Estimation of Covid-19 Prevalence Dynamics from Pooled Data

Abstract

Estimating the prevalence of a disease, such as COVID-19, is necessary for evaluating and mitigating risks of its transmission. Estimates that consider how prevalence changes with time provide more information about these risks but are difficult to obtain due to the necessary survey intensity and commensurate testing costs. Motivated by a dataset on COVID-19, from the University of Notre Dame, we propose pooling and jointly testing multiple samples to reduce testing costs. A nonparametric, hierarchical Bayesian model is used to infer population prevalence from the pooled test results without needing to retest individuals from pools that test positive. This approach is shown to reduce uncertainty compared to individual testing at the same budget and to produce similar estimates compared to individual testing at a much higher budget through simulation studies and an analysis of COVID-19 infections at Notre Dame.

Reconstructing prevalence dynamics of wildlife pathogens from pooled and individual samples

“…Parameter π can then be used to model Z ∼ Bernoulli ( π ), where Z is a binary observed variable indicating whether or not the pooled sample is positive. This implementation has, for example, been proposed as a way to model prevalence dynamics over time for SARS-CoV-2, in combination with individual data [19]. This approach has two key limitations however.…”

“…Here, we used a relatively simple Gaussian Process (GP) smoothing function, which uses a Gaussian kernel to model prevalence over time. This approach was based on the one used in [19].…”

“…Most studies on the analysis of pooled samples focus on testing protocols for cost reduction, with the goal of eventually identifying the positive individuals [3, 14, 15]. Perhaps for this reason, few methods have been developed for explicitly using pooled samples to estimate prevalence in the population [10, 12, 16–19], and even fewer have attempted to use the actual concentration of the infectious agent in the pooled sample to estimate how many of the contributing individuals are positive [12, 13]. Furthermore, to our knowledge no methods exist that are able to estimate the proportion of positive individuals using concentration when the underlying distribution of test values does not follow a standard-family (e.g.…”

“…The approach presented here is particularly useful when the goal is not to identify which specific individuals are positive but to determine prevalence in the population, because there is no need to re-test depooled samples. Examples include monitoring SARS-CoV-2 prevalence [12, 19], estimating prevalence in wastewater if the number of contributing individuals can be estimated [5], assessing pathogen prevalence in the animal production industry [22], or estimating pathogen prevalence in wildlife populations [23]. Note that while the example presented here focuses on infection prevalence, the model can also be applied to other biomarkers such as antibodies.…”