Estimating SARS-CoV-2 seroprevalence and epidemiological parameters with uncertainty from serological surveys

Daniel B Larremore; Bailey K Fosdick; Kate M Bubar; Sam Zhang; Stephen M Kissler; C. Jessica E. Metcalf; Caroline Buckee; Yonatan Grad

Age-structured simulations and inference. For simulated sampling and inference (Fig. 3) , n = {n i } serological samples were allocated to subpopulations with heterogeneous seroprevalence values θ (Supplementary Table S2) , shifted upward or downward to achieve the targeted overall seroprevalence. Simulated test outcomes included false positive and negative results as dictated by the test being modeled (see Supplementary Table S1 ). Test allocations {n i } were done in proportion to age demographics of blood donations, delivering mothers, uniformly across subpopulations, or according to a variance reduction strategy, MDI; see below. Given per-subpopulation test outcomes, 1, 000 or more samples were drawn from the posterior distribution Eq. (2) using MCMC (Supplementary Materials). Posterior distributions of overall seroprevalence were produced by a demographically weighted average of age-specific seroprevalence samples. Posterior distributions of R eff were produced by using samples of age-specific seroprevalences in the age-structured model, described below. For both overall seroprevalence and R eff , 90% equal-tailed credible intervals were recorded. Average posterior 90% CI widths were calculated using 250 technical replicates per pixel/point ( CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.