Risk assessment in a changing world

I have noticed that when people think about risk, they only think about probability: How likely is it that x will happen?

For example, the UK Government’s Flood Risk service tells you that Trinity College, Cambridge has a “very low risk” of flooding, by which they mean that this area has a chance of flooding of less than 0.1% each year.

Similarly, and closer to the point I want to make later on, in a recent Nature review on long COVID, Hannah E Davis et al. quantify the risk of developing long-term damage after COVID purely in terms of the percentage of people that develops the condition, reporting for instance that “MRI studies revealed cardiac impairment in 78% of 100 individuals who had a prior COVID-19 episode”.

Chances, probabilities, likelihoods. We’ve heard a lot about those during the last few years. However, risk assessment is not only about probability or chance. If you want to determine the risk of something, you should be thinking just as much about the severity of the consequences: How bad would it be if x happened?

Umbrellas and seatbelts

Compare the following two examples. When I step outside to go for a long walk, there often is a chance higher than one in five (>20%) that I’ll be caught in bad weather and get soaked. Yet, even with those odds I normally don’t go through the hassle of carrying an umbrella.

On the other hand, when I get into a car I always wear a seatbelt, even though the chance of getting involved in an accident is less than 1%. Why do I leave the umbrella home, but insist on wearing the seatbelt? For keep in mind, the chance of the negative outcome the umbrella protects against happening is much higher than the chance I’ll be in an accident.

The answer is of course that it doesn’t matter much if I get wet. It’s inconvenient, but when it happens it’s not much more than an uncomfortable but interesting adventure. In stark contrast, the consequences of a road accident are likely to be severe, possibly even very severe. It could easily lead to hospitalisation or death.

Given this, getting caught in a shower is in fact a lower risk than the risk of a car accident. Risks are about consequences as well as chances. For similar considerations I don’t smoke, even though I know that the majority of smokers will reach old age. With smoking, for me it’s not about the chance of being fine, but about what happens when you’re not.

Calculating risk

Generally, we should calculate risk with the following formula:

Risk = Probability x Consequences

When we let our assessments of risks be guided by this formula, we will see that the more severe the consequences of an outcome are, the less the probability of those consequences matters.

When Sir Christopher Wren designed the famous library at Trinity College (1676–84), in his design he factored in flooding. The library is raised on pillars high above the ground. Was there a high chance of flooding? No. But the consequences of damage to the precious library would be so devastating, that even the low probability of the river Cam’s leaving its banks was enough to build in flood defences. The chance was low, but the risk was high.

Risk in an ongoing pandemic

All this matters for risk assessment in the ongoing pandemic. In the Nature review I quoted above the authors observe that “Long COVID is an often debilitating illness that occurs in at least 10% of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections” (this figure includes vaccinated people). Not only do we here get a quantified expression of the chance of developing long COVID (>10%), we also get a qualitative estimation of its consequences. The consequences are described as ‘often debilitating’. To my mind, this puts them somewhere close to car accidents or the kind of damage done by smoking.

Can we now assess the risk of developing long COVID? No, because there’s more. Consequences involve contextual factors as well. Getting wet on a sunny day is less bad than finding yourself soaked when it’s just above freezing. Similarly, the conditions under which we lead our lives have changed significantly since the start of the pandemic.

In part due to the way the pandemic has been mismanaged, in many countries the health care system is under severe strain. This means, for example, that in the UK more patients than ever are on waiting lists for treatment (7+ million people), and that 500 people die every week simply because they cannot get access to urgent care. In The Netherlands, the care watchdog has announced that with the changed situation access to care is no longer a given.

In other words, if you are ill and in need of treatment, it is becoming increasingly less likely you will have access to that treatment. This in turn means that the consequences of accidents and of falling ill are becoming increasingly severe. This increase in severity can happen even if, strictly speaking, the medical facts about accidents or illnesses don’t change, or even become more favourable.

This matters for risk assessment in a COVID-19 pandemic. The higher the disease burden on societies, the more unreliable the health care system becomes for many of us. And that in turn means that the consequences of getting COVID, long COVID or some other COVID sequelae are becoming worse and worse. Falling chronically ill in a society that has a well-functioning social healthcare system is bad, but falling ill in a society where the social healthcare system is at the brink of collapse is much, much worse.