Key questions for modelling COVID-19 exit strategies

Robin N. Thompson, T. Déirdre Hollingsworth, Valerie Isham, Daniel Arribas‐Bel, Ben Ashby, Tom Britton, Peter Challenor, Lauren H. K. Chappell, Hannah Clapham, Nik J. Cunniffe, A. P. Dawid, Christl A. Donnelly, Rosalind M. Eggo, Sebastian Funk, Nigel Gilbert, Paul Glendinning, Julia R. Gog, William S. Hart, Hans Heesterbeek, Thomas House, Matt J. Keeling, István Z. Kiss, Mirjam Kretzschmar, Alun L. Lloyd, Emma S. McBryde, James M. McCaw, Trevelyan J. McKinley, Joel C. Miller, Martina Morris, Philip D. O’Neill, Kris V. Parag, Carl A. B. Pearson, Lorenzo Pellis, Juliet R.c. Pulliam, Joshua V. Ross, Gianpaolo Scalia Tomba, Bernard W. Silverman, Claúdio J. Struchiner, Michael J. Tildesley, Pieter Trapman, Cerian Ruth Webb, Denis Mollison, Olivier RestifProceedings of the Royal Society B Biological Sciences 2020 · DOI

TL;DR

This paper outlines critical questions and a roadmap for developing better mathematical models to guide COVID-19 exit strategies. It emphasizes the need for improved epidemiological parameter estimation, understanding population heterogeneity, and better data collection, especially in low-resource settings, to balance public health with socio-economic needs.

What Problem Does This Paper Address?

The COVID-19 pandemic forced governments worldwide to implement unprecedented non-pharmaceutical interventions (NPIs), commonly known as lockdowns, to curb transmission. As case numbers stabilized or declined, the challenge shifted to designing "exit strategies" – ways to relax these restrictions while controlling the risk of resurgent outbreaks. Mathematical modeling has been crucial for informing these interventions, but the complexity of ongoing transmission, coupled with the unprecedented nature of these strategies, presents a significant challenge. This paper addresses the need for a structured approach to identify and answer the key modeling questions that will enable more accurate predictions of the effects of different exit strategies, moving beyond immediate crisis response to sustainable pandemic management.

Methods

The paper doesn't present a novel modeling technique or algorithm. Instead, it synthesizes the discussions and consensus from a workshop of diverse mathematical modelers who advise governments. Their "method" is a collaborative, expert-driven identification of critical knowledge gaps and a proposed roadmap for future research. This roadmap is structured around three core pillars:

  • Improving Epidemiological Parameter Estimates: This involves refining our understanding of key metrics like the reproduction number ($R(t)$) and herd immunity thresholds, accounting for factors like viral transmissibility and seroprevalence.
  • Understanding Population Heterogeneities: This focuses on how differences within and between populations (e.g., age, household structure, social networks, health vulnerabilities) impact transmission and intervention effectiveness.
  • Focusing on Data Requirements: This highlights the need for better data collection strategies, particularly in low-to-middle-income countries (LMICs), and how to effectively communicate model uncertainty to policymakers and the public.

Datasets

The paper discusses the types of data needed and used for modeling, rather than presenting specific datasets for training or evaluation of a new model. These include:

  • Case notification data: Used for estimating reproduction numbers.
  • Seroprevalence survey data: To estimate population immunity.
  • Mobility data: To understand contact patterns.
  • Contact diaries and social media data: For network analysis.
  • Data on deaths: Often used for model fitting.
  • Environmental data: Such as temperature and humidity, to explore their influence on transmission.
  • Metadata associated with testing protocols: Crucial for understanding biases and improving parameter estimates.

The paper emphasizes the need for better data collection, especially in LMICs, and highlights the challenges of data availability, reliability, and comparability across different regions.

Key Insights and Ideas

  • The "exit strategy" is not a single event, but a complex, iterative process: Relaxing restrictions requires continuous monitoring and adaptation, making real-time modeling crucial.
  • Herd immunity is more complex than a simple threshold: Population heterogeneity significantly lowers the herd immunity threshold, especially when immunity is gained through infection rather than uniform vaccination.
  • Seroprevalence surveys are valuable but have limitations: They offer a snapshot of past infections but can be affected by test sensitivity/specificity and the unknown duration of immunity.
  • Children's role in transmission is still unclear: While they may be less susceptible and infectious, their reopening of schools has significant societal and epidemiological implications that need careful modeling.
  • Agent-based models (ABMs) offer potential for capturing social complexity: While computationally intensive, ABMs can represent individual behaviors and interactions, offering a richer understanding of societal responses to interventions.
  • Data limitations in LMICs are a major hurdle: Addressing these requires creative approaches to data collection and utilization, alongside international collaboration.
  • Communicating uncertainty is paramount: Presenting single "best guess" figures can create a false sense of confidence; understanding the range of possible outcomes is vital for informed decision-making.

Key Results

This paper is not about quantitative results from a specific model but rather a qualitative synthesis of expert opinion and a proposed research agenda. The "results" are the identified key questions and the structured roadmap to address them. The paper argues that answering these questions will lead to:

  • More accurate predictions of exit strategy effects: Enabling better balancing of public health and socio-economic considerations.
  • Improved understanding of disease dynamics: Particularly concerning heterogeneity and the impact of interventions.
  • A framework for global collaboration: Fostering a coordinated scientific response to future pandemics.

Limitations and Future Work

The authors acknowledge several limitations and suggest future directions:

  • Data Availability and Quality: Significant challenges remain in obtaining reliable, comparable data, especially in LMICs. Future work needs to focus on developing methods for data-limited settings and value-of-information analyses to prioritize data collection.
  • Model Complexity vs. Parsimony: Striking the right balance between model realism and tractability for timely predictions is an ongoing challenge.
  • Uncertainty Quantification and Communication: Developing better methods to quantify and communicate uncertainty to policymakers and the public is essential.
  • Interdisciplinary Collaboration: The paper stresses the need for continued collaboration between modelers, epidemiologists, social scientists, and policymakers.
  • Developing Robust Models: Future work should focus on creating flexible, iterative modeling frameworks that can be continuously updated as new data and knowledge emerge. The development of a "digital twin" of society for virtual policy experiments is also proposed.

Summary updated

Abstract

Combinations of intense non-pharmaceutical interventions (lockdowns) were introduced worldwide to reduce SARS-CoV-2 transmission. Many governments have begun to implement exit strategies that relax restrictions while attempting to control the risk of a surge in cases. Mathematical modelling has played a central role in guiding interventions, but the challenge of designing optimal exit strategies in the face of ongoing transmission is unprecedented. Here, we report discussions from the Isaac Newton Institute 'Models for an exit strategy' workshop (11-15 May 2020). A diverse community of modellers who are providing evidence to governments worldwide were asked to identify the main questions that, if answered, would allow for more accurate predictions of the effects of different exit strategies. Based on these questions, we propose a roadmap to facilitate the development of reliable models to guide exit strategies. This roadmap requires a global collaborative effort from the scientific community and policymakers, and has three parts: (i) improve estimation of key epidemiological parameters; (ii) understand sources of heterogeneity in populations; and (iii) focus on requirements for data collection, particularly in low-to-middle-income countries. This will provide important information for planning exit strategies that balance socio-economic benefits with public health.
BibTeX
@inproceedings{thompson2020key,
  author = {Robin N. Thompson and T. Déirdre Hollingsworth and Valerie Isham and Daniel Arribas‐Bel and Ben Ashby and Tom Britton and Peter Challenor and Lauren H. K. Chappell and Hannah Clapham and Nik J. Cunniffe and A. P. Dawid and Christl A. Donnelly and Rosalind M. Eggo and Sebastian Funk and Nigel Gilbert and Paul Glendinning and Julia R. Gog and William S. Hart and Hans Heesterbeek and Thomas House and Matt J. Keeling and István Z. Kiss and Mirjam Kretzschmar and Alun L. Lloyd and Emma S. McBryde and James M. McCaw and Trevelyan J. McKinley and Joel C. Miller and Martina Morris and Philip D. O’Neill and Kris V. Parag and Carl A. B. Pearson and Lorenzo Pellis and Juliet R.C. Pulliam and Joshua V. Ross and Gianpaolo Scalia Tomba and Bernard W. Silverman and Claúdio J. Struchiner and Michael J. Tildesley and Pieter Trapman and Cerian Ruth Webb and Denis Mollison and Olivier Restif},
  title = {{Key questions for modelling COVID-19 exit strategies}},
  booktitle = {Proceedings of the Royal Society B Biological Sciences},
  year = {2020},
  doi = {10.1098/rspb.2020.1405},
}
Figure