March 22, 2020

expert reaction to comparisons of UK and Italy death rates from COVID-19

There have been questions in the media about the differences between the death rates from COVID-19 in the UK and in Italy.

Dr James Gill, Honorary Clinical Lecturer, Warwick Medical School, said:

 “There are going to be several interplaying points looking at Italy and comparing with the rest of Europe and the UK.  The first is simply time.  COVID19 has had longer to affect the Italian health system and spread within the population compared to other countries.  So there is an element of everywhere else being the curve from a time perspective. If we look at the cumulative case trajectories, China dropped below the 33% daily increase marker at around 21 days after their 100th case.

“The UK pulled below that same marker 5 days after the 100th British death, whereas Italy pulled below that same 33% case increase line at 16 days, notably as did Germany. Hopefully these changes indicate the sharing of knowledge and best approaches to tackling the outbreak. Japan, Singapore and Hong Kong have all had much shallower trajectories than the rest of the world and may be able to provide insights for countries who are still earlier in their outbreaks.

“As for the differences seen across Europe, it is easy to draw conclusions about the different EU countries case trajectories based on demographics. Such as the age of populations, with Italy having 22.5% of its population over 65yrs of age, compared to 18.3% in the UK, or per previous per capita spend on health care, with Germany spending €4500 per person compared to €2500 per person in Italy.

“At this stage with the data we have on COVID 19 changing on a daily bias, it is exceptionally difficult to be able to draw accurate conclusions between different countries currently battling the infection. What we must ensure though is that countries learn from and adapt with one another when effective strategies are demonstrated. Which may mean being prepared to change course if an attempted approach to cutting the infection rate is shown to inferior to another

“One thing is for certain, good data collection and it’s analysis is going to play a pivotal role in the story of how COVID 19 progresses.”

Prof Duncan Young, Professor of Intensive Care Medicine, University of Oxford, said:

 “In terms of whether we (the UK) have a different case fatality rate (what the chance of dying is if you test positive) compared with Italy depends on the testing policy. Testing more patients increases the denominator (people who have a positive coronavirus test) and so for a given number of deaths reduces the case fatality rate.

“One way of gauging the quality of care (or the effects of overload of the healthcare system) is to compare death rates for known cases of equal severity. Thus assuming that patients in Italy and the UK are equally sick when they are admitted to hospital or to an ICU, the in-hospital or in-ICU death rates for COVID positive patients will give a VERY crude estimate of the quality/overload of the care system. Any difference could, however, also be due to differences in patient characteristics (age, smoking status, etc) between UK and Italy.”

Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:

“If other countries have varied their COVID-testing regimes as much as the UK has, then I would not suggest making direct comparisons on the basis of deaths per number of COVID-19 positive cases reported. Rather, consider aligning countries on the basis of the evolution of their epidemic (e.g. UK two weeks behind Italy’s COVID-19 deaths) and, for each age-group, comparing COVID-19 deaths per million of population.

“The starkness of the increase in relative COVID-19 death rates by age-group that we have seen in other countries is why the UK seeks to protect those aged over 70 years from becoming exposed to COVID-19. For example, based on 1625 COVID-19 deaths in Italy (reported as 0.25% aged 30-39 years, 0.62% aged 40-49 years, 2.65% aged 50-59 years, 8.55% aged 60-69 years, 35.57% aged 70-79 years, 42.71% aged 80-89 years and 9.6% aged 90+ years), and considering the age-distribution of the Italian population, the COVID-19 death-rate per million of population appears to be 4 times higher at 50-59 years than at 40-49 years; 19 times higher at 60-69 years; 105 times higher per million of population at 70-79 years than at 40-49 years; and around 210 times higher per million of population at 80+ years than at 40-49 years. In just a few weeks, sadly, UK will be able to estimate these rates for its own population.

“More encouragingly and importantly, in a few further weeks, the UK will also be able to assess how effectively we have implemented social distancing as these relativities would be expected to change in a manner that safeguarded the elderly.

“COVID-19 will also have indirect effects – on other causes of mortality. A fuller reckoning will therefore be to compare each age-group’s annual mortality in 2020 with expected mortality, based on national life-tables. Ordinarily, there are around 11,000 per week in the UK as we head into March. Some deaths from other causes may be counted as COVID-19 deaths if the patient had tested positive: hence, our need to track all-cause mortality by age-group.”

Prof Robert Dingwall, Professor of Sociology at Nottingham Trent University, said:

“Interpreting mortality rates is not as easy as it sounds because different countries have very different systems of death registration and coding. I did some work* on this in relation to England, France and some US states about ten years ago. In the case of Covid-19, it is important to know whether deaths are being reported as due to Covid-19 or to the underlying medical condition with Covid-19 as a secondary cause. Different countries may vary in this respect – within the US it may vary by state or even by county.

*”This work, which was prompted by the 2003 heatwave in Europe and the associated mortality crisis, was mainly published in French (C. Vassy, R. Keller and R. Dingwall) Enregister les morts, identifier les surmortalités: une comparaison Angleterre, États-Unis et France, Presses de l’école des hautes etudes en santé publique, Rennes, 2010.”

Prof Sir David Spiegelhalter, Chair, Winton Centre for Risk and Evidence Communication, University of Cambridge, said:

“To compare rates of exponential growth early on in the epidemic, it is best to fit a line to the logarithm of the observed number of deaths each day, using a technique called Poisson regression.  This reveals that, over the 12 days following each country reaching 5 deaths, the daily increase in underlying death rate in the UK (37%) was only marginally higher than that of Italy (35%).  A difference this size could easily occur by chance, were the countries following a similar underlying trajectory.”

All our previous output on this subject can be seen at this weblink: www.sciencemediacentre.org/tag/covid-19/

Declared interests

None received.