Variability in temperature-related mortality projections under climate change

Area of Science:

• Environmental health
• Climate change impacts
• Epidemiology

Background:

• Existing studies on future temperature increases and mortality often use limited simulations, neglecting variability and uncertainty.
• Assessing the full impact of climate change on public health requires robust projections that account for diverse climate models and emission scenarios.

Purpose of the Study:

• To evaluate the variability in temperature projections and their impact on future mortality estimates.
• To utilize a comprehensive panel of climate simulations to understand uncertainty in mortality projections.

Main Methods:

• Employed Poisson regression models with historical Montreal data (1990-2007) to determine relative risks (RR) for daily mortality linked to temperature metrics (mean, min, max) during summer.
• Utilized 32 climate simulations (global and regional models) for June-August 2020-2037 to project attributable deaths, incorporating varied RR estimates and bootstrap samples for uncertainty analysis.
• Applied analysis of covariance to assess the influence of simulation type, projected year, and RR sets on mortality estimates.

Main Results:

• Simulations showed minimal variability (<1%) in projected June-August temperatures for 2020-2037.
• Estimated attributable deaths per summer ranged from 34 to 174.
• A significant portion (38%) of the variability in mortality projections stemmed from the chosen temperature-mortality relative risk (RR) estimates.

Conclusions:

• The selection of relative risk (RR) estimates is a critical factor in reducing uncertainty for future mortality projections.
• Accurate estimation of temperature-related mortality requires careful consideration of the RR used in the models.