top of page

From crisis support to COVID-19 pandemic preparedness

Escape_Original_No-BG (1).png

The COVID-19 pandemic shone a spotlight on mathematical and statistical modelling like never before. Suddenly, the scientific community, and the models they produced, were subject to unforetold levels of pressure – from those in positions of decision-making to the media and the general public. Experts in mathematical epidemiology and biostatistics from across Europe joined forces in the European-funded projects EpiPose and ESCAPE. Strong ties dating back to the POLYMOD project1 in 2006 and even earlier, enabled the success of model-driven research in ‘EpiPose2‘. The project was submitted to the EU for funding in just ten days and with limited knowledge of the virus at the time, with the aim of providing epidemic intelligence to combat SARS-CoV-2.

Through the course of the COVID-19 pandemic, EpiPose researchers and many other prominent modellers encountered similar difficulties centred around three themes: getting access to the right data at the right time, flexibly adapting models and analytical frameworks, and producing rapid yet trustworthy advice for decision-makers. At many levels, society was not prepared for such a crisis as that posed by COVID-19. Important lessons have been learned but these are only valuable if we as a community proactively shape them into actions, agreements and guidelines for future response plans. ESCAPE will work towards a science-based blueprint for pandemic control, a game plan with data, modelling and public-science-policy interactions as key ingredients.

Rapid access to high-quality data: a holy grail or just a first step?

Models to inform public health decisions are fed with various types of data. Traditionally, small yet high-quality data such as number of cases, hospitalisations and deaths provide a basic understanding of how an epidemic is evolving. However, it is a common misunderstanding that mere access to data provides all the answers. True, data holds information, and in the context of COVID-19, data-driven modelling saves lives, but only when looked at thoroughly and with the right analytical framework. Data complexities like missing values, censoring, truncation or reporting delays should be accounted for. A trend in the data, e.g. an ominous exponential growth in the number of infections, does not always speak for itself. This causes a bias in risk perception and decisions to put control measures into place. There is a clear task ahead of the ESCAPE team to train decision-makers in understanding basic statistical and epidemiological concepts.

Another learned lesson is that numbers collected in different settings or countries may have different meanings. Heterogeneous definitions for critical care beds3 and outdated data made it difficult to compare COVID-19-induced healthcare pressure across countries. EpiPose facilitated this through the EU COVID-19 Healthcare Pressure platform4 yet it was clear that even when considering rather simple datasets, society was not ready. Likewise, COVID-19-induced deaths were counted very differently across countries.5 Ignoring this heterogeneity set the scene for sensational reports on Belgian COVID-19 mortality but using excess mortality as a richer metric, the story becomes more nuanced. Numbers, although seemingly objective and of high quality, always need to be put in the context of healthcare and surveillance traditions and interpreted with caution.

Besides traditional data, COVID-19 modellers from all over the world increasingly exploited the value of other ‘big’ data sources related to people’s health, mobility, sentiments and economic behaviour. Although not collected to serve public health, non-traditional data helped modellers tremendously to inform response plans. Early on in the COVID-19 pandemic, mobile phone trajectories were used to quantify the reduction in mobility and related contacts at work and on public transport in China.6 EpiPose and ESCAPE researchers rapidly used such large-scale data to assess the efficiency of a lockdown, e.g. in France7,8 and Italy.9 The importance of private sector data cannot be stressed enough. Hence, ESCAPE will improve protocols for sharing data and analytical tools across sectors and countries. Another example is how sensor-based contact tracing in schools – tracking interactions between children and teachers — enabled the fine-tuning of plans concerning COVID-19 screening and school closure.10,11 Such data complements the European-wide survey-based monitoring of contact behaviour in EpiPose’s CoMix study.12 Behavioural data is crucial in studying how a pandemic affects peoples’ lives in various dimensions and in assessing adherence to measures. Various countries represented in Influenzanet,13 deployed web-based surveillance platforms to monitor symptoms, health-seeking behaviour and risk perceptions of volunteers. Last, social media analytics unravelled the public’s perception and response to health risks and interventions.14

ESCAPE will address data readiness at all levels. This is a much-needed step, but just the first in the model-based combat against a pandemic.

Learning form the COVID-19 pandemic

Data is core to the mathematical models which are used to inform an outbreak response. During the COVID-19 pandemic, much attention was focused on the use of models to predict the virus’ spread, project the demand for hospital beds and provide estimates of what might happen to case numbers were a certain intervention measure (e.g. lockdown) enacted. Members of EpiPose and ESCAPE among many others were at the forefront of these modelling efforts. They used mathematical modelling to assess how fast the epidemic was spreading,15 forecast future case numbers,16  and provided scenarios to highlight the potential impact of different policy interventions and estimate the transmissibility of newly emerging COVID-19 variants.17 Their collective efforts provided decision-makers with evidence on which to act in response to an ever-evolving virus.

For a model to be useful to those in decision-making roles, the results must be to a scientific standard such that the results can be easily reproduced by other research teams. They must satisfy the basics of scientific rigour. They must perform well (e.g. forecasts should reflect observed outcomes), have the ability to inform other models – be they simpler or more complex – and be explainable to those in decision-making positions.

Vitally, in the context of a public health crisis, the results must be generated rapidly. At times, when data is not readily available, this can mean relying on estimates and sensitivity analyses rather than certainties. Modelling during a crisis is therefore by necessity a different exercise to modelling during ‘peacetime’. And while scientific standards must be upheld regardless, during a public health emergency, developing a simple model, which can be flexible to incorporating new information as it emerges, is often the most effective approach.

The situation is constantly changing throughout a pandemic – under influence of intervention measures or new variants – and many new questions arise. Models must be flexible in adapting to this and teams willing to adapt their methods in response to the evolving context. During EpiPose, we learnt what worked and where we needed to reset our initial expectations. For example, the team initially proposed using a meta-population model to understand spatial patterns of SARS-CoV-2 transmission. However, due to the fast-evolving pandemic, the time, data and resource constraints involved in meta-population models did not outweigh the benefits compared to alternative approaches, such as compartmental and individual-based models.

In reality, the very nature of an outbreak means that not all of these criteria can be satisfied easily. During the COVID-19 pandemic – and before it – modelling teams across the globe developed their own tools and models in order to answer the specific questions being posed by decision-makers in their jurisdictions. The result can be multiple teams redeveloping the same tools, or errors made from individuals writing code in isolation.18 Through EpiPose, we were fortunate to be afforded funding to work collaboratively and share code, methods, challenges and solutions across countries.

We know how crucial it is that the results of modelling efforts be easily translatable across jurisdictions. Facilitating this exchange is one of ESCAPE’s core tasks. As part of a wide suite of activities focused towards improving pandemic preparedness, ESCAPE will produce a range of digital tools to facilitate the comparison of information across countries. Additionally, in order to demonstrate how teams can use and adapt the concepts underlying the models to their own local contexts, ESCAPE researchers will also develop reusable code, digital tools and associated guidance documents which can easily be implemented in and shared across countries.

Where science meets policy

From the first brainstorm sessions about EpiPose and ESCAPE, it was clear that translating model-based insights into health policy decisions would be a driving factor of the projects’ success.

Throughout the pandemic, members of EpiPose provided evidence-based support to national governments and public health institutions. Various dimensions of the COVID-19 pandemic impact were considered: disease burden and health costs, healthcare pressure, cost effectiveness of vaccines and NPIs as well as the assessment of mental health and the wider socio- and macroeconomic impact. EpiPose members participated in 24 expert committees in Europe and exchanged results with ECDC and WHO, creating a strong societal impact in their respective countries and beyond.

However, science-policy interactions are not without difficulties. Scientists and policymakers need to create a common understanding of the kind of answers models can and cannot provide, as well as the underlying uncertainty inherent to modelling. During this process, scientists not only have a pressing responsibility to create and choose the most appropriate methods, but to interpret their models in terms of useful insights in the political decision-making process. Being advisors but not decision-makers however, scientists remain independent actors in the political arena. For this reason, their media presence and the visibility and ‘explainability’ of their work have a great influence on public acceptance of the introduced measures.

Even though the baseline characteristics of evidence-based decision-making are universal, countries have developed various mechanisms in order to harmonise scientific evidence with local political considerations. It is therefore not surprising that the results of data-based modelling have been incorporated differently in decision-making processes during the COVID-19 pandemic. In some countries, like Germany, the UK or the Netherlands, the discussion between scientists and policymakers benefitted from an institutionalised framework. During the pandemic, among their many responsibilities, the Center of Infectious Disease Control at the Dutch National Institute of Public Health and Environment (RIVM) collected and analysed data to support national health policies. For Switzerland, on the other hand, the pandemic provided a chance to reflect on opportunities for scientific policymaking. Before the crisis, the country typically incorporated expert opinion on specific health policies from external actors through short-term mandates. This approach, however, was not sufficient under the rapidly evolving pandemic crisis, and by the end of March 2020, the government appointed the Swiss National COVID-19 Science Task Force.

During the first two years of the pandemic, the multidisciplinary team of the Task Force advised the government on various aspects of the pandemic response, including epidemiological, public health and economic considerations, contributing to more than a 100 policy briefs. Nevertheless, the interaction between science and policy in Switzerland has not been without challenges, and the rich experience of this ad hoc advisory group has forced the government to consider various ways in which scientific advice can best be incorporated into policy-making during future crises.

What became possible in crisis mode needs to be maintained and optimised in peacetime. Creating a blueprint for pandemic response will not be an isolated exercise among ESCAPE researchers. Other scientists, policymakers and representatives of the media and general public will be consulted along the way such that Europe, as a connected, intelligent community, is ready to respond fast and efficiently when the next pandemic hits.


This work was funded by the UK Research and Innovation (UKRI) under the UK Government’s Horizon Europe funding guarantee grant number 10051037.

EpiPose and ESCAPE received funding from the EU Horizon 2020 (project 101003688) and Horizon Europe (project 101095619) programmes respectively.

On behalf of the EpiPose and ESCAPE consortium members:

Sarah Vercruysse, Lisa Hermans, Zita Zsabokorszky, Anna Carnegie, Vittoria Colizza, Jacco Wallinga, Christian Althaus, and Niel Hens.


1. Mossong et al., PLOS Med, 2008.

2. EpiPose project:

3. Rhodes et al., Intensive Care Med, 2012


5. Molenberghs et al., Euro Surveill, 2022

6. Lu et al., Health Data Sci, 2021

7. Domenico et al., BMC Med, 2020

8. Pullano et al., Lancet Digit Health 2020

9. Pepe et al., Sci Data, 2020

10. Colisi et al., Lancet Digit Health, 2022

11. Tornei et al., eLife 2022



14. Crupi et al.,2022

15. Gressani et al., PloS Comput Biol, 2022

16. Davies et al., Lancet Public Health, 2020

17. Barnard et al., Nat Commun, 2022

18. Kucharski et al., PLoS Biol, 2020

bottom of page