Decision-making at a time of great uncertainty
Continued uncertainty around COVID-19 demands more qualitative modelling, says Dr Ayham Fattoum, Lecturer in Disaster Operations Management at HCRI.
In the wake of countries adopting such a broad range of measures to control COVID-19 in recent months, there are growing calls for global consistency when it comes to supporting decision-makers responding to the virus. The thinking is that by trying to identify correlations between different response strategies from different countries, so we can reach best practice guidelines.
However, although such correlations can be informative, quantitative approaches to decision-making are not reliable nor compatible with the great uncertainty one sees in a pandemic. The confused and panicked responses to COVID-19 around the world is a manifestation of this systemic weakness.
Why pandemics are different
So just why is decision-making during pandemics so different to other crises, and so challenging? A distinction here between different types of problems is very relevant.
‘Puzzle-like’ problems are complicated and difficult to solve but everyone knows that a solution exists, and when obtained everyone agrees that this is the solution. However, another type of problem is ‘mysteries’ – complex problems which evoke questions that can have only contingent answers and hence no one ‘single’ right solution. Even after the event, there will be no agreement that this was, or any other option would have been, the right solution.
Pandemics such as COVID-19 are mysteries because they are biological and social phenomena. Biological because they are influenced by the bio-behaviours of the virus, humans, and maybe animals (e.g. ability to infect and mutate and the immunity response of the host). Social because their impact depends on people’s behaviours (e.g. habits and cultural norms that increase the possibility of transmission) including their response to the pandemic.
To tackle the COVID-19 problem, we have seen governments across the world turning to a vast array of different models for help. The scientific approach to making decisions has been dominant with political leaders depending on the predictions of mathematical models that use assumptions and data (inputs) to generate numbers, words, or figures (predictions). Decisions to allocate resources, purchase PPE (Personal Protective Equipment), build new hospitals, and determine the length of lockdowns have all been driven by these predictions.
Although such models are scientifically robust and can be enhanced by artificial intelligence (AI), algorithms, big data, and sophisticated formulae, they have been criticised in some quarters for their susceptibility to users’ subjectivity and significant reliance on historic data. Indeed the risks of using models in decision-making were highlighted in a document published by the UK government in 2013.
In fact, predicting the occurrence and course of a pandemic requires a model that includes all relevant parameters and all the possible relationships among them, which is practically impossible given our current technology and knowledge.
For instance, a Chinese team has recently discovered that the virus’s ability to mutate was underestimated and reported 33 mutations since December 2019, which may partly explain the varying deadly impact of the virus in different countries. Still, it is not possible to predict if further mutations will occur or, if they do, to what form.
Indeed, almost all governments around the world have been surprised by the virus despite having experience of previous biological events (e.g. SARS, HIV, and Ebola) and the many predictions of a pandemic by scholars, scientists, and business figures.
Planning for the future
Faced with these daunting challenges, what course should governments take now in their wider approach to the pandemic threat?
One could argue that the decision not to prepare for a pandemic may be justified from a traditional decision-making perspective. The probability of a pandemic happening is very small (if known at all), and hence it could be argued that investing resources into preparing for it is simply not feasible. Even during the current pandemic, knowing that there is a 25% chance that COVID-19 will mutate in London in August is almost meaningless because it does not inform on the consequences of this mutation (it may result in killing the virus).
Also, it will not be possible to pin down best practices, or accurately evaluate decision choices, even after the pandemic. In practice, this means that decision-makers cannot depend solely on models of probabilities to make decisions, and they may need to consider innovative decision-making strategies to address such uncertainty
Uncertainty around COVID-19 is likely to continue in the recovery period and beyond. In the short-term, I would argue that more qualitative techniques should be integrated to avoid further surprises and mitigate against the consequences of reliance on incompatible and inaccurate predictions.
In the long-term, emergency policies should inform decision-makers at the local and national levels on how to use qualitative approaches in disaster preparedness, response, and recovery. Further, to ensure that whatever our ‘new normal’ is after COVID-19 is sustainable, qualitative approaches must be ready to be used too.