Estimating deaths from heatwaves

Credit: istock/Marccophoto

On 26 July, ‘The Guardian’ published an article by its excellent environment editor, Damian Carrington, on a newly published report on heatwaves by the House of Commons Environment Audit Committee.

The article included the following quote from me: “It is likely that more than 1,000 people have died in the UK as a result of the extended heatwave conditions this summer”.

I made this estimate based on the extra deaths that occurred during hot weather in 2016 and 2017.

For instance, the report on annual mortality figures published by the Office for National Statistics in November 2017 noted a spike in deaths on 19 June 2017. It stated: “The peak in number of daily deaths on 19 June 2017 coincided with a period of warmer weather where the daily temperatures were substantially higher than any other days in the preceding weeks and were also higher than the five-year average temperature for these days. Central England daily temperatures from the Met Office Hadley Centre show that the highest temperature of the year was observed on 19 June 2017 with a mean temperature of 21.9 degrees Celsius and maximum temperature of 29.1 degrees Celsius.”

This was a notable increase in deaths from one hot day. In 2016, there were three short periods of hot weather between 19 and 21 July, 23 and 25 August and 14 and 16 September. An analysis carried out by Public Health England and published in July 2018 concluded that there had been 612 “excess deaths” during the first period and 296 in the second period, a total of 908.

The UK has experienced two short periods of uncommonly high temperatures this summer, in late June and late July, when some parts recorded daily maxima in excess of 30 Celsius degrees, and I estimated that these might have had a comparable impact to the two periods of hot weather in summer 2016.

However, Professor Richard Tol, a former adviser to the Global Warming Policy Foundation, asserted earlier this week on Twitter: “Bob Ward’s prediction did not come true”. Unfortunately, he arrived at his conclusion through poor quality analysis, and he apparently did not understand that it is not yet possible to calculate how many deaths have occurred due to the hot weather this summer.

Professor Tol based his claim on provisional figures published by the Office for National Statistics for deaths registered weekly in England and Wales. Professor Tol produced a graph which purported to compare the weekly figures for 2018 with the average for the period between 2010 and 2017.

When I attempted to reproduce Professor Tol’s figures, I discovered that he had made a number of unexplained mistakes in calculating the averages. The real figures show that the weekly number of registered deaths in 2018 has been higher than average every week from the week ending 8 June. However, Professor Tol’s graph inexplicably shows lower than average numbers for several weeks.

But even if Professor Tol had used the correct figures, he could still not tell whether there have been any excess deaths associated with the hot weather for a number of reasons, many of which have been explained in a blog by Nick Stripe, Joint Deputy Director of Health Analysis and Life Events at the Office for National Statistics.

The first reason is that the not all deaths are registered on the day that they occur. The second is that weekly figures do not always allow daily peaks to be identified. The Office for National Statistics painstakingly analyse each registered death to establish the actual date on which it occurred, and publish the results as daily death counts on a quarterly and annual basis. The first set of figures for this summer is due to be published in September.

While Professor Tol would have benefited from heeding Mr Stripe’s warnings, his blog does provide some slightly misleading guidance. For example, Mr Stripe compared the total number of registered deaths during the seven weeks between 2 June and 20 July 2018 and observed that it was lower than the comparable figure for the corresponding period in 2017 and 2016. However, this is not an indication that there have not been any heat-related deaths in 2018, as those periods in 2016 and 2017 both included bouts of hot weather that we now know were associated with excess mortality.

Furthermore, Mr Stripe pointed out that the weekly all-cause mortality surveillance carried out by Public Health England has not identified any statistically significant increase this summer. This again is not an indicator of whether there have been any excess deaths caused by the hot weather. The surveillance report for the week including 19 to 21 July 2016 also did not pick up any statistically significant increase, even though we now know that there were 612 excess deaths during that period.

The truth is that we will not know the full extent of the excess deaths associated with the hot weather this summer until Public Health England carries out a detailed analysis that takes account of the heat-related deaths associated each year in order to calculate an appropriate baseline. This may take some time as the analysis for summer 2016 was only published in July 2018.

Until then, Professor Tol and others would be wise not to jump to conclusions.

 

Bob Ward is policy and communications director at the ESRC Centre for Climate Change Economics and Policy and the Grantham Research Institute on Climate Change and the Environment at the London School of Economics and Political Science.