Twenty years, 28,000 adults
Researchers drew on the Health and Retirement Study, following 28,512 community-dwelling adults aged 50 to 89 who had no dementia at the start. Average age was 59.8, follow-up averaged 10.5 years, and the data ran from 2000 to 2020. During that period 3,773 participants developed dementia, or 13.2%, and 7,887 died without a dementia classification.
Heat exposure was measured with the daily heat index, which combines temperature and humidity, estimated for each person's census tract from a 4 km climate dataset. Days were sorted into two National Weather Service categories: Extreme Caution, at 90F to 102F, and Danger and above, at 103F or higher. Participants averaged 5.8 Danger-plus days a year. Exposure was counted over the preceding five years and recalculated at each survey wave.
Risk rises at the top of the heat scale
Milder heat days showed no statistically detectable link to dementia at any exposure level. The highest category did. Compared with people who had 0.2 or fewer Danger-plus days a year, those with 0.4 to 1.8 days had a 14% higher hazard of dementia, and those with 7.2 or more days had a 23% higher hazard. The middle band showed no rise, so the dose-response curve is not a clean straight line.
Neither heat category was linked to mortality in the main analysis. The authors argue that a five-year cumulative measure captures something different from the short heat spikes that cause heat deaths, and may reflect slower damage rather than acute crisis.
Where you live matters
The association was much stronger in disadvantaged neighbourhoods. Among people with the most Danger-plus days, the predicted dementia hazard was 40% higher in the most socially vulnerable areas, against 11% higher in less vulnerable ones.
Cooling access mattered too. In neighbourhoods with lower estimated central air conditioning, the predicted hazard was 36% higher for the most exposed group, compared with 13% where cooling was more common. The authors, from the University of Southern California, say prevention planning should treat local heat exposure as a modifiable risk factor and target communities with less capacity to stay cool.