Why is the daily number of deaths not falling?

Brendan Burgess

Brendan Burgess

Founder
Messages
52,117
It takes about 19 days on average from infection to death.

We introduced restrictions on 24 March, so the spread should have been slowed down as that was 22 days ago.

We went into lockdown on 27 March , which was 19 days ago.

So the underlying true rate of infection should have slowed down, should it not?

Were there not more people infected on one day, 29 days ago than 19 days ago?

A confounding factor would be the care homes.

The rate of deaths in the community excluding care homes, could well have fallen dramatically, but these would be obscured by a rise in deaths in care homes, where the lockdown wouldn't have had much impact.

Brendan
 
Here are the numbers

4444

4445

4446
 
You would wonder how the care homes got so badly affected by this virus in the first place though
 
I would imagine it is still spreading in nursing homes, care settings, hospitals and hospices. Only 10% (40 people) of deaths have been in ICU. They won’t intubate the very old or gravely ill so I assume most of the remaining 90% fall into that category.
 
Professor Philip Nolan spoke last week about the reproductive or r value of the virus. If this is greater than 1 the rate of infection will increase and therefore number of deaths. I think he said that before restrictions were introduced in March the r value was about 3.5, so everybody who was infected passed it on to an average on 3.5 more.

Last week his group estimated that the r value was close to 1 but they could not be exact about it. Once it is less than 1 the rate of infection will drop and they can look at easing the restrictions. Obviously there will be a time lag before the number of deaths start to drop.

@irishdataviz shows that there has been a large drop in the growth rate of deaths over the past two weeks.

IMG_20200415_094503.jpg
 
Brendan Burgess said:
It takes about 19 days on average from infection to death.

We introduced restrictions on 24 March, so the spread should have been slowed down as that was 22 days ago.

We went into lockdown on 27 March , which was 19 days ago.

So the underlying true rate of infection should have slowed down, should it not?

Were there not more people infected on one day, 29 days ago than 19 days ago?

A confounding factor would be the care homes.

The rate of deaths in the community excluding care homes, could well have fallen dramatically, but these would be obscured by a rise in deaths in care homes, where the lockdown wouldn't have had much impact.

Brendan
Click to expand...

Brendan,

There needs to be a caveat on looking at day to day numbers. The data will be really noisy. Some deaths may not be reported / confirmed / tallied on day of death. So it's possible some numbers are from a few days ago. I know the UK numbers have some deaths from up to a week or two prior.

It's the problem with raw data like this - the noise can be confusing. I think the rolling average and trend is better (cf the FT analysis). But it is something to look for over the coming days - we should be seeing some movement by end of week
 
llgon said:
there has been a large drop in the growth rate of deaths over the past two weeks.
Click to expand...

That is very interesting and I am trying to get my head around it.

If 100 people are infected and there are no measures in place, the growth rate could be 35% a day, but that would be only 35 new cases in the first day.

If 1,000 people are infected, but due to measures in place, the growth rate is down to 10%, that is still 100 cases.

So the number of cases is increasing while the growth rate is falling.

But I don't think that the growth rate is that important. It should be the absolute number of new cases.

Brendan
 
Brendan Burgess said:
That is very interesting and I am trying to get my head around it.

If 100 people are infected and there are no measures in place, the growth rate could be 35% a day, but that would be only 35 new cases in the first day.

If 1,000 people are infected, but due to measures in place, the growth rate is down to 10%, that is still 100 cases.

So the number of cases is increasing while the growth rate is falling.

But I don't think that the growth rate is that important. It should be the absolute number of new cases.

Brendan
Click to expand...

I think they key risk they look at is the growth rate (or number of days to double) as that causes the overwhelming of critical care i.e. 20% growth as the infection spreads causes new critical cases exponentially increase causing the crisis. A reducing growth rate even if it is an increase in absolute numbers means the spread is slowing. Which I think is why the daily absolute numbers can be confusing. Now at some point the absolute number does matter when the beds are full and new hospitalisations exceed "people released plus dead". But more broadly, 10% critical cases from a large infection which is slowing is better than 35% critical cases from a smaller but rapidly growing infection population

Again - going back to the FT data analysis (which I find very useful and are freely available on twitter including explanations - @jburnmurdoch), they use exponential scales to visualise the growth & spread. It provides a clear indication of direction of travel
 
Brendan Burgess said:
But I don't think that the growth rate is that important. It should be the absolute number of new cases.
Click to expand...

The rate is more useful in modelling and determining next steps. It was hinted in the briefing yesterday that they believe the R0 number is still above one despite the current measures, so until that drops below 1, we can expect the numbers to continue rising.
 
? clusters in nursing homes and institutional settings given some of this population have underlying health conditions, added with ? inappropriate environments and inability to enforce social distancing of residents, just a thought for death rate not falling
 
llgon said:
Last week his group estimated that the r value was close to 1 but they could not be exact about it. Once it is less than 1 the rate of infection will drop and they can look at easing the restrictions.
Click to expand...

This.

On average each infected person is still infecting more than one other person. So the actual number of infected is growing. So daily deaths are growing.

Remember actual infections is much higher than confirmed cases
 
Last edited:
I wouldn't expect number of deaths a day to start falling till few weeks after number of infected a day started falling.

Our confirmed cases a day is a proxy for number of new infections a day.
 
We have slowed the daily growth rate, but as the total number is now bigger, the number of new dailys is still higher.
 
There are a few caveats to the reported daily cases and we haven't clarity on;
The new daily cases when were these tests taken? These could have related to swabs taken from say 31st March, person could have had symptoms for a few days before looking for a test and the wait for the test could mean an additional few days so its no unreasonable to assume that a further 3 weeks would pass (including 14 days incubation period) between time of contracting virus and getting tested which would mean that results announced this week could relate to early March.
Take for example Mary Lou McDonald she got her results on 13th April, so we can assume she was in the case numbers for Monday, this test was taken on 28th March, she said she had a head cold on 26th March which means she had symptoms meaning she could have contracted the virus up to 14 days prior to showing symptoms, which puts date of contraction at 12th March. I would think she is also an exception from the short period from showing symptoms to getting tested, most are waiting longer that this.
Therefore the daily case numbers we are receiving this week are from infections occurring 4 weeks ago. So by simple maths and if the restrictions are effective you should see a drop off in case numbers from the latter end of the month.
 
You must log in or register to reply here.
Back
Top