Giorgi, Emanuele and Schlüter, Daniela K and Diggle, Peter John (2017) Bivariate geostatistical modelling of the relationship between Loa loa prevalence and intensity of infection. Environmetrics. ISSN 1180-4009
loaloa_paper_GDS_v3.pdf - Accepted Version
Available under License Creative Commons Attribution-NonCommercial.
Download (14MB)
Abstract
Loiasis is a neglected tropical disease (NTD) caused by the parasitic roundworm Loa loa. A challenge faced by current multinational programmes to control two other diseases, namely, lymphatic filariasis and onchocerciasis, by mass administration of prophylactic medication to at-risk communities is that individuals highly coinfected with Loa loa are at risk of developing serious adverse reactions to the medication. For this reason, understanding the geographical distribution of Loa loa prevalence and the distribution of microfilarial loads in communities has become of crucial importance. In this paper, we develop methodology to analyse data on microfilariae counts per millilitre of blood whilst allowing for spatial correlation. One feature of the data is the excess of zero counts, which makes the use of standard geostatistical methods for prevalence data inappropriate. This phenomenon, also known as zero inflation, is typical of count data from NTDs, whose endemic boundaries are often unknown, thus leading to the inclusion of disease-free communities in the sampling frame. We introduce a bivariate geostatistical model in order to study the relationship between the distributions of prevalence and intensity of Loa loa infections at the community level. We show through a simulation study that the spatial model leads to more precise spatial predictions than the nonspatial approach used by Schlüter et al. (2016), and accordingly provide a geostatistical reanalysis of the Loa loa data.