The Models Behind NPHET's Concern About Covid-19 Surge

Professor Simon More, Professor of Veterinary Epidemiology and Risk Analysis within the UCD School of Veterinary Medicine, recently contributed to RTÉ Prime Time report on the models behind NPHET's concern about Covid-19 surge. 

What exactly will NPHET modellers be looking at as they consider tighter Covid-19 restrictions in the coming days?

With detected cases of Covid-19 rising sharply in recent days, the Cabinet is due to meet on Tuesday to decide on what sort of restrictions will be imposed after Christmas – and exactly when they will come into force.

By now, we've all heard the refrain that NPHET advises and the government decides. But Minister for Health Stephen Donnelly said on Sunday that we needed to do everything we could to ensure that hospitals were protected.

And when it comes to protecting our health system, the reasoning from NPHET is nearly always reliant on the now-infamous models.

But how do they work? And what exactly will the modellers be looking at as we head into Christmas and swing into the New Year?

Back in March, as the medical community learned how to treat Covid-19, public health and pandemic experts tried to understand how exactly it spread.

Epidemiologists and researchers worked to get answers to very specific questions – answers that continue to underpin the population-level models that predict the spread of disease.

"Initially, there was no information available, given that it was a completely new disease," Dr John Griffin told RTÉ's Prime Time Explained.

Dr Griffin is an epidemiologist who works on a UCD team that advises NPHET about some basic Covid-19 parameters. These include the incubation period and the latent period, which is the period from a person becoming infected until they become infectious.

"The parameters that I was looking at were the serial interval and the generation time," said Dr Griffin. "And that's really about the spread of disease from one individual to another."

Dr Simon More, who leads the UCD team, emphasised that the models are only useful if they can, as best as possible, reflect reality.

"How long is it from when a person first becomes infected through to when they first show clinical signs? Or what percentage of people are likely to be asymptomatic? Of those that are infected, what percentage will be symptomatic? All of these things were important to know," said Dr More.

These parameters feed into the calculations used to determine the reproduction number – or the number of cases that that are caused by one particular person who gets Covid-19.

The reproduction number is crucial to the population-level modelling, which is then used to inform something else: the potential impact that Covid-19 will have on our hospital system at any given time. This is known as capacity modelling.

And capacity modelling is about two main things: time and space.

"It takes a while for them to become ill, to get tested, to turn up at hospital. So, there's already a lag between the disease onset and turning up in hospital," said Dr Seán Lyons, an ESRI economist who works on capacity modelling for NPHET.

"Once they get to hospital, if they're going to get severe illness, there's probably a further lag. We assume, say, three days between going into a normal hospital bed and needing critical care," he told Prime Time Explained.

If you’ve a lot of people on the road to hospital, you’ll eventually run out of space on the wards.

NPHET now operates on the assumption that, for every 1,000 cases in people under 45 years old, there’ll be 12 hospitalisations and a single admission to ICU.

Older cohorts are more likely to experience severe illness. For those in the 45-to-64 bracket, there will be 50 hospitalisations, 10 admitted to ICU, and three deaths.

1000 cases of people over 65? More again. 100 hospitalisations, 30 ICU admissions, and 30 deaths.

NPHET is pretty confident about these numbers, but they’re not perfectly precise – as the modellers themselves tend to underline.

"Models are just a representation of what might happen, and there is simplification inevitably and especially with as many elements feeding into it," said Dr Lyons.

"Of course, it's accepted that the model is only as good as the the data, the expertise that is used to create them," said Dr More.

Knowing how many people will need ICU treatment is important, but to project capacity, NPHET also analyses how long those beds will be needed.

Somewhat counterintuitively, younger patients who do end up in ICU can have a greater impact on resources. Since they’re more likely to survive, they can end up spending a long time in critical care.

"This is a severe disease in patients, who get quite sick and don't get out of ICU for quite a while. Some patients have an even longer stay – up to 30 days," said Dr Michael Power, an intensive care specialist at Beaumont Hospital.

"And we had one patient, who thankfully has gone home, but was in the ICU for 60 days."

All that means, by looking at the case data as it comes in, analysing the age profiles, and examining the capacity in the system, the modellers believe they can confidently predict the likely impact of last week’s cases on next month’s hospital wards.

In the coming days, as the Government weighs up when and how to introduce new restrictions, NPHET and its modelling sub-group will be looking at the case data as it comes in, analysing the age profiles, and examining the capacity in the system, to try to understand the likely impact of last week's cases on next month’s hospital wards.

The models might be a simplification, but they’re relied upon and trusted by NPHET. They will inform the key decisions that the Government will make in the coming days – decisions that will undoubtedly change lives and livelihoods after a very strange Christmas.

This report is by Edmund Heaphy and Mark Coughlan and was originally published as part of a Prime Time feature on