Interpreting result of infectious disease models

The COVID-19 epidemic provides a good example of how a rapidly evolving situation generates also an exponentially increasing amount of research – including mathematical models – with often conflicting and even contradictory results. This confuses not only the public, but also the decision-makers who need to develop prevention policies based on the available evidence. This raises often the question: how can we know which models are right and which wrong, and what conclusions should we make from the available evidence?

The gold standard of evidence synthesis is a systematic review – a systematic search of all existing evidence on a particular topic, possibly followed by a meta-analysis, which attempts to synthesize and present all available evidence in a concise and comprehensive way. Whereas the value of systematic reviews of clinical studies in decision-making is well acknowledged, the same cannot be said about mathematical models. Far too often, policy-makers adapt only one single model to support their decisions. This may be done based on previous positive experience of a similar model, but the main problem remains – any model can reflect only what follows from its assumptions, and they may well be incorrect.

Governments have the responsibility to make decisions to control the epidemic and protect their citizens and residents, and they should thus have access to all the available evidence. A single model or study cannot provide this. Instead, to understand the true situation, policy-makers should have broad knowledge about all studies and models, their findings – and in particular for the reasons that the studies provide differing results