Germany's heat alerts show no average benefit — until the cities are told apart

Heat health warning systems are the cheapest adaptation measure a country can deploy, and among the least evaluated. A study published in Environment International on 23 August assessed Germany's across 15 major cities, and produced two different answers depending on what the analysis controlled for [s1]. Both belong in any honest account of it.

The study

Researchers linked daily all-cause mortality during the warm-season months of May to September, from 1993 to 2020, with heat alert data and meteorological information for 15 major German cities [s1]. They used a difference-in-differences approach to estimate city-specific effects of heat alerts on mortality during heat episodes, then pooled those estimates with meta-regression models to examine heterogeneity across cities [s1].

Difference-in-differences is a design that compares the change in outcomes before and after an intervention against a comparison that did not receive it — an attempt to isolate the effect of the alert system from the general trend.

The two results

Substantial variation appeared across cities: some showed significant reductions in mortality during heat episodes after the warning system was implemented, and others showed no significant effect [s1].

The pooled relative risk, based on meta-variables averaged across all studied cities, was 1.00 (95% CI, 0.98 to 1.01) — no overall significant protective effect of heat alerts on mortality [s1].

When the analysis controlled for three city-level variables — recreational area per person, total population, and population density — the pooled estimate became 0.85 (95% CI, 0.75 to 0.97), described by the authors as a significant but small protective effect across all studied cities [s1].

Those two numbers are not in conflict; they answer different questions. The first asks whether heat alerts, on average across German cities as they actually are, are associated with fewer deaths. The second asks whether they are associated with fewer deaths once you account for how much green and recreational space a city has and how densely its people live. The gap between them is the finding.

The authors' conclusion follows from that: effectiveness varied considerably across cities, and city-specific factors — population size, density, and the presence of blue and green urban infrastructure — matter for whether a warning does anything [s1]. Understanding those factors, they argue, can help tailor heat-health action plans to the specific characteristics of different urban areas [s1].

The obvious limit

A warning is an instruction to do something. If a city's residents have nowhere cool to go, no shaded park within walking distance, and housing that holds heat overnight, the instruction has less to act on. A difference-in-differences study cannot show that mechanism, but the pattern in the results is consistent with it, and the authors' emphasis on green and blue infrastructure points the same way [s1].

The study also cannot separate the alert itself from everything else that changed when it was introduced. Warning systems tend to arrive alongside media attention, care-home protocols and public messaging.

Why the question is getting more urgent

A modelling study published in Nature Communications on 11 August projected future heat-related mortality across Europe using a distributed lag nonlinear model that, unusually, incorporates humid heat and compound day-night heat extremes — periods when both daytime and night-time temperatures are high [s2]. Night-time heat matters because recovery overnight is what makes a hot day survivable.

Under adaptation held at 2010-2019 baselines, the study projects heat-related mortality in Europe increasing annually by 103.7 to 135.1 deaths per million people by 2100 for every degree of global warming, across a range of population and climate scenarios, with Western and Eastern Europe suffering the most [s2].

Two further findings from that paper deserve attention. If global warming exceeds 2 °C, climate change would dominate the projected increase in heat-related mortality — accounting for 84.0% to 96.8% of it, rather than demographic ageing [s2]. And across all socioeconomic pathways, even a 50% reduction in heat-related relative risk through physiological adaptation would be insufficient to offset the climate-driven escalation [s2].

That last point is a direct constraint on what warning systems and behavioural adaptation can achieve. The German study asks how well an alert works; the projection says that even large improvements in how well people tolerate heat do not close the gap on their own.

What these are and are not

The German study is observational, covers 15 cities rather than the country, and spans a period in which the warning system itself evolved [s1]. The European projection is a model — its outputs are conditional on emissions scenarios, population pathways and assumptions about adaptation trajectories, and it says so [s2]. Neither tells any individual anything about their own risk.

Together they frame a specific policy question: whether heat-health action plans should be evaluated and funded as information campaigns, or as urban infrastructure programmes. The German data suggest the alerts do more where the built environment gives people somewhere to act on them [s1]. The projection suggests the amount of adaptation required is larger than behaviour alone can supply [s2].

Sources

• Assessing the effectiveness of the heat health warning system in preventing mortality in 15 German cities: A difference-in-differences approach — Environment International, 2025-08-23
• Future heat-related mortality in Europe driven by compound day-night heatwaves and demographic shifts — Nature Communications, 2025-08-11