Infectious diseases co-occurrence and co-infection: effects on epidemic dynamics.
Lead Research Organisation:
Lancaster University
Department Name: Medicine
Abstract
Nowadays, the co-occurrence of infectious diseases is poorly understood in terms of both geography and time. As many as 30% of infectious diseases may result from co-infections, climbing as high as 80% in certain populations. Several methodological challenges remain when it comes to understanding co-occurrences, in particular, detecting interaction among associations and understanding the complexity of host-pathogen-environment interactions. Existing patterns of infectious disease co-occurrence could thus play a critical role in resolving or anticipating current and future disease threats. Therefore, understanding the ecology and transmission of co-occurring pathogens within populations is crucial to design suitable disease prevention programmes, but methods to gain insights into host-pathogen-environment interactions are currently underdeveloped. By modelling the prevalence of viruses in areas where pathogens co-occur and determining the risk for pathogens to co-occur and co-infect, the project aims to develop statistical frameworks for public health surveillance and control.
The aim of this project is to create a platform capable of detecting the potential of infectivity in an area where co-infection of vector populations occurs and to develop a model that can determine co-occurrence attributed risk of infectious diseases. This could then lead to identification of hotspots or areas of high risk attributed to co-occurrence and inform policies.
The project will start by working with two distinct types of data. The first is remote sensing data (satellite data) from MODIS (Moderate Resolution Imaging Spectroradiometer) and IRI Maproom (map data). The second is secondary data (mosquitoes sampling) pulled from the OIE animal health information database for Europe, since no (open) human health database has the same level of detail in the geographic and epidemiological characteristics of outbreaks. With spatial and temporal information about 116 animal diseases in the world (available from 2005 and updated in real time), the OIE dataset is one of the largest database freely available to use (containing hundreds of thousands disease locations). However, the methods will be equally applicable to human diseases should suitable data become available.
The project will start by reviewing current methods, mainly applied to species co-occurrence, which often focus on bi-co-occurrence, and only implicitly consider interactions between species, and do not incorporate non-independence among hosts, and more importantly do not account for the biology inherent the disease transmission. Therefore, the project aims to develop a new statistical framework to model pathogens co-occurrence for transmission risk mapping by taking into account intra-relationships within pathogens and hosts, between pathogens and hosts and between pathogens, hosts and environment.
The initial method will be a bivariate Gaussian process model, which is a geostatistical model, as it will allow us to combine the secondary data which contains surveys on mosquitoes sampled at arbitrary locations and the spatial data. Furthermore, the objective is to predict, over a certain location, the prevalence of viruses in the mosquitoes (presence/absence/abundance). This would help us to find out if an association exists between co-infection in mosquitoes and their potential of disease transmission to humans.
The aim of this project is to create a platform capable of detecting the potential of infectivity in an area where co-infection of vector populations occurs and to develop a model that can determine co-occurrence attributed risk of infectious diseases. This could then lead to identification of hotspots or areas of high risk attributed to co-occurrence and inform policies.
The project will start by working with two distinct types of data. The first is remote sensing data (satellite data) from MODIS (Moderate Resolution Imaging Spectroradiometer) and IRI Maproom (map data). The second is secondary data (mosquitoes sampling) pulled from the OIE animal health information database for Europe, since no (open) human health database has the same level of detail in the geographic and epidemiological characteristics of outbreaks. With spatial and temporal information about 116 animal diseases in the world (available from 2005 and updated in real time), the OIE dataset is one of the largest database freely available to use (containing hundreds of thousands disease locations). However, the methods will be equally applicable to human diseases should suitable data become available.
The project will start by reviewing current methods, mainly applied to species co-occurrence, which often focus on bi-co-occurrence, and only implicitly consider interactions between species, and do not incorporate non-independence among hosts, and more importantly do not account for the biology inherent the disease transmission. Therefore, the project aims to develop a new statistical framework to model pathogens co-occurrence for transmission risk mapping by taking into account intra-relationships within pathogens and hosts, between pathogens and hosts and between pathogens, hosts and environment.
The initial method will be a bivariate Gaussian process model, which is a geostatistical model, as it will allow us to combine the secondary data which contains surveys on mosquitoes sampled at arbitrary locations and the spatial data. Furthermore, the objective is to predict, over a certain location, the prevalence of viruses in the mosquitoes (presence/absence/abundance). This would help us to find out if an association exists between co-infection in mosquitoes and their potential of disease transmission to humans.
| Title | Targeted spatial sampling design |
| Description | This study design targets randomly selected households across a study region. We extended the spatial inhibitory design with close pairs principle (Chipeta and Diggle, 2017) to allow sampling within sites with spatially heterogeneous populations. This addresses the practical problem of efficiently identifying households on the ground, reducing the number of locations sampled where in fact no household exists. Different implementations of the algorithm provide multiple ways to identify appropriate households in diverse areas. The resulting design methodology will be written up as a scientific paper, and published with software alongside my PhD thesis. The resulting data collection from this design methodology falls under the DRUM data sharing agreement (MR/S004793/1). |
| Type Of Material | Database/Collection of data |
| Year Produced | 2019 |
| Provided To Others? | No |
| Impact | Impact will be determined at the end of this PhD thesis. |
| Description | DRUM Consortium |
| Organisation | Liverpool School of Tropical Medicine |
| Country | United Kingdom |
| Sector | Academic/University |
| PI Contribution | The research team will use the data captured to develop an agent-based model which will improve our understanding of the drivers of AMR movement between humans, animals, and the environment in Uganda and Malawi. Identifying drivers of AMR transmission is predicated on an ability to understand acquisition of AMR bacteria. Agent-based models (ABMs) may be used to 'caricature' individuals in terms of physical characteristics, social interaction, and behaviour relevant to AMR acquisition and loss. An important feature of the ABM approach is that it allows us to admit qualitative data in a coherent fashion and consequently, to study how movement of specific AMR markers within the community may be affected by intervention strategies. To model AMR movement, we will superimpose a transmission model onto the ABM. Outcomes characterised by our model will be identified and discussed with the project field teams and incountry partners. Suitable intervention strategies will be chosen qualitatively and in collaboration with the DRUM partners to reflect cost-feasibility, cultural acceptance, and practical implementation. The model will be packaged, together with detailed documentation, as an application aimed at educational and operational use by public health clinicians and key decision makers. My contribution will focus on the sampling design of the study areas and the investigation of possible co-occurrence of malaria and AMR in these areas. It will also include looking at some of the water, sanitation and hygiene (WASH) data collected by our partners to investigate potential correlations between the WASH habits in the study areas in Uganda and Malawi and how to better include them in the agent-based model. |
| Collaborator Contribution | DRUM will transform our understanding of the drivers of AMR in Eastern Africa, and enable the design of interventions to mitigate antimicrobial resistance (AMR) spread by determining specific drivers of transmission. The DRUM consortium strives to address how human behaviour and antibacterial usage in urban and rural Africa leads to the transmission of AMR amongst E. coli and K. pneumoniae in humans, animals and the environment and influences the clinical impact of drug-resistant bloodstream infection (DR-BSI) in humans. Our vision is to establish Ugandan and Malawian sites as sustainable model settings for interdisciplinary study and mitigation of AMR by embedding a One Health strategy at the heart of a consortium that will generate outputs applicable to similar communities throughout East and Southern Africa and beyond. The multi-disciplinarity of this consortium will allow us to get a better understanding of what is now seen as one of the most serious global threats to human health in the 21st century. |
| Impact | This collaboration is multi-disciplinary, and the disciplines involved are : - Microbiology - Infectious disease surveillance - Sociology - Environmental health - Statistics and epidemiological modelling - Health economics - Policy |
| Start Year | 2018 |