YOUR NUMBER’S UP

It’s killed millions, but how deadly is Covid-19?

Just how lethal is the novel coronavirus? This is a question almost all of us have asked. It’s also a question with no easy answer.

A plethora of underlying factors means a simple one-sentence synopsis is out of reach. Men are more likely to die from Covid. People with comorbidities, like diabetes or heart disease, face a greater risk – illustrating the dangers to Ma¯ ori and Pacific people.

The stability of a country’s health service (do doctors actually have time to treat someone?) factors in. Are those health services easy to reach? What are Covid treatments like?

There are two variables that tower over the others.

Age is by far the most important factor in determining someone’s risk to Covid-19; more so than sex, occupation, ethnicity and even underlying health status.

According to US data more than 75 per cent of Covid-19 deaths have occurred in people over 65. People aged 75 to 84 were 150 times more likely than those in the 18-to29-year old age group.

You can see a hint of this in New Zealand’s (relatively low) Covid numbers: 32 of 42 deaths (as of yesterday) were in people over 60.

And then, full vaccination reduces the risk of death by 94 per cent across all age groups in the highly influential Te Pu¯ naha Matatini (TPM) Covid modelling, Professor Michael Plank said.

This assumption is consistent with that of UK health officials which suggest the Pfizer vaccine is effective at preventing 90 to 99 per cent of mortalities.

How do you actually work this out in the first place?

Scientists use two measures to determine how deadly a disease is. The first is case fatality rate (CFR) or the proportion of deaths among confirmed cases – a blunt, incomplete measure. As of yesterday, the CFR in New Zealand was about 0.4 per cent. There were 10,789 cases and 42 Covid deaths. A new death, of a patient in their 80s, was reported yesterday.

Infection fatality rate (IFR), or the proportion of deaths among all infections (not just diagnosed cases), offers a more complete picture. There have been some 250 million confirmed Covid-19 cases reported across the world, but this number only includes those who have had a positive test. There are clearly many more people who suffered mild (or no) symptoms and have never been tested.

So to accurately determine the number of infections, researchers use seroprevalence surveys (blood testing), to work out the percentage of people in the population with Covid-19 antibodies.

Researchers may, for example, take a sample of 1000 people and find out how many have had the virus, and then extrapolate out across the population as a whole. Once they have that number, they take official death rates and work out how deadly Covid-19 is.

IFR is clearly much better than CFR, but even then it varies significantly from country to country. Our tool is based on a study that compiles a range of other studies on fatality rates.

So, it’s tricky?

Yes, very much so. A good IFR estimate still requires accurate reporting of Covid-19 deaths.

Accurate antibody tests to determine how many people in total have, or have had, Covid-19. There were concerns last year that tests were not reliable.

What’s more, it’s becoming ever harder to put your finger on how deadly Covid is. The data that underpins our tool is from early to mid-2020. Working out the IFR at the time was arguably easier because there were simply fewer variables. There was one dominant Covid-19 variant and there were deaths. There were no vaccines and doctors were still learning how to treat the disease.

The Daniel Herrera-Esposito study has recently been updated to acknowledge that fact.

When I asked Herrera-Esposito about these uncertainties, he also volunteered that improved Covid19 treatments would reduce how dangerous the disease is.

‘‘My takeaway in the paper is that there is quite some uncertainty around how the severity rates have changed. It seems clear that better treatments did reduce disease severity, but it’s not clear to what extent (maybe reduced deaths ~50%, maybe less).’’

To illustrate that fact, he shared two studies, one from Norway and one from Denmark, which showed a drop-off in IFR from the first Covid-19 wave to the second. These studies too have significant limitations and can’t be applied universally.

OK. So what do we know?

A very early New Zealand model assumed an IFR of 1 per cent, and therefore projected tens of thousands of deaths, assuming we enacted no restrictions.

Clearly, our understanding of the virus has evolved since early 2020. The current TPM modelling on the Auckland August outbreak also includes a breakdown of Covid-19 risk in a range of different age brackets.

The probability of death in these numbers (and this is before someone is vaccinated) is informed by two studies. The first was published in late March 2020 and is based on data from mainland China. The paper estimated an overall infection fatality ratio for China was 0.66 per cent.

I asked one of that paper’s authors, Dr Robert Verity, about its ongoing use. He pointed out that the IFR figure it landed on was consistent with a Disease Control and Prevention (CDC) estimate of 0.65 per cent, which came months later.

The other study was published in October 2020, and brings together a range of research on Covid’s lethality. It found the death rate was very low for children and young adults, and that the virus kills 4.6 per cent of those aged 75 and 15 per cent of those aged 85. Verity said it would be better if Covid modellers used this study, or similar. While his study holds up, it’s just a touch out of date.

Plank told me TPM is updating its Covid modelling based on the pre-print Herrera-Esposito study. It includes data that represents 5 per cent of the world’s population and builds on the work of the October 2020 paper.

The Herrera-Esposito study finds (once again) that the risk of dying from Covid-19 increases exponentially with age, but also articulates the risk of serious illness and critical illness.

Interestingly, the authors note, the ‘‘rate of increase in the risk of severe and critical disease outcomes with age is less marked than the rate of increase in lethality’’. This example explains it: ‘‘According to our estimates, people in the 20-25 years old range are on average 836 times less likely to die from Covid-19 than 70-75 year old people, 146 times less likely to develop critical disease, and 41 times less likely to develop severe disease.’’

Isn’t Delta more deadly?

This is complicated but stick with me here, as this very much informs our understanding of how dangerous Covid-19 is. The Verity and Herrera-Esposito studies are based on Covid-19 data that predates the Delta variant.

This is an issue, as a number of other studies suggest the now dominant Delta variant is more dangerous than wild-type Covid-19 (and the Alpha or British variant). Without getting into the maths, essentially this has meant modellers like Plank need to factor in that severity in their calculations.

In the August modelling, for example, the TPM team state: ‘‘Infection fatality ratios are . . . adjusted by an odds ratio of 2.32 for the Delta variant.’’ The Doherty Institute in Australia includes a similar assumption in its modelling.

The TPM modellers cite two key studies on the severity of Delta. One, from Canada, found Delta increased the risk of death.

There is disagreement on this. In a recent piece in The Guardian, Gregory Dore, an epidemiologist at Australia’s Kirby Institute, suggested the variant’s severity may have been overestimated.

And Herrera-Esposito said: ‘‘I would say that taking IFR estimates such as mine, and correcting for variants of concern but not for improvements in treatment, is very likely to lead to an overestimation of the rates of severe disease/death. To what extent, or even in what direction those two forces changed the overall IFR from then to now, seems uncertain to me.

‘‘Although many studies point to an increase in disease severity with Delta, it doesn’t seem to me like there is a clear scientific consensus regarding how much severity may have increased because of variants. It is possible that it is less than what is found by Fisman et al [the Canadian study], which is

a single study with its limitations.’’

A paediatrician suggested the consensus was Delta is no more severe in children, there’s just more of it as it’s more infectious.

The TPM modelling also referenced a UK study that informs how the modellers determined the Delta hospitalisation risk.

I asked Verity about this. He told me: ‘‘There are other studies that come to similar conclusions but also based on small numbers. For this reason, I think it’s defensible for The Guardian piece to report that it’s possible that we overestimated the severity of Delta, as it is possible.

‘‘It’s also quite difficult to compare fatality rates between variants (at least based on raw numbers) due to times at which they peaked, combined with changing vaccination coverage.’’

What did Plank make of this? Could the Delta risk have been overestimated? He pointed me to recent Public Health England data on case numbers and deaths in unvaccinated people that appears to line up closely with fatality rates factoring in the Delta assumption. He urged caution (obviously not all unvaccinated infections are detected) but pointed out they broadly line up with Delta-adjusted IFR.

‘‘I’ve certainly seen nothing to suggest that making the adjustment for Delta virulence results in . . . overestimates.’’

Australian epidemiologist Tony Blakely also pointed to this Scottish study, which suggested a significant rise in hospitalisation risk from Delta.

I asked another Australian epidemiologist, Prof Alexandra Martiniuk,

about this. She shared the Australian Agency for Clinical Innovation’s guidelines on Delta with me. They state there is an ‘‘increased risk of emergency care and hospitalisation, including higher odds of oxygen requirement, admission to an intensivecare unit, and possible increased risk of mortality.’’

These uncertainties are why, in our tool, we show two IFRs. One is based on 2020 data from the Herrera-Esposito study. The other bakes in that assumption TPM use for Delta.

If Dore is correct, and we’ve got the severity of Delta variant wrong, that particular IFR in our tool will be an overestimate. What’s more, we haven’t factored in the difference improved treatments for Covid make.

Herrera-Esposito makes this point: ‘‘I would say that improvements in treatment from early-mid 2020 to the present day likely had an effect comparable to the highest estimates of the effect of variants. To correct for the second but not for the first is likely to overestimate severity. And there still seems to be quite some uncertainty as to the magnitude of the effect of variants on disease severity.’’

Plank told me TPM would certainly consider building Covid-19 treatments into their modelling as the situation evolves.

What we do know is that Delta is much more infectious. A person with that variant will infect six others on average (assuming no public health mitigations like masks or vaccines). A person with wild-type Covid infected, on average, 2.5 others.

So even if Delta’s IFR isn’t higher, the fact that it’s more contagious means higher case numbers are likely. And more cases means more hospitalisations and deaths.

But can’t I just see how deadly Covid-19 is from looking at cases here?

As of yesterday there had been 11,114 confirmed cases of Covid-19 (including 145 reported yesterday), and 42 people had died with the virus. The ministry defines people who died of Covid as the following, which complicates matters further: ‘‘All cases that died who were classified as an active case of Covid-19 at the time of death. In some of these cases, the underlying cause of death may have been unrelated to Covid-19.’’

In a statement they said 10 of those fatalities were still under investigation to clarify the cause of death. This equates to a case fatality rate of 0.4, as mentioned. Remember, this is not an infection fatality rate, or IFR, as there will almost certainly be unknown cases not counted in the total.

This may seem low. But the dataset is too small to make sweeping statements about how deadly Covid-19 is. Most cases in Aotearoa have been in people under 50. Four people under 50 have died. Only 8 per cent have been in those older than 60; 32 people over 60 have died. Only 840 people (as of Friday morning) over 60 have had Covid-19 – a CFR in that age group of just under 4 per cent.

Age groups over 80 account for less than 1 per cent of cases but nearly half of the deaths.

Some 16 per cent of New Zealand’s population is 65 or older.

The other complicating factor is most Covid cases have emerged since August. Over that time, we’ve rolled out a huge number of vaccines, which will have neutered the severity of Covid.

The takeaway here is that the numbers in New Zealand are not representative of the risk to the population as a whole.

What are the other takeaways from all this?

The risk of death for young people is very low. However, young people with underlying conditions are still at risk, and death or ‘‘being OK’’ aren’t the only two Covid outcomes.

The Herrera-Esposito study looks at rates of severe illness and critical illness in people who catch the virus. It compared those with the risks of vaccine-induced myocarditis, a rare vaccine side-effect.

‘‘Using the [alternative] rate of myocarditis of 1 per 80,000 estimated by the CDC for the 16-to-30 age group, the rate of severe disease [from Covid-19] is 322 times larger, the rate of critical disease 42 times larger, and the risk of death is 4.4 times larger,’’ it noted.

So the maths support getting vaccinated even if you’re young. In the US state of Minnesota, the hospitalisation rate for vaccinated people between 18 and 48 is about 1 per 100,000, The New York Times has reported. And a healthy vaccinated person in this age group is less likely to pass on the disease to more vulnerable people.

The other standout is the harrowing increase in risk as people age. Older people – even if they’re vaccinated – are incredibly vulnerable to this virus. While vaccination hugely reduces the risk of older people dying, it doesn’t make it zero.

According to the data, the chance of a fully vaccinated person over 85 dying from Covid-19 is 1.4 per cent (that’s the optimistic scenario as it excludes the Delta assumption). That’s still a sizeable risk and speaks to two things: that older fully vaccinated are going to die with Covid-19 if it spreads widely, and that there’s an ongoing need to protect those people.

As Stuff has reported, worldwide, on average, the virus has cut short lives by an estimated 16 years per death, or about 10 years in high-income countries.

Just because someone is old doesn’t mean they’re about to die. It doesn’t mean they’re not worth protecting.

Which leads us finally to something that Plank calls the ‘‘25-year difference’’. Being fully vaccinated essentially makes a person 25 years younger to Covid-19. This is why an unvaccinated person in their 50s faces, broadly speaking, a similar risk of death to a vaccinated person in their mid 70s. An unvaccinated person in their late 50s, who may well be at considerable danger from the virus, is transformed into a person in their early to mid 30s after two jabs.