Introduction

Severe fever with thrombocytopenia syndrome (SFTS) is a tick-borne disease caused by SFTS virus, a newly identified phlebovirus in the family Bunyaviridae [1]. Cases with SFTS-like clinical manifestations, initially described as fever, thrombocytopenia and leukopenia syndrome, had been reported in Henan Province since 2007, while the causative virus was first identified during investigations of SFTS cases in Hubei and Henan provinces in 2009 [1,2]. ‌‌SFTS was subsequently confirmed in multiple provinces across central and northeastern China. The high case-fatality rate (CFR), which reached up to 30% in the early stage of the epidemic, together with the expanding endemic areas of SFTS, has contributed to its persistent public health threats since the identification of SFTSV [1,2]. Human cases are associated with wide clinical manifestations, varying from mild symptoms to life-threatening illness or death [3]. In view of continuing epidemic threats and insufficient preventive and curative solutions, SFTS is increasingly recognized as a prioritized emerging disease with the potential to generate public health emergencies [4]. This motivates researchers to support the urgently needed assessment of epidemic risks and ultimately unbiased intervention decisions.

As with other zoonotic diseases, SFTS risk assessment relies primarily on investigating how the ecological contexts modulate the transmission potential and severity of the disease [5,6]. It is widely recognized that tick-to-human transmission is the primary route of human infection with SFTS virus. Although the complete ecological cycle of SFTSV has not yet been fully elucidated, accumulating evidence identifies Haemaphysalis longicornis as the principal and confirmed vector, based on its high abundance in endemic areas and the frequent detection of viral RNA in field-collected specimens [1,3]. In addition, several other tick species have been proposed as potential contributors to SFTSV transmission, suggesting a more complex vector system that may vary across ecological settings. Vertebrate animals are thought to act mainly as amplifying hosts rather than true reservoirs, highlighting the central role of ticks in shaping human exposure risk [2,7,8]. The interplay of natural environment and tick population dynamics may shape the diverse ecological landscapes of local epidemic risk [9]. Crucially, meteorological conditions can shape local ecological contexts by influencing tick survival, development, and seasonal activity, thereby modulating opportunities for tick-human contact. Previous studies have shown that temperature, precipitation, and vegetation-related environmental factors are closely associated with the spatial distribution of SFTS risk [10,11]. Consequently, higher infection risks are frequently observed in mountainous or hilly rural areas where forested or shrub-dominated landscapes provide suitable habitats for ticks and where agricultural activities increase human exposure to tick bites [10–13]. However, existing epidemiological evidence has not consistently integrated ecological signature related to climate-associated variation in vector abundance and environmental context with the inference of SFTS dynamics and its severity.

The fundamental ecological dependence of vectored disease dynamics introduces an urgent need to contextualise mechanistic dynamics of SFTS in local ecological systems. Our recent research innovations have contributed to the development of eco-epidemiological models [5,6] that integrate natural environment, vector abundance and human cases. The interconnected nature of these models suggested that disease dynamics may be highly dependent on local vector dynamics, which, in turn, can be influenced by local climate conditions. Consequently, these models facilitate the concurrent analysis of vector responses to the natural environment and their impact on disease dynamics.

We ground our study in the context of the most recent increase in SFTS incidence and case fatality in Jiangsu Province, China [14,15]. We characterize the evolving epidemic profile of SFTS along two key dimensions: the spatiotemporal diffusion of infections and annual CFR. This perspective enables the identification of endemic counties, changes in severity and shifts in seasonality of the disease. We subsequently focus on an eco-epidemiological modelling framework to assess the dynamic risk of SFTS across these endemic counties. Within this framework, climate-based vectored dynamics are used as a proxy for the transmission, linking SFTS ecology with its epidemiological dynamics. Specifically, we test the hypothesis that (1) climate-associated increases in estimated tick abundance are associated with higher transmission potential () across endemic counties, and (2) variation in estimated vector abundance is independently associated with differences in case fatality risk among reported SFTS cases after accounting for demographic, ecological and behavioural factors. The findings of our study highlight the profound influence of ecological contexts on disease dynamics and severity, offering valuable insights applicable to a broad spectrum of zoonotic diseases.

Methods

Ethics statement: This study was approved by the Ethics Committee of Jiangsu Provincial Center for Disease Prevention and Control, which waived the requirement for informed consent due to the retrospective nature of the study and the use of anonymized data. All methods were performed in accordance with the relevant guidelines and regulations. All experimental protocols were approved by the aforementioned institutional committee.

Eco-epidemiological model

In line with the fundamental ecological dependence of vectored disease dynamics, we developed an eco-epidemiological model that connects natural environment, vector abundance and human cases. Specifically, we employed a statistical model to establish climate-abundance association; and sequentially developed a mechanistic epidemiological model to derive SFTS transmission among humans. The climate-abundance association was formed by incorporating local meteorological conditions and the long-term surveillance data of ticks into a generalized additive model (GAM) Eq (1). With the associations, we derived the monthly estimates of climate-based vectored abundance. We assumed that variations in vectored abundance is proportional to the seasonal transmission rate of SFTS in human populations. Therefore, vector abundance was used as a proxy for SFTS transmission rate in the seasonal susceptible–infected–recovered (SIR) model (Eqs (2)–(4)). The coupling model is described by the following equations:

(1)(2)(3)(4)

where is the vector abundance in month in county . is a two-dimensional component characterizing spatial heterogeneity. Given the roughly 5-week growth and development period of ticks [17], the model includes a 1-month lag between the local climate conditions and vector population dynamics. Such that and captures the monthly accumulated temperature and number of rainy days in the previous month, respectively. The parameters and denote the overall intercept and model error, respectively. Model diagnostics indicated an adequate fit of the GAM, with approximately 36% deviance explained and no evidence of overfitting based on estimated degrees of freedom. A quasi-Poisson specification was used to account for overdispersion. Moreover, mechanistic model is consisted of three components, i.e., the susceptible (), infected (), and recovered () individuals in human populations. Thus, the population size is calculated as . represents the estimates of tick abundance using Eq (1) and is used as a proxy to introduce seasonal and spatial variation in transmission intensity in the SIR model. Transmission rate among humans is collectively determined by and using . In this case, represents a time-varying transmission efficiency. It is modelled as a scaling factor that links estimated tick abundance to the transmission rate in the SIR model and captures multiple unresolved influences on human transmission, such as human behavior, tick infection prevalence, or habitat factors. The average infectious period () is modelled as a random constant varying from 10–20 days [3,18]. Consequently, the effective reproductive ratio was calculated as , where .

We calibrated the GAM using monthly tick abundance and mean climatic data from 2017 to 2023. We additionally calibrated the SIR model to the recorded cases during significant outbreak years (defined as years where reported cases exceeded the county-specific median during the study period) in endemic counties. We assume homogeneous susceptibility within human populations, justified by the low background immunity in the general population, and reinitialised the model 500 times with varying values at the beginning of each outbreak year. The median estimates and 95% confidence intervals of cases from all simulations using varying values are presented. Additionally, we estimated the annual average magnitudes of to assess the county-specific transmission potential. County-specific median estimates of vector efficacy were derived, and then rescaled as relative vector efficacy as compared with the overall efficacy using , where is vector efficacy in county and is average vector efficacy among counties, years and infectious period, ensuing the relative vector efficacy fluctuates around a mean of 1.

Results

The Results are structured to follow the analytical workflow described in the Methods, beginning with descriptive epidemiological analyses, followed by modelling of the association between climate variables and estimated tick abundance, eco-epidemiological transmission modelling, and finally case-level analysis of case fatality risk factors.

Epidemiological dynamics of SFTS

Surveillance of human SFTS infections reveals distinct spatial heterogeneity across counties (Fig 1A). Precisely, we observed a typically larger number of human infections in six counties: Xuyi, Lishui, Jiangning, Pukou, Luhe and Jurong, which collectively account for approximately 80% of the total infections (S2 Table). Additionally, these endemic counties are characterized by a longer persistence of infections than others, exhibiting over six years of continuous disease circulation among humans (S2 Table). Focusing on these endemic counties, we show a substantial increase in the annual number of cases, rising by 572% from 25 cases in 2017–168 cases in 2023, alongside an increase in the number of infected towns (Fig 1B). Importantly, this uptick coincides with a growing severity of the disease (Fig 1B), with the annual CFR increasing from 0% to roughly 10% over the past years. Furthermore, the dynamics of SFTS exhibit strong seasonality within endemic counties (Fig 1C). It is also worth noting that such seasonality varies, yielding shifts in the peak occurrence from July to May since 2021.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Epidemiological dynamics of SFTS in Jiangsu in 2017-2023.

(A) County-level number of laboratory-confirmed SFTS human cases presents the spatial heterogeneity of disease risks (https://cloudcenter.tianditu.gov.cn/administrativeDivision/, Map Review No. GS (2024) 0650). (B) Among endemic counties, epidemiological dynamics of the disease is characterized by (B) the annual number of human cases, CFR and affected towns; as well as the (C) seasonality.

https://doi.org/10.1371/journal.pntd.0014408.g001

Climate-based vectored dynamic risks

By incorporating climate-based vectored transmission rate in our seasonal epidemiological model, we operationalize a quantitative association between estimated vector abundance and SFTS transmission. The GAM revealed a significant non-linear association between accumulated temperature (lagged by one month) and estimated tick abundance, whereas the effect of precipitation days (lagged by one month) was weaker and not statistically significant (S1 Fig). Overall, the model reproduced the main temporal patterns of observed disease dynamics across years and counties (Figs 2 and S2). Therefore, our model adeptly portrays the seasonal dynamics of human SFTS across diverse climatic and ecological contexts.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Model inference of SFTS human cases among endemic counties.

The observed number of laboratory-confirmed human cases (black points) in outbreak years (the grey shaded area) is used for model simulation. The model was reinitialized by using a plausible range of infectious periods at the beginning of each outbreak year. The median estimates of human cases (red lines) and corresponding confidence intervals are presented.

https://doi.org/10.1371/journal.pntd.0014408.g002

Further investigation indicates heterogeneity in the transmission potential of SFTS among endemic counties. For instance, the median estimates of in Xuyi and Lishui, depending on outbreak years, is 0.10 (95% CI: 0.001–0.224). This contrasts with a median of < 0.01 in the other four counties (Figs 3A and S3A). Consistent with these disparities, our estimates shows that the median estimates of the peaking magnitude of over the years is around 0.3 in Xuyi and Lishui, while it remains below 0.1 in other counties (S3B Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Estimates of the transmission potential of SFTS among endemic counties.

(A) The median estimates and confidence intervals of in outbreak years capture the overall transmission potential of SFTS among humans. (B) Monthly estimates of vector efficacy characterize the relative capacity across counties.

https://doi.org/10.1371/journal.pntd.0014408.g003

It is noteworthy that these spatial differences in coincide with the higher vector efficacy in Xuyi and Lishui. As compared with the overall transmission efficacy spanning various years and counties, there is evidence of a notably higher efficacy from March to September in these two counties (Fig 3B). Given the relatively limited variation in estimated vector abundance across counties (S4 Fig), the higher efficacy and in Xuyi and Lishui suggest additional heterogeneity in transmission intensity beyond that captured by modeled abundance alone. Investigating the tick positivity and exposure history, unveils higher tick positivity and a larger fraction of patients with tick exposure or biting history in Xuyi and Lishui (S5 Fig). This implies a possible collective contribution to the heightened vector efficacy observed.

Multifaceted determinants of case fatality after infection

After adjusting for occupation, onset-diagnosis duration and tick abundance, older age correlates with significantly increased odds of death after infection (Table 1). Our estimates show that people over 65 years would have 197.9% (43.3%–582.5%) higher odds of death than those under 65 years. Notably, patients living in the neighbourhood with higher estimated tick abundance showed higher odds of deaths after infection. It is estimated that higher level of tick abundance is associated with approximately 202.8% higher odds of death. Comparatively, we found no evidence of a significant effect from occupation and onset-diagnosis duration on the odds of case fatality.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Potential factors associated with the risk of fatal outcomes. Three sets of factors were included: demographic factors (age, sex, occupation), tick abundance in the neighbourhood environment, care-seeking behaviour (or the duration from onset to diagnosis). Median estimates and corresponding 95% CIs of each factor are provided. Numbers with “*” indicate statistically significant (p < 0.05).

https://doi.org/10.1371/journal.pntd.0014408.t001

Discussion

Our study aligns with the observed increase in SFTS cases in most recent years [14,15]. Crucially, we observed shifts in the epidemic profile of the disease, with a substantial increase in human infections, the number of affected areas, and disease severity. These findings are broadly consistent with other studies done across a variety of settings [19–21]. Importantly, the seasonality of the disease is not static, and can shift over time. While the incidence of human cases might typically be higher from May to August, there has been a notable peak around May starting in 2021. This shift may be associated with earlier seasonal tick activity and changes in human outdoor behaviour in recent years, including altered activity patterns following the COVID-19 pandemic, which could advance the timing of human exposure to ticks [15]. Therefore, we provide the first empirical evidence of the evolving epidemiological dynamics of SFTS. These insights emphasize the need to investigate the potential changes in future disease dynamics among broader ecological settings.

A primary challenge in assessing SFTS risk involves contextualizing epidemiological dynamics within local ecological context. The novelty of our study lies in its inherently multidisciplinary nature and cohesiveness, harnessing a suite of evidence from disparate perspectives across the fields of ecology and epidemiology. By integrating extensive data streams to eco-epidemiological model, we demonstrate how local weather conditions, through associations with estimated tick abundance, may contribute to observed variations in SFTS dynamics in China (see S1 Text). While these climate–abundance associations highlight the role of meteorological conditions in shaping seasonal variation in tick abundance, they represent only the component of variability that can be explained by climate. Tick abundance is influenced by a broader set of interacting biotic and abiotic factors, including host availability, habitat structure, and land-use patterns, which were not explicitly modelled in this study. This framing first allows us to gain insight of vector population dynamics in response to local weather condition; it then aids in assessing the non-linear effects of vector abundance on SFTS dynamics. Therefore, our predictive framework offers a route to placing tick-borne disease dynamics in an ecological system.

Although the estimated values are consistently below one, this does not contradict the observed persistence of SFTS cases across multiple years. Humans are considered dead-end hosts for SFTSV, and sustained transmission within the human population is not expected. Instead, recurrent seasonal spillover from infected ticks can generate observable outbreaks each year, even when the effective reproduction number remains below one. In this context, should be interpreted as a relative indicator of transmission potential shaped by ecological and seasonal forcing, rather than a threshold parameter for epidemic persistence. While provides a useful relative indicator of transmission potential within counties over time, comparisons of absolute values across counties should be interpreted cautiously due to differences in ecological context, surveillance intensity, and reporting.

It is notable that vector efficacy and thus transmission potential is typically higher in Xuyi and Lishui compared to other counties, coinciding with higher tick positivity and increased likelihood of human exposure. Recent years have witnessed crucial developments in understanding of how the spatiotemporal dynamics of SFTS depend on a multitude of interacting ecological factors. It has become evident that mountainous region, especially at the border of Jiangsu and Anhui provinces, pose a heightened risk of SFTS infections [15,22]. However, such models tend to rely heavily on the statistical interplay with ecological factors without fully encapsulating the inherent epidemiological processes. We argue that opportunities exist to more deliberately adopt an eco-epidemiological approach in SFTS mechanistic transmission and risk assessment. The success of our model inference, combined with large-scale vector surveillance and lab experiment, opens up the possibility of applying eco-epidemiological approach to the real-time risk assessment of zoonotic diseases.

Notably, the key to improving public health decisions is to recognize the determinants of the epidemiological trajectories of the disease. Most of the existing models have tended to improve risk assessment by relying on multiple parameters [23]. However, the complexity of such models may inevitably yield biases when identifying key determinants. To address this, we specifically modelled the rate of transmission using vector efficacy and abundance and linked the epidemiological interpretation of the efficacy to tick positivity and human exposure history. Within this framework, vector efficacy should be understood as a composite, population-level indicator that reflects transmission efficiency under local ecological and behavioural contexts, rather than a direct biological property of ticks or a causal determinant at the individual level. Our model simulations highlight that tick positivity and human exposure are tightly tied to vector efficacy and thus seasonal transmission potential of SFTS. Attention to this joint signature of ticks and human behavioural risk factors are especially relevant to designing targeted strategies for SFTS risk management. Existing evidence have suggested that preventive measures should be primarily directed towards to people exhibiting high-risk behaviours, such as those with occupational exposure and those engaging outdoor activities [24–26].

Beyond ecological observations, our findings offer crucial insights for clinical studies. Historically, clinical scoring systems have contributed to the prediction of fatal outcome [27–29]. Of the factors included in such scoring system, age of patients has been recognized as the key to determining their vulnerability to death. Consistent with this, we identified that patients over 65 years old associate with higher risk of fatal outcomes. This means that senior patients should be prioritized for diagnosis and treatment in clinical settings. It is worth noting that our findings not only support the well-recognized signature of age on the risk fatal outcomes, but also propose the significance of considering tick abundance in the neighbourhood of residence in clinical activities. We show a considerable excess risk for patients living in the neighbourhood of higher tick abundance. Therefore, tick abundance in the residential area may provide contextual information for population level risk assessment and targeted prevention strategies, rather than serving as a direct clinical predictor. Tick abundance was estimated from a climate-informed statistical model rather than directly observed, and the associated uncertainty is implicitly carried forward into subsequent analyses. However, uncertainty from the statistical model was not explicitly propagated into the fatality model, and measurement error in estimated abundance may attenuate the observed association with fatal outcomes. Accordingly, associations involving tick abundance and vector efficacy should be interpreted cautiously as population-level, non-causal relationships reflecting broader ecological and potentially socio-demographic contexts, such as rurality, healthcare access, or unmeasured comorbidities.

Our study highlights opportunities for future development. We demonstrate essential role of vector efficacy in transmission. However, some ecological studies suggest vector population expansion would be the key in other settings [30]. Thus, one focus for future research would be assessing the relative role of vector efficacy and abundance in diverse settings. In addition, while the present analysis focused on overall tick abundance, future studies incorporating species-specific tick surveillance data may provide further ecological and mechanistic insight, and help refine the interpretation of transmission heterogeneity in eco-epidemiological modelling. Explicitly incorporating tick positivity into mechanistic transmission model may have the potential to implicating the interacting processes across scales and risk assessment. Furthermore, the increase in SFTS incidence and case fatality over the most recent years might partially attributable to the improved health system, changing care-seeking behavior, and the increased outdoor activities. It is crucial to evaluate their alterations on disease epidemiology.

The key messages of this study are of great public health implications. We make critical contributions to understanding how epidemiological profile may have evolved over time. Crucially, this study helps reveal how ecological contexts has disproportionately affected SFTS epidemiological dynamics across empirical settings. This represents a promising juncture between disease ecology and epidemiology. ‌‌Such evidence can guide health authorities to grasp the intricacies of diverse epidemic risks, and ultimately facilitating the prioritization of resources to the most at-risk regions and populations.

Supporting information