Group Testing Large Populations for SARS-CoV-2 (preprint)

Group testing, the testing paradigm which combines multiple samples within a single test, was introduced in 1943 by Robert Dorfman. Since its original proposal for syphilis screening, group testing has been applied in domains such as fault identification in electrical and computer networks, machine learning, data mining, and cryptography. TheSARS-CoV-2 pandemic has led to proposals for using group testing in its original context of identifying infected individuals in a population with few tests. Studies suggest that non-adaptive group testing - in which all the tests are determined in advance - for SARS-CoV-2could help save 20% to 90% of tests depending on the prevalence. However, no systematic approach for comparing different non-adaptive group testing strategies currently exists. In this paper we develop a software platform for evaluating non-adaptive group testing strategies in both a noiseless setting and in the presence of realistic noise sources, modelled on published experimental observations, which makes them applicable to polymerase chain reaction (PCR) tests, the dominant type of tests for SARS-CoV-2. This modular platform can be used with a variety of group testing designs and decoding algorithms. We use it to evaluate the performance of near-doubly-regular designs and a decoding algorithm based on an integer linear programming formulation, both of which are known to be optimal in some regimes. We find savings between 40% and 91% of tests for prevalences up to 10% when a small error (below 5%) is allowed. We also find that the performance degrades gracefully with noise. We expect our modular, user-friendly, publicly available platform to facilitate empirical research into non-adaptive group testing for SARS-CoV-2.

Understanding the effectiveness of government interventions in Europe’s second wave of COVID-19 (preprint)

As European governments face resurging waves of COVID-19, non-pharmaceutical interventions (NPIs) continue to be the primary tool for infection control. However, updated estimates of their relative effectiveness have been absent for Europe’s second wave, largely due to a lack of collated data that considers the increased subnational variation and diversity of NPIs. We collect the largest dataset of NPI implementation dates in Europe, spanning 114 subnational areas in 7 countries, with a systematic categorisation of interventions tailored to the second wave. Using a hierarchical Bayesian transmission model, we estimate the effectiveness of 17 NPIs from local case and death data. We manually validate the data, address limitations in modelling from previous studies, and extensively test the robustness of our estimates. The combined effect of all NPIs was smaller relative to estimates from the first half of 2020, indicating the strong influence of safety measures and individual protective behaviours--such as distancing--that persisted after the first wave. Closing specific businesses was highly effective. Gathering restrictions were highly effective but only for the strictest limits. We find smaller effects for closing educational institutions compared to the first wave, suggesting that safer operation of schools was possible with a set of stringent safety measures including testing and tracing, preventing mixing, and smaller classes. These results underscore that effectiveness estimates from the early stage of an epidemic are measured relative to pre-pandemic behaviour. Updated estimates are required to inform policy in an ongoing pandemic.

How Robust are the Estimated Effects of Nonpharmaceutical Interventions against COVID-19? (preprint)

To what extent are effectiveness estimates of nonpharmaceutical interventions (NPIs) against COVID-19 influenced by the assumptions our models make? To answer this question, we investigate 2 state-of-the-art NPI effectiveness models and propose 6 variants that make different structural assumptions. In particular, we investigate how well NPI effectiveness estimates generalise to unseen countries, and their sensitivity to unobserved factors. Models that account for noise in disease transmission compare favourably. We further evaluate how robust estimates are to different choices of epidemiological parameters and data. Focusing on models that assume transmission noise, we find that previously published results are remarkably robust across these variables. Finally, we mathematically ground the interpretation of NPI effectiveness estimates when certain common assumptions do not hold.

The effectiveness and perceived burden of nonpharmaceutical interventions against COVID-19 transmission: a modelling study with 41 countries (preprint)

Background: Governments are attempting to control the COVID-19 pandemic with nonpharmaceutical interventions (NPIs). However, it is still largely unknown how effective different NPIs are at reducing transmission. Data-driven studies can estimate the effectiveness of NPIs while minimizing assumptions, but existing analyses lack sufficient data and validation to robustly distinguish the effects of individual NPIs. Methods: We collect chronological data on NPIs in 41 countries between January and May 2020, using independent double entry by researchers to ensure high data quality. We estimate NPI effectiveness with a Bayesian hierarchical model, by linking NPI implementation dates to national case and death counts. To our knowledge, this is the largest and most thoroughly validated data-driven study of NPI effectiveness to date. Results: We model each NPI's effect as a multiplicative (percentage) reduction in the reproduction number R. We estimate the mean reduction in R across the countries in our data for eight NPIs: mandating mask-wearing in (some) public spaces (2%; 95% CI: -14%-16%), limiting gatherings to 1000 people or less (2%; -20%-22%), to 100 people or less (21%; 1%-39%), to 10 people or less (36%; 16%-53%), closing some high-risk businesses (31%; 13%-46%), closing most nonessential businesses (40%; 22%-55%), closing schools and universities (39%; 21%-55%), and issuing stay-at-home orders (18%; 4%-31%). These results are supported by extensive empirical validation, including 15 sensitivity analyses. Conclusions: Our results suggest that, by implementing effective NPIs, many countries can reduce R below 1 without issuing a stay-at-home order. We find a surprisingly large role for school and university closures in reducing COVID-19 transmission, a contribution to the ongoing debate about the relevance of asymptomatic carriers in disease spread. Banning gatherings and closing high-risk businesses can be highly effective in reducing transmission, but closing most businesses only has limited additional benefit.