Background: A measles outbreak involving 60 cases occurred in Yamagata, Japan in 2017. Using two different mathematical models for different datasets, we aimed to estimate measles transmissibility over time and explore any heterogeneous transmission patterns.
Methods: The first model relied on the temporal distribution for date of illness onset for cases, and a generation-dependent model was applied to the data. Another model focused on the transmission network. Using the illness-onset date along with the serial interval and geographical location of exposure, we reconstructed a transmission network with 19 unknown links. We then compared the number of secondary transmissions with and without clinical symptoms or laboratory findings.
Results: Using a generation-dependent model (assuming three generations other than the index case), the reproduction number (R) over generations 0, 1, and 2 were 25.3, 1.3, and <0.1, respectively, explicitly yielding the transmissibility over each generation. The network data enabled us to demonstrate that both the mean and the variance for the number of secondary transmissions per primary case declined over time. Comparing primary cases with and without secondary transmission, high viral shedding was the only significant determinant (P < 0.01).
Conclusions: The R declined abruptly over subsequent generations. Use of network data revealed the distribution of the number of secondary transmissions per primary case and also allowed us to identify possible secondary transmission risk factors. High viral shedding from the throat mucosa was identified as a potential predictor of secondary transmission.
Keywords: Paramyxoviridae; epidemic; mathematical model; statistical estimation; transmissibility.