CIESIN Reproduced, with permission, from: Kalkstein, L. S. 1992. Impacts of global warming on human health: Heat stress-related mortality. In Global climate change: Implications, challenges and mitigation measures, ed. S. K. Majumdar, L. S. Kalkstein, B. Yarnal, E. W. Miller, and L. M. Rosenfeld, 371-83. Easton, PA: The Pennsylvania Academy of Science.

Chapter Twenty-Six



Center for Climatic Research
Department of Geography
University of Delaware
Newark, Delaware 19716


With the threat of a human-induced global warming, interest in climate/human health studies has dramatically increased, and at least three comprehensive reports summarizing most of this research have appeared in recent years.[1,2,3] The impact of weather on human well-being goes beyond mortality; even birth rates and sperm counts appear to be affected by climatological phenomena.[4,1] However, the majority of the climate/human health evaluations have concentrated on mortality, and most all of the studies correlate a number of climate variables with daily or weekly mortality statistics. These data are not difficult to obtain for the United States or Canada, and they are much easier to interpret than most sources of general morbidity data, such as hospital admissions tallies or visits to the doctor.

There is disagreement among researchers concerning the impact of weather on human mortality. For example, medical researchers have noted that mortality attributed to weather seems to vary considerably with age, sex, and race, but there is disagreement in defining the most susceptible population group. Other issues of contention exist, including the role of air conditioning in mitigating heat-related mortality and whether small, irregular aberrations in weather or very large fluctuations have a greater impact on mortality.

This essay will evaluate these questions and others. First, historical weather/mortality relationships will be discussed, as it is impossible to determine the impact of global warming on human health without understanding present-day and past relationships. Most of this work concentrates on the United States, but results from a pilot study in China will be briefly presented. Second, and most important, the potential impact of a large-scale global warming on human mortality will be discussed, with allowances for possible human acclimatization to the expected increasing warmth.


Although a few studies contend that mostly long term (i.e., monthly and annual) fluctuations in temperature affect mortality,[5] and only small aberrations can be explained by daily temperature variability,[6] most researchers insist that hot weather extremes have a more substantial impact on human mortality. Thus many "heat stress" indices have been developed to help assess the degree of impact.[7,8,2] Most research on heat/mortality relationships indicate that high temperature alone has a very dramatic effect on mortality, and other factors such as humidity and wind are essentially unimportant.[9,10] In fact, the notion of a "temperature threshold" is quite common in certain studies, and can be demonstrated when daily mortality in New York City and Shanghai, China are compared to maximum temperature (Figure 1a, b, c). Particularly interesting is the lack of a weather/mortality relationship at temperatures below the threshold, indicating that only the warmest 10-15 percent of all days in summer have an impact on human mortality.[11] For example, the five cities which demonstrate the greatest rise in mortality during very hot conditions are New York, Chicago, Philadelphia, Detroit, and St. Louis, respectively.[12] The threshold temperatures for these cities are: New York: 33deg.C; Chicago: 32deg.C; Philadelphia: 33deg.C; Detroit: 32deg.C; and St. Louis: 36deg.C. The magnitude of the threshold temperature appears to be related to its frequency across these cities; for example, a maximum temperature of 36deg.C in St. Louis occurs with approximately the same frequency as a maximum temperature of 32deg.C in Detroit. This strongly suggests that the notion of a "heat wave" is relative on an interregional scale, and is dependent upon the frequency of a given maximum temperature.

However, there is also evidence to refute this notion. Mortality rates in the southern and southwestern United States do not seem to be affected by weather in the summer, no matter how high the temperature. For example, the cities of Dallas, Atlanta, New Orleans, Oklahoma City, and even Phoenix show little change in mortality even during the hottest weather. A similar phenomenon is noted in China, where Guangzhou, with a summer climate similar to New Orleans, demonstrates much less day-to-day variation in mortality than Shanghai, which is located at a higher latitude.[13] In fact, threshold temperatures for these hotter cities are virtually impossible to define. One possible explanation may involve the variance in summer temperatures between the two regions. In the northern and midwestern cities, the very hot days or episodes are imbedded within periods of cooler weather. Thus, the physiological and behavioral "shock value" of a very high temperature episode is quite high. This point is further substantiated by the fact that most of the high mortality days occur during hot weather early in the summer season. Thus the first or second heat episodes of June or early July are much more critical than a comparative episode in August. In the southern cities, the hottest periods are less unique, as they do not vary as much from the mean. This seems to play a role in diminishing the impact of a very hot episode on human mortality.

Thus, a most interesting finding relating to heat-related mortality which potentially has a profound influence on such deaths in a warmer world is the large degree of inter-regional response. A recent EPA-sponsored study demonstrated that many cities in the northeastern and midwestern United States show a sharp rise in total mortality during unusually hot weather conditions, and in some cases, daily mortality can be more than double baseline levels when the weather is oppressive.[12]

The use of a new automated air mass-based synoptic procedure to evaluate weather/mortality relationships has supported and expanded the findings from the threshold temperature research (refer to Kalkstein et al., 1990[14] for a discussion of synoptic index development). The synoptic procedure is designed to classify days which are considered to be meteorologically homogeneous into air mass categories. Thus, the synergistic relationships that exist between numerous weather elements which comprise an air mass can be evaluated simultaneously, representing a significant improvement over an individual weather element approach.

The synoptic procedure was applied to ten U.S. cities in different climates to determine inter-regional mortality/weather sensitivities: Atlanta, Boston, Chicago, Dallas, Memphis, New York, Philadelphia, St. Louis, San Francisco, and Seattle. For many of these cities, a single offensive summer synoptic category was noted, which possessed a much higher mean mortality than the other categories (refer to Table 1; St. Louis example). Although category 6 in St. Louis was only slightly cooler than category 9, it appears that category 9 alone exceeded an oppressive threshold associated with very high mortality.

A comparison of results from the 10 cities is very instructive (Table 2). Most of the seven cities with a moderate or strong weather/mortality signal (as determined by the existence of an oppressive synoptic category associated with unusually high mortality) were located in the Northeast or Midwest (Memphis is a notable exception). The San Francisco area, which experiences infrequent but sometimes persistent hot weather, is also associated with this group. Of the three cities with no weather/mortality signal, two are located in the South, and Seattle possessed virtually no air mass types associated with very hot weather. Thus, this inter-regional disparity in response is supported within the synoptic evaluation.

Interestingly, a majority of studies have found that most of the excess deaths that occurred during periods of intense heat were not attributed to causes traditionally considered to be weather-related (such as heat stroke), but were attributed to a broad range of illnesses, accidents, and even adverse effects of medicinal agents.[15,16] In fact, heat stroke and heat exhaustion comprise a very small proportion of the increase in deaths which traditionally occurs during very hot episodes. An evaluation of individual causes of deaths which seem to rise during hot weather produces some surprises, with complications of pregnancy, ischemic heart disease, and various injuries ranking high on the list.[16]

Most research indicates that mortality rates during extreme heat vary with age, sex, and race. Oechsli and Buechley (1970)[17] found that mortality rates during heat waves increase with age; this is supported by more recent work.[16] The elderly seem to suffer from impaired physiological responses and often are unable to increase their cardiac output sufficiently during extremely hot weather. In addition, sweating efficiency decreases with advancing age, and many of the medications commonly taken by the elderly have been reported to increase sensitivity to the heat.[15] Certain researchers have determined slight rises as well in the mortality of infants during heat waves.[18,19]

Although there is general agreement concerning the impact of heat on the elderly, there are conflicting results involving mortality/gender and mortality/race relationships. Several studies have noted increased mortality rates among females during hot weather;[20] this may be attributed to differences in dress among the sexes.[21] However, at least two studies have found higher heat-related mortality rates among men.[18,22] A study on race/heat-related mortality have found that blacks are more susceptible in St. Louis and whites are more susceptible in New York.[23] However, two other studies have discovered that white mortality rates are higher than black's under almost all examined conditions.[18,19] Rather than race, socioeconomic status may have an influence on weather/mortality relationships, and large numbers of deaths during heat waves are found among poor inner-city residents who have little access to cooler environments.


A proper analysis which considers the impact of global warming on heat-related mortality should address all of the issues discussed above, including:

  1. an understanding of thresholds or tolerance limits of humans;

  2. the impact of possible acclimatization to warmer conditions;

  3. variations in response among the sexes and races;

  4. demographic changes expected to occur within the next century.

At present, we have some knowledge about the first of these issues and a lesser amount of information about the third. There is very little known about the second and fourth issues. Thus, any estimates of the impacts of global warming on human mortality must be treated with skepticism. However, there have been estimates offered for 15 U.S. cities, and they vary considerably based on scenarios using different assumptions about the issues listed above.

Assuming that people react to weather in a warmer world much as they do today (implying little or no acclimatization), it is quite likely that deaths from heat stress-related causes will increase dramatically (refer to Table 3a, b for the full range of estimates). Thus, EPA has reported to Congress that we might expect a sevenfold increase in heat-related deaths by the middle of the 21st century if acclimatization does not occur.[12,24] This would render heat-related deaths as a major killer, rivaling the present number of deaths from leukemia. The greatest brunt of this increase would be borne in the northeastern and midwestern U.S., and it is estimated that the number of heat-related deaths in New York City would rise from the present 320 during a typical summer today to 1,743 during a typical summer in the mid 21st century.[21] Under such circumstances, the number of days exceeding the threshold temperature would show at least a two-fold increase, contributing to this dramatic rise.

Of course, a more likely scenario is that some degree of acclimatization will occur, as the gradual increase in warmth over the next 60 years would permit time for certain social and physiological adjustments. A satisfactory means to account for this acclimatization has alluded the scientific community, although an attempt was made in the EPA Report to Congress on the potential effects of global climate change.[24] Analog cities, which possess present day weather most duplicative of predicted mid-21st century weather for each evaluated city, were established to account for full acclimatization. For example, the use of a climate change scenario developed from a global circulation model to estimate weather in Atlanta in the mid-21st century will produce a regime which approximates the present weather in New Orleans.[25] Since New Orleans residents are fully acclimatized to this regime, the weather/mortality relationships that exist today for New Orleans can be utilized for the mid-21st century estimated climate in Atlanta to account for full acclimatization.

Employing this acclimatization procedure, new estimates were developed for the number of heat stress-related deaths in the mid-21st century, yielding very different results from the unacclimatized estimates. Rather than the sevenfold increase developed for the unacclimatized model, less than a two-fold increase results when accounting for acclimatization. In fact, many cities actually show a decrease in heat-related mortality if the population fully acclimatizes to a climate change. For example, the number of heat-related deaths which occur in an average summer today in Los Angeles is 84. If residents in Los Angeles fully acclimatize to the increased warmth expected in the mid-21st century, the number of heat-related deaths is estimated to drop to virtually zero.[24] Astoundingly, if residents of Los Angeles don't acclimatize at all, the number of heat-related deaths estimated for an average summer in the mid-21st century rises to 1570! Similar disparities between acclimatized and unacclimatized estimates are noted for many northern and midwestern cities such as Detroit, New York, and St. Louis.

This wide variation has much to do with the technique used in this evaluation to account for acclimatization. Since the change analogs for the northern cities are other cities located further south, and considering the southern populations respond much less dramatically to heat, it is not surprising that the fully acclimatized mortality estimates are so low. This points to some major shortcomings within our acclimatization procedure. First, the only similarities drawn between the target and analog cities relate to climate alone, with no consideration of possible urban structural or architectural disparities. For example, much of the low income urban population in the South, who are especially vulnerable to heat stress due to a lack of air conditioning or other amenities, reside in small frame houses often referred to as "shotgun shacks".[26] Although far from luxurious, these residences are well-adapted to the rigors of the southern summer, as they are often light colored with metal (reflective) roofs and well-ventilated with windows or doors on four sides. In contrast, most northern urban poor live in tenement dwellings constructed of dark-colored brick, black tar roofs, and windows often on two sides associated with a row house motif. During very hot conditions, these northern dwellings suffer from poor ventilation and absorb considerably more radiation than their southern counterparts. These types of factors might partially explain the differential mortality response between northern and southern populations.

Assuming that people can physiologically adapt to the predicted increasing warmth, we can expect only a partial acclimatization to occur as it is highly unlikely that the architectural makeup of the urban area will change significantly over the next 50 to 75 years to account for warmer conditions. This is particularly true for dwellings of the poor, and it is unlikely that the tenements of northern cities will be leveled and replaced by other forms of housing, such as shotgun shacks, which are more amenable to hotter conditions.

An added shortcoming of the acclimatization procedure is underscored by the fact that the analog and target cities are also very different in their demographic composition, as the differential proportion of elderly (who are particularly vulnerable to heat-related mortality) and minorities has not been accounted for. Furthermore, no attempt has been made to determine what these proportions might be in the target city in the mid-21st century. The expected increase in the percentage of elderly people (greater than 65 years old) over the next 75 years might partially counter the dampening in heat-related mortality attributed to physiological acclimatization.

Thus, the Report to Congress also includes an estimate of mid-21st century mortality assuming partial acclimatization.[24] The assumption here is that the apparent full acclimatization which presently exists in the South will not occur in a warmer North for two reasons: (1) any architectural changes in dwellings will lag considerably behind the warming itself, rendering many residences unsuitable for the increasing warmth, and (2) the demographic composition of these cities will be comprised of a larger proportion of people more vulnerable to heat-related stresses.

Finally, it is possible that the expected 1.5deg.C to 4.0deg.C warming by the mid-21 century might actually increase temperatures in many southern and southwestern cities to levels which might be beyond the tolerances of our society. For example, the average annual number of days in Atlanta which exceed 32deg.C today is 17. This number is expected to reach 53 by the mid-21st century using the Goddard Institute for Space Studies (GISS) doubled CO2 scenario.[25] Even more troubling is the number of days with temperatures exceeding 38deg.C. Presently, cities such as Memphis, Jackson, and Birmingham experience three such days during an average summer. This will increase to about 20 days with the expected temperature increase.[27] Considering similar increases in the number of very hot days in Phoenix and other southwestern cities, it is possible that certain areas might simply become uninhabitable in the warmer world of the mid-21st century. Of course, this is highly speculative, as no examples presently exist to support this notion.

To improve upon present attempts to account for acclimatization, the Climate Change Division of EPA has developed a global warming/human health program which places some emphasis on an improved understanding of human acclimatization in a warmer world. The program will concentrate on the social and cultural adjustments expected to occur, including possible demographic alterations that could take place because of migration and other factors attributed to global warming.[28] Contacts have been made with the Institute of Medicine at the National Academy of Sciences to help in the formulation and coordination of such a study.

Of course, skeptics contend that global warming will probably not occur, that the earth and atmospheric system have feedback mechanisms to prevent a human-induced climatic change from occurring, and that time and money may be wasted in developing mitigating policies for a possibility that may never happen. There is no claim in this essay that global warming will definitely occur, but the risks are high enough to support a program that attempts to estimate impacts and tries to develop mitigating action. Thus, improved evaluations of the possible impacts of climate change on human mortality are needed to guide policy decisions and to foster the international cooperation that is vital to deal with the possible threats effectively.


1. White, M.R. and I. Hertz-Picciotto. 1985. Human Health: Analysis of climate related to health. Characterization of Information Requirements for Studies of CO2 effects: Water Resources, Agriculture, Fisheries, Forests and Human Health. Washington: Department of Energy.

2. Kalkstein, L.S. and K.M. Valimont. 1987. Climate Effects on Human Health, pp. 122-152. In: Potential Effects of Future Climate Changes on Forests and Vegetation, Agriculture, Water Resources, and Human Health, EPA Science and Advisory Committee Monograph #25389.

3. World Health Organization. 1990. Potential health effects of climatic change. Geneva.

4. Calot, G. and C. Blayo. 1982. Recent course of fertility in Western Europe. Population Studies 36:345-372.

5. Sakamoto, M.M. and K. Katayama. 1971. Statistical analysis of seasonal variation in mortality. Journal of the Meteorological Society of Japan 49:494-509.

6. Persinger, M.A. 1980. The Weather Matrix and Human Behavior. New York: Praeger.

7. Quayle, R., and F. Doehring. 1981. Heat stress: A comparison of indices. Weatherwise 34:120-124.

8. Steadman, R.G. 1984. A universal scale of apparent temperature. Journal of Climate and Applied Meteorology 23:1674-1687.

9. Driscoll, D.M. 1971. The relationship between weather and mortality in ten major metropolitan areas in the United States, 1962-1965. Journal of the International Society of Biometeorology 15:23-40.

10. Ellis, F.P., F. Nelson, and L. Pincus. 1974. Mortality during heat wave in New York City, July 1972 and August and Septemer 1973. Environmental Research 10:1-13.

11. Kalkstein, L.S. 1991. Bioclimatological Research and the Issue of Climatic Sensitivity. Physical Geography. 3:274-286.

12. Kalkstein, L.S. 1989. The impact of CO2 and trace gas-induced climate changes upon human mortality, pp. 1-12-1-35. In: J.B. Smith and D.A. Tirpak (Ed.) The potential effects of global climate change on the United States: Appendix G-Health. Washington, DC: US Environmental Protection Agency.

13. Tan, G. 1991. Weather-related human mortality in Shanghai and Guangzhou, China. Proceedings on the Tenth Conference on Biometeorology and Aerobiology. American Meteorological Society, In Press.

14. Kalkstein, L.S., P.C. Dunne and R.S. Vose. 1990. Detection of climatic change in the western North American arctic using a synoptic climatological approach. Journal of Climate 3:1153-1167.

15. Jones, T.S., A.P. Liang, E.M. Kilbourne, M.R. Griffin, P.S. Patriarca, G.G. Wassilok, R.J. Mullan, R.F. Herricek, H.D. Donnel, Jr., K. Choi and S.B. Thacker. 1982. Morbidity and mortality associated with the July, 1980 heat wave in St. Louis and Kansas City. Journal of the American Medical Association 247:3327-3330.

16. Kalkstein, L.S. and R.E. Davis, 1989. Weather and human mortality: An evaluation of demographic and inter-regional responses in the United States. Annals of the Association of American Geographers 79:44-64.

17. Oechsli, F.W. and R.W. Buechley. 1970. Excess mortality associated with three Los Angeles September hot spells. Environmental Research 3:277-284.

18. Bridger, C.A., F.P. Ellis and H.L. Taylor. 1976. Mortality in St. Louis, Missouri, during heat waves in 1936, 1953, 1954,1955 and 1966. Environmental Research 12:38-48.

19. Kalkstein, L.S. 1988. The Impacts of Predicted Climate Change on Human Mortality. In: J.R. Mather (Ed.) Publications in Climatology pp. 110.

20. Applegate, W.B., J.W. Runyan, Jr., L. Brasfield, M.L. Williams, C. Konigsverg and C. Fouche. 1981. Analysis of the 1980 heat wave in Memphis. Journal of the American Geriatrics Society 19:337-342.

21. Rotton, J. 1983. Angry, sad, happy? Blame the weather. U.S. News and World Report 95:52-53.

22. Ellis, F.P. 1972. Mortality from heat illness and heat-aggravated illness in the United States. Environmental Research 15:504-512.

23. Schuman, S.H. 1972. Patterns of urban heat wave deaths and implications for prevention: Data from New York and St. Louis during July, 1966. Environmental Research 5:58-75.

24. Smith, J. B. and D.A. Tirpak, eds. 1989. The potential effects of global climate change on the United States. Washington, D.C.: U.S. EPA Office of Policy, Planning and Evaluation.

25. Hansen, J.,I. Fung, A. Lacia, S. Lebedeff, D. Rind, R. Ruedy and G. Russell. 1988. Global climate changes as forecast by the GISS 3-D model. Journal of Geophysical Research 93:9341-9364.

26. Kniffen, F.B. 1965. Folk housing, key to diffusion. Annals of the Association of American Geographers 55:649-577.

27. Titus, J.G. 1989. Regional studies: southeast, pp. 3-23-3-58. In: J.B. Smith and D.A. Tirpak (Ed.) The potential effects of global climate change on the United States. Washington, D.C.: U.S. Environmental Protection Agency.

28. Kalkstein, L.S. 1990. A proposed global warming/health initiative. Environmental Impact Assessment Review. 10:383-392.