By P Gosselin
“Heat Deaths” According to the RKI
By Frank Bosse
There is a publication by the RKI [Robert Koch Institute] on “Heat and Excess Mortality in Germany.”
The methodology is also explained in it: calendar weeks with a mean temperature above 20°C are filtered out, and then the excess mortality figures provided by the Federal Statistical Office for these periods are declared to be “heat deaths.”
The “model” is simple and delivers the desired results: this year, up to calendar week 29, we have already seen nearly 10,000 heat deaths. A distinction between “from heat” or “with heat” is not even attempted (just as with the Covid-19 deaths), even though it could be very important for assessing the actual cause of a death. The headline is set and eagerly quoted.
The 3rd figure from the RKI shows deaths per year starting from 2016. They occur in the summer months. What could be more obvious than building an (equally) simple model that directly links annual death figures with average summer temperatures? This “model” also doesn’t care about real physiology and the actual impact of heat on human beings.
When is heat actually life-threatening?
For this, there is the “wet-bulb temperature.” You can determine it yourself: wrap a moist cloth around a thermometer. Evaporation follows, which cools. How much it cools depends on the humidity of the ambient air. This metric measures both, and starting at around 30 °C on a thermometer prepared this way, it becomes critical for humans because thermoregulation (sweating) can no longer function sufficiently.
What are realistic values? On July 30, 2026, at 4:00 PM, 37 °C was measured in Berlin with a dew point of 12 °C. The wet-bulb temperature came out to 21°C. That was about 9°C short of the critical threshold. The heat was NOT life-threatening because the air was too dry, allowing thermoregulation to work effectively. The RKI does not take any of this into account; it solely uses the “mean air temperature.”
Well then, the RKI “model” for annual figures can be recreated in just a few minutes.
The data for the deaths assumed by the RKI can be downloaded here. Only the numbers for Germany are interesting. Now, one just needs the “average summer temperatures” from the DWD [German Weather Service] from here.
A scatter plot is created in no time using the 2016–2025 data:
Figure 1: Heat deaths vs summer mean temperature.
Indeed, the number of deaths depends 81% on mean temperatures (the R-squared). From the line equation in the diagram, a tool emerges to replicate the RKI figures:
Robert Koch Institute Figures and Model
Figure 2: Heat deaths vs year.
The correlation is sufficient; the mean error is only 21%.
Now one can apply this “model” starting from 1950. According to it, no one dies “from” heat at a mean summer temperature below 17.4 °C. Above that, they do! However, mean temperatures say almost nothing about actual “heat” on individual days.
Figure 3: Heat Deaths Since 1950 (RKI)
Up to 2025, the year 2003 takes the cake with almost 8,700 heat deaths—except the RKI hasn’t noticed this yet. According to this logic, a total of around 72,000 people became “heat victims” since 1950.
So much for the “model” using simple linear regression. It simply links mean temperatures with “heat deaths” and aligns very well with the RKI’s model along with its stated uncertainties.
Pure statistical artefact
What both have in common is that they have nothing—absolutely nothing—to do with physiology or real health risks from heat. The “heat deaths” are a pure statistical artifact. It is simply “assumed” that excess mortality during a warm week is caused by the heat. The final image achieves the desired effect: “Look how many fatalities climate change is costing us!”
All of this from a “model” by an institute whose actual job is to provide valid risk analyses for human health. Its “heat death model” is a joke without any physiological background! Is the institute serving purely political interests once again? That would be truly tragic!
