Hacker News Reader

Comment by cmrdporcupine | Hacker News Reader

I'm counting ICU admissions among the unvaccinated separately, not against the population as a whole. When you consider that the majority of the population (65% total, 80% eligible) is vaccinated, then you have to separate the two groups because the outcomes are totally different.

But I did spot something wrong with the way I'm counting, so yes the 10% is probably wrong. The reason is that the daily new case count given by the province is incremental, while the ICU count is current # of cases, not new admissions. So it's not possible to do the comparison in this way. I'd have to take a look at the current active case count by vaccination status, which is something I don't think the province is reporting.

View on HN

Your biggest problem is that your data isn't accurate.

The page notes that it's case count isn't necessarily correct. This page shows the total case delta at 650 for today versus yesterday: https://covid-19.ontario.ca/data/case-numbers-and-spread

Whereas the spreadsheet from your page says 426. It's not even clear to me that the 650 number is accurate because not everyone may necessarily be reporting they have COVID. I know when my whole family got COVID we didn't report it to the government. I've seen estimates in the U.S. that actual case count is more than double reported case count: https://www.cidrap.umn.edu/news-perspective/2021/07/us-covid...

Secondly, ICU case counts are based on people who are in the ICU and happen to have COVID, not people who are in the ICU primarily because of COVID. So for yesterday the delta in ICU COVID case count was 15, but we don't know how many of those net new 15 are actually in the ICU for COVID symptoms primarily.

So taking the raw numbers for yesterday's delta, that's 15 / 650 as a rough estimate (dividing the deltas isn't really what we want but it's the best I can come up with), and that lands us at 2.3%. I also believe that number is far too high for the other reasons outlined above.

For comparison, the numbers here seem to indicate a 5% chance of hospitalization for the unvaccinated: https://www.wbay.com/2021/08/19/covid-19-wisconsin-dhs-compa...

They give no ICU numbers, so we can expect the ICU odds to be closer to 2% again, and their data suffers from the same problems.

Not sure which spreadsheet you're talking about, the page I linked to has a series of feeds, and if you download today's CSV and sum all 4 case count columns it adds up to 650. 426 is the unvaccinated count. 650 is the count of all cases.

  Date covid19_cases_unvac covid19_cases_partial_vac covid19_cases_full_vac covid19_cases_vac_unknown
  2021-08-20 426 64 103 57

BTW, it's not "my page"; it's the official gov't of Ontario COVID data API. It's where the other link you pasted gets its data. There's another feed that provides just testing numbers, but doesn't break down by vaccination status. It also reports yesterday as 650. So the two accord.

But please, go on. "Your biggest problem" is that you don't read. Just like the 90% in Israel stuff.

You're right. We both made an analytical mistake, although yours was much bigger. If that means I don't read, then you must not read well either. Congrats on devolving to personal insults about my character though. I guess that means you ran out of data-based arguments.

If we adjust for my mistake and make 426 the denominator, the odds go up to 3.5%. If we take into account that at least half of cases go unreported, the odds drop to 1.75% or lower. If we take into account that 25% or more ICU COVID cases aren't really COVID-related, (hard to find data on that but it was what was reported that way for Florida last week), the odds go down to 1.3%.

I apparently read well enough to instantly spot that 10% ICU conversion of infected unvaccinated is a bogus number. And my Israel data mistake didn't affect my conclusions at all. Maybe you should work on your reading?

I think both of you would benefit from a less "engaged" discussion mode, more focused on the data.

Not which one of you is right or wrong, but rather what data and conclusions are right or wrong.

Because otherwise I benefited very much from your discussion. Thank you both.

Reply on news.ycombinator.com