Fig 1.
Flowchart of methodological framework.
Our overall approach consisted of three steps, depicted in the three columns. The products of all three combine to yield our estimate of the relative contribution to overall force of infection from each infection class. Likewise, uncertainty in each component of the analysis is propagated into other components of the analysis through repeated Monte Carlo sampling, with Monte Carlo samples from one portion of the analysis flowing to others according to the arrows and providing the basis for credible intervals reported in the Results section. Details of how the sub-steps indicated in this figure were performed are described in subsections of the Materials and Methods section.
Fig 2.
Definitions of DENV infection classes.
Fig 3.
DENV viremia and infectiousness trajectories by infection class.
(A,D,G,J): DENV viremia since time of infection for different infection classes and pre-exposure histories (1°: primary infection; 2°: secondary infection). Lighter lines denote 3,000 replicates and dark lines means. (B,E,H,K): Infectiousness of humans to mosquitoes over time. (C,F,I,L): Probability density of net infectiousness as defined in Eq (4) based on curves from the middle column. The solid blue line denotes the median and the dashed line denotes the median for the reference group (1° symptomatic). The solid and dashed red lines denote the mean and 95% confidence interval of the net infectiousness of 1° symptomatic infections as measured empirically [63].
Fig 4.
Pre-exposure history (A and C) and infection class (B and D) stratification by age and for FoI values of 0.01 (top) and 0.1 (bottom).
An individuals’ susceptibility to infection and clinical outcome depend on pre-exposure history. Serohistory by age (A and C) is estimated using a system of ordinary differential equations with state variables denoting the proportion of the population pre-exposed to 0–4 serotypes. Transition to pre-exposure state i occurs at rate iFoI. Individuals entering a new pre-exposure state have temporary heterologous immunity (gray) to all serotypes before later becoming susceptible again to each serotype to which they do not have a history of exposure. After four infections with four different serotypes, individuals are assumed fully immune (black) to all serotypes.
Fig 5.
Mean contribution of each infection class to total force of infection (FoI).
The contribution to the total FoI of an infection class is derived from the ratio of FoI attributable to a given class and total FoI, as in Eq (1). The respective net infectiousness is derived from the 3,000 random samples displayed in Fig 3. The infections are further distributed according to the estimated proportion of net infectiousness to occur before and after symptom onset (pre-symptomatic (Eq (5)) and post-symptomatic (hatched lines) (Eq (6)). The histogram shows the distribution of FoI contributions by asymptomatic infections at FoI = 0.1, accounting for parameter uncertainty.
Fig 6.
Contribution to the total force of infection (FoI) of inapparent infections (As+IS) for different As:IS:AS ratios (FoI = 0.05).
Darker blue colors indicate a larger contribution to the FoI by As+IS infections in the population. The coordinates reflect the proportions per infection class with (0.09, 0.64,0.27) reflecting a default based on [16]. The white lines indicate the expected proportion of inapparent contribution if all types of infections had an equal net infectiousness.