The first point to make is that rule one of statistics is that correlation is not causation ‘Data fitting’ means finding an apparent relationship and then attributing a cause, which is frowned on because trends may appear to be in step, but not actually connected. The most famous example is ‘fitting’ the Dow Jones Index in the 1920s and 1930s to average skirt length in fashion magazines.
The test is the reverse relationship: if Chanel had decided to launch the miniskirt in 1935, would the Dow Jones have gone up, and why?
The second rule of statistics is to make sure you are measuring the right things. If you draw your data sources too narrowly, you may miss a factor. Rather than ‘fit’ your choice of factors to a trend, consider if there are other potential factors.
THE DATASET: ENGLAND, AND ENGLAND/WALES
The dataset has been chosen deliberately because the UK has a National Health Service (NHS), where the vast majority of hospital episodes and doctor’s appointments are logged into a central system, and where all deaths must both be registered by a coroner, and a cause of death must be attributed by a doctor. If there is a doubt over a cause of death, an autopsy will be ordered. Cause of death is attributed under the ICOD 22 system, an internationally agreed protocol. Data is published by the NHS, and all of the following health data is from the NHS dataset which is publicly available.
First: what does a normal winter look like?
There are seasonal flu outbreaks every year, which impact the elderly most heavily. On average, of those who die of respiratory problems, 76% are
over 80.
Some years are worse than others: 2018 was a bad flu season. The number of deaths in total can be mapped against the number of doctor’s consultations about flu or flu-like illnesses (‘ILI’ or ‘influenza-like illnesses’). In an average week in England, total deaths are about 10,000, and this does increase to reflect seasonal flu patterns, as shown by consultation numbers.
Since December 2019, coronavirus has been spreading, and is thought to have reached the UK in January 2020. Because of public concern, the NHS has been publishing daily figures on coronavirus deaths. However, other respiratory deaths have not stopped because of coronavirus. There was a peak at Christmas, as there usually is, but the current overall number of respiratory deaths as of mid-March 2020 is in line with figures for 2019.
DEATHS, WEEK BY WEEK, RESPIRATORY DISEASES, COMPARABLE TIME SERIES 2019 AND 2020
This does not deny the crisis which is occurring in intensive care units: a surge in cases, at the same time as large numbers of staff are not available because they are quarantined, and a lack of immediate availability of equipment such as ventilators, is a crisis. However, the NHS has been warning about a lack of winter preparedness by equipment and beds for at least 6 months.
Comparing available data for 2019 and 2020 to date, winter deaths from respiratory diseases of all kinds are continuing, 2019/20 was a mild winter in the UK.
