Fig 1.
Blood sample distribution across the simulated scenarios for the city of Accra using 5-year age groups.
Panel A shows the distribution of the available samples across the age-groups and compares them with the population age structure of Ghana (World Population Prospect [21]). Panels B to F illustrate the sample distribution under scenarios A to E compared to the available samples in the baseline scenario 0 (which in the panel is indicated as “samples”).
Table 1.
Summary of accuracy metrics obtained for Ghana with 5-year age groups under the optimal scenarios across models and assumptions.
Fig 2.
Summary of the accuracy metrics obtained for Accra, Kumasi and Tamale with 5-year age category across scenarios.
The four columns represent respectively the bias (in percentage), coverage (in percentage), number of tested samples and uncertainty obtained for each scenario. The scenario highlighted in green indicates the selected scenario. The median bias and uncertainty are reported with their 95% CrI (columns 1 and 4), while the median coverage is reported with its 95% exact binomial CI. The orange line represents the median (columns 1, 2 and 4) and the orange ribbon represents the 95% CrI of the baseline scenario 0 (columns 2 and 4). The pink ribbon in the first column represents the 15% tolerance around the upper bound of the 95% CrI of scenario 0, which was used in the first step of the selection criterion.
Fig 3.
Model fit for the three cities under the 5-year age categorization with model 1.
The first row represents the fit for scenario 0, where the median and 95% CrI of the estimated seroprevalence are reported in orange as a line and shaded area, respectively. The second row illustrates the model fit with the selected scenario in green. In both panels, the datasets used for model fitting are shown in black, with the error bars showing the exact binomial 95% Confidence Interval (CI). The third row shows a comparison of the estimated seroprevalence in scenario 0 (orange) and the optimal scenario (green).
Fig 4.
Model fit for Accra under the 5-year age categorization and a high and low FOI.
The first row represents the seroprevalence using the chosen scenario, where the median and 95% credible interval of the estimated seroprevalence are reported in green as a line and shaded area, while the simulated dataset is reported in black, with the error bars indicating the exact binomial 95% CI. The second row illustrates the age-distributions of the tests with the chosen scenario (in green) for each city, compared to the available sample distribution in orange. The figure reports the results obtained with the high FOI (first column) and low FOI (second column).
Fig 5.
Conceptual description of the workflow for the implementation of serological surveys in the absence of previous information (such as serosurvey or case-notification data) on the FOI.
Illustration of the two-step process proposed to identify the sample sizes and optimize their age-distribution when the study is conducted in locations with little or no prior information on dengue FOI.