COVID numbers

[Jason]

I discovered something interesting that makes me wonder what the number reported of COVID-19 cases actually means and raises another question.

Take a look at the numbers reported by Peterborough Public Health for local cases:

Confirmed positive: 78
Confirmed negative: 2156
Results pending: 502
Deaths: 1
Resolved: 40

The cases confirmed positive is since our first case was counted here. It's a cumulative total. You have to subtract the resolved cases and the death(s) from that number to get the actual number of active cases. So in this case, that's 37 active cases (as of today).

The reason I ask this is what number is the press reporting when they give the number of cases? Is it the number of active cases of cumulative cases since the first case? The result doesn't matter if we're thinking of flattening the curve but it gives a better idea of how many people actually have it now although it's obviously under-reported as to the actual number since we're not doing much testing. My belief is that they're reporting the cumulative number since that's the exact wording they use in in the second series of graphs in this article by the Toronto Star.

[fox]

In case you're interested in the latest update (as of May 25) for the City and County of Peterborough, Hiawatha and Curve Lake First Nations

Confirmed positive: 81
Individuals tested: >7,200
Deaths: 2
Resolved: 71
Active cases: 8

Best of all, no active cases in long term care institutions! I spoke to a friend yesterday in one of the institutions that had an outbreak. He told me that residents were kept in their rooms for more than a month after the outbreak (food was brought to them), and only now are they allowed to leave their rooms. They are still being served meals in their rooms, but the facility is making plans to open up the dining room with two shifts to allow for social distancing at the tables.

[Jason]

Note that the number of active cases has doubled from a few days ago, sadly. I think the province jumped the gun on re-opening businesses. Most provinces if you look at the numbers were at the bottom of the curve before they re-opened. Ontario started re-opening retail stores before we had gone down far enough. If you look at this Star page with the latest COVID-19 data and scroll down to the provinces, you can see the curves of new cases. We went down from about 41 cases/day decreased to 22 cases/day and now we're back at 28 cases/day. That's a 40% reduction followed by a 25% increase. It was expected we'd lose some ground but I don't think they expected this much. If you look at new deaths, it went from decreasing to flattening out. We should have waited at least a couple more weeks.

Btw, Local institutions don't just include LTC homes. It includes senior residences and the hospital. I'm glad there are no more outbreaks; I hope it lasts. We've either been very lucky that we weren't hit harder. Although we've had a few outbreaks, the numbers in each were small. There have been 2/3 of seniors who have died in some LTC homes related to an outbreak. Maybe it wasn't luck and our LTC homes and seniors residences are just better than most of the province at infection control. From what you describe, it sounds like they took it very seriously. We also have a couple of non-profit LTCs in the City at least. I think that matters.

[Jason]

Up 2 more active cases today - 10 total active cases.

[Jason]

I just discovered that Global News has a good graphical tracker page. They break out active cases, deaths and recovered cases, no silly total cases that many sites were just reporting along with deaths. It seems most news sites are now focusing on recovered cases and active cases and not just the other. The provincial breakdown is better than other trackers I've seen, too, in that it notes cases per 100k for each province which is useful info, not the total numbers.

It's really quite sad that the death rate is much higher than was first expected. It looks to be roughly about 7-8% at this point overall. Of course, what's with cases that have been confirmed. An Indiana randomized study looked at a sample of one or thousand people and found that the ratio of actual cases to confirmed cases was about 11:1. If that study is representative, we could be looking at a mortality rate of only under 1%. Still bad, but not as bad as we think. The flu seems to vary in it's supposed mortality rate depending on which study you look at. It varies from 0.2 to ~1%. I hope they do a lot of serological testing later to see how many actually had it or were exposed.


Update: As Fox points out below, the breakdowns below the chart are largely outdated. I was looking at only the chart at the top which is accurate and updated every 10 minutes they claim. I compared Health Canada's most recent numbers which are updated almost as often to the chart's and they were almost right on.

[fox]

It is a good site, but not all of the data are up-to-date - breakdowns and cumulative testing stats are not. I was interested in the Indiana randomized study; I found it here.

[buster]

"The flu seems to vary in it's supposed mortality rate depending on which study you look at. It varies from 0.2 to ~1%."

I think the flu statistics are used by some as a way to 'prove' a particular point of view. I don't remember a time when the flu overwhelmed the hospitals with horrendous growth rates as covid has around the world. That may be because it spreads itself out over months. Or it may be because of flu shots (which we've used from day day one - and have always avoided the flu, so far anyways), and this might create some community immunity.

With statistics, the devil is in the details.

[Jason]

It is a good site, but not all of the data are up-to-date - breakdowns and cumulative testing stats are not.

Which data are not up-to-date? And what is the source that you're comparing these to? They use Health Canada and each of the provincial health authorities for the numbers. Global News compared to Health Canada numbers: total confirmed cases is off by 12, confirmed cases by province is on in the 4-5 provinces I looked at, deaths are dead-on, tests are about 80,000 off (~5%). I can't check the accuracy of how many are cases are active as I don't know how many have recovered compared to Health Canada's numbers which I don't see on their page. Health Canada also notes under 'How Canada is monitoring COVID-19' regarding testing numbers:

Should there be any differences with the national case count compared with testing numbers reported by provincial and territorial public health officials, provincial data should be considered the most up-to-date.

That probably explains the difference between national numbers and provincial numbers of the number tested. Provinces have the information faster and report it to Health Canada and then Health Canada has to gather it and then report it. By the time they've done that the numbers have probably changed again. Global News is using the provincial numbers for that data, I would guess since they mention them as their sources and it's the only info that deviates that I can see. The federal numbers are lower than the provincial totals as would be expected since it takes longer for the federal government to post that information.

Had you heard of the Indiana study before I mentioned it? I'm just curious. I'm ignoring most of the news lately, even news of Canada so I wouldn't have noticed if it is getting much coverage. I learned of it from watching a Healthcare Triage episode here. The presenter/writer is pediatrician Aaron Carroll. His show is excellent and provides health information and news based on looking at studies and meta-studies and how solid or flawed they may be. His opinions sometimes fly against the mainstream views and he explains why and how. He's well worth watching.

[Jason]

I think the flu statistics are used by some as a way to 'prove' a particular point of view. I don't remember a time when the flu overwhelmed the hospitals with horrendous growth rates as covid has around the world. That may be because it spreads itself out over months. Or it may be because of flu shots (which we've used from day day one - and have always avoided the flu, so far anyways), and this might create some community immunity.

Statistics in general are often used this way. However, I wasn't trying to say it wasn't much worse than the flu. In relative terms even with the numbers we're probably looking at COVID-19 having been at least twice as deadly as the flu probably more but it depends on which number we use for the mortality rate of the flu. That number isn't agreed upon for a wide range of reasons. But either way, I believe being twice as many deaths as the flu is bad enough. If I thought it was the same as the flu, I wouldn't be wearing as mask. The mask provides little benefit to me, it's for others, especially the elderly.

As for not remembering a time when hospitals were overwhelmed, that's because COVID-19 is much more contagious. I believe the R0 of the flu is around 1.1. The RO of COVID-19 is believed to be between 2 and 3. All the measures we've taken isn't just because the mortality rate is believed to be much higher, it's because we don't want the system overwhelmed by controlling the spread of it. And it's working here.

The problem with measuring mortality rates is difficult as well because the primary cause of death isn't as cut-and-dried as we might think. I had a chat with a person on Facebook who has worked in a mortuary and what he said was fascinating:

in Australia, where I am, the death certification is broken into parts - so in the example you have above of a car accident could be listed as 1(a) intracranial haemorrhage - hours; (b) road traffic crash - hours 2) any other relevant history like diabetes and hypertension- years. Now if that patient had survived the brain bleed but suffered disability then died later of something related to that brain injury, even years later, then the car accident would also be listed.

Now people may die of a stroke or heart attack, but yes, they should be tested for COVID because we know it can cause these issues, and if positive it should be listed as a contributor. Same if they die at home, but could not access healthcare (like in peak times in Spain/Italy/NYC) the 1(a) will still be cardiac arrest/respiratory arrest - immediate, (b) pneumonia (COVID) - 9 days, 2) hypertension - 10 years, diabetes - 12 years etc.

Now in the event that a healthcare system was completely and utterly overwhelmed due to a pandemic and there was a road traffic crash and the ambulance was severely delayed and that person dies or the ambulance had difficulty getting them into a trauma service and they die and under normal operating services they could have survived - I believe they should have COVID listed under number 2 as a contributor. But we have tried exceedingly hard here in Australia to ensure that doesn't happen.

[fox]

Which data are not up-to-date? And what is the source that you're comparing these to?

Had you heard of the Indiana study before I mentioned it? I'm just curious.

The breakdowns of cases are not up to date. I believe it was by age but it might have also been by sex. I know that because it states the date in the report. Same goes for the cumulative testing stats. Both of these are at the end of the report.

No, I hadn't heard of the Indiana study before.

[Jason]

The breakdowns of cases are not up to date. I believe it was by age but it might have also been by sex. I know that because it states the date in the report. Same goes for the cumulative testing stats. Both of these are at the end of the report.

I was should have explained that I was just talking about the chart at the top; I didn't look at that as I wasn't particularly interested. That's why I was curious you thought was outdated. Now I see what you mean. I don't find a breakdown by age and demographics all that important because it's collecting information on who got it, not the ratio of who died from it by age or sex. Those stats are the equivalent of stat porn to me. While interesting, they don't really provide information. There are two many confounding variables such as activity level, number of friends, how close are those interactions. Until they've been adjusted for those things, what does it really tell us? However, the provincial breakdowns are outdated, too, and that's more important to know. So thanks for pointing that out since I never went that far in.

[Jason]

No, I hadn't heard of the Indiana study before.

I thought I had shown you something but instead, I led you astray. The study to which I'm referring to was done to see how prevalent the virus actually us using randomized antibody testing. Here's a snippet:

The tests are finding large numbers of people in the U.S. who were infected but never became seriously ill. And when these mild infections are included in coronavirus statistics, the virus appears less dangerous.

"The current best estimates for the infection fatality risk are between 0.5% and 1%," says Caitlin Rivers, an epidemiologist at the Johns Hopkins Center for Health Security.

That's in contrast with death rates of 5% or more based on calculations that included only people who got sick enough to be diagnosed with tests that detect the presence of virus in a person's body.

Here's an even more interesting one:

And 45% of the infected people reported having no symptoms at all.

I find that last stat fascinating and quite unexpected. Those people could have been passing it around and had no idea. I remember the head of the local health unit saying to assume you have it, so physical distance and wash your hands even if you have no symptoms. Imagine if we hadn't done that if this data are correct.

[fox]

All very interesting. The middle quote shows why I have said all along that the mortality rate based on number of diagnosed covid cases is meaningless. And when testing rates and testing strategies vary among jurisdictions, a comparison of those mortality rates in different jurisdictions is also meaningless. What is more meaningful is the covid mortality rate per 100,000 population. But even there, you will get variation based solely on how early a given jurisdiction was exposed to covid.

[Jason]

I think Ontario is moving to fast with lifting restrictions. Look at the chart at the bottom of this article showing new cases. The cases were going down, then retail opened and it jumped up and it's starting to go down again, but we're still at 2/3 of the level we were at the peak of the curve. I'm certainly no health expert but in other provinces, I've noticed that lifting of some restrictions began when the number of cases was much lower to even bottoming out.

I don't blame the jump all on retail or even mostly on retail because of Mother's day and that it can take 5-10 days before symptoms show and then people get tested. But based on the place in the curve, I think it's risky. And they're already talking about stage 2 that would involve opening restaurant patios and who knows what else as well as easing on the maximum size of groups. The new cases are going down again but I've also read somewhere that testing numbers are also down again.

[fox]

Testing has gone up in Ontario. There was a week where we were not testing at the rates the Premier indicated we should be, but if you look at the CBC report, Ontario is now 3rd in testing rate (5,033/100K), and is catching up to both Quebec (5,348/100K) and Alberta (5,547/100K). I'm pretty sure that the increased testing rate is responsible for some of the extra cases. As I have noted several times, the more you test, the more cases you find. Remember that Ontario is now encouraging anyone with any symptoms to go for testing, and they have made it a lot easier to get a test.