How many heat deaths?

“Portugal scrapped its initial estimate of 1,300 deaths and lowered it to just four,” from today’s cnn.com. Germany claims 40 heat-related deaths this summer.

The British claim 907 extra deaths across the span of a very hot week. Here are the Italians: “The Health Ministry said on Thursday 34,071 people over the age of 65 died between July 16 and August 15, compared with 29,896 in 2002 — a 14 percent increase.” French estimates range between 10,000 and 15,000 deaths.

How much of this was avoidable and how much was random movement in the numbers? In net terms, how many people actually died prematurely?

Consider the stability of mortality statistics. I checked the UK Office of National Statistics tables, for England and Wales, and was surprised how much death rates bounce around (at the link you need to go through some work to create the file in readable form, follow the instructions).

Take deaths over the time period 1985-2001. Rounding off the figures to the thousands, the median change in death numbers, from one year to the next, is 9,000 (the totals run from 530,000 to 590,000 deaths per year). The biggest change we see across a year is about 25,000. We shouldn’t expect those changes to be distributed perfectly evenly across the months. But if we divide by 12 “naively,” it would be very common for death tallies, on a monthly basis, to change by 700 to 800 people, when comparing one year’s August to the previous year’s August.

Now let us go back to the British figure. If the number of deaths jumps in a single week by 907, compared to an average change, this is out of the ordinary but not unthinkable. On the other hand, random noise plays a very small relative role when the monthly death rate jumps from 10,000 to 15,000.

These people all would have died anyway, the next question is when. Twelve months from now, will the yearly death rate stand above its average or will we now see fewer deaths for a while? To what extent did the heat redistribute deaths from September and October to August? (Eli Lehrer raises the further interesting question of whether some of these people died through a psychological effect, given that the media were reporting that a “heat death time” was upon us.)

Note that the richer and more technologically advanced a society, the fewer heat wave deaths we should expect, read here and here. For one thing, richer people are more likely to buy air conditioners. Everyone points the finger at negligent children, or the French hospital systems or August vacations. We can also blame the sluggish economic growth rates of the Continent. Social welfare states can be quite inhumane, once we examine secondary consequences.

The bottom line: something very bad did happen in France and Italy.
Something less bad happened in Britain. But we still have not gotten to the bottom of how bad or why.