Fig. 3

Graphs of infection and test prevalence estimates from 398 pupils in 8 schools. Separate models were run with CCA trace considered as negative (left) and positive (right), both from the original models (top) and models where the specificity of CAA was fixed to 100% (bottom). Infection prevalence estimates were calculated as a weighted average of the prevalence estimate in each school outputted from the Bayesian LCA (weighted for number of pupils sampled in each school). The distribution of estimated test prevalence was obtained by a weighted average of the number of children infected in each school, simulated by drawing from a Binomial distribution with n equal to the number of children tested in each school and p equal to the proportion of children positive by the focal test