**4. Interpretion of Temporal Trends in Temperature Effects**

What factors are responsible for the declining ERF for heat-related mortality? While attribution to specific factors remains elusive, studies have speculated that increasing wealth, enhanced heat-health awareness and prevention measures, general improvements in population health, and housing improvements all could play a role [22,23,32,33,42,43]. One leading explanation, at least in the U.S., is the increasing prevalence of A/C usage in recent decades. There is some empirical support for a role of A/C as a modifier of effects, but a great deal of evidence is cross sectional. For example, A/C prevalence can explain some of the city-to-city differences in effect estimates [7,8]. Longitudinally, the evidence remains incomplete, with some studies reporting a strong role for A/C [25] and others not [24]. It seems likely that the power to test for effect modification over time by A/C has been limited by the coarse temporal scale of A/C survey data (e.g., decadal), and also by the problem of not having A/C usage data that are specific to the vulnerable population subset (the ill, elderly and poor). While

potentially effective as an adaptive measure, A/C has several important limitations, including capital and energy cost, carbon- and pollution-generating energy demand, and potential for failure during power outages [19].

A related question is the extent, to which heat adaptation trends are being driven by climate warming itself. After analyzing this question, Christidis and colleagues [43] suggested that trends in heat- and cold-related mortality have more to do with economic and cultural trends than with adjustments to the changing climate. In addition, there has been considerable attention paid to reducing heat-health risks by a range of public actors in the U.S. and Europe in the past two decades. This view is supported by the fact that heat effects have declined quite rapidly over the past several decades during a time when climate has warmed only slightly. This is not to say that climate warming will not affect future adaptation trends, but rather that most of the trends observed to date are likely driven by non-climate factors such as those noted above.

Hondula and colleagues provided a thoughtful review of the role of climate in spatial and temporal trends in adaptation [15]. They reviewed ways in which adaptation has been modeled to date in climate and health projection studies. An important caveat to keep in mind is that the ERF, while declining, is only one component of future risk. Rising temperatures and ageing populations could lead to increasing risks in the future [39,44].
