Mathematical Modelling of Typhoid Fever Transmission Dynamics and Intervention Impact in Harare, Zimbabwe (2018–2020)

In recent years, the West South West and Southern districts of Harare in Zimbabwe experienced a significant surge in the number of reported typhoid fever cases. The problem was that every year there were reported increasing cases and recurring cases that catch health sector un prepared, ending up failing to prevent the disease to full capacity. This unexpectedly increased cases not only pose a serious risk to the health and well-being of the population but also places considerable strain on local healthcare resources and infrastructure. Despite previous efforts to manage and control typhoid outbreaks through public health interventions, the recent uptick indicated potential gaps in the existing prevention and response strategies. This study aimed to: 1: To predict typhoid fever outbreaks by predictive trends in Harare City typhoid hot areas. 2: To develop a mathematical model to simulate the spread of typhoid fever incidence under different intervention scenarios and recommend evidence-based strategies for reducing the disease burden in Harare City. By utilizing epidemiological data, this research seeks to provide a comprehensive Mathematical Model to analyze the future surge in typhoid cases, with the goal of informing actionable strategies and interventions that can effectively curb the outbreak and enhance public health resilience in the districts.

Figure 2.Trend of typhoid cases 2011 to January 2020

Source: Author

Figure 2 shows fluctuations of cases. In 2018 and 2019 the graph is showing the highest cases to be recorded in a decade. From 2019 to 2020 there was a sharp decrease in cases , mainly due to 2019 targeted vaccination.

Typhoid fever remains a significant public health concern in many urban settings, particularly in low- and middle-income countries. Harare City, Zimbabwe, has experienced recurrent typhoid outbreaks, largely attributed to poor sanitation, contaminated water supplies, and socio-economic disparities. Mathematical modelling provides a robust framework for understanding the disease dynamics and evaluating the effectiveness of intervention strategies. In Harare City, Zimbabwe, an increase in cases from 2018 to 2020 highlighted the need for a deeper understanding of the driving factors behind these outbreaks (Pope, 2020).

Historical data indicate that Typhoid Fever is endemic in Zimbabwe, typically associated with periods of heavy rainfall and inadequate sanitation infrastructure (Chakuvinga P. e., 2017). The resurgence of cases in Harare during the specified period can be attributed to several factors, including urbanization, population density, and infrastructural challenges (Mavindu, 2018). A number of studies have identified key drivers responsible for the rise in Typhoid Fever cases. Poor water quality and sanitation are crucial factors in Typhoid transmission. Studies show that 80% of Typhoid cases are linked to contaminated water supplies (WHO, 2018). Harare's aging infrastructure has frequently resulted in sewer leaks and water contamination, as observed in outbreaks (Sokhela, 2019).

Environmental, Socio-Economic, and Demographic Drivers of Typhoid Fever Transmission: These factors determine the risk of exposure, the spread of infection, and the effectiveness of intervention measures. In Harare, rapid population growth, inadequate infrastructure, and climate variability exacerbate typhoid outbreaks. Understanding these drivers is critical for developing targeted interventions that reduce transmission and improve disease management.

Environmental Factors: Environmental conditions play a crucial role in the persistence and spread of Salmonella Typhi, the causative agent of typhoid fever. Poor water quality, inadequate sanitation, and seasonal variations contribute significantly to typhoid transmission. Studies have shown that contaminated water sources, including boreholes and open wells, are primary reservoirs of S. Typhi in endemic regions (Luby et al., 2019). Furthermore, seasonal rainfall has been linked to an increased incidence of waterborne diseases, as heavy rains facilitate the contamination of water supplies with fecal matter (WHO, 2018)

Socio-Economic Factors: Rapid urban population growth in Harare, combined with economic challenges, exacerbates the situation. Access to clean water and proper sanitation is often limited in high-density informal settlements (Davies, 2019). Research by Mhlanga et al. (2020) illustrates a direct correlation between socioeconomic status and incidence rates of Typhoid. Socio-economic disparities significantly influence the risk of typhoid infection. High-density suburbs in Harare, such as Mbare and Glen View, experience higher transmission rates due to overcrowding and inadequate waste disposal (MoHCC, 2020). Additionally, limited access to healthcare and delayed treatment-seeking behavior exacerbates disease burden, as undiagnosed cases continue to spread

the infection (Kariuki S, 2015). Informal food vending, common in urban markets, also plays a role in short-cycle transmission, as contaminated food and utensils facilitate the spread of S. Typhi (Dewan A, 2013).

Demographic Factors: Demographic characteristics such as age distribution, population mobility, and immunity levels influence typhoid transmission. Children and young adults are at a higher risk of infection due to increased exposure in schools and public spaces ¹¹. Migration patterns in and out of Harare also impact disease persistence, as infected individuals may introduce new cases into different communities (Kim, 2019). Moreover, natural immunity following typhoid infection declines over time, necessitating booster vaccinations or repeated interventions (Andrews, 2019).

Behavioural Factors: Public WASH practices significantly influence the transmission dynamics of Typhoid Fever. A study by Ndlovu et al. (2019) found that inadequate handwashing and poor food handling practices are prevalent in affected communities, contributing to the persistence of the disease (Ndlovu, 2019).

Predicting Future Trends in Typhoid Fever Incidence: Mathematical modelling has been instrumental in predicting typhoid fever outbreaks and evaluating potential control strategies. Compartmental models such as the Susceptible-Infected-Recovered (SIR) framework have been widely used to simulate disease dynamics under different intervention scenarios (Pitzer et al, 2015). Recent studies have incorporated seasonal variations, antimicrobial resistance, and vaccination coverage to enhance model accuracy (Antillon et al, 2017). In the context of Harare, predictive models can be used to assess the impact of environmental and behavioural changes on future typhoid incidence.

One study conducted in sub-Saharan Africa demonstrated that improved water quality and sanitation could reduce typhoid incidence by up to 60% over a decade (Park et al., 2018). Similarly, vaccination campaigns targeting high-risk populations have been shown to provide significant long-term benefits, especially in endemic regions (Shakya et al, 2019). By integrating real-world data into mathematical models, policymakers can make informed decisions about resource allocation and disease control priorities.

Water, Sanitation, and Hygiene (WASH) Interventions: WASH interventions, including improved sanitation, water treatment, and hand WASH, have been widely recognized as effective measures for typhoid control. A systematic review by Balfour et al. (2021) found that access to clean water significantly reduced the risk of typhoid infection, with households using treated water experiencing a 40% lower infection rate compared to those relying on untreated sources. Furthermore, behavioural interventions such as community-led total sanitation (CLTS) programs have successfully reduced open defecation, thereby decreasing environmental contamination (Cairncross et al, 2010).

Vaccination Campaigns: Typhoid conjugate vaccines (TCVs) offer a long-lasting protective effect and have been recommended by the World Health Organization (WHO) for routine immunization in endemic areas. A randomized controlled trial in Nepal demonstrated that a single-dose TCV provided over 80% efficacy against typhoid fever over three years (Shakya et al, Efficacy of typhoid conjugate vaccine: A randomized controlled trial., 2021). Mathematical modeling studies have shown that incorporating TCVs into routine immunization schedules could lead to a substantial decline in typhoid incidence, particularly when combined with WASH interventions (Antillon et al, 2017). Between 2016 and 2019, Harare witnessed multiple typhoid outbreaks, with significant morbidity and mortality. These recurring outbreaks and the evolving problem of multidrug resistance (AMR) highlighted the urgent need for sustainable interventions to control the disease and prevent future epidemics. In response, the introduction of the Typhoid Conjugate Vaccine (TCV) was prioritized as part of a comprehensive strategy to combat typhoid fever by the Ministry of Health and Child Care (MOHCC) in February, 2019

Antibiotic Stewardship and Drug Resistance Management: The emergence of multidrug-resistant (MDR) S. Typhi strains poses a significant challenge to typhoid control efforts. In recent years, extensively drug-resistant (XDR) typhoid outbreaks have been reported in parts of Africa and South Asia, highlighting the need for improved antibiotic stewardship (Klemm et al, 2018). Surveillance data from Zimbabwe indicate a growing prevalence of resistant strains, necessitating the development of targeted treatment protocols and the promotion of rational antibiotic use (MoHCC, 2021). There is a need to understand the burden of typhoid fever in Harare and the effectiveness of TCV in mitigating outbreaks. Despite various interventions, typhoid remains a persistent health threat in the city, particularly in areas where there is inadequate sanitation.

Evaluating the trends and the effectiveness of TCV will provide critical insights to policymakers for refining vaccination strategies, enhancing public health interventions, and strengthening disease surveillance systems. This retrospective review aims to describe the epidemiology of typhoid outbreaks in selected suburbs of Harare from 2016 to 2024 and evaluate the effectiveness of the TCV campaign in reducing the typhoid disease burden in Harare City. By analyzing secondary data from health facilities and public health records, this study seeks to provide insights into the effectiveness of vaccination campaigns and inform future public health strategies. The findings will contribute to the growing body of evidence supporting the role of TCV in controlling typhoid fever in resource-limited settings

Conclusion: Mathematical modelling serves as a powerful tool for understanding typhoid transmission dynamics and evaluating intervention strategies. In Harare City, the integration of environmental, socio-economic, and demographic data into predictive models can enhance outbreak preparedness and response efforts. Evidence-based interventions such as WASH improvements, vaccination campaigns, and antibiotic stewardship programs offer promising pathways for reducing the disease burden. Future research should focus on real-time data integration and the development of adaptive models to address emerging challenges in typhoid control.

The study employed secondary data analysis

We utilized data on microbiologically confirmed cases of Salmonella Typhi infection among patients seeking care at Beatrice Road Infectious Disease Hospital ( BRIDH) in Harare, Zimbabwe. This data was obtained from the City Health Departmental report of the cases line listed. The line list data was triangulated with data captured in the District Health Information System (DHIS2) for the same period. BRIDH is one of the two Infectious diseases hospitals in the City of Harare, providing free care for patients from Harare district and the surrounding towns. We had data on the aggregate number of S. Typhi isolates from January 2018 to January 2020 by patient age ( in 5 year age grouping). We also obtained demographic data on the population of Harare district (Harare Urban + Harare Rural) for the same period (Zimstats, 2022)

This refer to the proportion of individuals who become ill during an outbreak or specific time period, and it is calculated as:

Attack Rate = Number of new cases of disease in a population/Total population at risk during the specific time period* 100,000

This rate allows for easy comparisons between populations. Attack rates aids in identifying the severity of an outbreak and can be used to assess the effectiveness of interventions, such as vaccination campaigns

The study was conducted in Harare City. Harare City has a population of 1 846 437 (ZIMSTATS, 2022). During the time of the outbreak Harare City was made up of eight (8) districts. The outbreak was mainly confined to the West South West and Southern districts. In that regard this study will only focus on these two hotspot districts. The areas sampled for the study purposes are Mbare, Waterfall, Glen View, Mufakose and Budiriro suburbs. The total population for the sampled suburbs is 453 245 and a total of 2341 confirmed cases reported. The study was carried out using data collected electronically from linelists from the City’s Health Information Unit.

We developed and modified a dynamic compartmental SIR-based model to simulate typhoid transmission through both direct human-to-human contact and environmental pathways. The methodology involved the following approaches:

· Model Calibration: The model was calibrated using surveillance data from Harare (2018–2020) by adjusting transmission parameters to fit observed case trends.

· Model Validation: Validation was performed by comparing model outputs against independent data, ensuring the model accurately captured the disease dynamics.

· Model Fitting: The model was fitted using statistical techniques to minimize discrepancies between simulated and reported case counts, verifying accuracy through goodness-of-fit metrics.

· Scenario Analysis: Different intervention scenarios, including targeted vaccination in high-risk areas, were simulated to assess their impact on transmission reduction.

· Sensitivity Analysis: Key model parameters were varied systematically to determine their influence on disease dynamics and to assess model robustness.

· Case Prediction: The model projected future case trends under different intervention strategies, allowing for comparative assessment of potential outbreak control measures.

We adapted a previously developed mathematical model for the transmission dynamics of typhoid in Malawi ⁸ by fitting to the data from BRIDH and modifying the model to explore hypotheses for the increase in typhoid fever cases beginning in 2018. The model is illustrated in Figure 1 below. The model assumes that individuals are born susceptible to infection and disease caused by Salmonella Typhi (S1). The susceptible persons are infected at a rate (λp + λw) that is mediated through contaminated food and water in the immediate environment (“short-cycle transmission,” λp) as well as bacterial contamination of water supplies and the broader environment (“long-cycle transmission,” λw). Individuals experiencing primary infection (I1) remain infectious for a period 1/δ, after which we assume a fraction (α) experience disease-induced mortality, a fraction (θ) develop an infection of the gallbladder and become chronic carriers (C), and the remaining individuals (1-θ-α) recover and are temporarily immune (R). We assume that θ depends on the age of the infected person, consistent with epidemiological data ⁸. Immunity to reinfection wanes at a rate ω, leaving the individual partially susceptible (S2). If reinfection occurs prior to waning of partial immunity (at a rate ϵ), we assume these reinfections (I2) are subclinical and are not reflected in our data. Sub-clinically infected individuals can again recover or become chronic carriers. Additional dynamics include susceptible individuals that get vaccinated (V) at a rate τ. The infectious individuals shed bacteria into the environmental reservoir (W). We assume all infectious individuals shed bacteria into the environmental reservoir (W) at a rate γ, and the bacteria remain infectious for a period 1/ξ. Both chronic carriers and subclinically infected individuals have infectiousness reduced by a factor r. For the compartmental mathematical model for the transmission dynamics of typhoid, we assumed that short-cycle transmission was contact frequency dependent and mixing was homogeneous while long-cycle transmission depended on both contact frequency and population density. Frequency-dependent transmission was applied in settings where water sources are shared by a fixed number of individuals, such as households or small communities. In modeling the transmission dynamics of typhoid, we assume that short-cycle transmission (e.g., within households or small communities) follows a frequency-dependent pattern with homogeneous mixing. This means that infection risk is based on how often individuals come into contact with contaminated food, water, or fomites, rather than the total population size. Since these interactions occur in smaller, well-connected groups, every susceptible individual has a similar probability of exposure. This assumption aligns with real-world scenarios where water sources are shared by a fixed number of individuals, making contact frequency the key determinant of transmission.

In contrast, long-cycle transmission (e.g., through urban water supplies, rivers, or reservoirs) is modeled as density-dependent, meaning infection risk increases with population size. This reflects the reality that larger populations contribute more bacteria to shared water sources and have a greater likelihood of exposure. The choice between frequency-dependent and density-dependent transmission depends on the scale of bacterial contamination and epidemiological context. By incorporating both transmission pathways, the model provides a more accurate representation of typhoid spread, helping to design effective public health interventions such as household water treatment and large-scale sanitation improvements.

In contrast, density-dependent transmission was used to model scenarios where larger populations contribute to and are exposed to a common water reservoir, such as urban water supplies or rivers. The choice of transmission mode depended on the specific epidemiological context and scale of bacterial contamination. Finally, there are three model assumptions which are homogeneous mixing within each population, Constant recruitment rate (B) into the susceptible compartments, Natural mortality occurs at rate μ in all compartments (Figure 3).

Figure 3.Compartmental diagram of model structure. Source: (Adapted from Pitzer VE, 2014)

Table 1. Model Variables

C	Chronic disease (career) population.
W	environmental reservoir
X	susceptible individuals that develop cross-protective immunity
V	Represents the flow of individuals from the vaccinated (immune) compartment back to the susceptible class due to waning immunity. It models waning vaccine-induced immunity over time, whereby vaccinated individuals gradually become susceptible again.
Observed cases	Reported cases derived from infected compartments.

Table 2. Model Parameters

Parameter	Parameter	Estimate	Source	Estimation Method Used
B	Birth rate	27/1000	ZIMSTATS (Zimbabwe National Demographic Data)	Direct demographic data (empirical)
μ	Natural mortality rate	7.4/1000	ZIMSTATS (Zimbabwe National Demographic Data)	Direct demographic data (empirical)
1/δ	Rate of leaving the Infectious compartment	1/7 days = 0.1429	Estimated from clinical typhoid data	Empirical estimation from clinical recovery duration
α	Fraction of infected individuals who die from the disease	0.0029	Estimated from clinical typhoid data	Empirical estimation from clinical cas fatality data
θ	Fraction of infected individuals who become chronic carriers	0.02	Estimated from clinical typhoid data and literature	Literature- informed empirical estimate
1/ω	Rate of waning immunity	0.33	Estimated based on immune waning studies and literature	Bayesian estimation calibrated withliterature priors
ϵ	Rate of becoming fully susceptible again	0.1	Estimated from literature and localdata (WHO, 2023)	Literature- informed empiricalestimate
γ	Rate of bacterial shedding into the environment	1 infectiousunit per week	Estimated frommicrobiological studies of carriers	Literature-based empirical estimate
r	Reduction factor in infectiousness for chronic carriers and subclinical infections	0.3	Estimated from studies on chronic carriers’ infectivity	Literature-based empirical estimate
ξ	Rate of bacterial decay in theenvironment	1/3 per week	Pitzer et al. (2014)	Literature-based fixed estimate
βp	Short-cycle transmission coefficient	0.0001	Estimated via model fitting to case data	Maximum Likelihood Estimation (model calibration)
βw	Long-cycle transmission coefficient	0.0005	Estimated via model fitting; calibrated to observed trends	Maximum Likelihood Estimation (model calibration)
q	Seasonal variation amplitude	0.5	Estimated from seasonal case data	Model fitting using Maximum Likelihood Estimation
φ	Phase shift for seasonal transmission	0	Estimated during model fitting	Model fitting (MLE)
e	Education coverage	0.3	Estimated and assumed from intervention coverage data	Empirical assumption based on program data
h	WASH interventions	0.5	Estimated from programimplementation data	Empirical estimate from intervention reports
ρ	Rate of loss of vaccine immunity rate	0.33	Estimated based on vaccine efficacy studies	Bayesian estimation based on vaccine immunity decay literature

Parameter values were derived from a combination of empirical data, published literature, and model calibration procedures to ensure biological and epidemiological realism. Demographic parameters such as the birth rate (B) and natural mortality rate (μ) were obtained directly from ZIMSTATS national demographic data (2023). Clinical parameters including the infectious period (1/δ), disease fatality rate (α), and fraction progressing to chronic carriage (θ) were empirically estimated from hospital and surveillance data and cross-validated with published clinical studies of typhoid in sub-Saharan Africa (Table 1).

Parameters associated with immunity dynamics, such as the rate of waning immunity (ω), return to full susceptibility (ε), and loss of vaccine-induced immunity (p), were derived from immunological and vaccine efficacy studies, supported by WHO (2023) reports and regional serological data. These were fine-tuned through Bayesian posterior estimation and Maximum Likelihood Estimation (MLE) techniques to improve model fit to observed typhoid trends.

Environmental and transmission-related parameters including short-cycle (βp) and long-cycle (βw) transmission coefficients, bacterial decay rate (ξ), and reduction factor in infectiousness (r) were estimated via model fitting and calibration using reported typhoid case time series. Literature sources such as Pitzer et al. (2014) informed plausible ranges. Intervention coverage parameters (education e, WASH h) were estimated from program implementation data and public health reports in Zimbabwe.

This mixed approach integrating empirical data, literature-informed priors, and formal parameter estimation ensured that model parameters reflected both local epidemiological conditions and globally recognized biological constraints, thereby improving the robustness and interpretability of the simulation outcomes (Table 3).

Table 3. Intervention parameters

Intervention parameters	Description	Estimate
edu_start_time	Health education starts after 1 year	365 * 1
edu_coverage	70% coverage	0.7
edu_efficacy	50% reduction in transmission	1
wash_start_time	WASH intervention starts after 2 years	365 * 2
wash_coverage	80% coverage	0.8
wash_efficacy	60% reduction in environmental transmission/shedding	0.6
v	Vaccination rate	0.05

Differential Equations:

Susceptible individuals (S₁)

dS₁/dt=B−(λp+λw)S₁−μS₁+ϵS₂ - v S₁ +ρV

B: Birth rate of susceptible individuals.

λp: Rate of short-cycle transmission.

λw: Rate of long-cycle transmission.

μ: Natural mortality rate.

ϵ: Rate at which partially immune individuals become fully susceptible again.

v : Vaccine rate

ρ: Rate at which individual lose vaccination immunity

Primary infected individuals (I₁)

dI₁/dt=(λp+λw)S₁−(δα+μ)I₁−δ(1−α)I₁−δθI₁

δ: Rate at which infected individuals leave the infectious compartment.

α: Fraction of infected individuals who experience disease-induced mortality.

θ: Fraction of infected individuals who become chronic carriers.

Recovered individuals (R)

dR/dt=δ(1−α−θ)I₁ +(1−θ)δII₂−ωR−μR

ω: Rate at which immunity wanes.

Partially susceptible individuals (S₂)

dS₂dt=ωR−(λp+λw)S₂−μS₂−ϵS₂-vS₂

Subclinically infected individuals (I₂)

dI₂/dt=(λp+λw)S₂ − (1−θ)δII₂ - δθI₂ - μI₂

Chronic carriers (C)

dC/dt = δθI₁ + δθI₂−μC

Environmental reservoir of bacteria (W)

dWdt = γ(I₁ +rII₂+rC)−ξW

r: Reduction factor in infectiousness for chronic carriers and subclinically infected individuals.

ξ: Rate at which bacteria in the environment lose infectivity.

γ: Rate at which infectious individuals shed bacteria into the environment.

Vaccination (V)

dV/dt = v S₁ + vS₂ - ρV- μV

Transmission Rates

· Short-cycle transmission (λp)

λp=βp(I1+rI2+rC)

βp: Transmission coefficient for short-cycle transmission.

· Long-cycle transmission (λw)

λw=βwW(1+qcos((2πt−φ)/365))

βw: Transmission coefficient for long-cycle transmission.

q: Amplitude of seasonal variation.

Model Scenarios for Typhoid Emergence at BRIDH

Scenario 1: Treatment Only

This scenario explores the impact of clinical treatment alone on reducing typhoid burden. All infected individuals who seek care are assumed to receive effective antibiotic treatment. We model reduced infectious duration and decreased transmission, assuming treatment success is uniform across the population.

Scenario 2: Treatment + Vaccination

This scenario incorporates both treatment of clinical cases and typhoid conjugate vaccine (TCV) rollout. Vaccination reduces susceptibility among the immunized and indirectly lowers transmission. We assume a given vaccine efficacy and coverage rate, with parameter estimation including vaccine-induced immunity duration.

Scenario 3: Treatment + WASH (Water, Sanitation, and Hygiene)

We evaluate the joint impact of clinical treatment and improved WASH infrastructure. WASH interventions reduce long-cycle transmission by lowering environmental contamination. This scenario assumes partial coverage of WASH improvements across the population and estimates corresponding reductions in transmission parameters.

Scenario 4: Treatment + Typhoid-related Health Education

This scenario assesses the effect of treatment combined with public health education focused on typhoid prevention behaviors (e.g., handwashing, safe food practices). The education component is modeled as reducing the effective contact rate for short-cycle transmission.

Scenario 5: Treatment + WASH Only

Here, we re-express the Treatment + WASH scenario assuming only partial or localized WASH coverage, to reflect implementation in limited settings such as schools or high-risk neighborhoods. We assess whether small-scale WASH improvements, together with treatment, still meaningfully impact transmission.

Scenario 6: Treatment + WASH + Vaccination

This scenario explores a comprehensive intervention involving treatment, vaccination, and WASH improvements. We estimate the synergistic effects of reducing susceptibility, infectiousness, and environmental contamination.

Scenario 7: Treatment + WASH + Typhoid-related Health Education

In this scenario, we assess the combination of treatment, WASH, and behavior change communication campaigns. The model incorporates changes in both environmental and behavioral drivers of transmission, estimating their joint impact on typhoid dynamics.

Scenario 8: Full Integrated Intervention (Treatment + WASH + Vaccination + Typhoid- related Health Education)

This scenario models the full suite of interventions implemented concurrently. It provides an upper-bound estimate of potential typhoid control, integrating medical, infrastructural, and behavioral approaches. We evaluate the combined effects on both short-cycle and long-cycle transmission pathways.

We fitted our model to the data on culture-confirmed cases of S. Typhi at BRIDH from 2018 to 2020 by maximum a posteriori estimation. We assumed uniform prior distributions for all model parameters (Table 2). We calculated the log-likelihood of each model assuming that the weekly number of observed cases in each age group is Poisson-distributed with a mean equal to the

model-predicted number of clinical infections over the duration of infectiousness times the reporting fraction, f. Because we only have data on the age of cases beginning in 2018, we fit the model to the data up to 2018 by summing the model-predicted number of cases each week across all age groups. We will calculate the Bayesian information criterion (BIC) for each best-fit model to compare models with a different number of estimated parameters across the different scenarios (Table 2).

Figure 4.Sensitivity Analysis (Interventions)

From the Graphs (Figure 4) above it seems as Vaccine, Education and wash start to make significant changes or start to be more sensitive at 87% however treatment shows to be very sensitive from small changes.

Figure 5.Average Fitted Histogram (Observed and fitted)

According to the graph (Figure 5) after average fitting Most cases recorded are from September 2018 to January 2019.Then from that cases were decreasing. The red dotted line shows that on average cases were decreasing from January 2019. The epidemic curve shows cases by date of onset from 2018 to January 2020. The highest peak of cases was recorded in December 2018. Mass vaccination started on February 15, 2019 for two weeks. Just over half of the cases (56%) were reported before TCV.

Figure 6.Fitted Model & Average Forecasting

Using the data from 2018 to 2020, the forecast predicted a decrease and constantly low number of cases, however as of 2025 we have experienced increasing cases. This could have been attributed to the waning immunity of the Typhoid Conjugate Vaccine (TCV) administered in 2019 in these hotspots.

Figure 7.Health education and Wash

In the primary infection model, the scenario with treatment alone results in a dramatic increase in cases, reaching approximately 300 within less than a year. However, when typhoid-related health education is introduced, the number of infections drops significantly to around 223. The addition of WASH interventions further reduces the burden, bringing primary infections down to approximately 140 cases. A similar trend is observed with secondary infections: without any intervention, secondary cases reach around 165. Following the introduction of health education, this number decreases sharply to 50, and further declines to 20 after the implementation of WASH interventions. These results highlight the substantial impact of layered public health interventions in reducing both primary and secondary typhoid infections (Figure 7).

Figure 8.Vaccination, Health education and Wash

In the primary infection model, the introduction of vaccination leads to a sharp and rapid decrease in infections, dropping from approximately 1,500 to around 500 cases within a few days. Following this, the implementation of typhoid-related health education results in a more gradual decline, with cases falling to about 245 within less than a year. The subsequent introduction of WASH interventions further reduces the number of primary infections to approximately 125. A similar pattern is observed for secondary infections: under vaccination alone, cases drop from 485 to 125. With the addition of health education, secondary infections fall to fewer than 25, and after the implementation of WASH interventions, the number declines even further to fewer than 10. These findings underscore the cumulative benefits of combining vaccination with health education and WASH in effectively controlling typhoid transmission (Figure 8).

Figure 9.Environmental Pathogen and Chronic infection with Wash Interventions

The graph (Figure 9) demonstrates a significant decline in both environmental pathogens and chronic infections following the implementation of each intervention. The number of chronic carriers drops markedly, primarily due to the impact of improved WASH interventions, which effectively reduce long-cycle environmental transmission (λw). This emphasizes the vital role of environmental sanitation in breaking the transmission cycle and reducing the reservoir of chronic carriers in the population. In the absence of interventions, chronic infections reach approximately 1,100 cases. With the introduction of health education, this number decreases to around 650, and further declines to about 400 after implementing WASH. Similarly, environmental pathogen levels show a dramatic reduction—nearly 75%—once both health education and WASH interventions are introduced (Figure 10).

Figure 10.Optimum Scenario: Projected intervention coverages > 70% for all

The dashboards are capable of displaying all scenarios, with adjustments made according to the selected interventions. This particular graph demonstrates the simulation of various interventions to determine their optimal impact. Additionally, the same dashboards can predict the total number of infections over time under different intervention strategies. In this case, the model predicts only three infections 360 days after the start of 2020, which closely aligns with the average trend depicted in Figure 6.

This study set out to predict future trends in typhoid fever cases in the hotspot areas of Harare City and to develop a mathematical model that simulates the spread of typhoid under varying intervention scenarios. The analysis of data from 2018 to 2020 revealed that the majority of typhoid cases occurred between September 2018 and January 2019. Following this peak, the fitted trend line showed a consistent decline in cases, particularly after January 2019. Forecasts based on this

historical data predicted a continued decrease and a persistently low number of cases into the future. However, real-world observations in 2025 show a concerning resurgence of cases, suggesting that immunity from the Typhoid Conjugate Vaccine (TCV), widely administered in 2019, may have waned over time, leaving the population vulnerable once again.

The simulation of intervention scenarios using the mathematical model further highlighted the differential impact of various strategies on primary and secondary infections. When treatment alone was applied in the primary infection model, there was a sharp increase in cases, reaching approximately 300 within less than a year. The inclusion of typhoid-related health education, however, resulted in a substantial reduction in cases to around 223. The addition of WASH interventions had a further suppressive effect, lowering infections to about 140. A similar trend was observed for secondary infections: from an estimated 165 cases without intervention, health education reduced this figure to 50, and WASH interventions brought it down even further to 20.

Moreover, the model that incorporated vaccination demonstrated a rapid and significant drop in infections. Primary infections fell from approximately 1,500 to 500 within days following vaccination. Health education and WASH interventions layered on top of vaccination continued to drive infections downward, reaching around 245 and 125 respectively. The same was seen with secondary infections, which decreased from 485 to 125 with vaccination alone, then declined to under 25 with health education, and ultimately fell to fewer than 10 with full intervention layering.

The simulations also assessed the impact of interventions on environmental pathogens and chronic infections. Without any interventions, chronic infections rose to around 1,100 cases. Health education led to a drop to 650, while WASH interventions further reduced this to 400. Environmental pathogen levels followed a similar trajectory, with a near 75% reduction when both health education and WASH strategies were implemented. Dashboards developed in this study allowed real-time scenario simulation and visualization of intervention impacts, predicting as few as three cases 360 days after the start of 2020 in a full-intervention scenario—closely aligning with trends depicted in the models.

The model described in the manuscript offers a structured approach to understanding typhoid transmission dynamics and intervention impacts. However, like all models, it has certain limitations that could affect its accuracy and applicability. Some key limitations have been identified:

In terms of Homogeneous Mixing the model likely assumes that individuals mix uniformly within the population, which may not reflect real-world social, geographic, or behavioral heterogeneity. Many parameters are assumed constant over time, ignoring seasonal or temporal variations in transmission, behavior, or intervention coverage. The model does not explicitly account for age-specific differences in susceptibility, exposure, or immunity, which are important in typhoid epidemiology. Some parameters were estimated from literature and limited local data, which could introduce inaccuracies. Typhoid case data usually suffer from

underreporting, misclassification, and diagnostic limitations, affecting model calibration and validation.

The model assumed static coverage levels for interventions like vaccination, WASH, or education, whereas in reality these change over time. The model also simplified immunity waning and vaccine efficacy, which can vary among individuals and over time. While carriers were included, the complexity of chronic carriage, including variable infectivity and potential for reactivation, may be oversimplified. The environmental transmission was modeled with a decay rate, but spatial heterogeneity and environmental complexities are not explicitly represented. The model did not incorporate spatial heterogeneity or social networks, which can significantly influence transmission dynamics, especially in urban vs. rural settings. Finally the model may not account for population movement, importation of cases, or cross-border transmission, which can affect local dynamics or potential changes in pathogen characteristics that could influence transmission or treatment outcomes.

While the model provides valuable insights into typhoid dynamics and potential intervention impacts, its limitations stem mainly from simplifying assumptions, data uncertainties, and lack of detailed heterogeneity. Recognizing these limitations is crucial for interpreting the results and guiding real-world decision-making. Future enhancements could include incorporating spatial structure, age stratification, behavioral factors, and stochasticity to address some of these constraints.

· Booster Vaccination Programs: Given the observed decline in immunity over time and the resurgence of cases in 2025, it is recommended that booster vaccination campaigns be implemented in hotspot areas. Continuous monitoring of vaccine efficacy should inform the need for additional doses, particularly in regions with high transmission rates. Future vaccination efforts should be integrated into routine immunization schedules to maintain high levels of community immunity over the long term.

· Integrated Public Health Interventions: The study highlights the effectiveness of combining vaccination, health education, and WASH interventions to reduce the burden of typhoid fever. It is recommended that health education campaigns focus on promoting hygiene practices, safe water use, and proper sanitation, particularly in urban slums and informal settlements. These educational efforts should be coupled with the widespread availability of safe water and improved sanitation infrastructure to address environmental transmission risks.

· Strengthening WASH Infrastructure: The reduction in chronic infections and environmental pathogen levels following WASH interventions underscores the need to invest in improved water, santation, and hygiene infrastructure. Local governments should prioritize water treatment, sewage systems, and waste disposal improvements, particularly in high-risk communities, to break the transmission cycle of typhoid fever. Additionally, community-based sanitation programs should be supported to ensure sustainability and community ownership of WASH facilities.

· Enhanced Surveillance Systems: Continuous surveillance of typhoid fever cases is essential to monitor trends and identify potential outbreaks in real-time. Surveillance systems should integrate data from hospitals, clinics, and environmental health programs to allow for a comprehensive understanding of the disease’s dynamics. This data can inform public health responses and allow for timely adjustments to intervention strategies.

· Research on Long-Term Vaccine Efficacy: Further research is needed to assess the long-term effectiveness of the Typhoid Conjugate Vaccine (TCV) in different populations. Longitudinal studies should focus on tracking immunity over time and identifying potential gaps in protection. Research should also explore alternative vaccines or formulations that may offer longer-lasting immunity.

· Community Engagement and Participation: Community involvement in health education and disease prevention efforts is critical for the success of typhoid control programs. Local leaders and community groups should be engaged to help disseminate information and encourage participation in vaccination and hygiene initiatives. This can help overcome barriers to intervention acceptance and ensure that public health campaigns are culturally appropriate and widely adopted.

· Intervention Modeling for Policy Decision-Making: Policymakers should utilize mathematical modeling tools, such as those employed in this study, to evaluate the potential impact of different intervention strategies and predict future disease trends. These tools can assist in optimizing resource allocation, ensuring that public health interventions are both cost-effective and tailored to the specific needs of affected communities.

All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.