Fig. 1.

Model fit to carriage prevalence and IPD incidence (a) and prior and posterior parameter estimates (b). We assumed that serotyping methods would only pick up the predominant serotype and that in case of co-colonisation this was always the vaccine serotype. Points with 95 ` confidence bounds represent data, and lines with ribbons represent median model estimates with 95% credible intervals. In b the grey line indicates the prior density distribution and the bars the posterior sample