Timely detection of excess mortality to support heatwave response: insights from Hesse, Germany

Abstract Background Heatwaves are intensifying due to climate change and significantly increase mortality, particularly among older adults. In 2023, the Hessian Heat Action Plan (HHAP) was launched to reduce heat-related deaths. Timely identification of excess mortality is crucial for guiding adaptive responses, yet reporting delays often hinder real-time detection. We developed and assessed an early warning system to identify excess mortality from the previous week during summer months. Methods We analyzed weekly all-cause mortality in Hesse, Germany (population 6.4 million; 22% aged ≥65, 7% ≥80) and applied a Bayesian nowcasting model to adjust for reporting delays of 0-12 weeks. Weekly mortality estimates with 95% credible intervals (CIs) were compared to a baseline (mean deaths during five spring weeks of the same year). “Heat weeks” were defined as those with mean temperatures >20 °C. Excess mortality was flagged using three criteria: (1) lower CI > baseline, (2) lower CI > upper baseline CI, and (3) median estimate >5% above baseline. Performance was evaluated using historical heat events (summers 2003, 2015, 2018, 2019) defined by the EuroMOMO method. Results Baseline mortality ranged from ∼1,000 to 1,200 deaths/week. All criteria detected substantial excess mortality (>1,400 deaths/week) during 2003 and 2015. Criterion 3 also identified moderate excess mortality (>1,300 deaths/week) in 2018 and 2019. Criterion 2 detected the 2019 event with a two-week delay. Criterion 1 did not detect moderate excess mortality events. Conclusions Nowcasting can support heatwave preparedness by enabling near real-time detection of excess mortality. Criterion 3 showed optimal sensitivity and is recommended for routine surveillance. Rapid feedback to HHAP stakeholders may inform timely mid-season interventions and reduce preventable deaths. Key messages • Nowcasting excess mortality enables faster response during heatwaves, potentially saving lives. • A sensitive 5% threshold improves early detection of moderate but meaningful mortality increases.