COVID statistics confusion
July 5, 2020 9:59 AM Subscribe
This handy stats site shows that 11% of COVID cases with an outcome end in death. However, at the same time, only 8% of COVID cases are "serious." Other sources give the morbidity rate at 1%-3%, however. That could square with the 11% number if there were more new cases without outcomes than old cases with outcomes. But there aren't, in Canada anyway. Can anyone make sense of these numbers?
Meanwhile, occasionally other sources (e.g. the CDC) says there might be as many as 10x as many unknown cases as known ones. And some sources estimate that as many as 25% of people in New York State might have had COVID. Presumably random testing would get a handle on that number, no?
And, I keep reading that you need to be exposed to a certain "viral load" to come down with this plague. But, being as viruses spread, well, virally within the body, why would that be the case? It makes sense to me that a high viral load might lead to a more serious case, as the body doesn't have time to react before its defenses are overwhelmed; but it doesn't make sense that there is a threshold for any infection whatsoever.
Meanwhile, occasionally other sources (e.g. the CDC) says there might be as many as 10x as many unknown cases as known ones. And some sources estimate that as many as 25% of people in New York State might have had COVID. Presumably random testing would get a handle on that number, no?
And, I keep reading that you need to be exposed to a certain "viral load" to come down with this plague. But, being as viruses spread, well, virally within the body, why would that be the case? It makes sense to me that a high viral load might lead to a more serious case, as the body doesn't have time to react before its defenses are overwhelmed; but it doesn't make sense that there is a threshold for any infection whatsoever.
Do you have a definition for "cases with outcome?" Because what that says to me is "cases where we followed the patient far enough to know how it ends," rather than "cases where the person is done being sick." Which means not total diagnosed cases, but those that were cared for by a doctor (and who followed up with info on the conclusion) or hospitalized. Most people would have gotten a diagnosis (which was recorded), went home to be sick, got better eventually, and no outcome was recorded.
As for viral load, a key piece of info is the two kinds of immune system you have. Your innate (generic) immune system starts fighting the virus when it enters your body. If you get only a small dose of the virus, your immune system kills it faster than it can replicate and you don't even know it happened. With a higher does, it replicates faster and you reach the point of getting sick before the innate immune system can destroy the illness.
After a certain amount of time, your adaptive immune system (which fights each specific illness differently by producing antibodies) takes over the fight, and then it struggles to beat back the infection. That's the duration of the experience of being sick. But with a low viral load, your innate immune system prevents you from "getting it" even though it entered your body.
posted by gideonfrog at 10:36 AM on July 5, 2020 [2 favorites]
As for viral load, a key piece of info is the two kinds of immune system you have. Your innate (generic) immune system starts fighting the virus when it enters your body. If you get only a small dose of the virus, your immune system kills it faster than it can replicate and you don't even know it happened. With a higher does, it replicates faster and you reach the point of getting sick before the innate immune system can destroy the illness.
After a certain amount of time, your adaptive immune system (which fights each specific illness differently by producing antibodies) takes over the fight, and then it struggles to beat back the infection. That's the duration of the experience of being sick. But with a low viral load, your innate immune system prevents you from "getting it" even though it entered your body.
posted by gideonfrog at 10:36 AM on July 5, 2020 [2 favorites]
And, I keep reading that you need to be exposed to a certain "viral load" to come down with this plague. But, being as viruses spread, well, virally within the body, why would that be the case? It makes sense to me that a high viral load might lead to a more serious case, as the body doesn't have time to react before its defenses are overwhelmed; but it doesn't make sense that there is a threshold for any infection whatsoever.
What We Do and Do Not Know About COVID-19’s Infectious Dose and Viral Load (Discover Magazine, April 18, 2020) ("Intuitively it might make sense to say the more virus, the worse the disease. But in reality the situation is more complicated."), Reducing risks from coronavirus transmission in the home—the role of viral load BMJ 2020; 369 doi: https://doi.org/10.1136/bmj.m1728 (Published 06 May 2020).
posted by katra at 10:46 AM on July 5, 2020
What We Do and Do Not Know About COVID-19’s Infectious Dose and Viral Load (Discover Magazine, April 18, 2020) ("Intuitively it might make sense to say the more virus, the worse the disease. But in reality the situation is more complicated."), Reducing risks from coronavirus transmission in the home—the role of viral load BMJ 2020; 369 doi: https://doi.org/10.1136/bmj.m1728 (Published 06 May 2020).
posted by katra at 10:46 AM on July 5, 2020
Yeah the confirmed case death rate is artificially high, mostly because of a lack of testing in the initial part of the pandemic. For the first few months only seriously ill people were getting tested at all so 11% is definitely not the real rate. The real rate you want is Infection Fatality Rate, which is estimated by scientists to be around around 1%, which is still very high for a rapidly spreading disease.
I don't have a cite for this handy, but my understanding about the viral load is that the main way Covid is deadly is due to autoimmune overreaction causing a cytokine storm. This is why it's so much more deadly for people with issues like asthma. But also a lower viral load means you are much less likely to get an autoimmune overreaction because the normal slow reaction works against a lower viral load.
posted by JZig at 10:49 AM on July 5, 2020
I don't have a cite for this handy, but my understanding about the viral load is that the main way Covid is deadly is due to autoimmune overreaction causing a cytokine storm. This is why it's so much more deadly for people with issues like asthma. But also a lower viral load means you are much less likely to get an autoimmune overreaction because the normal slow reaction works against a lower viral load.
posted by JZig at 10:49 AM on July 5, 2020
You are looking at the stats for Canada. There are two reasons why the fraction of diagnosed cases ending in death in those stats are significantly higher than the estimated actual case fatality rate of 1-3%.
The first is that, until some time around the end of May or beginning of June, limited testing capacity meant that many people with likely COVID infections were not being tested. Testing was limited to people who had traveled outside Canada within the preceding 14 days and were exhibiting a relatively narrow range of symptoms. Since then, increased test-processing capacity has allowed testing to be expanded.
The second is that COVID burned through a large number of long-term care homes in Ontario and Quebec, killing a lot of elderly and already-frail people. A disproportionate infection rate among a high-fatality-rate demographic will lead to a higher case fatality rate overall.
The Ottawa Public Heath COVID-19 dashboard has a nice plot of daily confirmed infections with institutional outbreaks separated out, as well as an age distribution of positive cases.
posted by heatherlogan at 11:08 AM on July 5, 2020 [3 favorites]
The first is that, until some time around the end of May or beginning of June, limited testing capacity meant that many people with likely COVID infections were not being tested. Testing was limited to people who had traveled outside Canada within the preceding 14 days and were exhibiting a relatively narrow range of symptoms. Since then, increased test-processing capacity has allowed testing to be expanded.
The second is that COVID burned through a large number of long-term care homes in Ontario and Quebec, killing a lot of elderly and already-frail people. A disproportionate infection rate among a high-fatality-rate demographic will lead to a higher case fatality rate overall.
The Ottawa Public Heath COVID-19 dashboard has a nice plot of daily confirmed infections with institutional outbreaks separated out, as well as an age distribution of positive cases.
posted by heatherlogan at 11:08 AM on July 5, 2020 [3 favorites]
The 8% is the proportion of cases that are currently regarded as serious not the proportion of overall cases, so that if the demographics of the people sick now are different then you would expect a different outcome. For example, if there are significantly more younger people who have tested positive, then it’s not unlikely that the prognosis is better.
As a comparison, the RKI in Germany is currently reporting 4.6% death rate for the whole population, but they mention in their report that "The median age was 82 years. Of all deaths, 7,716 (86%) were in people aged 70 years or older, but only 18% of all cases were in this age group." So age pays a huge role in the outcome. (At least in Germany).
As others have pointed out, the testing criteria have changed, so probably more of the milder cases have been picked up, and/or it’s spreading through a younger population.
posted by scorbet at 1:03 PM on July 5, 2020
As a comparison, the RKI in Germany is currently reporting 4.6% death rate for the whole population, but they mention in their report that "The median age was 82 years. Of all deaths, 7,716 (86%) were in people aged 70 years or older, but only 18% of all cases were in this age group." So age pays a huge role in the outcome. (At least in Germany).
As others have pointed out, the testing criteria have changed, so probably more of the milder cases have been picked up, and/or it’s spreading through a younger population.
posted by scorbet at 1:03 PM on July 5, 2020
> 11% of COVID cases with an outcome end in death. However, at the same time, only 8% of COVID cases are "serious."
Besides the other factors people upthread have mentioned:
- Current COVID cases include all current known COVID cases, including those who have been ill for only 2 or 3 or 7 days or whatever. Quite a number of those are "not serious" now but will move from the "not serious" to "serious" category as time moves forward.
For example, Person A is feeling a bit ill today so they go to the doctor and get a test. It comes back positive a day later and now they are in the stats as a "not serious" case. This continues for another 9 days. On day 10 they take a serious turn for the worse and are admitted to the hospital. Now they are counted in the "serious" statistics--but for the previous 9 days or so, they were not.
For the first 1/2 or 1/3 or 1/4 or so of their illness, almost everyone with COVID is in the "not serious" category.
As a result, the final tally of "serious" cases--meaning people whose case was serious at any point during their illness--is going to be quite a bit higher than the percentage of cases who are "serious" among the currently active caseload. The final tally could be as much as double what you see in the current cases.
- Doctors who work with COVID patients regularly are now getting quite a bit better at dealing with the disease throughout its course. Thus, outcomes are considerably better now than they were just a few weeks/months ago.
For those places where the medical system is not under extreme pressure due to massive outbreaks, the chances of surviving the disease if you do reach the point where you must enter the hospital, are probably quite a good bit better now than they were in March or April.
To your question, the point is that fatality rates for seriously ill patients have improved by some degree now, compared with the beginning of the outbreak. I don't think anyone knows exactly what the improvement rate is, but of those 8% of patients currently in the "serious" category, some amount more of them will survive now than would have back in March-April-May.
That doesn't explain the entire discrepancy but it helps account for the fact that the death rate now--and going forward--is lower than it was previously. Even if the improvement is quite small, say enough to save the lives of 10% of serious cases, that would be enough--by itself!--to move the death rate from 11% down to 10%.
I can't provide backup for either of the above assertions without doing some searching, but generally they are tidbits gleaned from listening to the This Week in Virology podcast, which I highly recommend.
They spend a considerable amount of time and detail discussing just the sort of good but complicated questions you have. In fact, you could submit your question to them--they spend a good bit of time in each episode answering listener questions and yours is a good one.
If I recall correctly, it was in Episode 632 where they discuss both the now well-understood typical course of COVID, which includes a very predictable long prelude where nearly all patients don't feel that sick, followed by a "day of decision"--a very short period of a day or two where things either take a turn for the better or worse. One of the improvements in care is that doctors are now well aware of when this happens and what to look out for to predict whether patients need to come into the hospital or not, so they can make a point of monitoring stay-at-home patients carefully during those critical few days, and telling the patient and family/friends what to look out for as well as monitoring via in-person or telemedicine appointments.
Also in that same episode they discuss the vastly improved practices and outcomes for treatment of COVID. You can see some of that in the show outline--dexamethazone, putting serious patients in a prone position, Tocilizumab, etc.
posted by flug at 3:49 PM on July 5, 2020 [1 favorite]
Besides the other factors people upthread have mentioned:
- Current COVID cases include all current known COVID cases, including those who have been ill for only 2 or 3 or 7 days or whatever. Quite a number of those are "not serious" now but will move from the "not serious" to "serious" category as time moves forward.
For example, Person A is feeling a bit ill today so they go to the doctor and get a test. It comes back positive a day later and now they are in the stats as a "not serious" case. This continues for another 9 days. On day 10 they take a serious turn for the worse and are admitted to the hospital. Now they are counted in the "serious" statistics--but for the previous 9 days or so, they were not.
For the first 1/2 or 1/3 or 1/4 or so of their illness, almost everyone with COVID is in the "not serious" category.
As a result, the final tally of "serious" cases--meaning people whose case was serious at any point during their illness--is going to be quite a bit higher than the percentage of cases who are "serious" among the currently active caseload. The final tally could be as much as double what you see in the current cases.
- Doctors who work with COVID patients regularly are now getting quite a bit better at dealing with the disease throughout its course. Thus, outcomes are considerably better now than they were just a few weeks/months ago.
For those places where the medical system is not under extreme pressure due to massive outbreaks, the chances of surviving the disease if you do reach the point where you must enter the hospital, are probably quite a good bit better now than they were in March or April.
To your question, the point is that fatality rates for seriously ill patients have improved by some degree now, compared with the beginning of the outbreak. I don't think anyone knows exactly what the improvement rate is, but of those 8% of patients currently in the "serious" category, some amount more of them will survive now than would have back in March-April-May.
That doesn't explain the entire discrepancy but it helps account for the fact that the death rate now--and going forward--is lower than it was previously. Even if the improvement is quite small, say enough to save the lives of 10% of serious cases, that would be enough--by itself!--to move the death rate from 11% down to 10%.
I can't provide backup for either of the above assertions without doing some searching, but generally they are tidbits gleaned from listening to the This Week in Virology podcast, which I highly recommend.
They spend a considerable amount of time and detail discussing just the sort of good but complicated questions you have. In fact, you could submit your question to them--they spend a good bit of time in each episode answering listener questions and yours is a good one.
If I recall correctly, it was in Episode 632 where they discuss both the now well-understood typical course of COVID, which includes a very predictable long prelude where nearly all patients don't feel that sick, followed by a "day of decision"--a very short period of a day or two where things either take a turn for the better or worse. One of the improvements in care is that doctors are now well aware of when this happens and what to look out for to predict whether patients need to come into the hospital or not, so they can make a point of monitoring stay-at-home patients carefully during those critical few days, and telling the patient and family/friends what to look out for as well as monitoring via in-person or telemedicine appointments.
Also in that same episode they discuss the vastly improved practices and outcomes for treatment of COVID. You can see some of that in the show outline--dexamethazone, putting serious patients in a prone position, Tocilizumab, etc.
posted by flug at 3:49 PM on July 5, 2020 [1 favorite]
This thread is closed to new comments.
Looking at the comparable figures for the UK, I suspect that site is showing 11% of confirmed COVID cases. But the total number of cases is estimated to be significantly higher than that, which is where the lower morbidity rate may be coming in. I don't know how 'serious' is measured but it sounds more plausible that it is using the same count of number of cases as the 1-3% morbidity rate.
posted by plonkee at 10:34 AM on July 5, 2020