Link between deaths and climate weakening over time

Published February 1, 2000

Proponents of catastrophic global climate change theory predict increased deaths as temperatures warm—by now a familiar litany:

On a warmer planet, intense heat waves alone are by 2050 likely to result in increases in death by cardiac arrest and respiratory ills of several thousand a year—especially in urban areas and among the elderly and very young.
—The Wall Street Journal, October 19, 1999

[Based upon data from several North American cities,] the annual number of heat-related deaths would approximately double by 2020 and would increase several-fold by 2050.”
—U.N. Intergovernmental Panel on Climate Change, 1995

Yet U.S. death rates have been falling and life expectancy rising for some time now. Will a global warming “apocalypse” reverse this trend? History says no, as four colleagues and I recently discovered.

Indeed, such doom-and-gloom pronouncements are usually based on extrapolating historic data given a future climate. For example, if an extra 50 New Yorkers die per heat wave of a given strength and duration (based on past data), and if climate models “forecast” three times as many heat waves by 2050, then you would predict an extra 150 deaths.

We carefully examined existing trends using an established approach I developed with climate/mortality expert Larry Kalkstein more than a decade ago. A plot of daily deaths (standardized for demographic changes) in New York City and environs vs. afternoon apparent temperature (an index that combines air temperature and humidity) results in a flattened U-shape that reveals that deaths are a tad higher than average when apparent temperatures are very low, and much higher during the most uncomfortable days of summer.

Although the overall trend is negative, few climate/mortality experts claim weather is a primary factor in mortality across the entire apparent temperature spectrum. Weather is most important when it’s extreme. In general, summer deaths are more closely linked with weather than winter deaths in the United States.

But how has the weather/death response been changing over time? In other words, are death rates more or less linked to climate now than they were in the past? To examine this question, we subdivided the data by decades and averaged them over apparent temperature classes.

Death rates clearly have been declining with each new decade; that is, more deaths occurred in the 1960s than in the 1990s—no matter what the temperature. But more important, notice at the right-hand side of the graph that when apparent temperatures exceed 30°C, the rapid mortality increase so prevalent in the 1960s has decreased over each decade, to such an extent that the 1990s’ curve is nearly flat. In other words, on any given day in the 1960s, every degree above 30°C would mean an extra 1.2 deaths. In the 1990s, the death increase per degree is only about 0.1.

But is this relationship somehow unique to the Big Apple? Thus far, our research team has examined five other major metropolitan areas of the United States: Detroit, Chicago, Houston, Dallas, and Atlanta. The cities fall into two groups: northern-tier cities, where the summer mortality response to weather has diminished over time (although the response is a little bit weaker in Chicago and Detroit than in New York City); and southern-tier cities, where the population and infrastructure have adapted to uncomfortable summer weather, and where there has never been a mortality response (at least from the 1970s onward).

Which brings us back to the original global warming pronouncements. The weather/mortality models used to make these scary forecasts obviously have not taken existing trends into account. For had the 1980s and 1990s data alone been used, no more than a handful of extra deaths would become evident no matter how warm it gets. The reasons for these trends are both obvious and numerous—improvements in medicine, for example, and human and infrastructure adaptations to extreme climates, including more access to air-conditioning, better weather forecasts and building design, and so on. In short, technological improvements have reduced deaths.

Air-conditioning saves lives. Of course, air-conditioning requires electricity, which when created releases CO2 into the atmosphere. In which of the following scenarios would the population be better off? A) If we increase global access to air-conditioning? or B) If we raise the cost of electricity to stave off global warming (for example, via the Kyoto Protocol)?

This study addresses only the response side of the ledger—death rates. But what about the climate change aspect? Without question, more people in U.S. cities have been exposing themselves to higher temperatures over the past 30 years—primarily because of the urban heat island effect, in which concrete-and-pavement cities warm relative to the surrounding rural areas. This effect far exceeds any climate change, whether related to greenhouse gases or not. Yet despite placing themselves in the “peril” of these higher temperatures, death rates have dropped dramatically. In short, there is no basis for these scare stories about high heat-related death rates from greenhouse gases.

Robert E. Davis is an associate professor of environmental science at the University of Virginia.


Davis, R.E. et al., 1999, Decadal changes in weather/human mortality relationships in U.S. cities, Proceedings of the 15th International Congress of Biometeorology and International Conference on Urban Climatology, Sydney, Australia, Nov. 10.