Mathematical models and simulations are driving the world’s response to COVID-19. Epidemiological modelling is by no means the only source of intelligence that politicians are taking into account when decision-making about our health, but articles like these: “Behind the Virus Report That Jarred the U.S. and the U.K. to Action” and “The battle at the heart of British science over coronavirus” make it particularly eye-catching.
In order to respond to the demands of agile governance, mathematical models have become a dominant source of policy evidence and a standard decision-support tool during various stages of disease management at strategic, tactical and operational policy levels. Other types of normative (value-driven) exploratory approaches may better identify and control the influence of longer-term drivers of disease spread. The latter include bio-social, political and economic determinants like poverty, food security, universal healthcare, climate change, urbanization, which are particularly important in developing countries where there are already challenges meeting sustainable development goals (SDGs). However, these approaches have not gained the same international policy foothold as quantitative mathematical models because they cannot be expediently deployed in response to a ‘securitised’ emergency and as qualitative instruments, may be perceived (incorrectly) by the public and policy-makers to provide insufficient or less convincing evidence in the face of scientific uncertainty.
Simulation models have been used to explore disease transmission and disease control strategies in other emergency outbreak events - but not without some controversy: the Foot-and-Mouth Disease (FMD) outbreak in 2001 was one of the first; models for SARS, Ebola, Zika subsequently followed. Indeed, the WHO contingency plans for avian influenza and advice for countries regarding national pandemic preparedness response plans are based on a mathematical model of H5N1 - purportedly the “largest-scale detailed epidemic micro-simulation” ever developed. This model is not dissimilar to any other mathematical model in that it inescapably incorporates numerous critical uncertainties. Not least of these are the assumptions about capabilities to detect, report and trace new cases, and compliance with effective disease mitigation measures. Until 4 weeks ago (March 17), implementing a disease control strategy which involves large-scale quarantine and social distancing measures was a purely theoretical exercise; until now, the assumptions and projected outcomes of this model have remained unchallenged and unverified in the absence of other “hard” ground-truthed evidence.
Images - The usually bustling Royal Mile and Princes Street in Edinburgh, following the declaration of a “lockdown” to control the spread of COVID-19.
The Imperial College team, responsible for this model, has come under unenviable and intense scrutiny at a time when they are working under unprecedented pressure to help inform better decisions. It is very difficult to model disease spread at an early stage in the outbreak, when uncertainty is high because data (on location and number of infected people, and their contact networks) are absent, scarce or not specific to the local context.
There are no agreed or formal standards by which models may be judged to be “fit-for-purpose” for decision-making. Until the epidemiological modelling community takes the lead, the same challenges are likely to continue to (re)emerge about the use of models during outbreaks. Among the most important are:
- Needing more data, but the right kind: Quantity, quality and speed of data access/sharing are key challenges; real-time data collection and sharing is therefore important. In this outbreak, testing and contact tracing of infected cases would be valuable in reducing uncertainty. When data are scarce, specific model assumptions may be particularly influential on model outcomes, so modellers need to communicate these assumptions explicitly to decision-makers and take steps to ensure there is mutual understanding. This is difficult. Models are complex and the language is foreign, so they are sometimes presented as “black boxes”. Many decision-makers will understandably not have the time, nor the tools to hand to scrutinise these critically or independently.
- Overlooking and underestimating the importance of communicating uncertainty: The identification of uncertainty or ambiguity around scientific outcomes generates a responsibility for scientists and policy-makers to close or communicate the knowledge gaps. Ignoring the effects of uncertainty on model results could lead to less than optimal decision-making. Alternatively, decisions to intervene (or not) derived from modelling advice which communicates uncertainty poorly may merely be equivalent to those arising out of the precautionary principle, in which case the modelling effort would be largely superfluous.
- Failing to mind the gap in ethical oversight: There is no ethical guidance for model-making. However, offering models as evidence for policy means that modellers’ in silico explorations result in real-world outcomes which positively and/or negatively affect people’s lives and livelihoods. Developing a useful ethical framework for future use will depend on: (i) the creation of science–policy partnerships to mutually define policy questions and communicate results; (ii) harmonized international standards for model development; (iii) strong data stewardship and (iv) improvement of traceability and transparency via searchable archives of policy-relevant models.
- Falling into the legitimacy trap: It is easy for models to acquire an over-stated sense of “legitimacy” as evidence. However, modellers themselves acknowledge that models are approximate summaries of our knowledge of the real world and shouldn’t be used in isolation, and definitely not in the absence of other, diverse, sources of expert opinion and intelligence obtained “on the ground”, which are vital in order to identify and challenge blindspots in thinking.
After the UK FMD outbreak in 2001, the animal health sector learned some important lessons about the use of models in outbreak disease emergencies. EPIC, Scottish Government’s Centre of Expertise on Animal Disease Outbreaks was created as a result of a wider recognition that the UK needed to improve disease outbreak preparedness. EPIC exemplifies a model of research provision for policy-making that utilizes academic partners, working closely with Government policy-makers, to contribute to evidence-based decision-making for outbreak science. This framework allows scientists in EPIC to pivot quickly to respond to emerging and novel animal/zoonotic disease emergencies in a timely way, potentially as soon as they arise. Transparent routes of communication with government underpin EPIC’s functionality. A unique, but important feature of EPIC is its specific role for knowledge-brokers who are embedded in government and who are fluent in different disciplinary expertise in order to bridge translation gaps between modellers and policy-makers. Impacts go beyond academic metrics and include broader conceptual changes about how scientists and policy-makers interact, moving further toward co-production.
Strengthening multi-sectoral and multidisciplinary links through academic-policy-industry consortia such as EPIC exemplify a “One Health” approach. The latter is vital to de-risking “human, animal and environmental health”. Indeed, One Health was borne out of a "global anxiety" about the emergence of a zoonotic pandemic – and it provided a framework in which the intergovernmental organisations (FAO, OIE, WHO and World Bank) could work closely together to address this threat. More than 15 years have passed since then, and we are now in the midst of that well-anticipated global health emergency. Yet in the UK at least, a coordinated plan to operationalise One Health tools and expertise is conspicuously absent from the current policy response. Scientists working to address “public and animal health threats and the ecosystems that interlink them”, could prove to be critical allies and sources of expert and experienced support for our public health colleagues. There is no better time than now, to lean on us and let us share the load.
Dr Lisa Boden (email@example.com).
Acknowledgments to Charles Bestwick, Iain McKendrick and Philip Skuce for their contributions and ideas.
Images (Edinburgh) - Sue McKendrick, 2020 and (Lisa Boden) - Michelle Wilson-Chalmers, 2018.
Article links, in order of appearance:
Walker, P. UK government response to coronavirus 'led by science' – Grant Shapps. The Guardian Published 16 March 2020. Available at: https://www.theguardian.com/world/2020/mar/16/government-response-to-coronavirus-led-by-science-grant-shapps
McCoy, D. Faith in coronavirus modelling is no substitute for sound political judgment. The Guardian, Published 10 April 2020. Available at: https://www.theguardian.com/commentisfree/2020/apr/10/modelling-pandemic-politicians-decisions-science
Adam, D. Special Report: The simulations driving the world’s response to COVID-19. How epidemiologists rushed to model the coronavirus pandemic. Nature News Feature, 03 April 2020. Available at: https://www.nature.com/articles/d41586-020-01003-6
Landler, M and Castle, S. Behind the Virus Report That Jarred the U.S. and the U.K. to Action. New York Times. Published 17 March, updated 2 April 2020. Available at: https://www.nytimes.com/2020/03/17/world/europe/coronavirus-imperial-college-johnson.html
Ford J. The battle at the heart of British science over coronavirus. Financial Times, Published 15 April 2020. Available at: https://www.ft.com/content/1e390ac6-7e2c-11ea-8fdb-7ec06edeef84?shareType=nongift
Gov.UK. Agile delivery: Governance principles for agile service delivery. Available at: https://www.gov.uk/service-manual/agile-delivery/governance-principles-for-agile-service-delivery
United Nations Sustainable Development Goals. Available at: https://sustainabledevelopment.un.org/?menu=1300
Keeling, M.J. Models of foot-and-mouth disease. Proc Biol Sci. 2005;272(1569):1195–1202. Available at: https://royalsocietypublishing.org/doi/10.1098/rspb.2004.3046
Han, X-Z. et.al. Mathematical modelling of SARS and other infectious diseases in China: a review. Tropical Medicine and International Health. 07 October 2009. Available at: https://onlinelibrary.wiley.com/doi/full/10.1111/j.1365-3156.2009.02244.x
Coltart, C.E, Lindsey, B., Ghinai, I., Johnson, A.M., Heymann, D.L. The Ebola outbreak, 2013-2016: old lessons for new epidemics. Philos Trans R Soc Lond B Biol Sci. 2017;372(1721):20160297. Available at: https://royalsocietypublishing.org/doi/10.1098/rstb.2016.0297
Suparit, P., Wiratsudakul, A., Modchang. C. A mathematical model for Zika virus transmission dynamics with a time-dependent mosquito biting rate. Theor Biol Med Model. 2018;15(1):11. Published 01 Aug 2018. Available at: https://tbiomed.biomedcentral.com/articles/10.1186/s12976-018-0083-z
Ferguson, N., Cummings, D., Fraser, C. et.al. Strategies for mitigating an influenza pandemic. Nature, 442, 448–452 (2006). Available at: https://doi.org/10.1038/nature04795
Cookson, C. and Payne, S. Coronavirus: why the UK stepped up its emergency response. Alarming new evidence led to rapid rethink but critics ask if it goes far enough. Financial Times. Published 17 March 2020.
Hamzelou. J. UK's scientific advice on coronavirus is a cause for concern. New Scientist Published 23 March 2020. Available at: https://www.newscientist.com/article/2238186-uks-scientific-advice-on-coronavirus-is-a-cause-for-concern/
Boden, L.A., McKendrick, I.J. Model-Based Policymaking: A Framework to Promote Ethical "Good Practice" in Mathematical Modeling for Public Health Policymaking. Front Public Health. 2017;5:68. Published 05 April 2017. Available at: https://www.frontiersin.org/articles/10.3389/fpubh.2017.00068/full
Vlek, C. A precautionary principled approach towards uncertain risks: review and decision-theoretic elaboration. Erasmus Law Review 02/02. Available at: https://core.ac.uk/download/pdf/18520439.pdf
Boden, L.A., Voas, S., Mellor, D. and Auty, H. (2020) EPIC, Scottish Government's Centre of Expertise in Animal Disease Outbreaks: A Model for Provision of Risk-Based Evidence to Policy. Front. Vet. Sci. Available at: https://www.frontiersin.org/articles/10.3389/fvets.2020.00119/full
Rivers, C., Chretien, J., Riley, S. et.al. Using “outbreak science” to strengthen the use of models during epidemics. Nat Commun 10, 3102 (2019). Available at: https://doi.org/10.1038/s41467-019-11067-2
Voegele, J. Evan, T. Kemper, K.E. One Health Approach is Critical to De-risk Human, Animal and Environmental Health 12 February 2018. Available at: https://blogs.worldbank.org/health/one-health-approach-critical-de-risk-human-animal-and-environmental-health
Gibbs, E.P., The evolution of One Health: a decade of progress and challenges for the future, Vet Record Volume 174, Issue 4. Available at: https://veterinaryrecord.bmj.com/content/174/4/85
WHO. One Health. Published 21 September 2017. Available at: https://www.who.int/news-room/q-a-detail/one-health
Webster, K. One Health is integral to APHA science 3 November. Available at: https://aphascience.blog.gov.uk/2019/11/03/one-health-day-2019/