFCC’s strategy towards global cholera control. Additionally, it aligns with the need for healthy markets outlined in the Immunization Agenda 2030 [33]. Specifically, the model scope relates to strategic priority 5 (outbreaks & emergencies) and 6 (supply & sustainability). It also aligns with the period and objectives defined in GAVI’s Market Shaping Roadmap for OCV published in 2023 [15]. Finally, it responds to a recommendation by the Strategic Advisory Group of Experts (SAGE) on Immunization to model OCV supply scenarios to inform timing of pOCV [34].

Global OCV demand comes predominantly from cholera-endemic countries facing outbreaks and/or humanitarian crises. Most are low- and middle-income countries (LMICs), with vaccine procurement facilitated by UNICEF and funded by GAVI. Demand in reactive settings translates to country requests to use available doses from the revolving global OCV stockpile. These requests are reviewed and decided on by the ICG. In preventive settings, GTFCC’s Country Support Platform (CSP) provides technical assistance for countries to apply for funding to support pOCV. These applications are reviewed by GAVI’s Independent Review Committee (IRC). Once approved, decision letters (DLs) outline expected doses to be delivered to countries. UNICEF plays a critical role in operationalizing these DLs by procuring vaccines from qualified suppliers via purchase orders. GTFCC provides an allocation framework for fair distribution of doses destined for preventive use.

2.2. Data collection & stakeholder engagement

Data was collected from academic and grey literature, including public data sets. Additionally, semi-structured interviews with key stakeholders were conducted in English, lasting 1–2 hours, to inform this research. Interviewees were required to be actively involved in OCV supply and/or demand related activities, including research and funding, or engaged in processes that inform vaccine market shaping strategies. Through purposive and snowball sampling, interviewees were contacted between September 2024 and February 2025. In total, 36 interviews were conducted. Of these, 16 took place during early stages of model development, while 20 focused on model validation and to solicit feedback on the user interface. All had 10 + years of relevant work experience. Interview recordings and transcripts were securely stored and deleted post-transcription to ensure data privacy and confidentiality.

Stakeholder engagement throughout the study was critical to ensure model relevance and validity. Stakeholders contributed to the following: 1) understanding challenges related to OCV supply and demand; 2) identifying root causes of historic OCV shortages, including barriers and enablers to achieving healthy vaccine markets; 3) defining model boundaries for the research aims and ensuring inclusion of the most dominant feedback dynamics expected to drive future behavior; 4) clarifying publicly available data, estimating parameters, and validating model assumptions, structures, and behavior; 5) defining priority interventions and policies for scenario design, as well as KPIs to measure their relative performance; and 6) providing feedback on the model user interface. Wide representation of organization types (Table A in S1 Appendix) was important to learn from diverse perspectives and reduce bias.

2.3. Model structure & analysis

SD models are comprised of four interlinked building blocks: stocks, flows, auxiliary variables, and parameters/constants [35]. These building blocks are often organized into subsystems, representing different parts of a larger system. Connections are made between building blocks within and across subsystems, creating feedback loops that give rise to dynamics behavior. Arrays are used to efficiently replicate model structures across multiple elements within a dimension (e.g., product) and reduce visual complexity of the resulting model. During the simulation, at each time step (e.g., 1 day in this study), coupled differential equations are solved to approximate model values. Stella Architect (version 3.3, Isee Systems) was used to build the model and run simulations. Model output was processed to measure KPIs at each time step, annually, or at the end of the simulation.

Our SD model is comprised of several subsystems, each with a well-defined scope and relevant stakeholders (Table 1). The model reflects the most important dynamics in the OCV market, as found in literature and elicited through extensive stakeholder discussions. The model considers reactive demand aggregated at a global level, while preventive demand is estimated for 47 endemic countries based on the time-dependent population at risk of cholera. Preventive demand also depends on the rate of pOCV adoption across countries and GAVI IRC approval. The model also captures the evolving supply landscape over time, defined as the number of suppliers and their corresponding production capacities. Once country requests are approved by the ICG or GAVI IRC, the decision to submit purchase orders to suppliers depends on policies that guide vaccine use across reactive and preventive settings. These policies are based on the available or future expected global OCV inventory. As a result, market dynamics that emerge provide insights on the extent to which supply aligns with demand. Additionally, three subsystems (labeled as ‘exogenous’) only include static parameters that inform dynamics captured in the model. A full description of model structures (stocks, flows, parameters), assumptions, and initial conditions can be found in S1 Appendix.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. High-level overview of model subsystems.

https://doi.org/10.1371/journal.pntd.0013995.t001

2.4. Structural & behavioral validity of model

Model validation is essential to build confidence in simulation results. This study’s focus on informing OCV market shaping strategies means validation centered around the directionality and relative magnitude of change under different settings, rather than specific predictions or optimization. Since the goal of SD models is to explain behavior through causal structures, validation standards are typically more rigorous than many other methodologies [42]. Unit consistency of the model was checked. When published data was limited, parameter values and ranges, relationships, and behavior were validated by stakeholders with relevant domain knowledge. Since future OCV market dynamics are expected to be significantly different from past behavior, it was particularly important to validate model behavior under likely supply and demand scenarios. Additionally, the large-scale preventive use of OCV is new for many endemic countries. Therefore, learnings were taken from vaccine markets similar to OCV, including those for yellow fever and meningococcal vaccines. Model improvements were made iteratively following validation sessions with stakeholders.

Multiple structure and behavior tests were conducted, starting with each subsystem individually and repeated for the full model [43,44]. This involved sensitivity analysis of key parameters, testing extreme conditions to capture dynamics not seen in the baseline scenario, and comparing simulated and observed behavior. Behavior reproduction was done for selected KPIs such as doses requested and produced, including testing the individual and combined effect of feedback loops. Additionally, weekly reported data on the global OCV stockpile level was compared to simulation results between 2021 and 2024. Boundary adequacy tests, largely through stakeholder interviews, helped ensure that the level of aggregation is appropriate for the research aims defined and that dynamics most responsible for driving behavior are captured in model structures. Taken together, these validation tests ensure the model matches observed behavior for the right reasons, increasing interpretability of results.

2.5. Interactive learning environment

In this study, a computer-based user interface was developed to provide an opportunity for key stakeholders to define, test, and compare model scenarios. The interactive learning environment (ILE) was developed with Stella Architect, allowing easy interaction with and adaptation of variables in the model. First, stakeholder interviews helped define user needs, requirements, and preferences. Second, a mockup of the ILE was designed using sketches and wireframes to ensure a clear user experience. Third, various prototypes were made with increasing functionality and unique user flows. Input (e.g., knobs, sliders, switches) and output (e.g., graphs, numeric display) options were explored. Fourth, user testing was done with stakeholders to solicit feedback on the ILE, especially around variables to be defined by users and KPIs to be displayed. Fifth, the ILE was further refined. Finally, a user guide, including a test case, was developed to facilitate independent use. Considering evolving OCV dynamics, the ILE was built in a modular way to rapidly adapt to new user requirements and realities.

3.1. Supply and demand assumptions

Three possible future supply settings are defined (Table 2). For each scenario, only one supply setting is selected. We assume each supplier only produces one OCV product. In the baseline supply setting (BSS), of the five OCV products with WHO PQ (Dukoral, Shanchol, Euvichol, Euvichol-Plus, Euvichol-S), we assume only Euvichol-S is produced as of 2025. The incumbent supplier (EuBiologics, South Korea) stays in the market throughout the entire simulation period (2025–2035), with a maximum capacity of 75 million doses per year as of 2025 [45]. For the two other supply settings, the model is supplier agnostic, aiming to capture diverse supplier profiles.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Definition of potential future supply settings.

https://doi.org/10.1371/journal.pntd.0013995.t002

In the optimistic supply setting (OSS), three new suppliers enter the market by 2028. To account for gradual capacity phase-in, the maximum production capacity (measured in OCV doses) for product 1, 2, and 3 is reached by 2028, 2030, and 2031, respectively. We model product 1 supplier as the most risk-averse and likely to exit the market, while product 3 supplier is least likely. Market attractiveness is determined based on annual review of the OCV market size and continuous review of excess inventory across all suppliers. In the pessimistic supply setting (PSS), three new suppliers enter the market by 2030, reaching their maximum capacity after three years. Although long-term production capacity is the same for OSS and PSS, they differ in their ability to rapidly remedy supply gaps, especially in anticipation of pOCV adoption. Assumptions on the number of new suppliers and their relative settings is based on publicly reported OCV development and technology transfer efforts. Capacity phase-in reflects historically observed behavior. Since OCV is a high-priority vaccine for WHO PQ [36], the majority of the reported scenarios consider the OSS setting.

The first four scenarios focus on different OCV supply and demand assumptions (Table 3). The ‘baseline’ scenario (scenario 1) assumes minimal intervention and a BSS. Vaccination is only considered in reactive settings, with an annual increase in OCV demand driven entirely by risk factors (e.g., climate) for cholera outbreaks and humanitarian emergencies. We assume the effect of climate and conflict to be the same in all scenarios. Scenarios 1–3 consider slow market entry of suppliers (PSS). The ‘fixes that fail’ scenario (scenario 2) assumes one-time pOCV with no re-vaccination, in the absence of multisectoral interventions. Adoption of pOCV in endemic countries is considered slow: 1–3 countries apply for funding each year between 2024–2033. For each scenario, the exact timing is based on a probability distribution. In the ‘multisectoral approach’ scenario (scenario 3), slow pOCV adoption is simulated with re-vaccination every 3 years. Also, WASH investments lead to a 50% reduction in the at-risk population between 2028 and 2035. These expected delays reflect challenges around capital-intensive investment requirements and need for regular maintenance. [46]. Scenario 3 considers the impact of health system strengthening and outbreak preparedness. Comparatively, the ‘accelerating investments’ scenario (scenario 4) assumes greater reduction (80%) of the at-risk population due to improved WASH and a more ambitious timeline for suppliers (OSS). It is defined as the ‘preferred’ demand scenario to accelerate progress towards GTFCC objectives.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Key parameter assumptions for OCV supply and demand in scenarios 1-9.

https://doi.org/10.1371/journal.pntd.0013995.t003

Building on assumptions in the preferred demand scenario, scenarios 5–9 further explore the extent to which two specific strategies, i.e., adapting the rate of pOCV adoption and vaccine roll-out (phases), influence market dynamics. The number of phases in a given preventive OCV program determines the size of each campaign until the target population is fully vaccinated. Scenarios 5 and 6 test the impact of shifting to a 2-phase or 4-phase program, respectively, compared to the 3-phase assumption in scenarios 1–4. Scenarios 7–9, investigate the impact of fast pOCV adoption: 3–5 countries apply for funding each year between 2024–2030. Several assumptions are made across all scenarios. In line with current practice, there are 12 months between each pOCV phase. Additionally, only GAVI-eligible countries classified as high- and medium-likelihood adopters, based on GAVI’s strategic demand scenarios for OCV, are considered for preventive vaccination. Starting in 2025, we assume pOCV resumes if the available inventory across all suppliers is at least 15 million doses. No constraint is placed on the number of shipments required to fulfill an order. The current dose sparing policy (1-dose vaccination) remains in effect in reactive settings, while S^* is maintained at 5 million doses.

3.2. Market shaping strategies under budget constraints

In previous scenarios, costs associated with vaccine procurement and implementation support are calculated, with no budget constraints. However, economic challenges and austerity measures across countries have led many GAVI donors to rethink funding commitments [47,48]. Therefore, several scenarios are defined to capture the impact of key feedback mechanisms and strategies on market health (Table 4). Scenarios 10–20 have the same supply, demand, and programmatic settings as scenario 4. However, they assume a total available programmatic budget of USD 2.75 billion between 2013 and 2035, compared to no budget constraint in scenarios 1–9. As a result, approved DLs cannot be fulfilled once the budget threshold is passed. Given this budget constraint, the model considers the individual effect of 1) adopting rapid diagnostics tests (RDT) (scenario 11), 2) price competition in procurement (PCP) as new suppliers enter the market (scenario 12), and 3) ordering behavior (OB) of countries (scenario 13) on the number of doses requested. Scenario 14 looks at the combined effects of all three.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Definition of interventions under budget constraints, including the use of rapid diagnostic tests, price competition during procurement, and ordering behavior.

https://doi.org/10.1371/journal.pntd.0013995.t004

Since 2023, GAVI provides support to improve diagnostic testing capacity for cholera [49]. Scenarios with RDT assume 16 countries have adopted rapid diagnostic testing by 2025, reducing pOCV demand by 25% in each. Price competition emerges as new suppliers enter the market, leading to a lower average awarded price per dose during UNICEF procurement, assumed to be USD 1.35 per dose as of 2025. We assume each new supplier leads to a 10% decrease in the awarded price, with a minimum of USD 1.00 per dose. Finally, historical data shows an effect of supply levels on ordering behavior [50]. The model considers this effect separately for reactive and preventive settings. In reactive settings, when the inventory level is below S^*, the number of ICG-approved doses is on average 35% lower than the volume requested without supply constraints. In preventive settings, the model assumes that the accumulation of pending DLs due to insufficient supply leads countries to further prioritize target populations for pOCV. They are unlikely to entirely cancel requests given investments made to develop NCPs and identify PAMIs.

As the supply situation improves, a recurring question highlighted by stakeholders is how to design policies to resume both pOCV and 2-dose reactive vaccination (BOOST), as well as anticipate tradeoffs between decisions. Therefore, scenarios 15–20 test 6 inventory policies, while accounting for the combined effect of budget constraints, RDT, PCP, and OB (as in scenario 14). Three decision criteria are considered for resuming pOCV: available inventory in stock, batches to be released within ~6 weeks (DP WIP), and batches to be released within ~10 weeks (total WIP). The WIP includes available inventory in stock. For BOOST, the inventory level in stock is the only criterion used. A policy is defined by combining one decision criterion with a minimum threshold to be met or exceed.

4.1. OCV market dynamics

Key dynamics and processes observed in the OCV market are captured in a high-level conceptual framework shown in Fig 1. It captures interactions between model subsystems defined in Table 1, including demand in both reactive and preventive settings, as well as OCV supply and order fulfillment. One example is the link between order fulfillment and market attractiveness. Suppliers are likely to stay in the market when the rate of order fulfillment is high, unless future expected demand is low or the market is saturated. This complex behavior is captured through multiple, endogenous relationships that evolve over time. Moreover, the model allows decision-makers to identify the root cause of dynamics, for example if a supplier leaves the market due to insufficient demand or excess inventory. The impact of supply levels on global OCV policies is another important dynamic captured. The decision to resume pOCV or BOOST has direct impact on doses requested and delivered. It also influences predictability and materialization of future orders and overall market size. A shift towards preventive vaccination attenuates demand in reactive settings and allows suppliers to better anticipate capacity requirements for pOCV. Some relationships are captured exogenously, including the impact of climate and conflict on OCV demand and that of complementary multisectoral interventions such as WASH and health system strengthening.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Conceptual stock-and-flow model of key subsystems, interactions between them, and feedback dynamics in the model.

Variables from exogenous subsystems are shown in red. Stocks (boxes) capture quantities (e.g., doses) in the system over time, based on the rate of inflow and outflows. Interactions between different system elements are captured by arrows with polarities that refer to a direct or inverse relationship. Non-linear behavior emerges from delayed effects between variables, as well as feedback loops – when an initial change in a variable causes a chain reaction that affects the starting variable.

https://doi.org/10.1371/journal.pntd.0013995.g001

Several KPIs (Table B in S1 Appendix) are defined to analyze simulated (quantified) system behavior over time, under different supply and demand settings. KPIs focus on OCV market health and the ability for vaccines to accelerate cholera control efforts (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Overview of key performance indicators across model subsystems and their relationship with overall objectives defined by the Global Taskforce on Cholera Control (GTFCC).

https://doi.org/10.1371/journal.pntd.0013995.g002

4.2. Model validation

Stakeholders expressed that the model conceptually and structurally captures the main issues currently seen in the OCV market. Furthermore, the model replicates historical UNICEF-reported inventory levels of the global stockpile (Fig B in S1 Appendix), which emerges from complex supply and demand dynamics. Statistical measures such as the root mean squared error, R², and Theil inequality coefficient are reported in S1 Appendix. As the OCV market evolves, model adaptations can be made to reflect learnings.

4.3. Scenario analysis: tradeoffs across OCV supply and demand assumptions

Between 2013 and 2024, 158 million and 47.3 million doses were delivered to countries for reactive and preventive use, respectively [37]. Fig 3 presents OCV supply and demand, as well as the number of qualified OCV suppliers in the market, for scenarios 2–4. In the baseline (scenario 1), OCV demand is solely driven by outbreaks and humanitarian crises, leading to a total of 881 million doses requested for use in reactive settings between 2013 and 2035. Annual production of 75 million doses in the BSS is sufficient to fulfill demand, but the limited predictability of demand means suppliers continue producing at risk. Stakeholders point to the lack of visibility on future demand as the biggest failure of the OCV market. Continued dependence on outbreak response will likely translate to recurring outbreaks and associated socioeconomic disruptions. Re-vaccination combined with multisectoral interventions leads to a significant reduction in the cumulative number of doses requested in reactive settings relative to one-time pOCV. Specifically, a 27% and 30% reduction is seen in scenario 3 and 4, respectively, relative to scenario 2. However, the cumulative number of doses requested in preventive settings is significantly higher with re-vaccination. These dynamics present important tradeoffs. First, demand predictability increases when pOCV is considered, with at least 50% of annual doses delivered reaching preventive settings by 2030 in scenario 3 and 2028 in scenario 4 (Fig C in S1 Appendix). Second, to effectively reduce the supply gap throughout the simulation period, new suppliers need to enter the market earlier and scale-up capacity faster. For scenario 4, reaching an annual supply capacity of ~200 million doses by 2031 helps close supply gaps in 2032. However, as supply accumulates and the annual OCV market size declines (Fig D in S1 Appendix), some suppliers will leave the market prior to 2035 due to reduced market attractiveness. Third, although scenario 4 is preferred to maximize timely availability and programmatic use of OCV, it comes at a higher cost (Fig E in S1 Appendix). The estimated ~USD 3.3 billion needed for OCV procurement and implementation support throughout the simulation period is 19% and 6% higher than expected costs for scenario 2 and 3, respectively. Since public procurement with GAVI funding depends on donors with competing priorities over time, understanding the distribution of costs across GAVI’s 5-year funding cycles is important. For scenario 2 and 3, the highest cost period is between 2031–2035 compared to 2026–2030 for scenario 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Key performance indicators for scenario 2 (grey), 3 (purple), and 4 (blue).

A-C) Gap between OCV supply and demand over time; D-F) Cumulative OCV supply (solid line) and demand (dotted line); G) Cumulative doses requested in reactive settings; H) Cumulative doses requested in preventive settings; I) Number of OCV suppliers with WHO pre-qualification supplying to the global stockpile through UNICEF public procurement. Note: The range for X-axes is the same for all figures (2013-2035), while the range for Y-axes depends on the KPIs displayed.

https://doi.org/10.1371/journal.pntd.0013995.g003

4.3.1. Implications of programmatic design.

Multi-year preventive vaccination programs, especially for endemic countries with large target populations, lead to spikes in demand. Therefore, programmatic design can play an important role in smoothing demand, especially when switching from 2 to 4 phases (Fig 4). In a supply-scarce environment, with a long wait time until new suppliers enter the market, extending pOCV over more years reduces delays between country requests and order fulfillment by suppliers. Better aligning demand with future expected supply also leads to improved management of the stockpile, as seen by longer periods when inventory is above S^* (Table C in S1 Appendix). However, unintended consequences can emerge from prolonged pOCV, as seen in scenario 6 and 9. When pOCV is stretched over too many years, countries are less likely to initiate second and third round vaccination in PAMIs. In all scenarios, 24 countries are expected to initiate and complete at least one pOCV by 2035. Assuming a 4-phase pOCV with slow country adoption, 17 countries initiate second round preventive vaccination compared to 24 countries in scenarios with fast adoption. For both slow and fast adoption, 50% less countries are able to initiate third round preventive vaccination compared to a 3-phase pOCV. Stretching pOCV over more years also means supply increases faster than demand, while the annual and overall market size is smaller. As a result, suppliers may leave the market if attractiveness falls below a threshold. In scenario 6, these dynamics are expected to renew delays in fulfilling OCV demand in preventive settings starting in 2033. Regardless of the number of phases in a pOCV, more time is needed to close the supply gap when country adoption is accelerated.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Key performance indicators for scenarios 5-9, relative to scenario 4 (blue).

Approved OCV requests waiting to be delivered to countries are shown under slow (A) and fast (B) adoption of prevention vaccination in endemic countries. Also, differences can be seen in the number of countries initiating a second (C) and third (D) multi-year phased preventive vaccination program.

https://doi.org/10.1371/journal.pntd.0013995.g004

4.4. Impact of market shaping strategies

Although scenario 4 is defined as the ‘preferred’ demand scenario to accelerate cholera control efforts, the behavior of OCV supply and demand depend on multiple dynamics not previously accounted for. These have an impact on the volume and timing of OCV requests, influencing market dynamics and thus the ability to deliver sufficient doses in a timely manner.

4.4.1. Feedback dynamics under budget constraints.

The preferred demand scenario, with accelerated market entry of suppliers and ambitious multisectoral interventions, has a programmatic cost of ~USD 3.3 billion. As seen in scenario 10, reducing available funds through 2034 by ~15% to USD 2.75 billion could translate to a 20% gap in doses delivered to preventive settings (Table 5). Since emergency requests for OCV are prioritized and budget constraints are only felt late (~2032) in the simulation when outbreaks have significantly declined, a 0.5% gap is expected in doses delivered to reactive settings. However, this can be concerning if new transmission chains spread due to the inability to rapidly deploy OCV during outbreaks. Several interventions could help remedy funding gaps to meet country needs around pOCV. Under budget constraints, RDT and PCP are effective strategies to reduce programmatic gaps. By targeting the most at-risk populations within PAMIs, deployment of RDT (scenario 11) leads to a 16% decrease in preventive demand and 47% reduction in unfilled DLs relative to scenario 10. In scenario 12, accounting for PCP leads to an even greater reduction in unfulfilled DLs, while delivering more doses overall (Fig F in S1 Appendix). In scenario 13, the effect of OB due to the accumulation of unfulfilled DLs leads to a smaller change than the individual effect of RDT or PCP. However, this is highly sensitive to the threshold for unfulfilled DLs that drives behavior, assumed to be 150 million doses. Combining all three interventions in scenario 14 provides the greatest alignment between supply and demand across the simulation period (Fig G in S1 Appendix).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Impact of different interventions to overcome budget constraints (BC) – including rapid diagnostic testing (RDT), price competition of procurement (PCP), and ordering behavior of countries (OB) – as well as effect of policies for resuming full programmatic use of OCV.

https://doi.org/10.1371/journal.pntd.0013995.t005

4.5. Bridging research and practice with an interactive tool

Development of a user interface had several benefits. The visual-based nature of the ILE (Fig 5) allowed validation of uncertainty in parameters and ranges. It also encouraged discussions on realistic and likely future scenarios, serving as a sounding board for interventions users might want to test. This led to refinement of both the model and interface. From a modeler’s perspective, it also helped to accelerate the validation of simulated behavior, including testing extreme conditions. Stakeholders were enthusiastic about the ILE and the ability to explore a broad range of scenarios, with a high degree of flexibility in defining parameters. Given the ILE’s potential to facilitate dialogue among stakeholders involved in market shaping activities, they recommended further pursuing its development.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Illustrative screenshots of the interactive user interface developed to facilitate learning through model simulation.

A) Landing page and menu options; B) Supply and demand settings; C) Interventions and policies; D) Dashboard of key performance indicators related to OCV market dynamics.

https://doi.org/10.1371/journal.pntd.0013995.g005