Data and Methodology


Data

Download data for 247 countries, territories, and dependencies and 940 global cities.

Methodology

Climate Central analyzed daily temperatures during the past five years (2020 to 2024) in 247 countries, territories, and dependencies (referred to as “countries” for simplicity) and 940 global cities, using two primary mechanisms:
  1. Pregnancy heat-risk days: We counted the number of days with temperatures warmer than 95% of temperatures observed at a given location (also referred to as temperatures above the 95th percentile). Research1 shows that this threshold can increase the risk of preterm birth. We define these extremely hot days as “pregnancy heat-risk days.”

  2. The Climate Shift Index (CSI) system: We calculated the number of pregnancy heat-risk days that would have occurred in a world without human-caused climate change (i.e., a counterfactual scenario) and compared that to the total number observed each year. This allowed us to count how many pregnancy heat-risk days were added by climate change annually.

The 940 global cities that we analyzed, drawn from simplemaps, include cities with populations exceeding one million people as well as various U.S. cities.

A full, detailed methodology is available in the report.

Note: Values shown in maps and tables across this website have been rounded for clarity. For exact data values, please download the full dataset.

While this report highlights days above the 95th temperature percentile over the past five years, the full analysis also includes data for the 90th and 99th temperature percentiles. Research1 indicates there is no single threshold for elevated preterm birth risk — it varies based on socio-economic factors such as national wealth, healthcare access, and race (see full methodology for more details). We also include data for 2024 — the hottest year on record — on its own. Download the full dataset to examine data that could be relevant for local populations with higher or lower vulnerability based on socio-economic factors.

1We chose this percentile based on peer-reviewed research from Kuehn et al. (2017), Wang et al. (2013), Wang et al. (2024), and McElroy et al. (2022).