Authors
M N Clements3; C Donnelly1; A Fenwick4; N Kabatereine4; A Miete2; E Muheki Tukahebwa6; E N'Goran5; Y Nalule4; S Nogaro4; F Fleming4; 1 Department of Infectious Disease Epidemiology, Imperial College London; 2 Ivory Coast Ministry of Health, National Program Against Filariasis, Schistosomiasis and Geohelminths; 3 Schistosomiasis Control Initiative; 4 Schistosomiasis Control Initiative, Imperial College London; 5 Unité de Formation et de Recherche Biosciences, Université Félix Houphouët-Boigny; 6 Vector Control Division, Ministry of Health Uganda Discussion
Performance of a new diagnostic can be assessed by calculating the sensitivity and specificity of the test. However, determining the sensitivity and specificity without a ‘gold standard’ is complicated as the true disease status of an individual is unknown, and latent class analysis (LCA) has been developed in order to address this issue. LCA categorises test results as either ‘negative’ or ‘positive’ and there is currently no method to analyse ambiguous results. The current most commonly used diagnostic for Schistosoma mansoni is the Kato-Katz method where stool samples are examined for the presence of eggs. Kato-Katz is known to be highly specific, as an egg is recognisable, but not very sensitive, as the method can fail to detect infection when it is present. A new test, CCA, which detects the presence of circulating cathodic antigen in urine has been developed and a number of studies have shown that is more sensitive that Kato-Katz. However, there is an ambiguous ‘trace’ result falling between ‘positive’ and ‘negative’ and previous analyses have elected to consider trace as either positive or negative with no clear consensus evident in the literature. We present a method for extending LCA to include ambiguous results and apply this method to studies from Uganda and Cote d’Ivoire. We compare the diagnostic performance in each country discuss implications of the results. 
