The goal of this research is to build upon the available literature on statistical methods in spatial epidemiology to devise a space-time disease mapping approach to evaluate disease control interventions addressing two main issues. First, the approach will be flexible enough to accommodate both longitudinal and cross-sectional data. Second, the approach needs to allow for the incorporation of a multitude of covariates to control for confounding. Specifically, we propose a novel spatio-temporal Bayesian hierarchical modeling approach where both individual and cross-sectional data are analyzed, considering alternative ways to address missing data. Our use of Bayesian Hierarchical models for the longitudinal data with spatial and non-spatial contextual effects allows us to include spatially contiguous incidence estimates, and help to reduce confounding effects on the individual infection status. In this model, the probability of malaria infection is modeled via a linear predictor that is a function of covariates (including intervention terms, as well as social, behavior, economic, and ecological aspects), and random effects. We will test and validate the proposed methodological approach using data collected during the Urban Malaria Control Program (UMCP) in Dar es Salaam, Tanzania. We chose the UMCP because the program has assembled a good quality database since its launch in March 2004. The UMCP covers 15 of the 73 wards of the city, encompassing a total area of 56 km2 and more than 610,000 residents. The program started with baseline data collection in 2004, in order to guide interventions and facilitate future program evaluation. A total of 64,537 individuals were interviewed between 2004 and 2008, and each was tested for a malaria infection through microscopy. Regarding interventions for malaria control, since March 2006, the use of microbial larvicides was introduced in three wards, expanded to nine wards in May 2007, and to all 15 wards in April 2008. Concurrent urban malaria control efforts included a pilot community-based environmental management intervention undertook in two drains in the city in 2007-8, and the introduction of Artemisinin-based combination therapy as first line drug for the treatment of malaria treatment in January 2007. We expect that upon the successful completion of this project we will deliver two outcomes. First, from a methodological point of view, we will have a statistical approach that could be used in a range of applications that have spatial and temporal components, and that combine different types of data. Second, from a public policy point of view, we will produce a comprehensive evaluation of the UMCP, and identify gaps in the program that could be improved, areas that impose challenges for vector control, and recommendations for program tuning and scaling-up. The first outcome will expand the current literature on spatial epidemiology, and facilitate the use of spatial statistics in a variety of applications (public health being just one of them). The second outcome will directly impact current activities of the UMCP, and ultimately provide evidence for other cities considering control programs similar to the Tanzanian effort.
The proposed study will build on the current literature of spatio-temporal disease mapping, considering an individual level longitudinal model for infection status, and then build on that with a linked joint cross-sectional model. It has the potential to allow a variety of applications in studies that combine individual and aggregated data, collected longitudinally but also in multiple cross-sectional surveys. An application of the model will also produce a comprehensive evaluation of the Urban Malaria Control Program in Dar es Salaam, Tanzania, and identify gaps that could be improved, areas that impose challenges for control, and recommendations for program tuning and scaling-up.
|Maheu-Giroux, Mathieu; Castro, Marcia C (2013) Impact of community-based larviciding on the prevalence of malaria infection in Dar es Salaam, Tanzania. PLoS One 8:e71638|