The death rate in Ireland is actually falling despite the reported numbers

[Eireog007]

Graph shows breakdown of recorded Covid-19 deaths

Chief Medical Officer Dr Tony Holohan has produced a new graph that indicates a different story about the number of daily deaths in Ireland from Covid-19 to date.

www.rte.ie

Very interesting report from Dr Holohan which states that the death toll actually peaked over a week ago and the deaths being announced each day since while being higher were not always from that particular day, similar to the additional cases added from the tests in Germany.

 

[llgon]

This graph shows that that were 20 deaths yesterday. Is Dr Holohan sure that all have been reported. Or could this be changed to 60 next week when all reports are in?

I didn't see the briefing yesterday but find these reports confusing.

 

[Eireog007]

This graph shows that that were 20 deaths yesterday. Is Dr Holohan sure that all have been reported. Or could this be changed to 60 next week when all reports are in?

I didn't see the briefing yesterday but find these reports confusing.

The facts behind the chart are the actual dates the deaths happened on, the dates we have been getting on the news the last few weeks were the days they were reported to the HSE to be included in the stats.

So the deaths are decreasing over the last 10 days. There may be another spike in the coming weeks or months but for now we are past the first peak according to the latest figures.

 

[Duke of Marmalade]

I was fascinated by that graph and I trust Dr Holohan is interpreting the figures correctly, but at first glance it is not clear to me that actual deaths are falling. For let us assume for sake of argument that there is on average a delay of 10 days between actual date of death and date of reporting. Then the deaths "by date of death" will always show falls over the recent few days as the figures are still incomplete for these days.
Put another way the graph shows there were 20 deaths actually occurred yesterday that we know of but if the reporting delays have not changed this could turn out to be 40 whenever the full figures are in.

 

[llgon]

Eireog, I think you've missed my point. Dr Holohan said that it can be a few days or even longer before a death is notified to them. So the figures on this graph representing the last week are liable to rise. It seems premature to me to plot these recent figures on the graph and to conclude that the figures peaked over a week ago.

So the deaths are decreasing over the last 10 days.

I don't see any evidence for this in the figures provided but maybe I missed something?

 

[Eireog007]

Eireog, I think you've missed my point. Dr Holohan said that it can be a few days or even longer before a death is notified to them. So the figures on this graph representing the last week are liable to rise. It seems premature to me to plot these recent figures on the graph and to conclude that the figures peaked over a week ago.



I don't see any evidence for this in the figures provided but maybe I missed something?

I understand where you are coming from and yes the figures for the last few days may rise but the trend for the last 10 days has been declining, I don't have all of the details other than what was in the report etc but what i believe has happened is that the reporting has become more streamlined so the lag between deaths and the reports is dropping.

Whatever about some politicians Holohan etc have been very reluctant to give any concrete pronouncements on the peak and whether cases for declining so i tend to give it more credence than other reports.

 

[mathepac]

Holohan stressed that the data-set he used above is incomplete and therefore no conclusions can be drawn from any analysis of it. The heading on the graph, the data-set title is, as his comments later in the link clarify, incorrect.

""That could be just a delay in the notification of those deaths. We have to wait to understand this. There is always a time lag between incident numbers, admission to hospitals and then mortality."

The Chief Medical Officer also pointed out that people have three months to notify the authorities of a death and not all deaths are notified until the end of that period."

George Lee's conclusions based on this data-set are incorrect, rubbish in fact, as are any other conclusions drawn.

The correct title for the data-set is "Number of Reported and Recorded Deaths to date by Date of Occurrence" and not "Number of Deaths by Date of Occurrence"

 

[Brendan Burgess]

OK, what can we say about actual deaths. Is the following correct?

While it can take up to a month to report a death, the vast majority of deaths are reported within 7 days.
Therefore the data up to 11 April as reported yesterday, is essentially complete.
The data since the 12 April is unreliable and the number of reported deaths will increase

On 7th April there were 40 deaths

On 11th April, the last date for which we have reliable data, there were 25 deaths.

4 days is not enough for a trend. As you can see from earlier on the graph there were similar falls over a few days, only for the rise to resume.

If it takes 19 days from infection to death, the graph should reflect the following:

24 March - restrictions introduced so infections should fall
12 April - 19 days later, deaths should start falling

27 March - Lockdown
15 April - deaths should fall further.

If it takes a week to report the actual deaths, then we will have to wait until Thursday or Friday of this week to see if the restrictions have worked.

Brendan

 

[llgon]

Given the fact that I didn't hear Dr Holohan speak about his graph on Friday, and Mathepac's account of his explanation, I was willing to give him the benefit of the doubt and attribute the interpretation of his graph to a journalistic error. However looking at his Twitter account from last night I think I was wrong to do this.

He tweeted a similar updated graph, the heading is effectively the same as the previous one. The interpretation he gives in the text of the Tweet appears to correspond with the RTE report in the original post. Very surprised at this.

https://twitter.com/CMOIreland/status/1252344273938219011?s=19

A new graph shows that of the deaths reported yesterday, the second most common date of death was 11th April.

While it can take up to a month to report a death, the vast majority of deaths are reported within 7 days.

Using yesterday's figures as an example, it's a small majority, 39-33 by my count, though it should total 77.

Therefore the data up to 11 April as reported yesterday, is essentially complete.

With the benefit of hindsight Brendan, this wasn't true. Not being critical of you in any way, you posed it as a question and I would have thought it to be true until I saw this new information.

 

[mathepac]

The death rate in Ireland is not falling. Headline from the Indo: - "Coronavirus Latest Updates: Biggest daily death toll as Covid-19 claims 77 more lives in Ireland" https://www.independent.ie/irish-ne...claims-77-more-lives-in-ireland-39140740.html

This headline from Monday 20th April makes Saturday's headlines and the graph even more nonsensical as none of the numbers stack up in retrospect, making projections based on the numbers Holohan & Co invent even more bizarre.

This announcement yesterday is further confusing the picture. No actual causes of death confirmed. https://www.rte.ie/news/coronavirus/2020/0421/1132779-covid19-coronavirus-ireland/

 

[Leo]

The death rate in Ireland is not falling. Headline from the Indo:

But the rate relates to the per day deaths. The 77 number is reported deaths and there's a variable lag there, have they published the breakdown of when the 77 occurred?

 

[john luc]

With the blanket testing now going on in the care homes we should expect a jump in the confirmed numbers. A care home I'm in contact with came back to me saying they have 3 cases identified since Sunday

 

[EmmDee]

With the blanket testing now going on in the care homes we should expect a jump in the confirmed numbers. A care home I'm in contact with came back to me saying they have 3 cases identified since Sunday

I think this could be true of testing in general. The number of "cases" is a very noisy stat to follow because it is a function of number of tests performed. I also suspect that when wide spread antibody testing happens, there will be a lot more cases discovered. There seems to be some anecdotal evidence in many countries of December / January "flu" cases that seem very similar to COVID.

The only "hard" data is deaths and hospital cases.