Written by Luke Paterson
Edited by Antony Fitzsimmons
26 May 2020
Reading time: 4 minutes
A brief insight into the meaning of the graphs related to the coronavirus deaths we see in the news each day.
One interesting issue the pandemic has brought to light is the idea that most people know extraordinarily little about statistics. Like many other subjects, economics has an Achilles heel or at least that’s what most economic undergraduates think. Econometrics is essentially statistics applied to economic data but gives a useful insight into understanding the world in general. Although one of my preferred economic subjects, I am not going to bother you with its technicalities, such as multivariate regression analysis and confidence intervals, as most would lose interest from this point onwards.
Throughout the crisis, graphs depicting the number of deaths by country, per million of the population, or above a 5-year period have been plastered, rather depressingly, all over the news. I provide a brief insight into the real meaning of such diagrams and why, in some cases, we should be careful in rushing to conclusions.
Death by country
Firstly, comparing cases and deaths by country. Obviously, by comparing with other countries you are effectively seeing how well your country is doing against others. The main issue is the way in which such figures are reported as this has a significant effect on the numbers published. For example, Covid-19 deaths in England and Wales only began to factor in care home deaths from 29 April onwards, compared to France and Germany which included daily care home deaths since reports began. It soon became apparent deaths in England and Wales had been significantly misjudged. Similarly, Germany’s care home deaths only included those who had tested positive for the coronavirus. The UK, conversely, has only a weekly figure where, without an official positive result, doctor’s suspect that Covid-19 contributed to the deaths.
This implies that countries, like Germany, may be underestimating Covid-19 deaths whereas other countries, like the UK, could be under or even over reporting deaths depending on the judgement of doctors; however, this doesn’t factor in the fact that Germany’s testing capacity was substantially greater than the UK’s. Secondly, when data is not reported daily, huge spikes in the daily death count are likely to occur. This is why, in most cases, a 7-day rolling average is used to smoothen out the curve.
Comparability is difficult
It is worth mentioning there is no international standard for the reporting of deaths, or even cases, during a pandemic! For example, GDP and the literacy rate are reported and checked based on an agreed international standard of publishing; maybe this is something we can learn from for the future? No international standard means some countries may be reporting data which is not comparable to others. Additionally, it must be considered that each country has different factors affecting the epidemiology – the spread of the virus – such as differing political ideologies, health care systems, testing capabilities, demographics and, crucially, population sizes. This is why comparing deaths per million is more sensible as it accounts for the fact the USA has nearly 5 times the number of people the UK has.
How testing is done matters!
Measuring and testing for the virus also massively affects the figures reported. Take the UK’s approach when the virus first started to spread; testing was only conducted on those ill and admitted to hospital. Why is this a problem? To put simply, it is biased. If someone has been admitted to hospital with Covid-19 they have, at minimum, aggressive symptoms of the virus. Cases like this were more than likely to become severe or result in fatalities. This implies that, initially, the death rate in the UK was overstated. As more testing slowly became more prevalent, the fatality rate decreased – it remains a constant moving picture.
In addition to this, the reliability of such tests and the time it takes individuals to receive their results are important things to consider. The true extent of the prevalence of the coronavirus will become fully apparent when enough people have had an antibody test. Only then will an accurate estimate of the amount of people who have had Covid-19 come to fruition!
Average deaths per year
Finally, one of the most striking graphs is deaths above the average of previous years. This is when we look at the number of deaths which have occurred in previous years and compare it to that of the current year. Rather effectively, this gives you a more complete picture and is arguably the best for cross-country comparison regarding how well each country responded to the pandemic. The diagram below gives us an idea of the true number of lives lost during the pandemic and highlights either a mass undercounting of Covid-19 deaths and/or a huge amount of indirect deaths which have been caused, in part, by NHS strains and people not seeking medical aid due to transmission fears.
Sadly, excess mortality is only available for a few select countries as it requires accurate data surrounding a country’s previous yearly fatalities. Our World in Data reports this is a major drawback of the much needed metric. As of 23 May, the UK’s Office for National Statistics reports the number of excess deaths to be somewhere in the region of 55,000 people (or 67% higher than normal). In other words, there are 18,325 unexplained deaths in addition to the currently confirmed 36,675 Covid-19 deaths. Unfortunately, comparing excess deaths to other countries, the UK is seen to have the most in Europe.
Two reliable sources of Covid-19 data in the UK: the Department of Health and Social Care (DHSC), who publish all deaths in hospitals at around 2pm daily, and the Office for National Statistics, who publish deaths in any location (including care homes) once a week at 9.30am on Tuesdays.
The Health Foundation
Office for National Statistics
Our World in Data