Associations were adjusted for the matching variable (age at selection), as well as selected demographic variables known to be associated with the risk of treated cholera in Matlab, including distance from residence to the Dhonagoda River and distance from residence to hospital. LOESS (locally estimated scatterplot smoothing), a non-parametric technique that uses local weighted regression to fit a smooth curve through points in a scatter plot, is described in the Methods. Smooth=0.9 (A) and 0.8 (B), which are optimally chosen values for the smoothing parameter, which ranges from 0.1 to 0.9 using the bias-corrected Akaike information criterion. The figure presents the fitted localised regression curve using 90% (A) and 80% (B) neighbouring data points. Proportionality test for waning of protection as described in the Methods.
