Review the data

I reproduce below my piece from April 11th. I am pleased others now have come to see that bad data and wrong forecasts are a problem for the government’s scientific advisers:

Review the data

The death rate is too high. Every death is a tragedy. We all want to see it going down. The nation mourns those who have lost their lives to this disease.

Soon the government needs to review progress with its object of flattening the curve of the virus spread, to consider how long we need to remain in lock down.

I am glad we are now privy to the figures the government relies on. In the daily briefings we are shown two graphs or “curves”, the trend in hospital deaths from CV 19 and the trend in hospital admissions for the infection.

It is presumably these curves that need to be sufficiently flattened to allow the government to transit to the third phase of its  advisers’ planned handling of the virus crisis.

There are several issues with the Death figures that need getting right. I think it would be good for some administrators and statisticians from government and or from the scientific community advising the government to spend some time ensuring accurate data. This should not involve medical and hospital staff time which is needed to handle the patients.

 There was a change in the basis of their compilation on March 26th, when they shifted from 8 hour to 24 hour reports, moving the numbers up.  Can they smooth the figures to allow for this?

There is the issue of whether the deaths are all recorded on the appropriate day. The day before yesterday we were told the higher number included deaths from earlier days which they thought had not been recorded at the time. Can’t the numbers to be reworked for all but the most recent by reference to the death date on the medical death certificate?

There is the possibility of double counting. If deaths are sometimes recorded promptly  before paperwork is completed, and other times when the paperwork is ready, there needs to be a check that they do not end up recording the same death twice.

The wider ONS figures are also of interest. These are higher as they include deaths not in hospital where CV19 was present. These include  some where  the deaths certified as with CV19  are based on statements  about symptoms with  no tests to confirm the presence of the virus. The figures include cases where  CV 19 is mentioned where other severe conditions mean the patient would probably have died without the virus anyway.

Hospital admission with the virus is an easier series to get right. Presumably all on admission for CV 19 treatment are tested to ensure they have it, to make the correct treatment available. Admission takes place at one stated time and date, so it should be relatively easy to get a clean series of numbers that are accurate. A simple check would be to compare bed numbers and occupancy rates by hospital and to examine any outliers.

This is such an important decision both to control the disease and for the jobs and livelihoods of the many, that the decision takers need the most accurate possible numbers.  (End of original)

I repeat again today these questions

What is a Covid 19 death? What are the numbers for dying of CV 19 and dying with CV 19?

Are the back numbers correctly attributed and compiled?

What is total NHS and private sector bed capacity and what is the current utilisation rate?

What proportion of total beds are currently taken by CV 19 patients?

How are the extra nurses recently recruited and the returners from early retirement being deployed?