Multiscale model for forecasting Sabin 2 vaccine virus household and community transmission

Fig 1. Model structure. A) Hierarchical demographic structure used in the model. Infected individuals are grouped into households, which are grouped into baris, which are grouped into villages that together define the Matlab Region. Map borders were obtained and used with permission from the International Centre for Diarrhoeal Disease Research, Bangladesh (icddr, b) (https: //github.com/InstituteforDiseaseModeling/community-structure-mediates-polio-transmission/tree/main/PopSim/data/shapefiles). B) Household evolution model. Households are represented as pedigree trees and their composition changes as individuals are born and removed either through death or marriage. The model is parameterized by the age- and sex-specific fertility, mortality, marriage of individuals and the rate with which newlywed couples move out after marriage (S1 Text). C) Poliovirus infection model. On infection, infected individuals receive a boost in OPV-equivalent antibody titer and assigned a shedding duration based on their titer prior to infection. Individual antibody titers are dynamic and declines over time. Viral shedding is measured in cell culture infectivity dose 50 (CCID50) units per fecal oral dose…

Fig 2. Simulating household structure. Household demography simulations (orange) compared against demography data from the 2014 BDHSS (blue). Error bars for simulation output represent middle 95 percentile as estimated from 20 random iterations. Error bars for the BDHSS or clinical trial data represent the 95% binomial confidence interval around the mean. A) Household size distribution. B) Housed contact age distributions. C) Age pyramid D) Village size. For D, only simulated confidence intervals are shown.

Fig 3. Modeled vs observed patterns of Sabin 2 following the mOPV2 vaccination campaign. A-C) Sabin 2 shedding in the eight infant and household contact cohorts after the mOPV2 challenge. Only data from the villages assigned with bOPV routine immunization was examined. The cohorts in A (left to right: cohorts one, four, seven) reflect direct vaccination with mOPV2. The cohorts in B (left to right: cohorts two, six, five) reflect transmission that most likely originated from a household or bari member. The cohorts in C transmission (left to right: cohorts three and eight) reflect transmission that most likely originated from a village or non-village member. The data and the 95% binomial confidence interval are shown in black. Simulated results with the multiscale (orange) and the two mass action models (purple and green) are shown. The multiscale model was calibrated to all the clinical trial data (the cohort shedding data and the tOPV data). In purple are the results from the mass action model after calibrating to the cohort shedding data alone, and in green after calibrating to all the clinical trial data. The solid line represents the simulated average and the shading indicates two…

Fig 4. Extrapolating Sabin 2 vaccine virus transmission risk. A) Transmission persistence following a Sabin 2 point importation event before (year 0), one, and five years after the Switch from 2000 simulations. Simulated results with the multiscale (orange) and mass action (purple) models are shown. This version of the mass action model was calibrated to the cohort shedding data alone (Fig 3, purple). Lasagna plot of showing the total persistence times and number of infected individuals following a point importation five years after the Switch using the multiscale (B) and mass action (C) models. Each line in the lasagna plot is a different simulation and 2000 iterations are shown. The length of each line indicates the persistence time and the color the number of shedding individuals at that timepoint. D) Immune profile of the first 300 days after a point importation using the multiscale (orange) and mass action (purple) model calibrated without the tOPV data during a severe outbreak. Severe outbreaks were defined as having a transmission duration of at least 300 days. Susceptibility was defined as having a OPV-equivalent antibody titer < 8. The increase in susceptibility over time is…

Fig 5. Simulating single-dose mOPV2 campaigns. The proportion of susceptible individuals with a serum neutralizing antibody titer under eight in A) children under five and B) the entire Matlab population during a single-dose mOPV2 campaign performed five years after the Switch. Simulation results using the multiscale model are shown in blue while those using the mass action model (calibrated without the tOPV) are shown in grey. C) Epi curves showing the number of infected individuals following the mOPV2 campaign before (orange) and five years after (blue) the Switch. For A-C), the darkness of the shading indicates the vaccination coverage in children under five. From darkest to lightest, the coverages were 80%, 40%, 20%, and 10%. 40% was the target coverage in the original mOPV2 campaign. Plotted are the mean expectations and the 95% prediction intervals obtained from 2000 simulation replicates. D) Violin plots showing the distribution of vaccine virus transmission duration before (orange) and five years after (blue) the Switch from 2000 simulation iterations. E) The expected shedding duration before (orange) and five years after (blue) the Switch. The solid line indicates the mean and…

Fig 6. Simulating multiple, single-dose mOPV2 campaigns five years after the Switch. The proportion of susceptible individuals with a serum neutralizing antibody titer under eight in A) children under five and B) the entire Matlab population following a 40% (light red) and 80% (dark red) coverage campaign. C) Epi curves showing the number of infected individuals following the mOPV2 campaign following an 80% (darkest), 40%, 20%, and 10% (lightest) coverage campaign. D) Violin plots of the total shedding durations following the second campaign. The colors used in C are the same as that for D. E) The expected shedding duration after the first (blue) and second (red) mOPV2 campaign. Both campaigns had 80% coverage in children under five. The solid line indicates the mean and the shading the bootstrapped 95% confidence interval of the mean.

Fig 7. Extrapolating cVDPV2 vaccine virus transmission risk. A) Transmission persistence following a Sabin 2 point importation event before (year 0, light purple) and five years after the Switch (dark purple) from 2000 simulations. B) Lasagna plot of showing the total persistence times and number of infected individuals following a cVDPV2 point importation five years after the Switch using the multiscale model.