Modeling an Outbreak: Limits and Lessons

Modeling is a tool we rely on heavily in epidemiology. Now more than ever, modeling is being used to address the global pandemic that is coronavirus disease 2019 (COVID-19). Using an array of variables, mathematical modeling can help predict potential outcomes of an outbreak. While there are many critical parts of an outbreak, from an infection preventionist perspective, epidemiological modelling can be immensely helpful.

Such models can help guide critical resources, predict surges in cases, and determine if interventions might help dent case counts. As the COVID-19 outbreak rages on, the use of these models is even more relevant, especially for leaders and policymakers.

Neil Ferguson, a mathematical epidemiologist at Imperial College London, is one such expert who works to drive outbreak intervention through modeling. Unfortunately, models are only as helpful as those who listen to them. While data points consistently change and those working to build and maintain COVID-19 models are feverishly working to update them, such tools of epidemiology and disease control must be used to be effective.

“When updated data in the Imperial team’s model indicated that the United Kingdom’s health service would soon be overwhelmed with severe cases of COVID-19, and might face more than 500,000 deaths if the government took no action, Prime Minister Boris Johnson almost immediately announced stringent new restrictions on people’s movements. The same model suggested that, with no action, the United States might face 2.2 million deaths; it was shared with the White House and new guidance on social distancing quickly followed (see ‘Simulation shock’),” the author of a news featured published in Nature wrote.

Perhaps one of the biggest challenges in modeling, and in disease control in general, is conveying that each outbreak is unique. This is especially true with a novel virus. We are running across the bridge as we build it. More often, I see people expecting models to be infallible and a permanent line in the sand, which they are not. They are used to help guide decisions, not make the decision itself.

Much data, early in the outbreak especially, is limited and often estimated. As more data is added into the models from different population groups, the models can also get increasingly more complicated. These are critical realities of modeling that many don’t realize, which can drive misunderstanding or a heavy dependence on the model.

Simulations are such a valuable tool during outbreaks, but they also rely on data, which in some outbreaks, can be more of a challenge to come by. As Vittoria Colizza, a modeler at the Pierre Louis Institute of Epidemiology and Public Health, notes “it depends on the question you want to ask”.

Accuracy is a challenging piece to this, which is why modelers work hard to make hundreds of separate runs and emphasize that projections are not perfect and set in stone. Ultimately, we must work harder to collect and provide better data. Understanding the implications of decisions, policies, and social practices, can improve outbreak response and also modeling efforts.