Understanding the spatial distribution of trichiasis and its association with trachomatous inflammation—follicular

Flueckiger, Rebecca Mann and Giorgi, Emanuele and Cano, Jorge and Abdala, Mariamo and Amiel, Olga Nelson and Baayenda, Gilbert and Bakhtiari, Ana and Batcho, Wilfrid and Bennawi, Kamal Hashim and Dejene, Michael and Elshafie, Balgesa Elkheir and Elvis, Aba Ange and Francois, Missamou and Goepogui, Andre and Kalua, Khumbo and Kebede, Biruck and Kiflu, Genet and Masika, Michael P. and Massangaie, Marilia and Mpyet, Caleb and Ndjemba, Jean and Ngondi, Jeremiah M. and Olobio, Nicholas and Turyaguma, Patrick and Willis, Rebecca and Yeo, Souleymane and Solomon, Anthony W. and Pullan, Rachel L. (2019) Understanding the spatial distribution of trichiasis and its association with trachomatous inflammation—follicular. BMC Infectious Diseases, 19. ISSN 1471-2334

Abstract

Background Whilst previous work has identified clustering of the active trachoma sign “trachomatous inflammation—follicular” (TF), there is limited understanding of the spatial structure of trachomatous trichiasis (TT), the rarer, end-stage, blinding form of disease. Here we use community-level TF prevalence, information on access to water and sanitation, and large-scale environmental and socio-economic indicators to model the spatial variation in community-level TT prevalence in Benin, Cote d’Ivoire, DRC, Guinea, Ethiopia, Malawi, Mozambique, Nigeria, Sudan and Uganda. Methods We fit binomial mixed models, with community-level random effects, separately for each country. In countries where spatial correlation was detected through a semi-variogram diagnostic check we then fitted a geostatistical model to the TT prevalence data including TF prevalence as an explanatory variable. Results The estimated regression relationship between community-level TF and TT was significant in eight countries. We estimate that a 10% increase in community-level TF prevalence leads to an increase in the odds for TT ranging from 20 to 86% when accounting for additional covariates. Conclusion We find evidence of an association between TF and TT in some parts of Africa. However, our results also suggest the presence of additional, country-specific, spatial risk factors which modulate the variation in TT risk.