Quantifying transmission of emerging zoonoses: Using mathematical models to maximize the value of surveillance data

Monique R. Ambrose; Adam J. Kucharski; Pierre Formenty; Jean-Jacques Muyembe-Tamfum; Anne W. Rimoin; James O. Lloyd-Smith

Because an estimate of the true number of localities under surveillance would help 293 determine the size of the population that could be detected for a given system, we assessed how 294 well we could approximate this value. Given the number of localities with one or more cases and 295 the mean parameter estimates, it is possible to calculate the expected total number of localities 296 under surveillance (see S1 Text). Estimates of the true number of localities calculated for the 297 simulated datasets center on the correct value (S2 Fig). The magnitude of estimate error is driven 298 by the spillover rate, which largely determines the proportion of localities that are observed by 299 surveillance. The mean percent error across simulations with spillover rate of 0.0001, 0.00036, 300 and 0.0007 were 25.4%, 7.9%, and 2.4%, respectively, while simulations with spillover rates of 301 0.004 and above almost always recorded at least one case in each locality during the five year 302 surveillance period and therefore tended to estimate the exact true number of localities. 303