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

In this work, we present model-based inference methods that allow us to infer R, the 144 spillover rate, and properties of spatial spread among humans from surveillance datasets with 145 non-localized spatial information and an unknown total number of surveilled localities. Our 146 approach builds on methods introduced by White and Pagano [33] and Kucharski et al. [34] , but 147 allows continuous spillover throughout the surveillance period and makes use of available spatial 148 information on case location. While the method could be readily adjusted to incorporate more 149 than with individuals from other localities, and therefore that infection is more likely to be 158 transmitted within a locality. However, the total number of localities under surveillance is 159 unknown because only localities with one or more cases appear in the dataset (the 'unknown 160 denominator' problem discussed above). We refer to the second spatial level as the 'broader 161 contact zone.' It describes a collection of localities that all occur within the same administrative 162 unit and likely share some amount of human movement. When multiple types of administrative 163 units of different sizes are reported in the dataset (e.g., districts, regions, provinces, etc.), the 164 ideal choice for broader contact zone is the smallest administrative unit that contains inter-165 locality human-to-human transmission events. If this scale is not known a priori, inferring the 166 appropriate scale of administrative unit is necessary.