To die with or to die from? That is the question. Australia’s federal tally clocks our COVID-19 death toll at 810 (seven deaths listed in Victoria this morning are yet to be included).
But recent questions around whether a person died from COVID-19 or with it has made many wonder just how accurate this tally is.
Just 9% of Australians who died from the virus had no co-morbidities.
How deaths get added to the tally
Epidemiologist and senior lecturer at the University of Melbourne Dr Alex Polyakov told Crikey that in Australia a death is attributed to COVID-19 if the person tested positive and there was no other reason for them to have died.
When filling out death certificates, doctors have to determine the main cause of death, followed by additional causes. Only those whose main cause of death was listed as COVID-19 were counted in the federal tally, he said.
“Everyone has to die of something,” he says. “Someone in their 90s who has a good death, they go to sleep and just don’t wake up. That could be attributed to a heart attack or a stroke.
“If they tested positive the day before they went to sleep, they would be classified as a COVID-19 death — even though [it’s] likely it’s just because their time has come.”
The classification is much easier in younger, healthier people. One man in his 20s was reported to have died with COVID-19 and a coroner investigated. His death was initially reported in Victoria’s daily death toll but wasn’t added to the federal death tally.
This, Polyakov said, was probably because he died of another clear reason unrelated to COVID-19: “Testing positive for the virus may have been incidental”.
How solid are the numbers?
In the early days of the pandemic, COVID-19 may have been deadlier than initially reported. Australia had more than 800 “excess deaths” in the first quarter of 2020 compared with the previous five-year average, although just 103 deaths were classified due to COVID-19.
But Polyakov said the data was solid now. “The deaths are a terrible tragedy for families, but they also show how well Australia’s response has been compared to the rest of the world,” he says.
The tally may fluctuate slightly when cases are reexamined.
While it’s a tragic toll, Australia’s death rate is much, much lower than most countries. The death rate in those infected with the virus has increased over the past month from 1.3% to 2.3%, partly due to delayed reporting in aged care home deaths.
Around the world, about 3.4% of those who tested positive for the virus have died. Mexico has the highest case-fatality ratio at 10.6%. The US has the highest death toll: more than 194,000 deaths and 6.5 million confirmed cases. Its case fatality rate is 3%.
Amber I’d like to acknowledge my appreciation for this article and for your efforts in bringing it to print.
The issue at question involves metrology: the science of measuring things. [https://en.wikipedia.org/wiki/Metrology] Good metrology depends on what information there is to measure, what processes and standards we can effectively use, the periods over which we apply them, and how we mean to eventually use the information we gather.
In an emerging disease like Covid-19 we’re still rapidly learning how it affects people, what acute and chronic symptoms are, how accurate infection testing is, and how Covid-19 interacts with other conditions (i.e. comorbidities.) Until we are confident in our understanding there will be more than one way of counting infections, deaths and Infection Fatality Rates. Also, when our knowledge changes it becomes hard to recount past measures, it’s confusing to policy analysts when we do, and it’s also hard to compare between jurisdictions that have different counting methods and reporting/reconciliation cycles.
I am confident that good health policy analysts are aware of the challenges and will adapt, but our politicans tend to be STEM-illiterate and are used to cherry-picking facts for rhetorical argument anyway: if they know about the issues they may not care; if they care I don’t think it changes much of what they do — especially when it’s politically expedient for the next news cycle.
And our journalists are almost as STEM-illiterate as our politicians and if they don’t get this stuff right then all we have left are armchair experts on Twitter, and most of them are useless.
I know that in past reporting you have skirted around the math and occasionally I’ve criticised you for that, but in this article you tackled it head on, and I believe it’s to significant public benefit.
Thank you, and my compliments.
Well writ, Ruv Drava.
All the current “statistics” are questionable in view of their means of collection as well as the huge raft of unknowns in respect of the virus itself.
On this topic an ample supply of salt for the myriad of pinches of it is required to consider the factors in detail.
Covid death reporting is far more complex than the article hints at.
For starters, any comparison of “case fatality ratios” is nonsensical – presently it is merely a reflection on the number of tests per capita conducted.
Given that the majority of infected will remain asymptomatic or have only mild symptoms, in most cases indistinguishable from the common cold, it is a certainly that the majority of infected do not get tested and hence will not enter the tally, unless the country or region is one of the very few that conducted systematic testing of communities (eg Germany, Sweden, California). To date, Australia has not conducted any systematic community testing, our incidence rate is guesswork.
Next question is what counts as a covid death. Sweden for example automatically matches its covid test registry with the register of deaths – any matches count as covid deaths, regardless of actual cause of death. This includes post mortem covid test results. In Norway, a doctor has to write on the death certificate that he is convinced that the cause of death was primarily covid and has to report it to a registry, before the nation will count it as covid death. Sweden would hence be more likely to over-report, Norway more likely to under-report covid deaths.
One way to analyse reported covid deaths is to investigate correlation with reported seasonal excess deaths. The economist ran such analysis, and showed that Sweden for example could explain 102% of their excess deaths with covid – meaning that their figures are likely very close to the truth. Other countries, eg Spain or the UK, reported far fewer covid deaths than their seasonal excess deaths suggested: they are most likely under-reporting. A few countries reported far more covid deaths than they had excess mortality – eg Belgium. They most likely seriously over-reported.
The only figure that becomes increasingly clear is that in all countries with a well functioning health system, deaths mainly occur in nursing home patients, the very elderly, and people with serious co-morbidities, eg the morbidly obese. In people below the age of 60, there seem to be less deaths compared to a bad influenza year. In countries with poorly functioning public health systems, death rates soar, including in younger people.
However, whatever measures governments undertake, liberal or draconian, seems to have little to no effect on the covid death tally in the end – all we seem to be able to do without a working vaccine or treatment is delaying inevitable deaths in the very elderly. The only other difference will be in the collateral damage caused by the various interventions,
“Given that the majority of infected will remain asymptomatic or have only mild symptoms“
I’ll pull you up there Horst. Asymptomatic cases, by definition, will be impossible to measure, and difficult to guesstimate, but the figures given by those studying this suggest 20%, not a majority.
Those with mild symptoms are largely getting tested, everyone I know who has had the slightest sniffle or sore throat has been getting tested. At best, or worst, perhaps a quarter of the population who has Covid aren’t being shown in the stats.
And while ‘excess deaths’ is an excellent analytical tool, it is just that. When you then factor in that flu deaths have been reduced by 90%+ this year, you have to add them back in to arrive at comparable figures. I’m not sure those ‘excess deaths’ methodologies have done that.
I have been an actuary for 40 years and mortality rates have therefore been an integral part of my professional life. It can seem a bit ghoulish, but counting the dead and understanding why they died is very important.
The debate about dying with versus of some infection is not new. It regularly gets dragged out when considering Flu, Measles, Malaria and a whole lot more. Covid is just the latest. The point in this debate is to remember that people can die with a heart condition, or with obesity, or with diabetes. There is nothing special about Covid just because some people would prefer its impact to appear less severe so that less severe containment measures are taken.
It is difficult in some circumstances to decide exactly what the cause of death is, but in others it isn’t. Someone with Covid killed in a car crash or a shooting clearly died with the virus. Someone with no previous diagnoses who develops severe symptoms consistent with the virus and dies, died of the virus.
Someone with one or other co-morbidity who contracts the virus, develops severe symptoms and dies also dies of the virus. The logic is reasonably simple. The virus is the precipitating event and therefore the cause. Just having another condition is not, of itself, a reason to blame that cause.
Someone with severe co-morbid symptoms who contracts the virus and has the symptoms worsen is possibly someone who died with the virus. The question again is the precipitating event. Was it likely that the symptoms would have got to that level in that time frame without the virus? If yes, they died with the virus. If no, they died of it.
The number of years of life expectancy lost due to a death is not the deciding factor. Many of those who die of the virus would have undoubtedly died reasonably soon, but not now.
Agreed Alun, this is not new, it was happening long before Covid came along and has always presented a problem statistically. And doctors aren’t all equally infallible when signing a death certificate. The ‘cause of death’ statistics must always be read with this in mind.
But it is always the Primary Cause, it is just that this isn’t always clear as you helpfully point out.
A person who has a heart attack or stroke while driving a car will be an automobile accident death, every time.
The effect of Corona virus and associated control measures on the elderly and their families has been devastating. However, in cold statistical terms, we need to compare the figures of nursing home deaths from January to now with the same period for the last two years to dissect out the net effect of the Corona virus vis-á-vis normal background mortality. Clearly in nursing homes many of the deaths brought forward by Corona virus would have occurred from other causes within a short time.
Many nursing home patients are already in palliative care with a dire outlook. The average survival for a ninety year old entering a nursing home in a Covid-free environment is measured in months rather than years. Some Corona deaths that would have been caused by the regular incidence of flu have been avoided by the current mild flu season, a side benefit of measures taken to contain Corona. For reasons that not clear, there aren’t as many people in Aged Care this year than last.
In summary it would appear that there fewer tears shed and fewer funerals held for people who have died in Victorian nursing homes in 2020 compared with 2019. Surely this cannot be true. Could someone please enlighten me.
There is also the added complication of the deaths caused or brought forward by medical resources being otherwise deployed.
Not to mention fewer road deaths and general ‘flu transmission and other communicable diseases by ceasing the millions mingling heedlessly.