Surveillance data of infectious diseases often constitute the first available source for studying the epidemiology of emerging diseases, and proper analysis of surveillance data can provide valuable insights on transmissibility and control strategies. However, proper analysis should address a few challenges. (1) there may be multiple transmission routes, e.g., from environmental reservoir to human and from human to human, and each route has its own spatio-temporal heterogeneity and correlation; (2) there may be multiple types of pathogens co- circulating and causing the same disease, but only a small proportion of cases are sampled to identify the responsible pathogen; and (3) a complex system including immunity, cross-immunity, unobserved asymptomatic infections, together with the birth and aging process of the population may have shaped the demographic, spatial and temporal structure of the surveillance data. Motivated by the Chinese surveillance data of the hand, foot and mouth disease (HFMD) that is caused by a family of enteroviruses (EV), we propose to address above challenges via the following Specific Aims: (1) To build and test a Bayesian modeling framework for surveillance data of co-circulating pathogens to assess pathogen-specific environment-to-human and human-to-human transmissibility and associated risk factors, while accounting for spatio-temporal heterogeneity and correlation in transmissibility; with this model applied to the surveillance data of HFMD in China, to estimate the transmissibility and effects of risk factors for two major enteroviruses, EV 71 and Coxsackie A16; and (2) To extend the modeling framework in Specific Aim 1 with asymptomatic infection, immunity and cross-immunity from previous exposure, and birth, aging and death of the population; to use this extended model to explain the long-term evolvement of the HFMD epidemics, to predict the effectiveness of EV 71 vaccination programs, and to investigate possible replacement of EV 71 by other enteroviruses. Our preliminary simulation study on a simplified model showed feasibility of simultaneous estimation of transmission rates, spatial effects and temporal effects with a moderate number of geographic units and a small amount of laboratory validation. Once the modeling frame work is validated in simulation studies, we will adapt it to multiple years of surveillance data of the HFMD epidemics to detect unknown features about the HMFD-related pathogens, in particular asymptomatic infection and cross-immunity, and to make recommendations about vaccination programs to public health policy-makers. The proposed analytic framework may be generalized, with appropriate customization, to a broad range of acute infectious diseases including influenza, adenovirus, cholera and dengue.
The proposed projects are relevant to public health for three reasons. First, they provide a joint analytic framework for using surveillance data to study transmissibility and associated risk factors for multiple co- circulating pathogens, while accounting for spatio-temporal heterogeneity and correlation as well as high- dimensional missing outcomes. Secondly, they will shed light on how asymptomatic infections and potential partial cross-immunity may be detected with surveillance data as well as how would these mechanisms shape the long-term dynamics of the epidemics. Finally, they provide a basis for evaluating the effectiveness of intervention programs in simulation studies which may inform decision-making on public health.
| Fisher, Leigh; Wakefield, Jon; Bauer, Cici et al. (2017) Time series modeling of pathogen-specific disease probabilities with subsampled data. Biometrics 73:283-293 |
| Fang, Li-Qun; Yang, Yang; Jiang, Jia-Fu et al. (2016) Transmission dynamics of Ebola virus disease and intervention effectiveness in Sierra Leone. Proc Natl Acad Sci U S A 113:4488-93 |
