Updated trend analysis of heat vulnerability in the United States using a Distributed Lag Nonlinear Model

The impacts of heat upon human health have been studied by myriad researchers. Several current themes have been identified: heat events in some areas are becoming more common; the impacts of heat on all-cause mortality seem to have declined over time; and there is a growing debate as to the impacts of mortality displacement (harvesting) on overall mortality totals during heat events. To build upon these themes, we use a recently extended US national database on mortality to assess heat vulnerability across the United States.

For 40 US metropolitan areas, we divide up the full mortality record into 4 sub-periods: 1975-1983, 1984-1992, 1993-2001, and 2002-2010. Using a distributed-lag non-linear model (DLNM), we assess spatio-temporal variability to heat using apparent temperature thresholds for heat-events.

Spatial and temporal variability is observed in the results, with an overall greater vulnerability in the northern and Midwestern cities than in southern cities and most western cities. Trends broadly show a decreased vulnerability over time, although statistically significant increases in mortality still occur with excessive heat events over many locations.