气候变化与人类健康Impact of regional climate change on human health

© 2005 Nature Publishing Group Impact of regional climate change on human health Jonathan A. Patz1,2, Diarmid Campbell-Lendrum3, Tracey Holloway1 & Jonathan A. Foley1 The World Health Organisation estimates that the warming and precipitation trends due to anthropogenic climate change of the past 30 years already claim over 150,000 lives annually. Many prevalent human diseases are linked to climate fluctuations, from cardiovascular mortality and respiratory illnesses due to heatwaves, to altered transmission of infectious diseases and malnutrition from crop failures. Uncertainty remains in attributing the expansion or resurgence of diseases to climate change, owing to lack of long-term, high-quality data sets as well as the large influence of socio-economic factors and changes in immunity and drug resistance. Here we review the growing evidence that climate–health relationships pose increasing health risks under future projections of climate change and that the warming trend over recent decades has already contributed to increased morbidity and mortality in many regions of the world. Potentially vulnerable regions include the temperate latitudes, which are projected to warm disproportionately, the regions around the Pacific and Indian oceans that are currently subjected to large rainfall variability due to the El Nin˜o/Southern Oscillation sub-Saharan Africa and sprawling cities where the urban heat island effect could intensify extreme climatic events. G lobal average temperatures are projected to increase between 1.4 and 5.8 8C by the end of this century1; an associated rise in sea level is also expected. The number of people at risk from flooding by coastal storm surges is projected to increase from the current 75 million to 200 million in a scenario of mid-range climate changes, in which a rise in the sea level of 40 cm is envisaged by the 2080s (ref. 2). Extremes of the hydrologic cycle (such as floods and droughts) are projected to increase with warmer ambient temperatures. Evidence is mounting that such changes in the broad-scale climate system may already be affecting human health, including mortality and morbidity from extreme heat, cold, drought or storms; changes in air and water quality; and changes in the ecology of infectious diseases3–5. We reviewed both empirical studies of past observations of climate-health relationships, and model simulation studies of pro- jected health risks and regional vulnerability associated with future climate change. Here we focus on the health implications of climate variability, past and present climate change impacts on human health, future projections and uncertainties. This review primarily examines relatively direct-acting temperature effects, while recogniz- ing that other major risk pathways exist, for instance, altered storm patterns, hydrologic extremes, and sea-level rise. Health implications of climate variability Non-infectious health effects. The summer of 2003 was probably Europe’s hottest summer in over 500 years, with average tempera- tures 3.5 8C above normal6–8. With approximately 22,000 to 45,000 heat-related deaths occurring across Europe over two weeks in August 2003 (refs 9 and 10), this is the most striking recent example of health risks directly resulting from temperature change. Judging from this extreme event, changes in climate variability associated with long-term climate change could be at least as important for future risk assessment as upward trends in mean temperature. The European heatwave in 2003 was well outside the range of expected climate variability8. In addition, comparisons of climate model outputs with andwithout anthropogenic drivers show that the risk of a heatwave of that magnitude had more than doubled by 2003 as a result of human-induced climate change3. The demonstration of a causal link between global warming and the occurrence of regional heatwaves indicates a potential for more frequent and/or more severe heatwaves in a future warmer world. On local and regional scales, changes in land cover can sometimes exacerbate the effect of greenhouse-gas-induced warming, or even exert the largest impact on climatic conditions. For example, urban ‘heat islands’ result from lowered evaporative cooling, increased heat storage and sensible heat flux caused by the lowered vegetation cover, increased impervious cover and complex surfaces of the cityscape. Dark surfaces such as asphalt roads or rooftops can reach tempera- tures 30–40 8Chigher than surrounding air11. Most cities show a large heat island effect, registering 5–11 8Cwarmer than surrounding rural areas12. But the effects of land cover change on climate are not limited to small areas: at the scale of the entire continental USA, Kalnay and Cai13 estimated that land-cover changes (from both agriculture and urban areas) caused a surface warming of,0.27 8C per century. Also, in southeast China, a warming of,0.05 8C per decade since 1978 has been attributed to land-use change from urban sprawl14. Exposure to both extreme hot and cold weather is associated with increased morbidity and mortality, compared to an intermediate ‘comfortable’ temperature range15. Heat mortality follows a J-shaped function with a steeper slope at higher temperatures16. The comfor- table or safest temperature range is closely related to mean tempera- ture, with an upper bound from as low as 16.5 8C for the Netherlands and 19 8C for London17, to as high as 29 8C in Taiwan18. Hot days occurring earlier in the summer season have a larger effect than those occurring later17. It should be noted that although the majority of temperature–mortality studies have taken place in developed REVIEWS 1Center for Sustainability and the Global Environment (SAGE), Nelson Institute for Environmental Studies, and 2the Department of Population Health Sciences, University of Wisconsin, 1710 University Avenue, Madison, Wisconsin 53726, USA. 3Department of Protection of the Human Environment, World Health Organization, Geneva, Avenue Appia, Geneva CH-1211, Switzerland. Vol 438|17 November 2005|doi:10.1038/nature04188 310 © 2005 Nature Publishing Group countries and in regions with temperate climates, the same pattern of temperature–mortality relationship found in European and North American cities occurs in Sa˜o Paulo, Brazil, a developing city with subtropical conditions19. In summary, although most studies to date show clear vulner- ability to heat in cooler temperate regions, tropical regions may well show a similar sensitivity as location-specific temperatures rise. Climatic influences on regional famines are another well- recognized climate–health association. Malnutrition remains one of the largest health crises worldwide and according to the WHO, approximately 800 million people are currently undernourished, with close to half of these living in Africa20. Droughts and other climate extremes have direct impacts on food crops, and can also influence food supply indirectly by altering the ecology of plant pathogens. Projections of the effect of climate change on food crop yield production globally appear to be broadly neutral, but climate change will probably exacerbate regional food supply inequalities21. Infectious diseases. Climatic variations and extreme weather events have profound impacts on infectious disease. Infectious agents (such as protozoa, bacteria and viruses) and their associated vector organ- isms (such as mosquitoes, ticks and sandflies) are devoid of thermo- static mechanisms, and reproduction and survival rates are thus strongly affected by fluctuations in temperature4,22. Temperature dependencies are seen in correlations between disease rates and weather variations over weeks, months or years23 and in close geographic associations between key climate variables and the distributions of important vector-borne diseases24,25. Malaria transmission has been associated with anomalies of maximum temperature in the highlands of Kenya26. Several studies of long-term trends in malaria incidence and climate in Africa, however, have not found a link to temperature trends, emphasizing instead the importance of including other key determinants of malaria risk such as drug resistance, human migration and immune status, inconsistent vector- or disease-control programmes, and local land-use changes27–30. However, in the highland Debre Zeit sector of central Ethiopia an association has been documented between increasing malaria prevalence and incidence with concomitant warming trends from 1968 to 1993 (ref. 31). Controlling for confounding factors, the association could not be explained by drug resistance, population migration, or level of vector-control efforts. In short, studies of the association of malaria and past climate in the African Highlands remains controversial in part due to varying quality of long-term disease data across sites in Africa, and in part due to the difficulty in adequately controlling for demographic and biological (drug resistance) data. A definitive role of long-term climate trends has not been ascertained. Dengue fever and the more serious form of this disease, dengue haemorrhagic fever (DHF), are caused by the world’s most prevalent mosquito-borne virus. All strains of the dengue virus are carried principally by the Aedes aegyptimosquito. This mosquito is strongly affected by ecological and human drivers, particularly the density of water-bearing containers, but is also influenced by climate, including variability in temperature, moisture and solar radiation. For rela- tively small countries with presumably some climate uniformity, a climate-based dengue model has been developed that strongly correlates with the inter-annual variability in dengue cases reported at the national level (Fig. 1)32. A few examples of other vector-borne diseases demonstrating variance with climate include the Ross River virus in Australia33,34, and plague35 in the American southwest. Bluetongue, a disease of livestock, has increased its northern range in Europe since 1998, paralleling trends in warming and controlling for many biological and socioeconomic factors36. Temperature has also been found to affect food-borne infectious diseases. For example, higher than average temperatures contribute to an estimated 30% of reported cases of salmonellosis across much of continental Europe37. In the UK, the monthly incidence of food poisoning is most strongly associated with the temperatures occur- ring in the previous two to five weeks38. El Nin˜o/Southern Oscillation and infectious diseases. With the exception of seasonal variability, the El Nin˜o/Southern Oscillation (ENSO) is the strongest naturally occurring source of climate variability around the globe39. Studies of malaria have revealed the health impacts of interannual climate variability associated with El Nin˜o, including large epidemics on the Indian subcontinent40, in Colombia41, Venezuela42 and Uganda43. Rift Valley fever epidemics Figure 1 | Correlation between simulated, climate-driven variations in Aedes aegypti mosquito density and observed variations in dengue and DHF cases. Using a computer model of mosquito physiology and development, estimated changes in the relative abundance of Aedes aegypti that were driven only by month-to-month and year-to-year variations in temperature, humidity, solar radiation and rainfall were analysed. The simulated, climate-induced variations in mosquito density were then compared to reported cases of dengue and DHF across many nations of the world that covered at least one degree of latitude and longitude and had at least five years of dengue caseload data. In many countries of Central America and Southeast Asia, the relationship is statistically significant (P , 0.05). For example, climate-driven fluctuations in Ae. aegypti densities appear to be related to annual variations in dengue/DHF cases in Honduras, Nicaragua and Thailand as shown. These represent relatively small-area countries; for larger countries endemic for dengue such as Brazil, China, India and Mexico, the association is not significant, as might be expected because the disease data was at the country level. Graphs adapted from ref. 32. NATURE|Vol 438|17 November 2005 REVIEWS 311 © 2005 Nature Publishing Group between 1950 and 1998 have coincided with unusually high rainfall in East Africa associated with ENSO-related Pacific and Indian Ocean sea surface temperature (SST) anomalies44. While more than three quarters of the Rift Valley Fever outbreaks between 1950 and 1988 occurred during warm ENSO event periods45, some epidemics have also occurred in years with no ElNin˜o, and the model has not been validated against new epidemics. A ‘wavelet analysis’ method was recently used to incorporate host immunity and pathogen population dynamics of DHF in Thailand. A spatial-temporal travelling wave explained a three-year period cycle in disease incidence, starting in Bangkok, moving radially at a speed of 148 km per month46. In a subsequent study that controlled for this intrinsic synchronization, El Nin˜o remained as a significant determinant of dengue epidemics that cycled every two to three years from 1986 to 1992 in Thailand47. Hantavirus pulmonary syndrome in the American southwest can be predicted on the basis of ENSO events; following the 1991–92 El Nin˜o, associated heavy rainfall led to an increase in the Table 1 | Global burden of climate-change-attributable disease Region CVD Diarrhoea Malaria Floods Mortality* Risk‡ Mortality* Disease† Risk‡ Mortality* Disease† Risk‡ Mortality* Disease† Risk‡ Inland Coastal AFR-D 1 1.007 5 154 1.08 5 178 1.02 0 1 1.36 1.64 AFR-E 1 1.005 8 260 1.08 18 682 1.14 0 3 1.48 1.18 AMR-A 0 1 0 0 1 0 0 1.51 0 4 4.93 1.19 AMR-B 1 1.004 0 0 1 0 3 1.15 1 67 2.13 2.27 AMR-D 0 1.005 1 17 1.02 0 0 1.08 0 5 1.78 4.64 EMR-B 0 1.003 0 14 1 0 0 1 0 6 2.67 1.75 EMR-D 1 1.003 8 277 1.09 3 112 1.29 1 46 3.05 3.91 EUR-A 0 0.999 0 0 1 0 0 1 0 3 3.55 1.14 EUR-B 0 0.999 0 6 1.01 0 0 1 0 4 1.82 6.31 EUR-C 0 0.998 0 3 1 0 0 1.48 0 1 2.35 1.04 SEAR-B 1 1.007 1 28 1 0 0 1 0 6 1.79 1.39 SEAR-D 7 1.007 22 612 1.09 0 0 1.01 0 8 1.12 1.04 WPR-A 0 0.999 0 0 1 0 0 1.48 0 1 1.76 1.04 WPR-B 0 1 2 89 1.01 1 43 1.42 0 37 1.62 1.05 World 12§ – 47 1,459 – 27 1,018 – 2 193 – – Region Malnutrition All causes Total deaths per million Total DALYs per million Mortality* Disease† Risk‡ Mortality* Disease† Mortality* Disease† AFR-D 8 293 1.02 19 626 66.83 2,185.78 AFR-E 9 323 1.02 36 1,267 109.4 3,839.58 AMR-A 0 0 1 0 4 0.15 11.85 AMR-B 0 0 1 2 71 3.74 166.62 AMR-D 0 0 1 1 23 10.28 324.15 EMR-B 0 0 1 1 20 5.65 147.57 EMR-D 9 313 1.08 21 748 61.3 2,145.91 EUR-A 0 0 1 0 3 0.07 6.66 EUR-B 0 0 1 0 10 1.04 48.13 EUR-C 0 0 1 0 4 0.29 14.93 SEAR-B 0 0 1 2 34 7.91 117.19 SEAR-D 52 1,918 1.17 80 2,538 65.79 2,080.84 WPR-A 0 0 1 0 1 0.09 8.69 WPR-B 0 0 0.99 3 169 2.16 111.36 World 77 2,846 – 166 5,517 27.82 925.35 *Estimated mortality in thousands attributable to climate change in 2000 (compared to baseline climate of 1961–1990). †Estimated disease burden in thousands of DALYs attributable to climate change in 2000. ‡Projected changes in relative risk for 2030. §Heat-related deaths without subtracting potential reductions in cold-related deaths; this value was therefore not included in the aggregate estimates of mortality due to climate change. The data in Table 1 are taken from ref. 57. The region key is taken from ref. 57. AFR-D: Algeria, Angola, Benin, Burkina Faso, Cameroon, Cape Verde, Chad, Comoros, Equatorial Guinea, Gabon, Gambia, Ghana, Guinea, Guines-Bissau, Liberia, Madagascar, Mali, Mauritania, Mauritius, Niger, Nigeria, Sao Tome and Principe, Senegal, Seychelles, Sierra Leone, Togo. AFR-E: Botswana, Burundi, Central African Republic, Congo, Coˆte d’Ivoire, Democratic Republic of the Congo, Eritrea, Ethiopia, Kenya, Lesotho, Malawi, Mozambique, Namibia, Rwanda, South Africa, Swaziland, Uganda, United Republic of Tanzania, Zambia, Zimbabwe. AMR-A: Cuba, Canada, United States of America. AMR-B: Antigua and Barbuda, Argentina, Bahamas, Barbados, Belize, Brazil, Chile, Colombia, Costa Rica, Dominica, Dominican Republic, El Salvador, Grenada, Guyana, Honduras, Jamaica, Mexico, Panama, Paraguay, Saint Kitts and Nevis, Saint Lucia, Saint Vincent and the Grenadines, Suriname, Trinidad and Tobago, Uruguay, Venezuela. AMR-D: Bolivia, Ecuador, Guatemala, Haiti, Nicaragua, Peru. EMR-B: Bahrain, Cyprus, Iran, Jordan, Kuwait, Lebanon, Libyan Arab Jamahiriya, Oman, Qatar, Saudi Arabia, Syrian Arab Republic, Tunisia, United Arab Emirates. EMR-D: Afghanistan, Djibouti, Egypt, Iraq, Morocco, Pakistan, Somalia, Sudan, Yemen. EUR-A: Andorra, Austria, Belgium, Croatia, Czech Republic, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Israel, Italy, Luxembourg, Malta, Monaco, the Netherlands, Norway, Portugal, San Marino, Slovenia, Spain, Sweden, Switzerland, United Kingdom. EUR-B: Albania, Armenia, Azerbaijan, Bosnia and Herzegovina, Bulgaria, Georgia, Kyrgystan, Poland, Romania, Slovakia, Tajikistan, Macedonia, Turkey, Turkmenistan, Uzbekistan, Yugoslavia. EUR-C: Belarus, Estonia, Hungary, Kazakhstan, Latvia, Lithuania, Moldova, Russian Federation, Ukraine. SEAR-B: Indonesia, Sri Lanka, Thailand. SEAR-D: Bangladesh, Bhutan (Democratic People’s Republic of), Korea, India, Maldives, Myanmar, Nepal. WPR-A: Australia, Brunei, Darussalam, Japan, New Zealand, Singapore. WPR-B: Cambodia, China, Cook Islands, Fiji, Kiribati, Lao, Malaysia, Marshall Islands, Micronesia, Mongolia, Nauru, Niue, Palau, Papua New Guinea, Philippines, Republic of Korea, Samoa, Solomon Islands, Tonga, Tuvalu, Vanuatu, Vietnam. REVIEWS NATURE|Vol 438|17 November 2005 312 © 2005 Nature Publishing Group rodent populations that preceded human cases of disease48. Based on these climate/ecology/disease relationships, a climate- and GIS (Geographic Information System)-based model was developed that predicted disease risk reasonably well for the following strong El Nin˜o event of 1997–98 (ref. 49). Waterborne diseases, such as childhood diarrhoeal disease, are also influenced by El Nin˜o, as was observed with the 1997–98 El Nin˜o event in Peru. During that unseasonable winter, the ambient temperature in Lima increased more than 5 8C above normal, and the number of daily admissions for diarrhoea increased bymore than twofold, compared to expected trends50. Cholera has varied with climatic fluctuations and SSTs affected by the ENSO phenomenon over multi-decadal time periods in Bangladesh51. In the Bay of Bengal, upward trends in cholera also have been linked to longer-term climate changes (that is, changes over approximately a century), with weak cholera/ENSO links found during 1893–1940, and strong and consistent associations occurring during themore pronounced ENSO fluctuations between 1980–2001 (ref. 52). One ecologically based hypothesis for this link involves copepods (zooplankton), which feed on algae, and can serve as reservoirs forVibrio cholerae and other enteric pathogens53; copepods bloom in response to the warming SSTs generally associated with El Nin˜o. Understanding interannual cycles of cholera and other infectious diseases (as seen above for dengue fever), however, requires the combined analyses of both environmental exposures and intrinsic host immunity to a disease. When these factors are considered together, interannual variability of cholera is strongly correlated to SSTs in the Bay of Bengal, ENSO, the extent of flooding in Bangladesh across short time periods (,7 years), and to monsoon rains and Brahmaputra river discharge for longer period climate patterns (.7 years)54. Although it is not clear whether and how ENSO dynamics will change in awarmer world, regions that are currently strongly affected by ENSO (for example, southeast Asia, southern and east Africa, the southwest USA, and various regions of South America) could exp