A GLOBAL HEALTH CRISIS
Stratify and shield: A better approach to lockdown
An interview with Helen Colhoun, AXA Chair in Medical Informatics and Life Course Epidemiology
People with pre-existing illnesses seem most at risk during this pandemic. What evidence do we have that this is the case?
We know that approximately three-quarters of people developing a severe form of COVID had at least one of what we call listed conditions. These include illnesses such as ischemic heart disease, diabetes, asthma, chronic obstructive airway disease (COAD) and liver disease. With asthma and COAD, for example, your risk is about double. For people with kidney disease, it’s around six times higher. However, these figures will be underestimates because, at the moment, we don’t have a complete picture. At the outset of the pandemic, we introduced shielding for vulnerable groups – so we don’t know the true risk had they been fully exposed to the pandemic. With less extreme conditions – people with diabetes, for example – we have a slightly better idea because they weren’t subject to full shielding.
To what extent is this also because we simply don’t know enough about COVID-19?
It’s true – this virus is not well understood. It takes time to acquire that knowledge. Unfortunately, we need an accrual of infections before we can get a proper handle on the epidemiology. In recent months, we have been prioritizing certain groups for testing. If we look at the rates for those who tested positive, the numbers are full of selection biases. The same is true even of hospitalization. That’s why, in our work, we have focused on patients who required critical care or who unfortunately died from the virus. Even so, those figures will reflect the probability of infection and, once infected, of developing a severe form of the illness. Until we have comprehensive diagnostic and serological testing, we won’t know for certain how much we are seeing of the iceberg – these severe and fatal cases are only the tip of that iceberg. There is clearly an enormous variance in risk, even among those with listed conditions. And the problem is that currently, in our public policies, we’re taking a binary approach to this risk: you are either high risk or you’re not. In Scotland, there are more than 300,000 people with diabetes. Of these, only 845 developed a severe or fatal form of COVID. Each case is a lost friend or family member, but overall, that represents less than 0.3% of the total population.
"We could follow a different policy if we understood more about this variance in risk, and who was really susceptible to this illness. We could have a stratify and shield approach, where vulnerable groups were much better ring-fenced”
“We need to build better risk prediction models, think more about the combination of different risks, and use the data we have more intelligently.”
Do we know what causes that variance in risk?
Risk is not binary – it’s actually a continuum. If you are a 55 year-old man with type one diabetes, do you really have the same risk as somebody who is 80 years old, without diabetes? Currently, we are putting those two people in the same risk basket. Having that kind of arbitrary cut-off is never going to be the best approach. This is what my AXA Chair is focused on – we need to build better risk prediction models, think more about the combination of different risks, and use the data we have more intelligently. We know there is substantial information in the clinical record that will help us do this. We could follow a different policy if we understood more about this variance in risk, and who was really susceptible to this illness. We could have a stratify and shield approach, where vulnerable groups were much better ring-fenced.
How easy would it be to implement a policy like that?
If we could properly identify vulnerable groups, we could allow infection rates to accrue in the rest of the population. That brings us to the idea of population immunity, where at least 65% of the population would need to be immune before there was a waning in the virus’ reproduction rate. So far, antibody tests have shown much lower positive rates than expected, at most 10%. This means either infection rates are really that low, or that there is another explanation – possibly that antibodies disappear or wane after a time, especially in those who have had only a mild infection, or even that some people are immune to the virus without that showing up in a serological test. My point is – there is scope here for a little more imagination. It is important that, whatever we do, it is subject to rigorous evaluation, and that it is based on evidence, not opinions. We have lost trillions in our economy already. In such circumstances, even impossibly expensive options might prove to be cost-effective. You have to ask – what is our end game? At the moment, we have a policy of suppression. Our aim is to keep the R rate – the reproduction rate – below 1. If we see an increase in cases, we impose a blanket lockdown. We need to think more about our approach to risk – and whether a more targeted policy of stratify and shield couldn’t get us out of this pandemic more safely and more quickly.
Readers also read
- COVID-19 will accelerate the trend toward more online health, by Claudio Gienal
- Lessons from malaria: more than test and trace, by Professor Gerry Killeen
Helen Colhoun
AXA Chair in Medical Informatics and Life Course Epidemiology, University of Edinburgh (UK)
Professor Helen Colhoun is Honorary Consultant in Public Health with the National Health Service (NHS) in the UK. Previously, she was professor in epidemiology at the University of Dundee, University College London and University College Dublin before joining the University of Edinburgh in 2016. Helen contributed to the UK’s national policy on diabetes. Her research has also been cited in international clinical guidelines.
