How deadly is covid-19?

September 2020 was the least deadly month in Swedish history, in terms of number of deaths per 100,000 population. Ever. And I don’t mean the least deadly September, I mean the least deadly month. Ever. To me, this is pretty clear evidence of two things. First, that covid is not a very deadly disease. And second, that Sweden has herd immunity.

When I posted this information on my twitter feed, the response from proponents of further lockdown was that the reason September was such an un-deadly month, was because everyone has already died earlier in the pandemic. To me, that seems like a pretty self-defeating argument. Why?

Because 6,000 people have died of covid in Sweden, a country with a population of 10,000,000 people. 6,000 people is 0,06% of the population. If it is enough for that tiny a fraction of a population to die of a pandemic for the pandemic to peter out so completely that a country can have its least deadly month ever, then the pandemic was never that deadly to begin with.

In August, I wrote an article where I proposed that the mortality for covid is only 0,12%, roughly the same as influenza. That number was based on a back-of-the-envelope calculation. I figured that, since the death rate had dropped continuously for months and was at very low levels, Sweden must have reached a point where it had herd immunity. And I figured that at least 50% of the population must have been infected for herd immunity to have been reached. 50% of Sweden’s population is five million people. 6,000 / 5,000,000 = 0,12%

At the beginning of October, one of the World Health Organisation’s executive directors, Mike Ryan, said that the WHO estimated that 750 million people had so far been infected with covid. At that point, one million people had died of the disease. That gives a death rate for covid of 0,13% . So the WHO said that the death rate is 0,13% . Not too far off my earlier back-of-envelope estimation. This of course begs the question why there are continued lockdowns for a disease that is no worse than the flu.

A short while later, the WHO released an analysis by professor John Ioannidis, with his estimate of the covid death rate. This analysis was based on seroprevalance data, i.e. data on how many people were shown to have antibodies to covid in their bloodstream at different times in different countries, which was correlated with the number of deaths in those countries. Through this analysis, professor Ioannidis reached the conclusion that covid has an overall mortality rate of around 0,23% (in other words, one in 434 infected people die of the disease). For people under the age of seventy, the mortality rate was estimated at 0,05% (in other words, one in 2,000 infected people under the age of 70 die of the disease).

As I’ve discussed before, I don’t think antibody data gives a very complete picture, since there are studies showing that a lot of people don’t produce measurable antibodies in their bloodstreams, but still have immunity, either thanks to a T-cell response, or thanks to local antibody production in the respiratory tract. So I think that the fatality rate is significantly lower than what the analysis by professory Ioannidis found, and more in line with what the WHO stated earlier in October.

But even if the antibody based number is the correct number, then covid still is not a very deadly disease. For comparison, the 1918 flu pandemic is thought to have had an infection fatality rate of 2,5%, i.e. one in forty infected people died. So the 1918 flu was 11 times more deadly than covid if you go by professor Ioannidis antibody based numbers, and 19 times more deadly than covid if you go by the fatality rate provided 12 days earlier by the WHO’s Mike Ryan.

And this is missing one big point about covid. The average person who dies from covid is over 80 years old and has multiple underlying health conditions. In other words, their life expectancy is very short. The average person who died in the 1918 pandemic was in their late 20’s. So each death in the 1918 pandemic actually meant around 50 years more of life lost per person than each death in the covid pandemic. Multiply that by the fact that it had a 19 times higher death rate, and the 1918 flu was in fact 950 times more deadly than covid, in terms its capacity to shorten people’s lives.

Ok, I’ve discussed the fatality rate of the 1918 flu pandemic, and compared that to covid. But what about the fatality rate of the common cold viruses that are constantly circulating in society? How does covid compare to them?

Many people think that the common cold viruses are harmless. But in fact, among elderly people with underlying health conditions, they are frequently deadly. A study carried out in 2017 found that, among frail elderly people, rhinovirus is actually more deadly than regular influenza. In that study, the 30 day mortality for frail elderly people admitted to hospital due to a rhinovirus infection was 10% . For frail elderly people admitted to hospital due to influenza, 30 day mortality was 7% .

What is my point? If you are old and frail, and have underlying health conditions, then even that most harmless of all infections, the so called “common cold”, can be deadly. In fact, it often is. Covid-19 is not a unique disease, and does not appear to have a noticeably higher mortality rate than the so called “common cold”.

There is one final aspect to all this that needs to be discussed. And that is the effect of covid on overall mortality. If it turns out that covid has no effect on overall mortality, then that really brings in to question why we are locking down, since we’re not actually preventing any deaths. So, what is the effect of covid on overall mortality?

Let’s look at Sweden, since that is perhaps the country that has taken the most relaxed approach of any to preventing spread, and which should therefore also reasonably be expected to have had the highest impact on its overall death rate. From January to September 2020, Sweden experienced 687 deaths per 100,000 population. The last time Sweden had a deadlier year was 2015. Personally, I don’t remember any big deadly pandemic happening in 2015.

In fact, 2020 is so far one of the least deadly years in Swedish history, and is largely in line with the average for the preceding five years. To be precise, it is 2,7% higher than the average for the preceding five years, which is well within the margin of error. In 2019, mortality was 6% lower than the average, so it should be expected that 2020 would have a slightly higher mortality than average, even without covid.

What does this mean? It means that covid, a supposedly deadly viral pandemic, has not killed enough Swedes to have any noticeable impact on overall mortality.

How can this be explained, when we know that 6,000 Swedes have died of covid?

As I see it, there are two possible explanations. The first is that most people who died “of” covid actually died with covid. In other words, they had a positive covid test and were therefore characterized as covid deaths, when the actual cause of death was something else. The second is that most people who died of covid were so old, and so frail, and had so many underlying health conditions, that even without covid, they would have died by now. There are no other reasonable explanations.

I am not saying that covid is nothing, or that it doesn’t exist. I am saying that it is a virus with a marginal effect on longevity. And yet, public policy in most countries has been driven by doomsday scenarios based on completely unrealistic numbers. To put it simply, we’ve acted like we’re dealing with a global ebola outbreak, when covid is much more like the common cold.

UPDATE (26th October 2020): After SCB updated their numbers it has become clear that September 2020 was in fact the second least deadly month in Swedish history, not the least deadly month. That award goes to June 2019.

You might also enjoy reading my article about why I think Sweden has herd immunity, or enjoy watching my conversation with Ivor Cummins of Fat Emperor about covid-19.

I am rolling out a ton of new science-backed content over the coming months, including:

- Analyses of the benefits and risks of all common supplements and medications
- The keys to a longer, healthier life (possibly quite different from what you may have heard)
- A long-term follow-up of the health consequences of the covid pandemic and global lockdown.

Please provide your e-mail address below and you will get all this content straight to your inbox the moment it is released.

Join 8,084 other subscribers

Author: Sebastian Rushworth, M.D.

I am a practicing physician in Stockholm, Sweden. My main interests are evidence based medicine, medical ethics, and medical history. I frequently get asked questions by my patients about health, diet, exercise, supplements, and medications. The purpose of this blog is to try to understand what the science says and to translate it in to a format that non-scientists can understand.

125 thoughts on “How deadly is covid-19?”

  1. I agree with your entire article except for herd Immunity. The confirmed infections in Sweden in the Worldometer data shows a huge spike. The deaths have not increased….which is great. Could you please comment.

  2. But infected people who are asymptomatic are contagious. So it is indeed relevant to count them from a epidemiological perspective.

  3. Dr Rushworth, some further brief comments and queries about some claims in your article where you assert the following: “From January to September 2020, Sweden experienced 675 deaths per 100,000 population. That is less than both 2017 and 2018”).

    I like to review relevant data and check things for myself wherever possible, so I went to to access relevant data and then did some quick calculations.

    I used the same population figures as were used by the author of the calculation you cited (i.e. the same population statistics used by and I accessed official Swedish death statistics from here:

    I then calculated the total number of deaths for Jan-Sep for each year during 2015-2020 and used these figures to calculate deaths per 100,000 people in Sweden for the Jan-Sep period for each year, which provided the following results:

    2020: 678 deaths per 100,000 people
    2019: 618 deaths per 100,000 people
    2018: 662 deaths per 100,000 people
    2017: 662 deaths per 100,000 people
    2016: 660 deaths per 100,000 people
    2015: 688 deaths per 100,000 people

    According this analysis, the mortality rate in 2020 is higher than 2017 and 2018, and the mortality rate in 2020 is 3% higher than the 2015-2019 average (which is 658 deaths per 100,000 people). The 3% increase in mortality (compared with recent years) may provide another (better?) way of estimating the epidemiological significance of the COVID-19 pandemic in Sweden?

    It might be a good idea to check the numbers yourself and see if you think these claims in your article need to be corrected.


    1. Hi Stephen,

      Thanks for your comments, and it’s great that you’re double-checking the data. I admit that I didn’t check the SCB numbers directly myself, I trusted that the numbers provided by Lars Wilderäng of the Cornucopia blog were correct. He is usually very careful to always make sure his numbers are correct. I’ve now gone through the numbers myself in the excel sheet.

      You are correct that 2020 is not less deadly than 2017 and 2018 so far, although it is less deadly than 2015. I will modify the article to reflect that. I get slightly different numbers from you though when I run the calculations, which I guess could be either due to the rounding errors if we are looking at different parts of the spreadsheet, or due to the fact that SCB has modified the numbers since you did your calculations. That might also explain why Lars Wilderäng reached still other numbers.

      Here are the numbers I get (looking at January-September):

      2020: 687 deaths per 100,000
      2019: 633 deaths per 100,000
      2018: 677 deaths per 100,000
      2017: 676 deaths per 100,000
      2016: 671 deaths per 100,000
      2015: 692 deaths per 100,000

      The average for the entire period is 673 deaths per 100,000, or 669 deaths per 100,000 if you exclude 2020. That means 2020 is 2,7% higher than the average for the preceding five years.

      I don’t think that can reasonably be called a covid effect, for two reasons. Firstly, it is within the margin of error, so I think it is completely accurate to say that 2020 is in line with recent years. Secondly, as is clear from these numbers, unusually few people died in 2019 (6% less than average), so you would expect more than average to die in 2020 even without covid.

  4. Thanks for reviewing this and doing your own calculations! Strange that we got slightly different results – the only thing I can think of is that you used different population statistics?

    I used the year-end population statistics for 2015-2019 (available from–the-whole-country/summary-of-population-statistics/), and the most recent estimate (August) for 2020 (available from–the-whole-country/preliminary-population-statistics-2020/).

    The formula I then used is: Total deaths (Jan-Sep) / Population x 100,000

    I downloaded the spreadsheet again and using this approach I get the same results as before for all years except 2020 because some additional deaths have been added to the 2020 column – I now get 679 deaths per 100,000 population (not 678).

    (NOTE: I used Table 1 in the death statistics spreadsheet and added up the deaths from January 1 – September 30).

  5. There’s lots of disagreement:

    India –
    Rathi et al. Lancet Infect. Dis. doi:10.1016/S1473-3099(20)30313-3,, Hydroxychloroquine prophylaxis for COVID-19 contacts in India, ICMR recommends chemoprophylaxis with HCQ for asymptomatic health-care workers and asymptomatic household contacts of confirmed cases, 4/17.
    Atikh Rashid,, Maharashtra expands use of hydroxychloroquine as preventive measure, 4/23.
    Ministry of Health and Family Welfare,, Advised all front line health care workers to take HCQ prophylactically, 3/28.
    Oneindia,, No COVID-19 death in Manipur, Mizoram, Nagaland, Sikkim so far: Govt, HCQ widely distributed, 6/26.
    Dr. Goldin,, Summary of HCQ usage in India from an MD in India, everyone diagnosed with COVID-19 gets HCQ, it is the standard of care country-wide. The only problem is some patients come very late, deaths are from those who wait too long before seeking medical help. Everyone at high risk, including policemen, firemen, healthcare workers, and nursing home patients, take HCQ, all contacts of positive cases also get HCQ prophylaxis. HCQ is OTC in India, so everyone else is also welcome to use it, 8/21.
    AFP,, India backs hydroxychloroquine for virus prevention, 5/27.
    AAPS,, Hydroxychloroquine Has about 90 Percent Chance of Helping COVID-19 Patients, many nations, including Turkey and India, are protecting medical workers and contacts of infected persons prophylactically, 4/28.
    The Indian Express,, Vadodara administration drive: HCQ helping in containing Covid-19 cases, say docs as analysis begins, used prophylactically in Vadora with positive results, 7/2.
    Barron’s,, Hydroxychloroquine: A Drug Dividing The World, used as preventative measure, 6/1.
    Government of India,, The caregiver and all close contacts of such cases should take HCQ prophylaxis, 7/2.
    The Australian,, India and Indonesia stand by antimalarials, 5/29.

  6. “Hi Ed, see my article about HCQ (link below). Sweden is in line with the current international consensus that hcq is ineffective.”

    Thanks.. I checked out your review.. The C19study site agrees with your first two assessments, but also it gives some relevant comments, listed below. The third study is simply declared wrong. Thus it looks like the issue is still unresolved for you. Finally, now we have this:

    #1 HC Q was significantly associated with reduction / elimination of viral load, which was enhanced with AZ. Updated 8/13: responses to this paper have raised methodological issues [1, 2, 3].

    #2 150 patients very late stage RCT showing no significant difference. Treatment very late, average 16.6 days after symptom onset.

    Data favorable to HC Q was deleted in the second version, see analysis [1].

    “[HC Q] accelerate[s] the alleviation of clinical symptoms”

    “More rapid alleviation of clinical symptoms with SOC plus HC Q than with SOC alone was observed during the second week since randomization”.

    “The efficacy of HC Q on the alleviation of symptoms, HR 8.83 [1.09-71.3], was more evident when the confounding effects of other anti-viral agents were removed”

    #3 Authors say NEGATIVE but the C19study evaluation ended up disagreeing..

    Post Exposure Prophylaxis study Source

    Boulware et al., NEJM, June 3 2020, doi:10.1056/NEJMoa2016638 (Peer Reviewed)

    A Randomized Trial of Hydroxych loroquine as Postexposure Prophylaxis for Covid-19

    COVID-19 cases are reduced by [49%, 29%, 16%] respectively when taken within ~[70, 94, 118] hours of exposure (including shipping delay). The treatment delay-response relationship is significant at p=0.002. PEP delayed treatment RCT.

    Currently this is the only study where we have evaluated the result as positive while the authors indicate it is negative. We provide a detailed explanation of why the results presented here are positive [1]. Note that author comments also differ from the published conclusion.

    6 independent analyses confirm efficacy: [2, 3, 4, 5, 6, 7, 8].

  7. You may have already addressed this elsewhere, but what can we do about this argument that we’re making other people ill even when we don’t know it that keeps going around? It seems to be a big driver of fear and guilt over here in North America and everyone says it’s backed by science and yet common sense and very little data tells me it feels bogus to a large degree. Thanks so much!

  8. As the famous Dr Fauci has said aysmptomatic people are not and never have been a major source of infection.

  9. Sweden is the perfect example of a Casedemic. Case have been climbing almost exponentially since since the beginning of September and yet the 7 day moving average on deaths is 1.

  10. Thanks for your insights! I would like your thoughts on:

    1. Excess deaths in the UK, which are reported to be significantly high.

    2. Countries like India have taken 6 months to peak, while European countries typically peaked in a month or so. Is there any difference in virus response by genetic differences in these people groups?

    3. How is China so well off even though the whole thing started there? No second wave with autumn either? Is it only a case of data manipulation?


    1. Hi Colin,
      1. I haven’t looked in detail at the UK numbers but so far there have been 45,000 covid deaths in the UK. Total annual deaths in the UK are about 600,000. That doesn’t seem vastly different from the proportion seen in Sweden.
      2. I haven’t looked at India enough to be able to comment.
      3. I don’t think we can trust any numbers coming out of China.

  11. Apparently, northern UK was not hit very hard in the Spring and therefore probably more susceptible people, lower levels of community immunity.

  12. A final query: I’m curious about data lags regarding the reporting of deaths in Sweden and, related to this, when death statistics in Sweden for a given month/year are considered to be accurate and complete. For example, the CDC in the United States caution that deaths data in recent weeks are incomplete. I also know from reviewing the death statistics released by Statistics Sweden a few times over the past week that the death statistics for September 2020 are still being updated.

    Given this, I wonder if your claim about September 2020 (being the least deadly month in Swedish history) could be premature Do you know how it will take for death statistics in September 2020 to be finalised?

    You may want to review your claim about September 2020 when the death statistics are finalised.

  13. Ivor Cummins recent YouTube video mentioned recent ideas that the vast quantities of virus including Covid are distributed by global air currents whose pattern matches the pattern of international outbreaks. This would explain why Peru which has a strict Lockdown has high infection rates and Spain that has a strict Mask mandate has high infection rates indicating these authoritarian measures have little impact on infection which will not please out authoritarian governments who will likely discourage the possibility of this infection route as much as they discourage any ‘cure’ that is not by vaccination?

  14. I don’t understand why some people didn’t see this: if a disease does not cause any symptom, then it should mean that the disease is not dangerous. Even common cold causes symptoms. Think of it….

  15. Hi

    Just a thought but it looks as though 2019 might have been anomalously low (significantly below average). If so, is it possible that Sweden (like UK) had a ‘soft’ flu year. This might mean that there was a far higher proportion of ‘vulnerable’ people in the population than normal.

  16. Sebastian
    In much of Europe presentations for heart attacks and strokes are well down on normal. With obvious implications re longer term adverse effects. Much of that is attributed to lockdowns and the fear of attending hospitals etc.
    what are the figures for Sweden in regard of that ?

    1. In spring we saw a similar effect. Here in Stockhol hospital admissions for heart attacks were down 40% . At present my feeling is that things are back to normal, people have realized that the danger from the virus is not greater than the danger from a heart attack.

  17. I think it’s also important to understand that immunity doesn’t necessarily mean that sterilizing immunity, ie prevents transmission. Immunity to a coronavirus may potentially mean that symptoms upon re-exposure/infection are significantly less severe or mitigated entirely, which is the cases with several other circulating coronaviruses. The vaccines also may lack sterilizing immunity, we don’t know as the clinical trial endpoints don’t address this, they were designed to determine a reduction in symptoms; in fact, the vaccine trials weren’t even necessarily designed to assess if mortality is reduced in the vaccinated group.

  18. I can’t speak for China but here in Australia extremely draconian liberty-destroying measures were employed to suppress the epidemic in Victoria, which had been running at nearly a thousand new diagnosed infections per day. The left-wing government in that state of Australia has close financial ties to the CCP and has signed up as a partner in the belt-and-road programme.

    It appears to be possible to stop transmission with tight controls over the population, including hideous police brutality and home invasions. Sad but true. It might even be possible to stop transmission completely with largely voluntary measures although that hasn’t been demonstrated anywhere yet to my knowledge.

    God bless.

  19. Thank you for this excellent, clear and easy to follow article. However, you say that Covid-19 definitely exists but do you know if anyone has actually managed to isolate it properly? As far as I have been able to establish no one appears to have done so. In this CDC document – – on page 39 under Performance Characteristics it says “Since no quantified virus isolates of the 2019-nCoV are currently available”. In the absence of isolation, which is the first required step of the Koch/Rivers postulates to establish the existence and pathogenicity of any virus, how can it be identified, or indeed identifiable, by any testing procedure with any certainty at all?

    Furthermore, they cannot have successfully proved that it causes any particular disease. In the same document on page 3 under Intended Use it says “Positive results are indicative of active infection with 2019-nCoV but do not rule out bacterial infection or co-infection with other viruses. The agent detected may not be the definite cause of disease.”

  20. In this article drawing from another CDC document it claims that they could hardly get the virus to infect Human cells at all!!! I do not pretend to understand some of the gobbledygook used in such scientific articles but perhaps you will.

    It seems to me that there is plenty of reason to doubt the certainty of claims that our current bout of illness and deaths is actually caused by this so-called new deadly virus and that it could in fact be nothing to do with it. Clearly, we have illness and deaths, and no-one is denying that, but, if it is not caused by SARSCov 2 then it must be something else and may well be any other cause of respiratory illness that we have had throughout time.

  21. Thank you for a good read. I’m sceptical about the causative correlation between the virus and the disease, even about the very existence of SARS-CoV-2 to begin with. See this:

    But out of curiosity, how would you explain the apperant increase in sick people i hospitals we had in Sweden during spring 2020? Where there really more of them or simply a mirage created by the rather questionable use of rt-PCR tests or what?!

  22. Carrying on from my last comment as it wouldn’t all fit into the box:

    In this article drawing from another CDC document it claims that they could hardly get the virus to infect Human cells at all!!! I do not pretend to understand some of the gobbledygook used in such scientific articles but perhaps you will.

    It seems to me that there is plenty of reason to doubt the certainty of claims that our current bout of illness and deaths is actually caused by this so-called new deadly virus and that it could in fact be nothing to do with it. Clearly, we have illness and deaths, and no-one is denying that, but, if it is not caused by SARSCov 2 then it must be something else and may well be any other cause of respiratory illness that we have had throughout time.

    This one in my opinion, however, has been made much worse by the unfounded claims and gullible acceptance of it being a “new deadly virus” which has exacerbated the incidence and death toll by creating fear, an extremely negative emotion that suppresses immune systems and increases the likelihood of getting ill and reduces the chances of recovery. Add to that, unwittingly inappropriate treatments being used in the form of overly toxic anti-viral drugs and excessive use of ventilators, that would never have been used had we simply treated it as a normal flu outbreak, and you have a recipe for disaster.

    As an aside, you mention the 1918 flu pandemic. I believe there is research that has suggested the flu itself was not that deadly, but the high death rate was actually caused by superbugs being generated because of the appalling conditions in which they were having to be treated, combined with excessive use of Aspirin. Also, although this may not be easily verifiable and no doubt has been debunked as a conspiracy theory, it has been suggested by some that vaccines were used at the time that may also have exacerbated the death toll.

    These may be rather more anecdotal and less acceptably scientific but nevertheless are food for thought:

    Is it true for instance that a number of vaccines were manufactured at that time and given to millions of people around the globe? Is it also true that Greece refused them and faired very much better than other countries? If so, could they be linked?

  23. As in the other post where you said Sweden has reached herd immunity, which you now said understand was incorrect (you mentioned that you did not think this was a seasonal virus). This post again boggles my mind. So once again, I ask, what is your null hypothesis?

    As an M.D., you are familiar with the multitude of independent studies that assess estimated population mortality rates above 0.50% (>0.65% in Bergamo, Italy, CEBM/OXFORD. This is an estimated value and is it of course not the infection fatality rate. In Canada the current (Nov 17) death rate for someone diagnosed with COVID by age:
    60-69 3%
    70-79 12%
    80-89 23%
    90-99 32%

    Beyond, that also an M.D. I would assume that a statement such as “that most people who died “of” covid actually died with covid” is wild. For example, you could then say that about anyone over 70 dying with cancer, “well they didn’t die of cancer, they died of old age, but they had Cancer”. In Canada, COVID is about 10-33 times more lethal than the flu. The data is available freely. The The flu also does not typically cause effects felt for months. I find it so hard to believe that you are not aware of this, or aware of the limited actual number of deaths due to influenza. Even the deaths typically discussed in the media are an estimate value as the actual influenza ‘causal’ deaths are but a fraction. Again, this data is readily available, in Sweden, Canada, UK, and many other higher quality reporting countries.

    You must also be familiar with the fact that “underlying health conditions” could mean things such as diabetes, being overweight, having heart or lung issues. These are not ‘rare’, in the US, around 40% of the population has “underlying health conditions”. I understand that Sweden may be healthier, due in some part the universal health care system (i.e. like Canada).

    So once again, I would love to know your null hypotheses on your ‘fact’: “in fact, it often is. Covid-19 is not a unique disease, and does not appear to have a noticeably higher mortality rate than the so called “common cold”.

    Actual 2018-19 Season –

    What is basis of evidence for flu related incidence, IFR and Population Fatality Rate? What % would you need to see to admit you were wrong?

    1. Hi AL,
      The data I point to is official WHO data, based on analysis of antibody prevalence. It is the highest quality data available at this time, and most likely it is an overestimate, since not everyone develops antibodies. Ioannidis has updated his estimate of infection fatality rate to 0,15-0,20% . As to flu not being able to cause long term effects, that is simply wrong. There is no evidence that covid causes more long term effects than flu.

  24. Thanks for your reply. Bit odd that no one apparently have purified the virus as of yet. And no one seems to be able to present a specific gene (genome?) sequence for it. But that’s common in virology I gather. Sorry for not using the correct terminology by the way, I’m no expert by any means.

  25. In Canada, the virus is less deadly than the flu in younger age groups.
    And the post viral syndrome is reported more frequently with flu and pneumonia than Covid at this time. However as there are more Covid symptomatic infections the gross numbers are higher.

  26. I think there is some confusion around terminology, which has been at the forefront of the challenges with much of the handing of the pandemic, as well as the public’s understanding of actual risk vs perceived risk. Beyond the issues & unprecedentedness of a positive tests being outright synonymous with a “case” in the US and other locales, regardless of presentation of symptoms or symptom severity, or where no test is conducted but instead a case is ruled probable (and likely counted, depending on local definitions/regulations) based upon presentation of symptoms, which are rather generalized and not distinct from many other disease/illnesses such as flu.
    The two other terms that direly needed clarification at the get go, especially among the media, but also by public health officials and modelers (see Imperial College’s model estimates & assumption), was CFR and IFR. That is ‘case fatality ratio’ vs ‘infection fatality ratio’ (not rate, as rate implies a time component). This difference here is quite significant and the implications of confusing these terms have been profound IMHO.
    Since “cases” are ill-defined, as I just mentioned, and since we don’t have a full picture of the true scope of infections, which the WHO has estimated to be 750M+, or more than 10x the number of “confirmed cases”, then these numbers will be different and both will be dynamic instead of static as the pandemic progresses (and as we have seen). The IFR has been estimated ~0.1-0.35%, as per Ioannidis et al and other studies, with variation across geographies and age distribution (a particularly important factor), which is a comparable IFR to a bad flu season (with quite different age distributed mortality between COVID vs flu).
    CFR is defined as the ratio of deaths of total cases; where cases are defined based upon diagnosis of a discrete, limited-time course disease; often with symptoms severe enough to necessitate treatment (as we have seen this is not how the term “cases” has been used however). On the flipside, IFR, is the ratio of deaths to all infections; including asymptomatic infections and those not diagnosed. With many SARS-COV-2 infections being asymptomatic, and also many undiagnosed, the IFR denominator is much larger (and thus the IFR value larger; by approx 10x or more in this case), giving a more complete picture of percentage of deaths resulting from SARS-COV-2 infections; and thus actual risk.
    Also, seroprevalence is known to be a poor determinant of infections or prevalence in a population, as some don’t produce Abs or high levels of detectable Abs, and Abs tend to wane with time (which is normal for infections; and thus making Ab testing time dependent) at which point cellular immunity takes over with long lived T & B cells being protective and present, providing immunological memory. Yet, testing for cellular immunity is relatively challenging logistically & technically compared to Abs.

  27. As mentioned, I would love to know your null hypotheses on your ‘fact’: “in fact, it often is. Covid-19 is not a unique disease, and does not appear to have a noticeably higher mortality rate than the so called “common cold”.

    Actual 2018-19 Season –

    WHO is one organization. There are many studies available and as a form of research, we should be mindful to include appropriate studies across multiple sources.

    Both and other data are now readily available.

    So again, what is your basis of evidence for flu related incidence, IFR and Population Fatality Rate? What % would you need to see in relation to COVID to admit you were wrong?

    1. Hi AL,
      This isn’t a number I’m pulling out of thin air, and so it isn’t something that I need to prove. It is well established that flu usually has an IFR of around 0,1%, sometimes we have a bad flu year and it goes up to 0,2 or even higher. And the article you link to at CEBM is talking about case fatality rate, which as mentioned already is distinct from infection fatality rate and usually much higher, since most infections don’t become cases.

  28. Thank you.

    So if we are to go with this notation, that the season flu has an IFR of 0.1% which is debatable – from the CEBM article “In Swine flu, the IFR ended up as 0.02%, fivefold less than the lowest estimate during the outbreak (the lowest estimate was 0.1% in the 1st ten weeks of the outbreak). In Iceland, where the most testing per capita has occurred, the IFR lies somewhere between 0.03% and 0.28%.”

    From another journal l(BMJ)l: “Research conducted in New Zealand (NZ) and internationally suggests that the IFR for COVID-19 is typically at least an order of magnitude higher than for seasonal flu. The most detailed study of seasonal influenza mortality in NZ to date estimated average annual mortality of 13.5 (95%CI 13.4, 13.6) per 100,000 population [1]. Furthermore, the proportion of the NZ population infected with influenza in a year has been measured from a seroconversion study at 35% (95%CI: 32%-38%) [2]. Combining these figures suggests an IFR for seasonal influenza of about 0.039% (ie, 13.5/35,000) in NZ. This seasonal influenza IFR is 17 times lower than that estimated for COVID-19 at 0.68% [3] and 0.65% [4], based on international data (there have been too few COVID-19 cases in NZ to produce an IFR estimate)”

    The CEBM review discusses IFR:
    Iceland – 0.3% (3x times)
    UK – 0.9% (9x times)
    Global Estimate: 0.4% (4x times)

    So ask you, what IFR for COVID would you accept to say “my fact was incorrect and was not actually a fact”?

  29. If I understand you correctly, you estimate a disease’s deadliness by IFR or CFR. IFR means how likely it is that one dies of the disease when being infected, correct? Shouldn’t then the likelihood of _becoming infected_ be another important factor? So if the virus/bacteria/whatever spreads easier and people are less immune, i.e. getting more easily infected, this makes it more dangerous in my opinion.

    Just to double-check: If you’re immune, it doesn’t count as an infection, does it? So if I am vaccinated against let’s say influenza and I acquired immunity so that getting in contact with (a significant number of) the virus does _not_ cause an infection, this would not cound as an infection in the denominator of the IFR, correct?

    To me the most worrying aspect of SARS-CoV-2 is that there is so litttle natural immunity in the population (I don’t know about influenza) and that it apparently quite easily spreads through the air. So in the large picture even a small chance of dying or getting severe symptoms is amplified by the chance of actually getting infected.

    1. Yes, like I said, the Ioannidis data is higher quality. It is normal for the IFR to seem higher at the start of an epidemic, and for the numbers to come down as new data comes in. And even though the overall IFR is 0,15-0,20, it will be higher in regions with a higher proportion of risk groups.

  30. In addition to add important emphasis: Flu IFR is also an estimate. The actual death certificate noted information for pneumonia is a faction of the estimated values.This information available by jurisdiction across dozens of countries.

    Typically those who die from the flu have other ‘underlying health conditions’ and are older. I hope to point to this to eliminate the inaccurate comparison of “people dying with COVID not from COVID” because this is the same for the flu. (CDC)

Leave a Reply