Why does the coronavirus fatality rate differ so much around the world?

Doug Hendrie

30/04/2020 5:02:13 PM

Inadequate testing, ageing populations and overwhelmed health systems may be key factors for higher death rates, experts say.

Global map
Why do case fatality rates vary so widely in different countries?

Australia has almost bent the curve.
A little over a month ago, the country was seeing more than 400 new cases a day. It is now down to single figures. By that metric, Australia is doing exceptionally well.
But, curiously, Australia is also doing well in terms of our case fatality rate (CFR), a widely used metric measuring the number of reported deaths per number of confirmed cases.
Australia has one of the world’s lowest, with 89 deaths from 6746 cases, giving it a CFR of 1.33%. That is similar to widely-praised Taiwan, which has kept the virus to a 1.4% rate.
That puts Australia at around the range estimated in a Lancet Infectious Diseases study, which found the overall CFR of the new coronavirus was likely to be 1.38%. That figure conceals a significant age gradient, however, with people far more likely to succumb the older they are.
But the CFR sits within a hugely broad range around the globe.
The world’s highest current CFR rate is now in France, where 18.65% of people known to be infected with the virus have died. Belgium and Italy also have double-digit CFR rates, while the US – which has the largest outbreak of anywhere in the world – has a CFR of 5.76% from more than one million cases. 
How can it be that a single virus with a low mutation rate can be more or less lethal in different countries? 

Speed and effectiveness of government responses is a vital factor. Stamping out an outbreak early – as Singapore and Taiwan have done – can keep death rates low through minimising cases and reducing impact on the healthcare system.
But experts say an often-overlooked cause is the fact testing is being done at widely differing rates around the world – and the fact that, in many countries, testing is unlikely to catch even a fraction of asymptomatic cases.
Belgium’s seemingly high death rate, for example, is likely due to the fact it includes deaths likely – but not proven – to be due to the coronavirus, with a government spokesperson claiming their approach is more accurate, and that other countries are undercounting fatalities.
University of Queensland virologist Ian Mackay told newsGP that testing is likely to be the key.
‘My crude presumption has been that in jurisdictions where rates of death are above 1%, there has been too little testing to capture the denominator,’ he said.
‘I think the majority of these examples can be traced back to too little testing or starting testing too late in that jurisdiction’s epidemic.’
The true fatality rate in many developing countries, where widespread testing has been unavailable, may never be known. But even in wealthy countries, the actual number of people infected may remain a mystery.
‘The true number of cases will never be known,’ Associate Professor Mackay said.
The UK-based Centre for Evidence-Based Medicine (CEBM) provides a list of reasons the CFR varies so greatly in an article published in March: 
  • The number of cases detected by testing will vary considerably by country
  • Selection bias can mean those with severe disease are preferentially tested
  • There may be delays between symptoms onset and deaths, which can lead to underestimation of the CFR
  • There may be factors that account for increased death rates such as coinfection, more inadequate healthcare and patient demographics
  • There may be increased rates of smoking or comorbidities amongst the fatalities
  • Differences in how deaths are attributed to coronavirus – dying with the disease (association) is not the same as dying from the disease (causation)
Professor of Evidence-Based Medicine at Bond University Paul Glasziou told newsGP the CEBM analysis showed these factors could skew the CFR by almost 100-fold.
‘The CFR is confounded by both missed cases from the denominator and insufficient time for deaths to occur missing from the numerator,’ Professor Glasziou said.
newsGP examines a number of these factors.
Testing not catching anywhere near every case
If someone dies of the virus without being tested, they aren’t counted in the CFR. Similarly, if someone contracts the virus and gets well by themselves without being tested, they aren’t counted as a survivor. 
That makes a second indicator – the Infection Fatality Rate (IFR) – more important, given this indicator captures the rate of deaths among confirmed cases, as well as those with undetected disease (asymptomatic and untested).
But the IFR is harder to pin down, given studies have estimated the actual number of infections is much higher than the number of people proven to have the virus through testing.
Ranges for the actual number of asymptomatic versus symptomatic infections vary significantly. One British Medical Journal study puts the asymptomatic rate at four times the symptomatic rate.
Meanwhile, a widely publicised pre-print study estimates the number of undetected cases to be as much as 50 times greater than confirmed cases, based on antibody testing. This higher figure has been contested, however.
Immunology Professor Peter Doherty told newsGP this study seems ‘wildly off’.
‘I doubt very much that’s right. It might be more like 2–3 times higher, based on South Korea’s testing regime,’ he said.
‘If that figure of 50 times higher was true, we would start to see the disease turning down with herd immunity. But we’re not seeing that at all.
‘The turndown is due to better social distancing.’
Professor Glasziou agrees.
‘There was a lot of interest in that study, as it could have taken into account infections missed by the testing regime,’ he said. ‘If it was correct, it meant the number actually infected was far larger than we thought.
‘But there were lots of flaws with that study – the volunteer bias and the fact it could be accounted for by false positives.
With a CFR of 3.19%, Germany’s success relative to many other nations in Europe may be linked to a decision to test early and broadly, to find emerging cases as quickly as possible.
Overwhelmed health systems
‘There are likely to be instances where very overwhelmed healthcare services have been unable to care for all severe COVID-19 cases or where excess deaths have occurred because COVID-19 reduced the resources needed to care for different acute illnesses and injuries,’ Associate Professor Mackay said.
The worst affected regions of Italy, Iran and the US have experienced periods where the numbers of serious and critically ill patients needing hospital care have overwhelmed the system. In these situations, larger numbers of deaths are likely – boosting the CFR.
Professor Glasziou said that, to date, we have not seen the worst-case predictions of deaths eventuate. But what we are seeing, he suggests, is geographic pockets where health systems have been swamped.
Professor Glasziou stresses that the pandemic is far from over, with second waves occurring in some Asian countries that had appeared to control rates.
New hotspots are emerging, such as Brazil, where low testing rates and denial of the severity of the virus at high levels have led to hospitals in major cities becoming overwhelmed by patients, with the death toll expected to surge.
The virus is more lethal to vulnerable groups
CEBM researchers suggest comorbidities and age may partly explain why some areas are particularly badly hit, such as Italy.
They note that Italy has the second-oldest population in the world, with the highest rates of deaths due to drug-resistant microbes, as well as a high rate of smoking, a factor associated with poor survival.
Hypertension, cardiovascular disease, chronic respiratory conditions, diabetes and cancer are also risk factors.
Nations with older populations – who are more likely to have comorbidities – are likely to be more at risk.
What can we learn?
Looking to the future, one option may be to use all-cause mortality to better gauge the impact of the coronavirus.
Official statistics are ‘vastly underestimating’ the true death toll across the world, according to Melbourne University epidemiologist Professor Alan Lopez.
‘The all-cause death rate is the very best way to assess the overall net impact of COVID-19 on mortality,’ he told the ABC.
Countries around the world are now turning to this method to gauge the impact of undiagnosed cases.
The Financial Times estimates there have been at least 122,000 more deaths than in a usual year across 14 countries, suggesting that the true death toll may be substantially higher.
While some of these deaths may be due to other causes than the coronavirus, the fact the death rate is so much higher than a normal year indicates the pandemic is playing a major role.
In the badly hit UK, for example, the number of excess deaths is surging.
One in 17 residents in aged care homes in the UK may have died in just five weeks, the New Stateman reports, while the ABC reports the number of ‘excess deaths’ – deaths above a usual year – have more than doubled in recent weeks, with 22,351 in the week ending 17 April. A normal year would have 10,000 deaths.
Professor Glasziou said that scanning all-cause mortality for the coronavirus in Australia would only show a small increase, given how few cases the nation has had.
‘But in New York now, you can see a huge increase in all-cause mortality,’ he said. ‘That’s suggesting either an underreporting of COVID-19 deaths, or increases in other causes of mortality because people are avoiding healthcare.
‘We don’t know which one that is.’
Log in below to join the conversation.

case fatality rates coronavirus COVID-19 death rate

newsGP weekly poll Would the reintroduction of MBS items for GPs to trace and report on electrocardiogram (ECGs) be beneficial to patients?

newsGP weekly poll Would the reintroduction of MBS items for GPs to trace and report on electrocardiogram (ECGs) be beneficial to patients?



Login to comment

Dr Horst Paul Herb   1/05/2020 10:35:06 PM

We should stop talking about "case fatality rates" when the number of cases remains unknown. It is just plain nonsense - especially in countries where no random community testing has taken place, such as in Australia.

Much fairer and more comparable to use population fatality rates, though even those are very hard to compare when countries use widely different methods or reporting deaths (eg in Norway a doctor has to fill in a death certificate stating that covid-19 was the most likely cause of death, while Sweden matches their death registry with the registry of covid-19 positively tested patients, and yet other countries reporting all nursing home and respiratory infection deaths currently as covid-19 deaths). It gets even more complicated when even with a country different states report differently, eg in Germany.

One of the very few countries providing reliable data based on testing is Iceland.

A.Prof Christopher David Hogan   3/05/2020 12:54:37 PM

Another possibility may be that other strains of Coronavirus apart from COVID-19 , MERS & SARS provide cross immunity eg from the many other causes of the common cold

Dr Sally Dunbar   11/07/2020 8:13:29 AM

There may be genetic factors at play too that predispose or protect certain ethnic groups. Eg there are suggestions that people with blood group A might have a higher mortality rate.