Assessing Risk to Heat Waves and Cold Spells using a Distributed Lag Non-Linear Model

In a changing climate, further understanding the physiological strain associated with heat waves and cold spells remains a topic of continued research. Acute, rapid responses to heat have been observed while the complexity surrounding cold-related mortality yields a more delayed response. The distributed lag non-linear model (DLNM) is commonly used in public health to assess the non-linear health responses following environmental conditions. Thresholds of daily mean apparent temperature are used to define anomalous temperature events (ATEs) of heat, extreme heat, cold, and extreme cold. Using daily mortality data, mortality responses following these heat and cold events are evaluated for 52 U.S. cities. The binary variables were incorporated into the DLNM to assess the impact of ATEs on daily mortality compared to days without anomalous temperatures. Geographic differences were observed with higher heat-related risk in northern locations. Similarly, elevated risk for cold-related mortality was found to be greater in more southern locations. This research supports the geographic variability found in other studies which suggest populations adapt or are better prepared for particular environmental conditions. Additionally, early season events showed higher risk values when compared to later occurring events. The impact of ATEs on human health outcomes is dependent on various factors including seasonal timing, duration, and intensity.