Effects of human mobility, temperature and mosquito control on the spatiotemporal transmission of dengue

https://doi.org/10.1016/j.scitotenv.2018.09.182Get rights and content

Highlights

  • A modeling framework with consideration of multi-scale factors and surveillance data is introduced to explore the spatiotemporal transmission dynamics of dengue.

  • Human mobility potentially sparks new dengue infections in virgin areas.

  • Temperature could be the causal factor leading to the explosive dengue outbreak in the PRD.

  • The adopted mosquito control strategies significantly reduced dengue incidence in the PRD.

Abstract

Dengue transmission exhibits evident geographic variations and seasonal differences. Such heterogeneity is caused by various impact factors, in which temperature and host/vector behaviors could drive its spatiotemporal transmission, but mosquito control could stop its progression. These factors together contribute to the observed distributions of dengue incidence from surveillance systems. To effectively and efficiently monitor and response to dengue outbreak, it would be necessary to systematically model these factors and their impacts on dengue transmission. This paper introduces a new modeling framework with consideration of multi-scale factors and surveillance data to clarify the hidden dynamics accounting for dengue spatiotemporal transmission. The model is based on compartmental system which takes into account the biting-based interactions among humans, viruses and mosquitoes, as well as the essential impacts of human mobility, temperature and mosquito control. This framework was validated with real epidemic data by applying retrospectively to the 2014 dengue epidemic in the Pearl River Delta (PRD) in southern China. The results indicated that suitable condition of temperature could be responsible for the explosive dengue outbreak in the PRD, and human mobility could be the causal factor leading to its spatial transmission across different cities. It was further found that mosquito intervention has significantly reduced dengue incidence, where a total of 52,770 (95% confidence interval [CI]: 29,231–76,308) dengue cases were prevented in the PRD in 2014. The findings can offer new insights for improving the predictability and risk assessment of dengue epidemics. The model also can be readily extended to investigate the transmission dynamics of other mosquito-borne diseases.

Introduction

The transmission of mosquito-borne infectious diseases (MIDs), such as malaria, Zika and dengue, is a spatiotemporal dynamic process, affected by multiple factors involved in vectors, pathogen and humans. These factors, including but not limited to environmental, demographic, socioeconomic, behavioural and genetic elements, largely determine the disease transmission patterns (Ostfeld et al., 2005, Lambrechts et al., 2011, Sanna and Hsieh, 2017, Tang et al., 2017). Previous studies claimed that temperature and human mobility are important driving force leading to the temporal and spatial transmissions of MIDs (Searle et al., 2017, Stoddard et al., 2009, Xu et al., 2017), in that temperature affects mosquito entomology and pathogen activity (Mordecai et al., 2017, Morin et al., 2013), and human mobility modifies their exposure to mosquitoes and pathogens (Stoddard et al., 2009). On the other hand, due to the lack of effective therapeutics and applicable vaccination for many MIDs, mosquito control has been the most commonly used strategy to fight against these diseases. The above-mentioned factors couple together, and play as activators or inhibitors in shaping the spatiotemporal distributions of MID incidences. In view of this, it would be desirable and essential to systematically assess the integrated impacts of related factors on the space-time variations of MID incidences (Liu et al., 2012). Doing so will help better understand the transmission patterns and further guide public health authorities to plan control strategies.

Technically speaking, many methods have been developed to analyze the heterogeneous distributions of MID incidences, and identify the risk factors underlying disease transmission (Racloz et al., 2012, Guzman and Harris, 2015, Reiner et al., 2013). Mapping techniques and statistical methods, such as geographic information systems (Wu et al., 2009), time-series regression (Xu et al., 2017), and spatial scan statistics (Li et al., 2012), are widely used to identify the hot spots and factors' effects, and estimate the dispersion process (Racloz et al., 2012). However, pure statistical and mapping analysis methods are limited in exploring the real dynamics of disease transmission (Shi et al., 2015). On the other side, following the classical Ross-Macdonald theory, various transmission models have been developed to describe the underlying epidemiological process (Mordecai et al., 2017, Reiner et al., 2013, Smith et al., 2014). Unfortunately, due to the intrinsic complexity of MID transmission, there are still some other factors, such as mosquito control and heterogeneity of mosquito-host behaviors, the combined effects of which have not been explicitly interpreted from the perspective of spatiotemporal dynamic mechanism (Reiner et al., 2013).

This paper aims to provide a modeling framework which systematically incorporates the impact factors allowed for exploring the spatiotemporal transmission patterns of MIDs. It focuses on the 2014 dengue outbreak in the Pearl River Delta (PRD) in southern China. One reason that dengue is chosen as a case study lies in that it is now regarded as the most prevalent and rapidly spreading MIDs among humans (WHO, 2012). Dengue is spreading across 120 countries and causes half the world's population being at risk of infection. The global incidence has increased 30-fold compared with 50 years ago (WHO, 2012), with an estimation of 390 million cases each year (Bhatt et al., 2013). In China, dengue infection in recent years also exhibited upward trend both in extent and severity (Lai et al., 2015). During 1991–2013, China recorded about 21,532 dengue cases with 620 deaths (Lai et al., 2015). The incidence reached an unprecedented level in 2014, where 46,864 cases were reported (Xiao et al., 2016). Another important reason is that the situation of dengue transmission in the PRD is complicated. First, the PRD is located in subtropical regions and is one of the most urbanized and densely populated regions in the world. Such conditions favor the survival and activity of mosquitoes as well as the transmission of dengue virus. Indeed, the number of dengue cases in the PRD always ranked first in China. In 2014, about 92.3% dengue cases were reported in the PRD. Second, the PRD has convenient transportation and is a transport hub in China, which potentially leads to the spatial spread of dengue in this region (Zhu et al., 2018, Qi et al., 2015). Third, mosquito control strategies have been widely employed in the PRD in 2014. Lin et al. (2016) claimed that the intervention program adopted in Guangzhou led to a 70.47% reduction of dengue cases. However, the control effects incorporating spatial connection by human mobility has not been systematically evaluated in the 7 cities of the PRD. Fourth, transmission potential of dengue virus in different regions of the PRD varies significantly due to the variations of influence factors (e.g., climatic, environmental, and socioeconomic elements) and demographic profiles (e.g., population density and human mobility) (Zhu et al., 2018, Qi et al., 2015, Li et al., 2017).

To clarify the underlying cause of geographic differences of dengue incidence and to identify the spatiotemporal dynamics of dengue transmission, in this paper, the following three technological challenges are addressed.

  • How to systematically integrate the vector-virus-host interactions with the impact of risk factors in a transmission model and thus to simulate the dynamic diffusion process?

  • How to characterize the roles of the dominant factors in activating or inhibiting the spatiotemporal transmission of dengue?

  • How to develop computational methods to quantify geographic variations by fitting model parameters to observed dengue incidences?

Here, mathematical model and machine learning method are employed to tackle these challenges. Specifically, a new dengue transmission model is built based on the notions of Ross-Macdonald theory (Reiner et al., 2013, Smith et al., 2014) and meta-population system (Bichara and Castillochavez, 2016, Ruktanonchai et al., 2016). The model characterizes dengue transmission potential and epidemic evolution by integrating the dynamical changes of virus activity, and the ecology and behaviour of both humans and mosquitoes. Thereinto, (1) temperature is incorporated by affecting the transmission elements involved in viruses and mosquitoes (namely, the rates of oviposition, mortality, aquatic transition, incubation, mosquito bites and transmission probability); (2) mosquito control effect (measured by the change of vector indices) is taken into account by reducing the abundance of mosquito population; and (3) human mobility (simulated by residence time matrix) modifies the force of infections. Furthermore, to quantitatively assess geographic variations of dengue incidences, machine learning methods (such as the Markov chain Monte Carlo (MCMC) method; Shi et al., 2015, Lopes et al., 2008 can be used by fit modeling results to real-world observations.

To evaluate the performance of the proposed model, a real-world study is carried out to investigate the spatiotemporal transmission of dengue among the 7 cities in the PRD, China. By implementing the proposed model, it is able to (1) identify the contributions of temperature and human mobility to the geographic variations of dengue incidences; (2) evaluate the effects of control strategies on stopping dengue transmission; (3) clarify the underlying transmission dynamics of dengue.

Section snippets

Modeling framework

Inspired by recently-developed mathematical models (Reiner et al., 2013, Smith et al., 2014, Bichara and Castillochavez, 2016, Ruktanonchai et al., 2016, Rodriguez and Torressorando, 2001, Iggidr et al., 2016), and based on the feature of dengue transmission as shown in Fig. 1, a new dengue transmission model is established in this section. In this model, certain geographic area (such as city, province or country) is viewed as a patch, and its permanent residents and vectors constitute

Application

In this section, the proposed model was applied to analyze the spatiotemporal transmission of dengue in the PRD between July and December in 2014.

Discussion

Dengue is posing an increasing threat to public health in the tropical and subtropical regions of the world. To control and prevent dengue transmission, it is widely recommended the development of effective and efficient surveillance systems for timely and accurate implementation of risk assessment and control strategies (Racloz et al., 2012, WHO, 2009). This requires clear information about epidemiologic features and transmission patterns of the disease. In reality, dengue spread is a hidden

Acknowledgments

This work was jointly supported by the National Key R&D Program of China (2018YFA0606202), the National Natural Science Foundation of China (11661026 and81773497), the Natural Science Foundation of Guangdong Province (2017A030313699), and Guangxi Natural Science Foundation (2017GXNSFAA198235), as well as Guangdong Provincial Medical Research Foundation (A2017474). The authors would like to thank the handling Editor and the anonymous reviewers for their constructive comments.

References (49)

  • H. Tian et al.

    Surface water areas significantly impacted 2014 dengue outbreaks in Guangzhou, China

    Environ. Res.

    (2016)
  • P.-C. Wu et al.

    Higher temperature and urbanization affect the spatial patterns of dengue fever transmission in subtropical Taiwan

    Sci. Total Environ.

    (2009)
  • G. Zhu et al.

    The spatiotemporal transmission of dengue and its driving mechanism: a case study on the 2014 dengue outbreak in Guangdong, China

    Sci. Total Environ.

    (2018)
  • D. Balcan et al.

    Multiscale mobility networks and the spatial spreading of infectious diseases

    Proc. Natl. Acad. Sci. U. S. A.

    (2009)
  • S. Bhatt et al.

    The global distribution and burden of dengue

    Nature

    (2013)
  • Q. Cheng et al.

    Climate and the timing of imported cases as determinants of the dengue outbreak in Guangzhou, 2014: evidence from a mathematical model

    PLoS Negl. Trop. Dis.

    (2016)
  • M.C. Gonzalez et al.

    Understanding individual human mobility patterns

    Nature

    (2008)
  • S. Lai et al.

    The changing epidemiology of dengue in China, 1990–2014: a descriptive analysis of 25 years of nationwide surveillance data

    BMC Med.

    (2015)
  • L. Lambrechts et al.

    Impact of daily temperature fluctuations on dengue virus transmission by Aedes aegypti

    Proc. Natl. Acad. Sci. U. S. A.

    (2011)
  • M.-T. Li et al.

    The driving force for 2014 dengue outbreak in Guangdong, China

    PLoS ONE

    (2016)
  • Q. Li et al.

    Ecological niche modeling identifies fine-scale areas at high risk of dengue fever in the Pearl River Delta, China

    Int. J. Environ. Res. Public Health

    (2017)
  • H. Lin et al.

    Community involvement in dengue outbreak control: an integrated rigorous intervention strategy

    PLoS Negl. Trop. Dis.

    (2016)
  • J. Liu et al.

    Malaria transmission modelling: a network perspective

    Infect. Dis. Poverty

    (2012)
  • H.F. Lopes et al.

    Spatial dynamic factor analysis

    Bayesian Anal.

    (2008)
  • Cited by (56)

    • Temporal trend and spatial clustering of the dengue fever prevalence in West Java, Indonesia

      2022, Heliyon
      Citation Excerpt :

      Environmental and social features of these areas are likewise seen to affect the number of dengue incidences. High population densities enable rapid urbanization and intensive human mobility that might support virus distribution and increase the possibility of contact between dengue vectors and humans [50]. Due to the lack of decent water infrastructure, congested human settlements can offer favorable breeding opportunities for mosquitoes using a wide range of natural and artificial water-filled containers for egg-laying [51].

    View all citing articles on Scopus
    View full text