The Current Wisdom: Temperature Variability And Human Mortality

April 24, 2012 • Commentary
This article appeared in Cato​.org on April 24, 2012.

The Current Wisdom is a series of monthly articles in which Senior Fellow Patrick J. Michaels reviews interesting items on global warming in the scientific literature that may not have received the media attention that they deserved, or have been misinterpreted in the popular press.

The Current Wisdom only comments on science appearing in the refereed, peer‐​reviewed literature, or that has been peer‐​screened prior to presentation at a scientific congress.

Prior to April, 2011, issues of this Wisdom, which began in 2010, are available at our blog Cato@Liberty (www​.cato​-at​-lib​er​ty​.org/).

PNAS has struck again. That’s the Proceedings of the National Academy of Sciences, which, according to bibliographic metrics, is the second‐​most influential science journal published in the U.S. Science, from the American Association for the Advancement of Science, is in first place.

As I have discussed previously (http://​www​.forbes​.com/​s​i​t​e​s​/​p​a​t​r​i​c​k​m​i​c​h​a​e​l​s​/​2​0​1​1​/​0​6​/​1​6​/​p​e​e​r​-​r​e​v​i​e​w​-and-…) the procedures for publishing in PNAS are not those typically associated with rigorous peer‐​review, and much more resemble “pal review”. As a result, sometimes some rather strange papers appear.

A textbook example is this paper by Antonella Zanobetti, Marie O’Neill, Carina Gronlund, and Joel Schwartz.

Zanobetti, from Harvard, and her colleagues attempt to answer the question as to whether or not summer temperature variability—not temperature itself—is associated with an increase in mortality in elderly people with chronic disease.

While the accepted wisdom is that heat waves can lead to elevated mortality (for those unprepared), what is not so widely known, but nonetheless obvious to those who think in economic terms, is that the more frequent high temperature extremes are, the lower the overall rate of heat‐​related mortality becomes. Pure and simple, people adapt.

(The EPA, in its zeal to regulate carbon dioxide, has it completely wrong on this association, claiming that increased heat wave frequency and magnitude constitute endangerment to the public health and welfare, when, in actuality, the data clearly show that more heat leads to a lower heat‐​related morality rate.)

This is obvious from Figure 1, which is taken from some heat‐​related mortality research that I was involved in (Davis et al., 2003). The heat‐​related mortality rate in each of 28 cities across the country for three different decades (1970s, 1980s, 1990s) is represented in Figure 1 by a set of three bars plotted at each city’s location on the map. The higher the bars, the greater the heat‐​related mortality.

Note that the cities along the southern tier of the U.S.—the ones with the hottest summers—have much lower rates of heat‐​related mortality than do the cooler cities in the Midwestern and Northeastern parts of the country. Where people are used to very high temperatures, they are adept at dealing with them. In the Midwest and the Northeast, where heat waves are rare, more people die.

Note also that the height of the bars in virtually every city is declining with time. In other words, more people died from heat waves in the 1970s than did so in the 1990s—in spite of generally rising summer time temperatures during that period (the mortality trend is indicated by the “+” or “-“ sign below each set of bars). In fact, in many cities across the southern and central portion of the country, there is no bar at all during the 1990s, indicating that heat‐​related mortality was statistically undetectable during that decade. Taken as a whole, the heat‐​related mortality in major U.S. cities declined by 75% across the last decades of the 20th century, despite rising summer temperatures. Clearly, we are adapting well to increased frequency and intensity of heat waves.

The one exception, which proves the adaptational rule, is Seattle—which has the coolest average summers and shows an increase in mortality.

Figure 1. Annual heat‐​related mortality rates (excess deaths per standard million population on days in which the decadal‐​varying threshold apparent temperature (AT) is equaled or exceeded) by city and decade, and long‐​term trend in summer afternoon AT. Each histogram bar indicates a different decade: from left to right, 1960s‐​1970s, 1980–1989, and 1990–1998. Decades without histogram bars exhibit no threshold ATs and no heat‐​related mortality. Decades with gray bars have mortality rates that are statistically significantly different from the decades indicated by black bars. The average excess deaths across all 28 cities is shown at the lower left. AT trends are indicated beneath each city abbreviation (Davis et al., 2003).

But there were some interesting questions raised concerning the interpretation of our results—in particular, about the role of temperature variability in shaping our findings. We noted (Davis et al., 2003):

“One current hypothesis is that individuals are stressed during the summer by significant temperature changes, particularly minimum temperatures. High minimum temperature variability has been linked to higher mortality rates in northeastern and northern interior cities.… This observation could partially account for the spatial pattern of decadal mortality declines across the United States, because mortality rates in the 1990s remain elevated in the Northeast and West Coast, where summer temperature variability typically is higher because of air mass changes associated with more frequent frontal passages.

So the idea that temperature variability may give rise to some of the patterns of heat‐​related mortality in the U.S. (also noted in the new Zanobetti paper) is not a new one.

In our work, we did a cursory examination as to whether changes in summer temperature variability over our study period may have influenced our results. But we found that in only three of the 28 cities we examined, was there a statistically significant trend in temperature variability (an increasing trend in two cities and a decreasing trend in another) over our study period (the early 1960s to the late 1990s). Thus we concluded that changes in temperature variability were largely absent in our data and therefore “played little role in the observed mortality declines.”

Zanobetti and colleagues report otherwise.

They used a fairly complicated methodology to determine that there is an association between high intraseason temperature variability and the survivorship of a cohort of elderly, previously ill persons. Greater day‐​to‐​day temperature variability during the summer led to more mortality in their subset of the elderly population.

However, Zanobetti et al. found that the association was only statistically significant in the portions of the country with the lowest average summer season day‐​to‐​day temperature variability (typically the southern portions of the country—their zones 4 and 5 in Figure 2).

Figure 2. Climate zones (in colors) used by Zanobetti et al. and the level of day‐​to‐​day variability in summer temperatures in each city studied (the degree of shading in each circle). Open circles indicate lower degrees of temperature variability than darkly shaded circles (from Zanobetti et al., 2012).

Table 1 is a breakdown of their results for the different climate zones when the temperature variability for a particular summer is 1°C above normal. “HR is the “hazard ratio”, including the 95% confidence intervals (CI) for the HR value. If the 95% CI includes the value of 1.0, then there is no statistically significant difference from the average mortality. The climate zones are numbered 1 to 5 with climate zone 1 typically characterized by more summer day‐​to‐​day temperature variability than climate zone 5 (see figure 2). The different elderly population cohorts with pre‐​existing medical conditions are listed as CHF (congestive heart failure), MI (myocardial infarction), diabetes, and COPD (chronic obstructive pulmonary disease).

The take‐​home message from these results is that only in climate zones 4 and 5 (the zones with the least amount of average daily summertime temperature variability) is there a statistically significant reduction in survivorship when the variability is 1°C above average. This suggests that people in cities that typically experience a high degree of day‐​to‐​day temperature variability during the summer are better adapted to extremes in temperature variability than those people living in cities where day‐​to‐​day temperature variability is low. Just as is the case with heat waves, more exposure leads to less susceptibility.

Pretty straightforward. Here is how Zanobetti et al. sum things up:

“Our data suggest that long‐​term increases in temperature variability may increase the risk of mortality in different subgroups of susceptible older populations.”


“Taken together, our present findings and previous evidence suggest that summer temperature variability could plausibly impair the health and shorten the life expectancy of older adults, particularly those with chronic medical conditions… A 1 °C increase in temperature [variability] is a plausible increase in some regions. Based on our findings, this increase in temperature [variability] would increase all‐​cause mortality in our MI cohort by 5%, for example. Based on an average of 270,000 deaths per year across all four cohorts, a 5% increase in mortality would correspond to ~14,000 additional deaths per year due to an increase in temperature variability in the United States.”

Come again?

“In addition to an increase in temperature, climate models predict an increase in summer temperature variability in mid‐​latitude Northern Hemisphere land areas, which may be a major public health concern.”

That is just the opposite to what a more enlightened view of their results suggest.

Zanobetti et al.’s data pretty clearly show that cities with a climate characterized by high summer temperature variability have lower rates of mortality in the cohorts they studied. This should lead directly to the conclusion that if climate change were to result in increasing summer temperature variability, then there should be a reduction in the mortality rate as the population becomes better adapted to higher temperature variability.

Instead, the authors conclude the opposite, the paper appears in the pal‐​reviewed PNAS, and headlines are made—global warming to kill more people.

Such is the state of climate “science” these days.


Davis R.E., P.C. Knappenberger, P.J. Michaels PJ, and W.M. Novicof, (2003), Changing heat‐​related mortality in the United States. Environmental Health Perspectives, 111, 1712–1718

Zanobetti A, M.S. O’Neill, C.J. Gronlund, and J.D. Schwartz (2012), Summer temperature variability and long‐​term survival among elderly people with chronic disease, Proceedings of the National Academy of Sciences, doi: 10.1073/pnas.1113070109

About the Author