Fig. 3

Top panel: Results of the ES. Red line indicates a positive finding of WPV1 in one of the four relevant sites (see ‘Methods’). Green line indicates no positive findings. Data shown here is up to the end of April 2014, after which there were no positive findings of WPV1 in any sewage sample. Bottom panel: The fit of the model to the stool samples data. The grey area marks the estimated 95 % credible interval of WPV1 prevalence in the modeled population of Bedouin children. The yellow area within the grey area presents a more restricted estimated range of WPV1 prevalence using parameter values whose log-likelihood is within 2 log-likelihood units of the best fit (a commonly used threshold for selection of the more probable fits to the data [34]). The blue x marks the proportion of stool samples positive for WPV1 in each of the days that samples were collected. The magenta dots present a weekly smoothing of the sampled data. For a description of the smoothing and the confidence intervals related to the stochasticity of the observation of the stool data see Figure S7 in Additional file 1. While the likelihood was calculated using the non-smoothed stool sample data, the smoothed data captures the trend of the estimated prevalence better, as it blends in the effect of days with zero positive samples, of which there were many in the second stool survey due to the low number of samples taken each day. The red lines show the cumulative vaccine coverage (right y-axis) of the first (solid line) and second (dashed line) OPV doses in the modeled population