For an overview of the BDI pathogens group's work on sustainable containment of COVID-19, including publications, please see our website. This repository contains detailed reports and preprints.

The epidemiological impact of the NHS COVID-19 App

Chris Wymant^1*, Luca Ferretti^1*, Daphne Tsallis², Marcos Charalambides³, Lucie Abeler-Dörner¹, David Bonsall¹, Robert Hinch¹, Michelle Kendall^1,4, Luke Milsom⁵, Matthew Ayres³, Chris Holmes^1,3,6, Mark Briers³, Christophe Fraser¹

Correspondence to christophe.fraser@bdi.ox.ac.uk
^*equal contribution
¹ Big Data Institute, University of Oxford, UK
² Zuehlke Engineering Ltd., London, UK
³ The Alan Turing Institute, London, UK
⁴ Department of Statistics, University of Warwick, Coventry, UK
⁵ Department of Economics, University of Oxford, Oxford, UK
⁶ Department of Statistics, University of Oxford, Oxford, UK

Abstract

Since its launch on 24 September 2020, the NHS COVID-19 app has been downloaded to over 21 million phones, and used regularly by approximately 16.5 million users in England and Wales, which is 49% of the eligible population with compatible phones, and 28% of the total population. The main epidemiological impact of the app to date has been from the exposure notification function, which automates contact tracing from confirmed cases based on digital measurements of proximity events between phones. The app sent 1.7 million exposure notifications: 3.2 per index case, or 4.4 per index case who consented to be contact traced. We estimated that 6.1% of app-notified individuals subsequently tested positive (the secondary attack rate, SAR), comparable to the SAR for manual contact tracing (7.3% for close contacts and 13.5% for direct contacts). We estimated infected cases averted by the app in October-December 2020 using two conceptually complementary approaches. Modelling based on the observed notifications and SAR yielded 284,000 (224,000-344,000) averted cases, whilst statistical comparison of matched neighbouring local authorities yielded 594,000 (317,000-914,000) averted cases, i.e. about one case averted for each case consenting to notification of their contacts through the app. Improvements in the app notification system gave increased epidemiological effectiveness of the app, further supporting evidence for an effect of the app. We estimated that for every 1% increase in app users, the number of infections can be reduced by 0.8% (from modelling) or 2.3% (from statistical analysis).

The supplementary material is available here.

Effective configurations of a digital contact tracing app: A report to NHSX

Robert Hinch¹, Will Probert¹, Anel Nurtay¹, Michelle Kendall¹, Chris Wymant¹, Matthew Hall¹, Katrina Lythgoe¹, Ana Bulas Cruz¹, Lele Zhao¹, Andrea Stewart¹, Michael Parker² Daniel Montero³, James Warren³, Nicole K Mather³, Anthony Finkelstein⁴, Lucie Abeler-Dörner¹, David Bonsall¹ Christophe Fraser¹

¹ Big Data Institute, University of Oxford, UK, ² Wellcome Centre for Ethics and the Humanities and Ethox Centre, University of Oxford, UK ³ IBM UK, ⁴ UCL / Alan Turing Institute.

Abstract

The overarching objective of this report is to present simulations that will support the deployment and optimisation of digital contact tracing within an established programme of epidemic mitigation and control, and specifically to explore the conditions for success as countries prepare for exit from lockdowns. A lockdown can be regarded as a quarantine applied broadly to most of the population, excluding only key workers for example, whereas digital contact tracing can limit quarantine requests to those most at risk of transmitting the virus.

Quantifying dynamics of SARS-CoV-2 transmission suggests that epidemic control and avoidance is feasible through instantaneous digital contact tracing

Luca Ferretti^1*, Chris Wymant^1*, Michelle Kendall¹, Lele Zhao¹, Anel Nurtay¹, David Bonsall^1,2 and Christophe Fraser^1,3†

^*contributed equally; ^† To whom correspondence should be addressed: christophe.fraser@bdi.ox.ac.uk

¹ Big Data Institute, University of Oxford, UK, ² Oxford University NHS Trust, University of Oxford, UK, ³ Wellcome Centre for Human Genetics, University of Oxford, UK

Abstract

Mobile phone apps implementing algorithmic contact tracing can speed up the process of tracing newly diagnosed individuals, spreading information instantaneously back through a past contact network to inform them that they are at risk of being infected, and thus allow them to take appropriate social distancing and testing measures. The aim of non-pharmaceutical infection prevention is to move a population towards herd protection, a state where a population maintains R₀ < 1, thus making it impossible for a pathogen to cause an epidemic. Here, we address epidemiological issues that affect the feasibility of an algorithmic approach to instantaneous contact tracing; ethical and implementation issues are addressed separately. First we quantify the parameters of COVID-19 in a framework that is consistent with the renewal equation formulation of epidemic spread. Second, we use an analytical solution to application of first-degree contact tracing in the renewal equation model to explore combinations of efficacy that can induce herd protection (R₀ < 1). With the emergence of the novel viral pathogen SARS-CoV-2, of clear potential for a global pandemic with high fatality rates and incapacitated health systems, the question of prevention has critical priority. We come to the conclusion that isolating symptomatic cases and tracing their contacts in a classical manner is not sufficiently fast to stop the spread of the epidemic and needs to be accompanied by measures of social distancing that are disruptive to a wide number of people. We show that first-degree instantaneous contact tracing, informing users when they can move safely or when to seek medical help and avoid vulnerable individuals, has the potential to stop the spread of the epidemic if used by a sufficiently large number of people with reasonable fidelity.

Sustainable containment of COVID-19 using smartphones in China: Scientific and ethical underpinnings for implementation of similar approaches in other settings

David Bonsall^1,2 Michael Parker³ and Christophe Fraser^1,4

¹ Big Data Institute, University of Oxford, UK, ² Oxford University NHS Trust, University of Oxford, UK, ³ Wellcome Centre for Ethics and the Humanities and Ethox Centre, University of Oxford, UK, ⁴ Wellcome Centre for Human Genetics, University of Oxford, UK

COVID-19 is a rapidly spreading infectious disease with pandemic potential, caused by the novel virus, SARS-COV-2. With intensive care support, the case fatality rate is approximately 2%, and around half of infections become cases [1]. More concerning is that the fraction of cases requiring intensive care support is 5%, and patient management is complicated by requirements to use personal protective equipment (PPE) and engage in complex decontamination procedures [2]. Fatality rates are likely to be higher in populations older than in Hubei province (such as in Europe), and in low-income settings where critical care facilities are lacking [3]. Even modest outbreaks will see fatality rates climb as hospital capacity is overwhelmed, and the indirect effects caused by compromised health care services have yet to be enumerated. Effective containment must be achieved.