Statistical analysis methods to count COVID cases


importance of statistics

Is the UK high death toll explained by the fact the UK are a nation of fatties?

I am not sure I can endorse the inference, but the question is a very good one.

Towards the end of July, the UK was told hospital admissions and deaths were trending downwards. In this article I will attempt to explore rising COVID cases in the UK and ask how worried we should be?

As hospital admissions and deaths are two key indicators of the spread of COVID and these indicators suggest things seem to be looking positive at least they do if you are living in Scotland. Nicola Sturgeon first minister of Scotland said, “I am very pleased to say in the past twenty-four hours no deaths of a person who had been confirmed through a test as having the virus.”

Scotland's daily COVID death count has recorded no deaths for five weeks. This seems to be very good news. Let us rip off the statistical sticking plaster. 

What is really happening.

Scotland daily death figures have been zero that is true. But the daily figures only record those who have tested positive for COVID nineteen within the past twenty-eight-days

That is a hard cut off. 

If someone dies more than twenty-eight-days after their positive test they are not included in the daily figures.

So, if someone either never got a test or they lived for twenty-nine-days or more, after getting a positive test, they would not be counted when they died.

There is another measure deaths’ in Scotland which comes from the national records of Scotland and they count any death certificates where COVID is mentioned as the cause of death.

The data from this lags slightly behind with daily data but we do have the numbers for the first four of these five weeks that Scotland has been reporting zero deaths from COVID nineteen and during that time the national records of Scotland have identified thirty four COVID related deaths

So, why did the government keep on announcing no deaths when the most credible source says there have been thirty-four?

What about the other nations in the UK?

Wales and Northern Ireland also have twenty-eight-day cut off point after that if someone with a positive test result in their medical history dies it is not counted as COVID virus related.  

But anyone waiting for zero daily deaths in England might be waiting a while. Professor Yoon Loke of Norwich University medical school and Professor Carl Heneghan at the University of Oxford highlighted the Public Health England has no such cut off point for the daily death figures. 

If you have a positive community test and you later die that death will appear in England daily COVID death toll even if tested positive in March and the person died in August after being hit by a bus.

So, while Scotland daily figures are underestimating their deaths. England’s are overestimating theirs.

English daily deaths are currently hovering around fifty.

What is interesting is to look at how these figures from Public Health England compared to those from the office for National Statistics. So just like Scotland, England has an organisations counting all the death certificates that mention coronavirus. In England this is being done by the Office for National Statistics (ONS).

At the beginning of the pandemic. Death figures from the ONS were much higher than the Public Health England figures. You can imagine a lot of people have not had the opportunity to have a test positive or otherwise. But their COVID symptoms were recorded on their death certificates. But now it has swapped.

Daily figures from Public Health England are coming in much higher.  Sometimes twice as high as those calculated by looking at registered deaths.

The latest ONS figures are hovering around twenty deaths a day rather than fifty. 

Let us suppose more people are being tested. As a bigger and bigger pool of people grows. Who once upon a time had a positive test result and at the same time with the virus currently at low levels?

The number of active infections is small. So, how do we know how much difference there is, in how we measure deaths. Is there a distorting difference between English and Scottish deaths?

So, the question is how many of the English deaths would not be included in the daily tally. If England was using the same cut off as the rest of the UK.

Mary Gregory is the deputy director at the office for statistics regulation. She has said Public health England has said that over the period of the pandemic. Until mid-July about ten percent of the current cases were published would not be included in the twenty-eight-day cut off. That is a relatively low number in the context of the deaths. But we do not know how much that is affecting the recent data compared to the old data. Therefore, what the scale of the impact is now.

So, ten percent over the whole pandemic.

That number is going to rise inexorably. 

Given the ever-large stock of people with positive tests.

It might be substantially more than ten percent of the more recent deaths. This could be a big deal and we really should know. Mary Gregory states she understands Public health England. Will publish some figures on that. The cut-off point is not the only reason. 

Why are Public Health England numbers higher than the office National Statistics numbers at the moment?

Mary Gregory said we do not have full details on what might be the differences. But it is likely that is not the only cause of the difference. Public Health England data is based on a lot of different data sources that are matched together. This allows us to get quite a board coverage of the data. This factor often reduces the delays in getting the data so some of it may be around the delays in registration. And they may well be other cause we do not yet know about. 

So, Public Health England are using multiple data sources. To try to put together the best possible information. Does that not introduce a risk of double counting?

Mary Gregory states Public Health England has put in place internal checks. Against the sources to try and avoid any duplication. It is not impossible with some duplication of individuals does get through. But they have worked hard to make sure that is avoided as far is as possible.

So, England is overestimating deaths by not having a cut off after a positive test. Scotland is underestimating deaths. Particularly as we get better for treating severe cases of the disease. Potentially delay deaths from that cut off.

Mary Gregory replies all the nations are reviewing as more evidence becomes available. What the appropriate cut off might be. She thinks we are likely to see all the nations. Move towards something that is more consistent with each other. She said she thought something around the sixty-day measure has been considered by all of them. This might be more appropriate at a point in which you will capture most of the cases that are related to COVID nineteen.

Perhaps we should all feel nervous. Big policy decisions are being made yet the data is still not quite as good as it should be.

Mary Gregory suggests the key. Is seeing more transparency about the data and understanding the data. If we had information available on what are the deaths. Within the twenty-eight-day cut off and the sixty-day cut off and what goes beyond that. Then those people who can understand that information could put it all in context. But we still need a clear headline measure that does not cause confusion. Because having to many figures. In the public domain makes it difficult for people to understand.  

Government has suggested they will be announcement as to how daily deaths will be reported.

Morale of this article apply the 5Ws Who, What, When, Where, Why to whatever you learn or told.

About the author 

Christopher Bird

Building your own Power App, BI solution, or automated workflow can be a mind-blowing experience. It can also be a nightmare. Particularly when you begin with a blank screen. My advice, get professional help as and when you need it. That's what successful people do.

{"email":"Email address invalid","url":"Website address invalid","required":"Required field missing"}
Subscribe to get the latest updates