Despite the successes of the Millennium Development Goals era, infectious diseases still remain a major threat to humankind with severe burden of mor- bidity and mortality especially in low and middle income countries (LMICs). Availability of accurate diagnosis constitutes an essential component in activities to combat these diseases such as within programs evaluating the effectiveness of interventions including verification of pathogens, elimination and detecting infections with markers of drug resistance. Molecular assays constitute the gold standard for the diagnosis of several infectious diseases but the lack of sufficient funds, suitably trained staff and laboratory supplies still hinders their use in LMICs. Schistosomiasis, is a long-lived, chronic, and highly debilitating tropical infectious disease caused by schistosome parasites, encountering some of these same challenges.
The aims of the current article are threefold. First, to study the effect that violations of model assumptions have on estimated parameters such as the prevalence of a disease, specificity and sensitivity of diagnostic tests. Second, to discuss extensions of the classical latent class model to account for local dependencies and zero-inflation (large non-pathological group). Third, to provide methodological guidance on the available modelling approaches for estimating test accuracy and disease prevalence in the absolute or partial absence of a gold standard for tropical infectious diseases using the paradigm of schistosomiasis by cautioning the practitioner not to blindly apply methods for estimating diagnostic error without a gold standard.