A major challenge facing mass drug administration (MDA) programmes targeting the control and elimination of lymphatic filariasis or onchocerciasis is the serious adverse – sometimes life threatening – reactions that this treatment can have on people who are also infected with loisasis. Commonly known as Eye worm, loisasis is caused by the parasitic worm Loa loa and is mainly found in forested regions in Africa and India. Those individuals with high levels of co-infection are the most likely to experience adverse reactions to ivermectin, the deworming drug used to target lymphatic filariasis and onchocerciasis.
The risk of severe adverse reactions is such that MDA isn’t recommended by the WHO in onchocerciasis-endemic communities that are likely to have a large number of highly infected individuals with loiasis. It is very important, therefore, to assess the level of loiasis endemicity in a community before initiating mass treatment.
To better quantify the burden of Loa loa infections in Western Africa, LASER‘s and CHICAS’s Emanuele Giorgi, together with colleagues from Lancaster University, have developed a new methodology to analyse the prevalence and intensity of infections in communities (measured by microfilia load per ml blood) alongside their spatial distribution.
This approach is highlighted in a recently published paper “Giorgi E, Schlüter DK, Diggle PJ. Bivariate geostatistical modelling of the relationship between Loa loa prevalence and intensity of infection. Environmetrics. 2017;e2447”.
One feature of the data analysed in the paper, is the high number 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 neglected tropical diseases, whose endemic boundaries are often unknown, thus leading to the inclusion of disease-free communities in the sampling frame.
The paper introduces 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. It shows through a simulation study that the spatial model leads to more precise predictions of the proportion of individuals in a village with more than 8000 microfilia per ml of blood, than previous non-spatial approaches and accordingly provides a geostatistical reanalysis of the Loa loa data.
This work shows how geostatistical methods allow public health researchers to make the best possible use of data on disease rates by exploiting their spatial correlation (“close things in space are more related than distant things”). This has the potential to improve the effectiveness of control programs by targeting more precisely communities that are likely to include individuals at high risk of experiencing a serious adverse reaction to ivermectin.