Tweet
<TABLE width="100%" xmlns="http://www.w3.org/1999/xhtml"><TBODY><TR><TD>The Lancet 2003; 362:1481-1489
DOI:10.1016/S0140-6736(03)14695-8
Review
El Ni?o and health
R Sari KovatsMSc
a 
, Menno J BoumaMD b, Shakoor HajatPhD a, Eve WorrallPhD c and Prof Andy HainesMD d
Summary
El Ni?o Southern Oscillation
El Ni?o and weather disasters
Infectious diseases
Problems with interpretation
Specific infections
Interventions: application of climate-disease associations
Conclusions
Search strategy and selection criteria
References
Summary
El Ni?o Southern Oscillation (ENSO) is a climate event that originates in the Pacific Ocean but has wide-ranging consequences for weather around the world, and is especially associated with droughts and floods. The irregular occurrence of El Ni?o and La Ni?a events has implications for public health. On a global scale, the human effect of natural disasters increases during El Ni?o. The effect of ENSO on cholera risk in Bangladesh, and malaria epidemics in parts of South Asia and South America has been well established. The strongest evidence for an association between ENSO and disease is provided by time-series analysis with data series that include more than one event. Evidence for ENSO's effect on other mosquito-borne and rodent-borne diseases is weaker than that for malaria and cholera. Health planners are used to dealing with spatial risk concepts but have little experience with temporal risk management. ENSO and seasonal climate forecasts might offer the opportunity to target scarce resources for epidemic control and disaster preparedness.
Published online May 20, 2003 http://image.thelancet.com/extras/02art5336web.pdf
Back to top
El Ni?o events have occurred for millennia but were probably first recognised in the late 19th century in Peru1 (panel 1). The name El Ni?o derives from the appearance of warm water off the coast of Peru and Ecuador, which was most noticeable around Christmas (El Ni?o meaning ?little boy? refers to the infant Jesus). From time to time, the warming is anomalous (ie, it exceeds expected variation) and persists for 12?18 months, severely disrupting local fish and bird populations. El Ni?o is consistently associated with heavy rainfall and flooding on the west coast of South America.2
Panel 1: El Ni?o events since 1899 Strong events are indicated
El Ni?o Southern Oscillation
Differences in air pressure across the Pacific basin were first identified by Gilbert Walker in the early 1900s as a contributor to monsoon rainfall in India, and an influence on world weather.3 The fluctuation in pressure difference between Darwin, Australia, and Tahiti is known as the Southern Oscillation. Only as recently as the 1960s were El Ni?o and the Southern Oscillation linked and identified as oceanographic and atmospheric components of the same phenomenon?ie, the El Ni?o Southern Oscillation (ENSO). The Southern Oscillation Index (SOI) is generally negative during an El Ni?o, or warm, event, and positive during ENSO's other extreme, the La Ni?a, or cold, event.
As a result of changes in global atmospheric circulation, ENSO events are accompanied by changes in storm activity, and effects on local climate are observed far from the Pacific region, a process known as teleconnection. Temperatures rise globally during El Ni?o by an average of 0?5?C. Precipitation anomalies, however, are less homogeneous: rainfall increases in some regions and decreases in others.4
Furthermore, El Ni?o and La Ni?a usually produce opposite anomalies. The patterns shown in figure 1 are constructed from averages of many El Ni?o events and mask the large variability between events with respect to intensity, duration, and geographical distribution of climate anomalies. Associations with drought are well described in North East Brazil, Southern Africa, South Asia, Indonesia, and Northern Australia.
Click to enlarge image
Figure 1. ENSO teleconnections and risk map for malaria Risk areas for drought and rainfall based on teleconnections associated with El Ni?o. M shows areas where there is a risk of epidemic malaria after the onset of an El Ni?o event.
The effect of global climate change on the future frequency, amplitude, or both, of El Ni?o is uncertain,5,6 but there are concerns that events might become more frequent or more intense.7 However, even with little or no augmentation, climate change is likely to lead to greater extremes of dry weather and heavy rainfall, increasing the risk of drought and flood that occur with El Ni?o in many regions.5
The effect of ENSO on crop production and weather disasters is such that global financial markets and the insurance industry now take seasonal climate forecasts into account.8 For example, seasonal climate forecasts of 3?6 months are now used by farmers in Australia and South America to plan for the planting of drought resistant crops when drier El Ni?o conditions are anticipated. In the health sector, application of ENSO and climate forecasts has lagged behind. Identification of causal associations between climate and disease, and the translation of these into a coherent public health policy remains a major challenge.
Previously, reports have been published about the effects of El Ni?o on mosquito-borne diseases9 including malaria10 and dengue,11 natural disasters,12 and diarrhoeal diseases.13 The improving ability of agencies to predict El Ni?o events and associated weather anomalies has raised the prospect of temporal risk assessment as a guide for public health policy and practice.14
Back to top
El Ni?o and weather disasters
ENSO is the most important climatic cycle that contributes to year-to-year variability in weather and the likelihood of extreme weather events such as heavy rainfall, droughts, and storms. Natural disasters have widespread implications for public health, and they interfere with the continuity of health care through damage to infrastructure, or because of shifting medical and political priorities.15 Some infectious diseases might be aggravated by malnutrition,16 and famine conditions are often associated with human migration; both factors might facilitate the spread of infectious diseases (figure 2).17
Click to enlarge image
Figure 2. Potential health effects of drought in developing countries
Worldwide, drought is twice as frequent in the year after the onset of El Ni?o than in other years.18 The risk is concentrated in Southern Africa and South Asia. However, disasters do not always occur during El Ni?o; in fact, there is much variability in climatic anomalies between events. For example, in 1997?98, the anticipated drought did not happen in the southern African region and some areas received above average rainfall.19 Famine was averted after the severe drought in South Africa in 1992, despite crop failure rates of 80% in some regions.20 Regional cooperation and external aid allowed the purchase and distribution of enough cereals to avert disaster.
The effect of El Ni?o on disasters is strong enough to be apparent at a global level.12 In an average El Ni?o year, around 35 per 1000 people are affected by a natural disaster?more than four times that in non-El Ni?o years, based on 1963?93 data. This difference in risk is much more pronounced for famine. El Ni?o's global disaster footprint is largely determined by the consequences of drought.12
Some major famines have been linked to El Ni?o: the event of 1876?78 was associated with ?the most destructive drought the world has ever known?21 in China, India, South Africa, Egypt, Ethiopia, Sudan, Java, and Brazil. In India, at least 7 million people died despite the presence of a modern railroad network and millions of tonnes of grain in commercial circulation. British imperial policies resulted in the relief reaching only one-tenth of those whose lives were threatened by food shortages.
Drought, in association with slash-and-burn methods of land clearance, can trigger uncontrolled forest fires. Every El Ni?o since at least 1982 has been associated with fires in Kalimantan, which have consequences for public health.22 Smoke from the 1997 forest fires on the Indonesian island groups of Kalimantan and Sumatra affected surrounding areas including Malaysia, Singapore, Philippines, and southern Thailand. Smoke from biomass burning contains pollutants harmful to health, including particulates (particles less than 2?5 μm in diameter that can penetrate human lungs).23
The relation between El Ni?o and intense rainfall is strong in many areas (figure 1). During the 1982?83 and 1997?98 events, intense rain and floods caused hundreds of deaths in Peru, Colombia, Ecuador, and Bolivia.19,24 Deaths associated with floods are also strongly associated with SOI in parts of Australia.25 On a global scale, ENSO is not associated with risk of flood-related disasters because floods are very localised and the risk is heightened during both El Ni?o and La Ni?a phases in different parts of the world.18
Hurricanes in the Caribbean, the Gulf of Mexico, and off the coast of northern Australia are less common than usual during El Ni?o, but more common during La Ni?a. However, typhoons are more likely to occur near the Marshall Islands, in the Pacific Ocean, during an El Ni?o26 event than at other times because storm tracks in the Pacific are shifted to the west during this time. For small islands that lie in their path, the shifting of storm tracks is of particular importance.
Seasonal climate forecasts are now used to mitigate the effects of drought and flood that are associated with ENSO events (see further reading).27 Climate forecasts are combined with other indicators (such as satellite data and food prices) to provide early warning of famine.14,28,29 El Ni?o forecasts could provide decision makers with the earliest possible warning of natural disasters linked to flood and drought.2
Back to top
Infectious diseases
Transmission of many infectious diseases can be affected by weather, especially for those pathogens that spend part of their lifecycle outside the human body. Pathogens carried by insects are exposed, with their flying hosts, to ambient weather. The transmission of vector-borne diseases typically occurs within seasonal patterns, in which the role of temperature and rainfall is well documented. Some vector-borne diseases display much year-to-year variation that can also be partly explained by meteorological factors. The ability to predict high or low transmission seasons would help target the timing and location of public health interventions.
We review studies that have identified associations between climate and disease risk based on El Ni?o or La Ni?a. Evidence for an association between disease risk and ENSO is more robust when analyses use a long time-series that incorporates more than one event and when there is appropriate geographical aggregation of data. Individual outbreaks of disease can be triggered by extreme weather. Such outbreaks are often attributed to ENSO if the weather pattern is consistent with the effects of this climatic event.30,31 However, in our opinion a true association between ENSO and disease in a given population can be confirmed only through analysis of several ENSO events with time series methods. Non-climatic explanations for an association should always be considered, although it is unlikely that environmental factors, such as vector control and changes in case detection, would vary within the same time patterns as ENSO.
We have identified 21 reports that quantify a relation between ENSO and human infectious disease in more than 18 countries, and which have used data series that incorporate more than one event (table).11,32?51 Most of these studies noted a significant association between disease and ENSO. A few used data from geographical areas where ENSO has little or no consistent effect on the weather: for example, malaria and Rift Valley fever in Kenya,36,41 and dengue fever in Bangkok.36 Workers have investigated a range of diseases in several regions, but few studies overlap information for the same areas. Consistent findings have been noted for an association between ENSO and malaria in the coastal regions of Venezuela and Colombia, and for evidence of an effect on cholera transmission in Bangladesh.
Table. Time series studies of ENSO and infectious disease
Back to top
Problems with interpretation
Several ENSO parameters have been used. The SOI shows more variability than do Pacific sea surface temperatures. The use of El Ni?o year as a time variable causes difficulty because El Ni?o does not run to the calendar year, and there is no official definition of what constitutes an El Ni?o event. To model the time-series data, some investigators have used spectral analysis to identify regular cycles.36,52
Measurements connected by time or location are probably correlated and not independent. Serial correlation (autocorrelation) refers to two adjacent observations being more alike than two randomly chosen observations.53 Autocorrelation should be accounted for before independence of variables can be assumed, but many studies that assess the relationship between ENSO and disease have not done so. The inclusion of autocorrelation terms in a regression model is thought to produce a more conservative estimate of the effect and reduced standard error45,54 than the one arrived at without accounting for autocorrelation. Future studies should report findings with and without adjustment (panel 2).
Panel 2: Guidelines for assessment and reporting of interactions between ENSO and health
?Test and report results of association between weather parameters and ENSO parameter in the data
?Report published assessments of ENSO teleconnections by climatologists in region of interest
?Describe the geographical area from which the health data are derived.
?Test and report results of association between weather parameters and disease outcome
?Use time series data with more than one ENSO event
?Remove any trend and regular seasonal patterns in the time-series data before assessing relationships
?Report associations both with and without adjustment for serial correlation
Both intrinsic factors (eg, changes in population immunity) and extrinsic factors (eg, climate variables) can affect the timing of disease epidemics. Such factors are not mutually exclusive and epidemics are caused by a complex interaction, with the balance of components varying between disease systems.55 Models of transmission dynamics that rely on population immunity have been developed for directly transmitted diseases such as measles.56 However, there is no good empirical evidence to show that changes in population immunity can account for malaria epidemics over periods of 3?5 years, although this possibility has been suggested by several authors.36,57 The El Ni?o cycle is irregular and varies in length from 2?7 years (panel 1). The extent to which a specific disease system is being driven solely by the replacement rate of people without immunity within a population should be assessed. The contribution of this mechanism has not yet been quantified for either dengue or malaria. Analysis of monthly cholera incidence in Bangladesh found a role for intrinsic factors (such as previous disease incidence) but also extrinsic ones such as ENSO in the dynamics of cholera transmission.48
Many El Ni?o time-series studies use aggregated national data. However, analysis of smaller geographical locations could help understanding of complex relations between outcome and local drivers such as temperature and rainfall.58 The association between climate variables (temperature, rainfall) and disease should be evaluated since these variables are the principal drivers of the biological processes by which ENSO affects health;59 however, few studies report such analyses. For example, Bouma and Dye38 investigated the complex association between rainfall and malaria epidemic years in Venezuela. Similarly, Pascual and colleagues48 explored the relation between cholera and ENSO, and cholera and local climate factors. Poveda and colleagues33 assessed the effects of El Ni?o on the annual cycle of both malaria and climatic indices, and showed that the association between malaria and climate was intensified during El Ni?o phases. The relation between local climate and ENSO should also be clearly reported and refer to published assessments by climatologists about teleconnections in the region of interest.
Back to top
Specific infections
Malaria
Public health ministries and institutions in countries where malaria transmission rates vary a lot between years, such as those in Colombia and Gujarat, India, have begun to appreciate the usefulness of forecasts. After a steep rise in malaria in Colombia in 1998, the National Public Health Surveillance System, Colombian Ministry of Health has reported the accuracy of earlier published malaria forecasts.60
Malaria epidemics occur in regions where transmission rates are not usually sufficient to provide protective immunity within the population. In some of these epidemic fringe regions, malaria transmission is restricted by climate?ie, conditions are either too dry, too wet, or too cold for vectors or parasites. Therefore, small changes in environmental conditions can trigger an epidemic. The latitudinal edges of malaria distribution are usually determined by the effectiveness of public health defences rather than climate.61
Malaria, through its variety of vectors and ecological conditions, has unique features in each location. Short-term atypical climate conditions (such as rainfall in arid regions and drought in humid climates) could cause epidemics. Therefore, generalisation of the effects of ENSO on epidemic malaria is not possible. There is evidence that in highland areas, raised temperatures associated with El Ni?o, especially during the autumn and winter months, might increase transmission of malaria. This effect has been shown in Northern Pakistan in 1981?91.62,37 Higher than usual temperatures and heavy rainfall have also been associated with short-term increases in highland malaria in Rwanda63 and Uganda.30 Conversely, increased rainfall in a highland area of Tanzania during the El Ni?o event of 1997 probably washed away breeding sites and lowered the number of malaria cases.64
In many desert fringe regions, rainfall and malaria transmission rates are often connected?eg, in the Punjab before 194035,65 and currently in Rajasthan bordering the Thar desert.66 Rainfall in this region is associated with ENSO,4 thus there is an increased epidemic risk during post El Ni?o and La Ni?a years.67
El Ni?o-related droughts have been associated with malaria outbreaks in Sri Lanka,35 Colombia,33 and Irian Jaya68 (figure 1). Epidemics can follow drought in very humid regions, when river flow decreases sufficiently to allow mosquito breeding.69 El Ni?o-related droughts could increase the risk of an epidemic by increasing population mobility, because non-immune people come into contact with infected populations who have moved in search of food.17,68
Post-drought malaria epidemics are associated with El Ni?o in Venezuela, particularly in coastal regions.38 When aquatic ecosystems are re-established after dry years, vector populations can increase to higher than usual numbers because predators of larvae have been reduced. Periodic droughts in coastal Venezuela have been associated with prominent changes in the vector distribution.70
Epidemic preparedness has an early precedent for malaria. In the Punjab between 1922 and 1947, the time window between rainfall and malaria was used to distribute the limited supply of quinine to areas most at risk. Workers estimated risk using rainfall, low recent malaria exposure (judged by low spleen rates) and high food prices, which can indicate lower nutritional status and immunity. This early warning system was used until the introduction of residual insecticides.57 Reduced reliance on residual insecticides and the recurrence of epidemic malaria raises the possibility of reintroduction of epidemic forecasting in parts of India.62,67,71 Epidemic early warning systems typically rely on various indicators that have different lead times.55 Important developments have been made in mapping the regions and timing of malaria risk, including the use of satellite data to provide proxy ecological variables, such as rainfall estimates or vegetation indices.72,73 In combination with climate forecasts, such techniques could provide a valuable tool for epidemic preparedness.14
Mosquito-borne viral diseases
Several studies have found an association between epidemic dengue and ENSO in populations in southeast Asia, South America, and the Pacific (table).31,39,40 However, links between climate, weather, and dengue are poorly understood because the disease is transmitted by container-breeding mosquitoes in urban areas. If climate is an important influence on current distribution of disease or vectors, then this can be considered as supporting evidence that the disease could be affected by past or future trends in climate.74 The global distribution of dengue has been mapped with models that use climate factors;75,76 however, these methods cannot provide insight into the drivers of changes from year to year in disease in a given location. Dengue haemorrhagic fever (DHF) has a tendency to recur in 2?5 year cycles and changes in dengue immunity seem to be important in predisposing a population to outbreaks.36 However, population immunity remains poorly characterised for dengue and DHF.
ENSO usually has a strong effect on the weather in parts of Australia and the relation between ENSO and indigenous arboviruses has been investigated. Infrequent but severe epidemics of Murray Valley encephalitis (Australian encephalitis) have arisen in Australia after above average rainfall and flooding associated with La Ni?a episodes.77,78 Some outbreaks of Ross River virus disease (epidemic polyarthritis) might be linked to weather patterns associated with ENSO, but associations with climate factors are highly localised and might not be detectable with aggregated data (table).79 Epidemics of Rift Valley fever in the dry grasslands in East Africa are triggered by heavy rainfall.80,81 However, no association between this disease in Kenya and SOI was noted80 and there is no consistent association between El Ni?o and rainfall variability in Kenya.
Rodent-borne disease
Rodents are reservoirs for several diseases and their numbers tend to increase after mild wet winters. Human cases of plague in New Mexico are more likely to arise following winter-spring seasons with above average rainfall.51 The emergence of hantavirus pulmonary syndrome in 1993 in southern USA was associated with an increase in the size of the local rodent population. Increased rainfall associated with the El Ni?o event was followed by drought conditions that increased rodent to rodent and rodent-human interactions.82,83 During and after the 1997?98 El Ni?o, the rodent population increased ten-fold to 20-fold, and reported cases of hantavirus pulmonary syndrome increased five-fold, despite strong public awareness about how to reduce exposure.84 This evidence suggests a mechanism by which climate factors affect hantavirus transmission in this region; however, a consistent association with ENSO has not been established.85
Diarrhoeal diseases and cholera
Heavy rainfall has been associated with an increase in outbreaks of enteric pathogens, usually as a result of a contamination of water supplies. In tropical regions, diarrhoeal diseases typically peak during the rainy season.86 Temperature is important in the seasonal and between-year variability of diarrhoeal diseases. Higher than usual daily temperatures during the warm event of 1997, adjusted for seasonal trend, were related to an increased number of admissions for diarrhoea in Peruvian children.13
The importance of temperature and other environmental factors in the epidemiology of cholera has been suspected since the 19th century.52,87 The bimodal seasonal pattern of cholera in Bangladesh mirrors sea surface temperatures in the Bay of Bengal and seasonal plankton abundance (a possible environmental reservoir of the cholera pathogen, Vibrio cholerae).47,88,89 Variability in cholera incidence between years in Dhaka, Bangladesh, between 1980 and 1998 was associated with ENSO.48 An analysis of historical data for Bengal (1890?1940) indicates that any effect of El Ni?o was confined to the coastal regions, where it might have triggered spring epidemics of the disease, which is a shift away from the usual seasonal pattern.47 The relation between cholera and SOI in Dhaka seems to have changed over time, becoming stronger in the last two decades.49 Some studies have found correlations between cholera incidence in Bangladesh and sea surface temperatures in the Bay of Bengal,52,47 which are also affected by ENSO. Climate factors, such as water temperature, could drive seasonality by their direct influence on the abundance or toxicity of V cholerae. The possible mechanisms by which increased sea surface temperatures affect disease transmission from year to year remain poorly understood.
Back to top
Interventions: application of climate-disease associations
Seasonal forecasts are used to predict major climate trends for the next few months to few seasons. They indicate areas where there is an increased probability of some deviation from the climatic mean, such as wet or dry, warm or cold conditions. Seasonal forecasts are issued on a regular basis for each season, but their accuracy is greater during ENSO events than at other times. The science of forecasting is developing rapidly and quite successful seasonal forecasts were issued for the first time during the El Ni?o of 1997?98.19Figure 3 shows various stages of forecasting with respect to public health. The forecast of an El Ni?o event per se has the longest lead time, but has limited geographical information. Prediction of an event is extremely difficult, but once the onset is confirmed, confidence in forecasts is increased and lead times of several months can be used to plan public health initiatives. Because of the differences in forecasts generated by various agencies, the NOAA Office of Global Programs and other agencies convene regional climate outlook fora that produce a consensus forecast and response plans for particular regions.
Click to enlarge image
Figure 3. ENSO and forward planning in public health As the accuracy of the forecast increases, the lead time and window of opportunity decreases.
Forecast of the risk of an epidemic is an essential component of epidemic control.90,91 Expensive control measures are sometimes implemented when transmission has already been naturally interrupted as a means of reassuring the population and showing political commitment.90 Such actions are unlikely to be cost effective. Seasonal climate and malaria forecasts on the other hand allow early intervention, which can mitigate effects of epidemics and improve the cost-effectiveness of control activities. To achieve these ends, forecasts must not only be accurate but they must also be trusted and used in a timely way.
Eight southern African countries now routinely participate in the Southern African Regional Climate Outlook Forum (SARCOF) to develop malaria forecasts specific to each region and country.92 As a direct result of these forecasts, several preparedness activities are carried out in southern Africa, including the establishment of epidemic stores, issuing of epidemic funds, identification of epidemic teams, and development of checklists. At a national level, epidemic plans, guidelines, and checklists have been developed and stocks, personnel, and funds are allocated. These activities are designed to strengthen and speed up response capacity once epidemics are identified.92
Decision-makers are often reluctant to invest resources in response to a forecast, when there might be more immediate demands on scarce resources. The potential economic benefit of malaria forecast has been simulated for one southern African country.93 The cost per malaria case prevented by residual spraying could be reduced by 25?56% (dependent on the severity of the transmission season) simply by bringing forward the spray programme by 3?4 months, to before the onset of the rains. Further improvements in cost-effectiveness could be achieved by use of a reliable seasonal climate forecast to determine the most appropriate level of activity each year, with most intense activity being carried out in high transmission years and least in low transmission years. Decision makers should carefully weigh these potential efficiency improvements against the risks and consequences of following incorrect forecasts. False positives could result in resources being wasted, while false negatives will represent a missed opportunity for malaria control. Both will result in a loss of confidence in malaria forecasts.
Back to top
Conclusions
The major effect of El Ni?o on health and society is mediated through an increased risk of natural disaster such as droughts, floods (figure 4), and tropical cyclones. There is also good epidemiological evidence that El Ni?o is associated with an increased risk of certain diseases in specific geographical areas where climate anomalies are linked with the ENSO cycle. The associations are particularly strong for malaria and cholera in some parts of the world, but only suggestive for other mosquito-borne and rodent-borne diseases. More research is needed to determine the mechanisms of these associations. We recommend that future studies should use time series approaches with robust statistical methods (panel 2). Additional research into the ecology of vectors, reservoirs, and pathogens are needed before the causal pathways from ENSO and climate variability to disease prevalence can be understood.
Click to enlarge image
Figure 4. Floods in Peru caused by heavy rains attributed to El Ni?o, 1998
In areas where El Ni?o can reliably be associated with regional or local climate variations such as droughts and floods, El Ni?o forecasts can provide decision makers with the earliest possible warning of an increased risk of such adverse climate conditions. Seasonal forecasts have already been included into many local and regional famine and drought early warning systems. The need to define predictive factors on which to base forecasts of epidemic risk94 is an enormous challenge for public health agencies working at the central and local level. Preparedness for epidemics is one of the focal points of WHO's Roll Back Malaria programme.14 Health planners are used to dealing with spatial risk concepts, but there is a lack of experience with temporal risk management. Furthermore, many developing countries lack the infrastructure for a prospective operational disease-forecasting system. Improved surveillance of diseases and health outcomes that seem to be influenced by the ENSO phenomenon will provide better quality data for research and enhance attempts to prevent adverse effects. The science of climate forecasting is developing rapidly. Health planners should be aware of similar advances in epidemic forecasting and use the best available evidence to apply scarce resources to reduce disease risk.
Back to top
Search strategy and selection criteria
We searched Medline, EMBASE, BIDS, and Web of Science using the search terms ENSO, El Ni?o, teleconnection, La Ni?a, and SOI. We searched for articles published between 1980 and April, 2002 in all languages, except with the terms El Ni?o and La Ni?a, to exclude Spanish language studies, because these terms resulted in a large number of articles about child health. We also used reference lists to identify additional articles, and we contacted authors who had published work on El Ni?o and health. We included articles that
?were published in peer reviewed journals
?were original research articles using epidemiological data.
?quantified any association with an ENSO parameter (eg, El Ni?o year, sea surface temperature, SOI or other index).
?had an outcome that was an infectious disease in human beings.
?had time series data that included more than one El Ni?o event.
We included 21 articles that met these criteria. Further information about the search strategy can be obtained from the authors.
Back to top
Conflict of interest statement
None declared.
Acknowledgments
Menno Bouma is partly funded by NOAA Office of Global Programs and DFID malaria programme. Eve Worrall was previously funded by the Liverpool Malaria Knowledge Programme. The authors would also like to thank the following for helpful comments: David Bradley, Eleanor Riley, Simon Hales.
Back to top
<!--start tail=-->References
1. Glantz MH. Currents of change: El Ni?o's impact on climate and society. Cambridge: Cambridge University Press, 1996:.
2. In: , Glantz MH, ed. Once burned, twice shy? Lessons learned from the 1997/98 El Ni?o. Tokyo: United Nations University Press, 2001:.
3. Walker GT, Bliss EW. World weather V. Mem R Met Soc 1932; 4: 53-84.
4. Ropelewski CF, Halpert MS. Global and regional scale precipitation patterns associated with the El Ni?o/Southern Oscillation. Monthly Weather Review 1987; 115: 1606-1626.
5. IPCC. Climate change 2001: the scientific basis. Contribution of Working Group I to the Second Assessment Report of the Intergovernmental Panel on Climate Change. New York: Cambridge University Press, 2001:.
6. Vinnikov KY, Robock A. Trends in moments of climatic indices.. Geophys Res Lett 2002; 29 (2): 10.
7. Timmerman A, Oberhuber J, Bacher A, Esch M, Latif M, Roeckner E. Increased El Ni?o frequency in a climate model forced by future greenhouse warming. Nature 1999; 398: 694. CrossRef
8. Workshop on applications of seasonal forecasting, 14?15 June 1999. Reading: European Centre of Medium-Range Weather Forecasts (ECMWF), 1999:.
9. Nicholls N. El Ni?o-Southern Oscillation and vector-borne disease. Lancet 1993; 342: 1284-1285. MEDLINE | CrossRef
10. Bouma MJ, Sondorp HE, van der Kaay HJ. Health and climate change. Lancet 1994; 343: 303. MEDLINE
11. Hales S, Weinstein P, Woodward A. Dengue fever epidemics in the South Pacific region: driven by El Ni?o Southern Oscillation?. Lancet 1996; 348: 1664-1665. Full Text | Full-Text PDF (19 KB) | MEDLINE | CrossRef
12. Bouma MJ, Kovats RS, Goubet SA, Cox JS, Haines A. Global assessment of El Ni?o's disaster burden. Lancet 1997; 350: 1435-1438. Abstract | Full Text | Full-Text PDF (70 KB) | MEDLINE | CrossRef
13. Checkley W, Epstein LD, Gilman RH, et al. Effects of El Ni?o and ambient temperature on hospital admissions for diarrhoeal diseases in Peruvian children. Lancet 2000; 355: 442-450. Abstract | Full Text | Full-Text PDF (443 KB) | MEDLINE | CrossRef
14. WHO. Malaria early warning systems, concepts, indicators and partners: a framework for field research in Africa.Roll back malaria/technical support network for prevention and control of malaria.. Geneva: World Health Organization, 2001:.
15. Noji EN. The public health consequences of disasters. New York: Oxford University Press, 1997:.
16. Zurbrigg S. Re-thinking the ?human factor? in malaria mortality. Parassitologia 1994; 36: 121-135. MEDLINE
17. Prothero RM. Forced movements of population and health hazards in tropical Africa. Int J Epidemiol 1994; 23: 657-664. MEDLINE
18. Dilley M, Heyman B. ENSO and disaster: droughts, floods, and El Ni?o/Southern Oscillation warm events. Disasters 1995; 19: 181-193. MEDLINE
19. WMO/UNESCO/UNEP/ICSU. The 1997?98 El Ni?o: a scientific and technical retrospective. Geneva: WMO, 1999:WMO No 905.
20. Yip R. Famine In: , Noji EN, ed. The public health consequences of disasters.. New York: Oxford University Press, 1997: 305-335.
21. Davis M. Late Victorian holocausts: El Ni?o famines and the making of the third world.. London: Verso Books, 2000:.
22. Nicholls N. ENSO, drought and flooding rain in south east Asia. In: Brookfield H, Byron Y, eds. South-east Asia's environment future: the search for sustainability.. Tokyo: United Nations University Press/OUP, 1993: 154-175.
23. WHO. Health guidelines for vegetation fire events: background papers. Singapore: Institute of Environmental Epidemiology, 1999:.
24. Cronica de Desastres: Fenomeno El Ni?o 1997?1998. Washington DC: PAHO, 2000:.
25. Kuhnel I, Coates L. El Ni?o southern oscillation: related probabilities of fatalities for natural perils in Australia. Natural Hazards 2000; 22: 117-138. CrossRef
26. Spennemann DH, Marschner IC. The association between El Ni?o/Southern Oscillation events and typhoons in the Marshall Islands. Disasters 1995; 19: 194-247. MEDLINE
27. Murphy SJ, Washington R, Downing TE, et al. Seasonal forecasting for climate hazards: prospects and responses. Natural Hazards 2000; 23: 171-196. CrossRef
28. Buchanon-Smith M, Davies S. Famine early warning and response?the missing link.. London: Intermediate Technology, 1996:.
29. Kovats RS, Bouma MJ, Haines A. El Ni?o and health.. Geneva: World Health Organization, 1999:.
30. Lindblade KA, Walker ED, Onapa AW, Katungu J, Wilson ML. Highland malaria in Uganda: prospective analysis of an epidemic associated with El Ni?o. Trans R Soc Trop Med Hyg 1999; 93: 480-487. MEDLINE | CrossRef
31. Corwin AL, Larasati RP, Bangs MJ, et al. Epidemic dengue transmission in southern Sumatra, Indonesia. Trans R Soc Trop Med Hyg 2001; 95: 257-265. MEDLINE | CrossRef
32. Gagnon AS, Smoyer-Tomic KE, Bush A. The El Ni?o southern oscillation and malaria epidemics in South America. Int J Biometeorol 2002; 46: 81-89.
33. Poveda G, Rojas W, Quinones M, et al. Coupling between annual and ENSO timescales in the malaria climate association on Colombia. Env Hlth Pers 2001; 109: 307-324.
34. Bouma MJ, Poveda G, Rojas W, et al. Predicting high-risk years for malaria in Colombia using parameters of El Ni?o southern oscillation. Trop Med Int Hlth 1997; 2: 1122-1127.
35. Bouma MJ, van der Kaay HJ. The El Ni?o Southern Oscillation and the historic malaria epidemics on the Indian subcontinent and Sri Lanka: an early warning system. Trop Med Int Hlth 1996; 1: 86-96.
36. Hay SI, Myers MF, Burke DS, et al. Etiology of interepidemic periods of mosquito-borne disease. Proc Natl Acad Sci 2000; 97: 9335-9339. MEDLINE | CrossRef
37. Bouma MJ. Epidemiology and control of malaria in northern Pakistan.. Dordrecht: ICG Printing, 1995:.
38. Bouma MJ, Dye C. Cycles of malaria associated with El Ni?o in Venezuela. JAMA 1997; 278: 1772-1774. MEDLINE
39. Gagnon AS, Bush ABG, Smoyer-Tomic KE. Dengue epidemics and the El Ni?o Southern Oscillation. Clim Res 2001; 19: 35-43.
40. Hales S, Wienstein P, Souares Y, Woodward A. El Ni?o and the dynamics of vectorborne disease transmission. Env Hlth Pers 1999; 107: 99-102.
41. Anyamba A, Linthicum KJ, Tucker CJ. Climate-disease connections: rift valley fever in Kenya. Cad Saude Publica 2001; 17: 133-140.
42. Tong S, Bi P, Parton K, Hobbs J, McMichael AJ. Climate variability and transmission of epidemic polyarthritis. Lancet 1998; 351: 1100. Full Text | Full-Text PDF (50 KB) | MEDLINE | CrossRef
43. Harley DO, Weinstein P. The southern oscillation index and Ross River virus outbreaks. Med J Aust 1996; 165: 531-532. MEDLINE
44. Bi P, Tong S, Donald K, Parton K, Hobbs J. Southern oscillation index and transmission of the Barmah Forest virus infection in Queensland, Australia. J Epidemiol Community Health 2000; 54: 69-70. MEDLINE | CrossRef
45. Maelzer D, Hales S, Weinstein P, Zalucki M, Woodward A. El Ni?o and arboviral disease prediction. Env Hlth Pers 1999; 107: 817-818.
46. Franke CR, Ziller M, Staubach C, Latif M. Impact of the El Ni?o southern oscillation on visceral leishmaniasis, Brazil. Emerg Infect Dis 2002; 8: 914-917. MEDLINE
47. Bouma MJ, Pascual M. Seasonal and interannual cycles of endemic cholera in Bengal 1891?1940 in relation to climate and geography. Hydrobiologia 2001; 460: 147-156. CrossRef
48. Pascual M, Rodo X, Ellner SP, Colwell R, Bouma MJ. Cholera dynamics and El Ni?o southern oscillation. Science 2000; 289: 1766-1767. MEDLINE | CrossRef
49. Rodo X, Pascual M, Fuchs G, Faruque ASG. ENSO and cholera: a nonstationary link related to climate change?. Proc Natl Acad Sci 2002; 99: 12901-12906. MEDLINE | CrossRef
50. Ebi KL, Exuzides KA, Lau E, Kelsh M, Barsnton A. Association of normal weather conditions and El Ni?o events with viral pneumonia hospitalizations in females, California, 1983?1998. Am J Public Hlth 2001; 91: 1200-1208.
51. Parmenter RR, Yadav EP, Parmenter CA, Ettestad P, Gage KL. Incidence of plague associated with increased winter-spring precipitation in New Mexico. Am J Trop Med Hyg 1999; 61: 814-821. MEDLINE
52. Pascual M, Bouma MJ, Dobson AP. Cholera and climate: revisiting the quantitative evidence. Microbes Infect 2002; 4: 237-246. MEDLINE | CrossRef
53. Schwartz J, Spix C, Touloumi G, et al. Methodological issues in studies of air pollution and daily counts of death or hospital admissions. J Epidemiol Community Health 1996; 50 (suppl 1): S3-S11.
54. Chatfield C. The analysis of time series: an introduction. London: Chapman and Hall, 1989:.
55. Myers MF, Rogers DJ, Cox J, Flahault A, Hay SI. Forecasting disease risk for increased epidemic preparedness in public health. Adv Parasitol 2000; 47: 330.
56. Anderson RM, May RM. Population biology of infectious diseases part I. Nature 1979; 280: 361-367. MEDLINE | CrossRef
57. Swaroop S. Forecasting of epidemic malaria in the Punjab. Am J Trop Med 1946; 29: 1-17.
58. Poveda G, Graham NE, Epstein PR, et al. Climate and ENSO variability associated to malaria and dengue fever in Colombia In: Diaz HF, Markgraf F, eds. El Ni?o and the southern oscillation, multiscale variability and global and regional impacts.. Cambridge: Cambridge University Press, 2000: 183-204.
59. Kovats RS. El Ni?o and human health. Bull World Health Organ 2000; 78: 1127-1135. MEDLINE
60. Anon. Sistema Nacional de Vigilancia en Salud Publica: sivgila number 39.. Colombia: Colombian Ministry of Health, 1998:.
61. Reiter P. Global warming and vector-borne disease in temperate regions and at high altitude. Lancet 1998; 351: 839-840. Full Text | Full-Text PDF (46 KB) | MEDLINE | CrossRef
62. Bouma MJ, Dye C, van der Kaay HJ. Falciparum malaria and climate change in the Northwest Frontier Province of Pakistan.. Am J Trop Med Hyg 1996; 55 (2): 131-137. MEDLINE
63. Loevinsohn ME. Climatic warming and increased malaria incidence in Rwanda. Lancet 1994; 343: 714-718. MEDLINE | CrossRef
64. Lindsay SW, Bodker R, Malima R, Msangeni HA, Kisinza W. The effect of 1997?98 El Ni?o on highland malaria in Tanzania. Lancet 2000; 355: 989-990. Abstract | Full Text | Full-Text PDF (61 KB) | MEDLINE | CrossRef
65. Yacob KBM, Swaroop S. Investigations of the long term periodicitiy in the incidence of epidemic malaria in the Punjab. J Malaria Inst India 1946; 6: 273-284.
66. Akhtar R, McMichael AJ. Rainfall and malaria outbreaks in western Rajasthan. Lancet 1996; 348: 1457-1458. Full Text | Full-Text PDF (24 KB) | MEDLINE | CrossRef
67. Bouma MJ, van der Kaay HJ. Epidemic malaria in India and the El Ni?o Southern Oscillation. Lancet 1994; 344: 1638-1639. MEDLINE | CrossRef
68. Bangs M, Subianto DB. El Ni?o and associated outbreaks of severe malaria in highland populations in Irian Jaya, Indonesia: a review and epidemiological perspective. Southeast Asian J Trop Med Public Hlth 1999; 30: 608-619.
69. In: Aron J, Patz JA, eds. Ecosystem change and public health.. Baltimore: Johns Hopkins University Press, 2002:.
70. Gabaldon A. The nation-wide campaign against malaria in Venezuela. Trans R Soc Trop Med Hyg 1949; 43: 113-160. CrossRef
71. Bouma MJ, van der Kaay HJ. Epidemic malaria in India's Thar desert. Lancet 1995; 373: 132-133.
72. Thomson MC, Connor SJ, Milligan PJ, Flasse SP. The ecology of malaria - as seen from Earth-observation satellites. Ann Trop Med Parasitol 1996; 90: 243-264. MEDLINE
73. Hay SI, Snow RW, Rogers DJ. From predicting mosquito habitat to malaria seasons using remotely sensed data: practice, problems, and perspectives.. Parasitology Today 1998; 14: 306-313. MEDLINE | CrossRef
74. Kovats RS, Campbell-Lendrum D, Reid C, Martens P. In: Climate and vector borne disease: an assessment of the role of climate in changing disease patterns.. Maastricht: University of Maastricht, 2000: 50.
75. Hopp M, Foley J. Global-scale relationships between climate and the dengue fever vector, Aedes Aegypti. Climatic change 2001; 48: 441-463. CrossRef
76. Hales S, deWet N, Maindonald J, Woodward A. Potential effect of population and climate changes on golbal distribution of dengue fever: an empircal model.. Lancet 2002; 360: 830-834. Abstract | Full Text | Full-Text PDF (242 KB) | MEDLINE | CrossRef
77. Broom AK, Lindsay MD, Johansen CA, Wright AE, Mackenzie JS. Two possible mechanisms for survival and initiation of Murray Valley encephalitis virus activity in the Kimberley region of Western Australia. Am J Trop Med Hyg 1995; 53: 95-99. MEDLINE
78. Nicholls N. El Ni?o Southern Oscillation impact prediction. Bull Am Meteorol Soc 1988; 69: 173-176.
79. Woodruff RE, Guest CS, Garner MG, et al. Predicting Ross River virus epidemics from regional weather data. Epidemiology 2002; 13: 384-393. MEDLINE | CrossRef
80. Linthicum KJ, Anyamba A, Tucker CJ, Kelley PW, Myers MF, Peters CJ. Climate and satellite indicators to forecast Rift Valley fever epidemics in Kenya. Science 1999; 285: 397-400. MEDLINE | CrossRef
81. Davies FG, Linthicum KJ, James AD. Rainfall and epizootic Rift Valley fever. Bull World Health Organ 1985; 63: 941-943. MEDLINE
82. Wenzel RP. A new hantavirus infection in North America. N Eng J Med 1994; 330: 1004-1005. MEDLINE | CrossRef
83. Engelthaler DM, Mosley DG, Cheek JE, et al. Climatic and environmental patterns associated with hantavirus pulmonary syndrome, four corners region, United States. Emerg Infect Dis 1999; 5: 87-94. MEDLINE
84. Hjelle B, Glass GE. Outbreak of hantavirus infection in the four corners region of the United States in the wake of the 1997?1998 El Ni?o-southern oscillation. J Infect Dis 2000; 181: 1569-1573. MEDLINE | CrossRef
85. Mills JN, Yates TL, Ksiazek TG, Peters CJ, Childs JE. Long-term studies of hantavirus reservoir populations in the southwestern United States: rationale, potential and methods. Emerg Infect Dis 1999; 5: 95-101. MEDLINE
86. Cairncross AM, Feachem RG. Environmental health engineering in the tropics: an introductory text.. Chichester: J Wiley, 1993:.
87. Hirsch A. Handbook of geographical and historical pathology.. London: New Sydenham Society, 1883:.
88. Colwell RR. Global climate and infectious disease: the cholera paradigm. Science 1996; 274: 2025-2031. MEDLINE | CrossRef
89. Lobitz B, Beck L, Huq A, et al. Climate and infectious disease: use of remote sensing for detection of Vibrio cholerae by indirect measurement. Proc Natl Acad Sci 2000; 97: 1438-1443. MEDLINE | CrossRef
90. Najera JA, Kouznetsov RL, Delacollette C. Malaria epidemics; detection and control, forecasting, and prevention.. Geneva: World Health Organization, 1998:.
91. Thomson MC, Palmer T, Morse AP, Cresswell M, Connor SJ. Forecasting disease risk with seasonal climate predictions. Lancet 2000; 355: 1559-1560. Full Text | Full-Text PDF (59 KB) | MEDLINE | CrossRef
92. WHO/AFRO/ICP. Presentation by WHO/AFRO/ICP for Southern Africa. Malaria epidemics in Southern Africa: a review of the progress in malaria epidemic forecasting, detection and response. In: At RBM/TSN for malaria epidemic prevention and control. Geneva: WHO, Dec 2001: 10-11.
93. Worrall E. An economic evaluation of malaria early warning systems in Africa: a population dynamic modelling approach (thesis).. Liverpool: Liverpool School of Tropical Medicine, 2001:.
94. Onori E, Grab B. Indicators for the forecasting of malaria epidemics.. Bull World Health Organ 1980; 58: 91-98. MEDLINE
Back to top
<!--end tail-->Affiliations
a. Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, UK
b. Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, London, UK
c. Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, London, UK
d. Department of Dean's Office, London School of Hygiene and Tropical Medicine, London, UK
Correspondence to: Sari Kovats, Centre on Global Change and Health, Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London WC1E 7HT, UK
</TD></TR></TBODY></TABLE><!--start eln:enhanced-links=--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--end eln:enhanced-links--><!--start hsp:referedArticles=--><!--start hsp:summaryCollection=--><!--end hsp:summaryCollection--><!--end hsp:referedArticles-->
DOI:10.1016/S0140-6736(03)14695-8
Review
El Ni?o and health
R Sari KovatsMSc
a , Menno J BoumaMD b, Shakoor HajatPhD a, Eve WorrallPhD c and Prof Andy HainesMD dSummary
El Ni?o Southern Oscillation
El Ni?o and weather disasters
Infectious diseases
Problems with interpretation
Specific infections
Interventions: application of climate-disease associations
Conclusions
Search strategy and selection criteria
References
Summary
El Ni?o Southern Oscillation (ENSO) is a climate event that originates in the Pacific Ocean but has wide-ranging consequences for weather around the world, and is especially associated with droughts and floods. The irregular occurrence of El Ni?o and La Ni?a events has implications for public health. On a global scale, the human effect of natural disasters increases during El Ni?o. The effect of ENSO on cholera risk in Bangladesh, and malaria epidemics in parts of South Asia and South America has been well established. The strongest evidence for an association between ENSO and disease is provided by time-series analysis with data series that include more than one event. Evidence for ENSO's effect on other mosquito-borne and rodent-borne diseases is weaker than that for malaria and cholera. Health planners are used to dealing with spatial risk concepts but have little experience with temporal risk management. ENSO and seasonal climate forecasts might offer the opportunity to target scarce resources for epidemic control and disaster preparedness.
Published online May 20, 2003 http://image.thelancet.com/extras/02art5336web.pdf
Back to top
El Ni?o events have occurred for millennia but were probably first recognised in the late 19th century in Peru1 (panel 1). The name El Ni?o derives from the appearance of warm water off the coast of Peru and Ecuador, which was most noticeable around Christmas (El Ni?o meaning ?little boy? refers to the infant Jesus). From time to time, the warming is anomalous (ie, it exceeds expected variation) and persists for 12?18 months, severely disrupting local fish and bird populations. El Ni?o is consistently associated with heavy rainfall and flooding on the west coast of South America.2
Panel 1: El Ni?o events since 1899 Strong events are indicated
Click to view table
El Ni?o Southern Oscillation
Differences in air pressure across the Pacific basin were first identified by Gilbert Walker in the early 1900s as a contributor to monsoon rainfall in India, and an influence on world weather.3 The fluctuation in pressure difference between Darwin, Australia, and Tahiti is known as the Southern Oscillation. Only as recently as the 1960s were El Ni?o and the Southern Oscillation linked and identified as oceanographic and atmospheric components of the same phenomenon?ie, the El Ni?o Southern Oscillation (ENSO). The Southern Oscillation Index (SOI) is generally negative during an El Ni?o, or warm, event, and positive during ENSO's other extreme, the La Ni?a, or cold, event.
As a result of changes in global atmospheric circulation, ENSO events are accompanied by changes in storm activity, and effects on local climate are observed far from the Pacific region, a process known as teleconnection. Temperatures rise globally during El Ni?o by an average of 0?5?C. Precipitation anomalies, however, are less homogeneous: rainfall increases in some regions and decreases in others.4
Furthermore, El Ni?o and La Ni?a usually produce opposite anomalies. The patterns shown in figure 1 are constructed from averages of many El Ni?o events and mask the large variability between events with respect to intensity, duration, and geographical distribution of climate anomalies. Associations with drought are well described in North East Brazil, Southern Africa, South Asia, Indonesia, and Northern Australia.
Click to enlarge image
Figure 1. ENSO teleconnections and risk map for malaria Risk areas for drought and rainfall based on teleconnections associated with El Ni?o. M shows areas where there is a risk of epidemic malaria after the onset of an El Ni?o event.
The effect of global climate change on the future frequency, amplitude, or both, of El Ni?o is uncertain,5,6 but there are concerns that events might become more frequent or more intense.7 However, even with little or no augmentation, climate change is likely to lead to greater extremes of dry weather and heavy rainfall, increasing the risk of drought and flood that occur with El Ni?o in many regions.5
The effect of ENSO on crop production and weather disasters is such that global financial markets and the insurance industry now take seasonal climate forecasts into account.8 For example, seasonal climate forecasts of 3?6 months are now used by farmers in Australia and South America to plan for the planting of drought resistant crops when drier El Ni?o conditions are anticipated. In the health sector, application of ENSO and climate forecasts has lagged behind. Identification of causal associations between climate and disease, and the translation of these into a coherent public health policy remains a major challenge.
Previously, reports have been published about the effects of El Ni?o on mosquito-borne diseases9 including malaria10 and dengue,11 natural disasters,12 and diarrhoeal diseases.13 The improving ability of agencies to predict El Ni?o events and associated weather anomalies has raised the prospect of temporal risk assessment as a guide for public health policy and practice.14
Back to top
El Ni?o and weather disasters
ENSO is the most important climatic cycle that contributes to year-to-year variability in weather and the likelihood of extreme weather events such as heavy rainfall, droughts, and storms. Natural disasters have widespread implications for public health, and they interfere with the continuity of health care through damage to infrastructure, or because of shifting medical and political priorities.15 Some infectious diseases might be aggravated by malnutrition,16 and famine conditions are often associated with human migration; both factors might facilitate the spread of infectious diseases (figure 2).17
Click to enlarge image
Figure 2. Potential health effects of drought in developing countries
Worldwide, drought is twice as frequent in the year after the onset of El Ni?o than in other years.18 The risk is concentrated in Southern Africa and South Asia. However, disasters do not always occur during El Ni?o; in fact, there is much variability in climatic anomalies between events. For example, in 1997?98, the anticipated drought did not happen in the southern African region and some areas received above average rainfall.19 Famine was averted after the severe drought in South Africa in 1992, despite crop failure rates of 80% in some regions.20 Regional cooperation and external aid allowed the purchase and distribution of enough cereals to avert disaster.
The effect of El Ni?o on disasters is strong enough to be apparent at a global level.12 In an average El Ni?o year, around 35 per 1000 people are affected by a natural disaster?more than four times that in non-El Ni?o years, based on 1963?93 data. This difference in risk is much more pronounced for famine. El Ni?o's global disaster footprint is largely determined by the consequences of drought.12
Some major famines have been linked to El Ni?o: the event of 1876?78 was associated with ?the most destructive drought the world has ever known?21 in China, India, South Africa, Egypt, Ethiopia, Sudan, Java, and Brazil. In India, at least 7 million people died despite the presence of a modern railroad network and millions of tonnes of grain in commercial circulation. British imperial policies resulted in the relief reaching only one-tenth of those whose lives were threatened by food shortages.
Drought, in association with slash-and-burn methods of land clearance, can trigger uncontrolled forest fires. Every El Ni?o since at least 1982 has been associated with fires in Kalimantan, which have consequences for public health.22 Smoke from the 1997 forest fires on the Indonesian island groups of Kalimantan and Sumatra affected surrounding areas including Malaysia, Singapore, Philippines, and southern Thailand. Smoke from biomass burning contains pollutants harmful to health, including particulates (particles less than 2?5 μm in diameter that can penetrate human lungs).23
The relation between El Ni?o and intense rainfall is strong in many areas (figure 1). During the 1982?83 and 1997?98 events, intense rain and floods caused hundreds of deaths in Peru, Colombia, Ecuador, and Bolivia.19,24 Deaths associated with floods are also strongly associated with SOI in parts of Australia.25 On a global scale, ENSO is not associated with risk of flood-related disasters because floods are very localised and the risk is heightened during both El Ni?o and La Ni?a phases in different parts of the world.18
Hurricanes in the Caribbean, the Gulf of Mexico, and off the coast of northern Australia are less common than usual during El Ni?o, but more common during La Ni?a. However, typhoons are more likely to occur near the Marshall Islands, in the Pacific Ocean, during an El Ni?o26 event than at other times because storm tracks in the Pacific are shifted to the west during this time. For small islands that lie in their path, the shifting of storm tracks is of particular importance.
Seasonal climate forecasts are now used to mitigate the effects of drought and flood that are associated with ENSO events (see further reading).27 Climate forecasts are combined with other indicators (such as satellite data and food prices) to provide early warning of famine.14,28,29 El Ni?o forecasts could provide decision makers with the earliest possible warning of natural disasters linked to flood and drought.2
Back to top
Infectious diseases
Transmission of many infectious diseases can be affected by weather, especially for those pathogens that spend part of their lifecycle outside the human body. Pathogens carried by insects are exposed, with their flying hosts, to ambient weather. The transmission of vector-borne diseases typically occurs within seasonal patterns, in which the role of temperature and rainfall is well documented. Some vector-borne diseases display much year-to-year variation that can also be partly explained by meteorological factors. The ability to predict high or low transmission seasons would help target the timing and location of public health interventions.
We review studies that have identified associations between climate and disease risk based on El Ni?o or La Ni?a. Evidence for an association between disease risk and ENSO is more robust when analyses use a long time-series that incorporates more than one event and when there is appropriate geographical aggregation of data. Individual outbreaks of disease can be triggered by extreme weather. Such outbreaks are often attributed to ENSO if the weather pattern is consistent with the effects of this climatic event.30,31 However, in our opinion a true association between ENSO and disease in a given population can be confirmed only through analysis of several ENSO events with time series methods. Non-climatic explanations for an association should always be considered, although it is unlikely that environmental factors, such as vector control and changes in case detection, would vary within the same time patterns as ENSO.
We have identified 21 reports that quantify a relation between ENSO and human infectious disease in more than 18 countries, and which have used data series that incorporate more than one event (table).11,32?51 Most of these studies noted a significant association between disease and ENSO. A few used data from geographical areas where ENSO has little or no consistent effect on the weather: for example, malaria and Rift Valley fever in Kenya,36,41 and dengue fever in Bangkok.36 Workers have investigated a range of diseases in several regions, but few studies overlap information for the same areas. Consistent findings have been noted for an association between ENSO and malaria in the coastal regions of Venezuela and Colombia, and for evidence of an effect on cholera transmission in Bangladesh.
Click to view table
Table. Time series studies of ENSO and infectious disease
Back to top
Problems with interpretation
Several ENSO parameters have been used. The SOI shows more variability than do Pacific sea surface temperatures. The use of El Ni?o year as a time variable causes difficulty because El Ni?o does not run to the calendar year, and there is no official definition of what constitutes an El Ni?o event. To model the time-series data, some investigators have used spectral analysis to identify regular cycles.36,52
Measurements connected by time or location are probably correlated and not independent. Serial correlation (autocorrelation) refers to two adjacent observations being more alike than two randomly chosen observations.53 Autocorrelation should be accounted for before independence of variables can be assumed, but many studies that assess the relationship between ENSO and disease have not done so. The inclusion of autocorrelation terms in a regression model is thought to produce a more conservative estimate of the effect and reduced standard error45,54 than the one arrived at without accounting for autocorrelation. Future studies should report findings with and without adjustment (panel 2).
Panel 2: Guidelines for assessment and reporting of interactions between ENSO and health
?Test and report results of association between weather parameters and ENSO parameter in the data
?Report published assessments of ENSO teleconnections by climatologists in region of interest
?Describe the geographical area from which the health data are derived.
?Test and report results of association between weather parameters and disease outcome
?Use time series data with more than one ENSO event
?Remove any trend and regular seasonal patterns in the time-series data before assessing relationships
?Report associations both with and without adjustment for serial correlation
Both intrinsic factors (eg, changes in population immunity) and extrinsic factors (eg, climate variables) can affect the timing of disease epidemics. Such factors are not mutually exclusive and epidemics are caused by a complex interaction, with the balance of components varying between disease systems.55 Models of transmission dynamics that rely on population immunity have been developed for directly transmitted diseases such as measles.56 However, there is no good empirical evidence to show that changes in population immunity can account for malaria epidemics over periods of 3?5 years, although this possibility has been suggested by several authors.36,57 The El Ni?o cycle is irregular and varies in length from 2?7 years (panel 1). The extent to which a specific disease system is being driven solely by the replacement rate of people without immunity within a population should be assessed. The contribution of this mechanism has not yet been quantified for either dengue or malaria. Analysis of monthly cholera incidence in Bangladesh found a role for intrinsic factors (such as previous disease incidence) but also extrinsic ones such as ENSO in the dynamics of cholera transmission.48
Many El Ni?o time-series studies use aggregated national data. However, analysis of smaller geographical locations could help understanding of complex relations between outcome and local drivers such as temperature and rainfall.58 The association between climate variables (temperature, rainfall) and disease should be evaluated since these variables are the principal drivers of the biological processes by which ENSO affects health;59 however, few studies report such analyses. For example, Bouma and Dye38 investigated the complex association between rainfall and malaria epidemic years in Venezuela. Similarly, Pascual and colleagues48 explored the relation between cholera and ENSO, and cholera and local climate factors. Poveda and colleagues33 assessed the effects of El Ni?o on the annual cycle of both malaria and climatic indices, and showed that the association between malaria and climate was intensified during El Ni?o phases. The relation between local climate and ENSO should also be clearly reported and refer to published assessments by climatologists about teleconnections in the region of interest.
Back to top
Specific infections
Malaria
Public health ministries and institutions in countries where malaria transmission rates vary a lot between years, such as those in Colombia and Gujarat, India, have begun to appreciate the usefulness of forecasts. After a steep rise in malaria in Colombia in 1998, the National Public Health Surveillance System, Colombian Ministry of Health has reported the accuracy of earlier published malaria forecasts.60
Malaria epidemics occur in regions where transmission rates are not usually sufficient to provide protective immunity within the population. In some of these epidemic fringe regions, malaria transmission is restricted by climate?ie, conditions are either too dry, too wet, or too cold for vectors or parasites. Therefore, small changes in environmental conditions can trigger an epidemic. The latitudinal edges of malaria distribution are usually determined by the effectiveness of public health defences rather than climate.61
Malaria, through its variety of vectors and ecological conditions, has unique features in each location. Short-term atypical climate conditions (such as rainfall in arid regions and drought in humid climates) could cause epidemics. Therefore, generalisation of the effects of ENSO on epidemic malaria is not possible. There is evidence that in highland areas, raised temperatures associated with El Ni?o, especially during the autumn and winter months, might increase transmission of malaria. This effect has been shown in Northern Pakistan in 1981?91.62,37 Higher than usual temperatures and heavy rainfall have also been associated with short-term increases in highland malaria in Rwanda63 and Uganda.30 Conversely, increased rainfall in a highland area of Tanzania during the El Ni?o event of 1997 probably washed away breeding sites and lowered the number of malaria cases.64
In many desert fringe regions, rainfall and malaria transmission rates are often connected?eg, in the Punjab before 194035,65 and currently in Rajasthan bordering the Thar desert.66 Rainfall in this region is associated with ENSO,4 thus there is an increased epidemic risk during post El Ni?o and La Ni?a years.67
El Ni?o-related droughts have been associated with malaria outbreaks in Sri Lanka,35 Colombia,33 and Irian Jaya68 (figure 1). Epidemics can follow drought in very humid regions, when river flow decreases sufficiently to allow mosquito breeding.69 El Ni?o-related droughts could increase the risk of an epidemic by increasing population mobility, because non-immune people come into contact with infected populations who have moved in search of food.17,68
Post-drought malaria epidemics are associated with El Ni?o in Venezuela, particularly in coastal regions.38 When aquatic ecosystems are re-established after dry years, vector populations can increase to higher than usual numbers because predators of larvae have been reduced. Periodic droughts in coastal Venezuela have been associated with prominent changes in the vector distribution.70
Epidemic preparedness has an early precedent for malaria. In the Punjab between 1922 and 1947, the time window between rainfall and malaria was used to distribute the limited supply of quinine to areas most at risk. Workers estimated risk using rainfall, low recent malaria exposure (judged by low spleen rates) and high food prices, which can indicate lower nutritional status and immunity. This early warning system was used until the introduction of residual insecticides.57 Reduced reliance on residual insecticides and the recurrence of epidemic malaria raises the possibility of reintroduction of epidemic forecasting in parts of India.62,67,71 Epidemic early warning systems typically rely on various indicators that have different lead times.55 Important developments have been made in mapping the regions and timing of malaria risk, including the use of satellite data to provide proxy ecological variables, such as rainfall estimates or vegetation indices.72,73 In combination with climate forecasts, such techniques could provide a valuable tool for epidemic preparedness.14
Mosquito-borne viral diseases
Several studies have found an association between epidemic dengue and ENSO in populations in southeast Asia, South America, and the Pacific (table).31,39,40 However, links between climate, weather, and dengue are poorly understood because the disease is transmitted by container-breeding mosquitoes in urban areas. If climate is an important influence on current distribution of disease or vectors, then this can be considered as supporting evidence that the disease could be affected by past or future trends in climate.74 The global distribution of dengue has been mapped with models that use climate factors;75,76 however, these methods cannot provide insight into the drivers of changes from year to year in disease in a given location. Dengue haemorrhagic fever (DHF) has a tendency to recur in 2?5 year cycles and changes in dengue immunity seem to be important in predisposing a population to outbreaks.36 However, population immunity remains poorly characterised for dengue and DHF.
ENSO usually has a strong effect on the weather in parts of Australia and the relation between ENSO and indigenous arboviruses has been investigated. Infrequent but severe epidemics of Murray Valley encephalitis (Australian encephalitis) have arisen in Australia after above average rainfall and flooding associated with La Ni?a episodes.77,78 Some outbreaks of Ross River virus disease (epidemic polyarthritis) might be linked to weather patterns associated with ENSO, but associations with climate factors are highly localised and might not be detectable with aggregated data (table).79 Epidemics of Rift Valley fever in the dry grasslands in East Africa are triggered by heavy rainfall.80,81 However, no association between this disease in Kenya and SOI was noted80 and there is no consistent association between El Ni?o and rainfall variability in Kenya.
Rodent-borne disease
Rodents are reservoirs for several diseases and their numbers tend to increase after mild wet winters. Human cases of plague in New Mexico are more likely to arise following winter-spring seasons with above average rainfall.51 The emergence of hantavirus pulmonary syndrome in 1993 in southern USA was associated with an increase in the size of the local rodent population. Increased rainfall associated with the El Ni?o event was followed by drought conditions that increased rodent to rodent and rodent-human interactions.82,83 During and after the 1997?98 El Ni?o, the rodent population increased ten-fold to 20-fold, and reported cases of hantavirus pulmonary syndrome increased five-fold, despite strong public awareness about how to reduce exposure.84 This evidence suggests a mechanism by which climate factors affect hantavirus transmission in this region; however, a consistent association with ENSO has not been established.85
Diarrhoeal diseases and cholera
Heavy rainfall has been associated with an increase in outbreaks of enteric pathogens, usually as a result of a contamination of water supplies. In tropical regions, diarrhoeal diseases typically peak during the rainy season.86 Temperature is important in the seasonal and between-year variability of diarrhoeal diseases. Higher than usual daily temperatures during the warm event of 1997, adjusted for seasonal trend, were related to an increased number of admissions for diarrhoea in Peruvian children.13
The importance of temperature and other environmental factors in the epidemiology of cholera has been suspected since the 19th century.52,87 The bimodal seasonal pattern of cholera in Bangladesh mirrors sea surface temperatures in the Bay of Bengal and seasonal plankton abundance (a possible environmental reservoir of the cholera pathogen, Vibrio cholerae).47,88,89 Variability in cholera incidence between years in Dhaka, Bangladesh, between 1980 and 1998 was associated with ENSO.48 An analysis of historical data for Bengal (1890?1940) indicates that any effect of El Ni?o was confined to the coastal regions, where it might have triggered spring epidemics of the disease, which is a shift away from the usual seasonal pattern.47 The relation between cholera and SOI in Dhaka seems to have changed over time, becoming stronger in the last two decades.49 Some studies have found correlations between cholera incidence in Bangladesh and sea surface temperatures in the Bay of Bengal,52,47 which are also affected by ENSO. Climate factors, such as water temperature, could drive seasonality by their direct influence on the abundance or toxicity of V cholerae. The possible mechanisms by which increased sea surface temperatures affect disease transmission from year to year remain poorly understood.
Back to top
Interventions: application of climate-disease associations
Seasonal forecasts are used to predict major climate trends for the next few months to few seasons. They indicate areas where there is an increased probability of some deviation from the climatic mean, such as wet or dry, warm or cold conditions. Seasonal forecasts are issued on a regular basis for each season, but their accuracy is greater during ENSO events than at other times. The science of forecasting is developing rapidly and quite successful seasonal forecasts were issued for the first time during the El Ni?o of 1997?98.19Figure 3 shows various stages of forecasting with respect to public health. The forecast of an El Ni?o event per se has the longest lead time, but has limited geographical information. Prediction of an event is extremely difficult, but once the onset is confirmed, confidence in forecasts is increased and lead times of several months can be used to plan public health initiatives. Because of the differences in forecasts generated by various agencies, the NOAA Office of Global Programs and other agencies convene regional climate outlook fora that produce a consensus forecast and response plans for particular regions.
Click to enlarge image
Figure 3. ENSO and forward planning in public health As the accuracy of the forecast increases, the lead time and window of opportunity decreases.
Forecast of the risk of an epidemic is an essential component of epidemic control.90,91 Expensive control measures are sometimes implemented when transmission has already been naturally interrupted as a means of reassuring the population and showing political commitment.90 Such actions are unlikely to be cost effective. Seasonal climate and malaria forecasts on the other hand allow early intervention, which can mitigate effects of epidemics and improve the cost-effectiveness of control activities. To achieve these ends, forecasts must not only be accurate but they must also be trusted and used in a timely way.
Eight southern African countries now routinely participate in the Southern African Regional Climate Outlook Forum (SARCOF) to develop malaria forecasts specific to each region and country.92 As a direct result of these forecasts, several preparedness activities are carried out in southern Africa, including the establishment of epidemic stores, issuing of epidemic funds, identification of epidemic teams, and development of checklists. At a national level, epidemic plans, guidelines, and checklists have been developed and stocks, personnel, and funds are allocated. These activities are designed to strengthen and speed up response capacity once epidemics are identified.92
Decision-makers are often reluctant to invest resources in response to a forecast, when there might be more immediate demands on scarce resources. The potential economic benefit of malaria forecast has been simulated for one southern African country.93 The cost per malaria case prevented by residual spraying could be reduced by 25?56% (dependent on the severity of the transmission season) simply by bringing forward the spray programme by 3?4 months, to before the onset of the rains. Further improvements in cost-effectiveness could be achieved by use of a reliable seasonal climate forecast to determine the most appropriate level of activity each year, with most intense activity being carried out in high transmission years and least in low transmission years. Decision makers should carefully weigh these potential efficiency improvements against the risks and consequences of following incorrect forecasts. False positives could result in resources being wasted, while false negatives will represent a missed opportunity for malaria control. Both will result in a loss of confidence in malaria forecasts.
Back to top
Conclusions
The major effect of El Ni?o on health and society is mediated through an increased risk of natural disaster such as droughts, floods (figure 4), and tropical cyclones. There is also good epidemiological evidence that El Ni?o is associated with an increased risk of certain diseases in specific geographical areas where climate anomalies are linked with the ENSO cycle. The associations are particularly strong for malaria and cholera in some parts of the world, but only suggestive for other mosquito-borne and rodent-borne diseases. More research is needed to determine the mechanisms of these associations. We recommend that future studies should use time series approaches with robust statistical methods (panel 2). Additional research into the ecology of vectors, reservoirs, and pathogens are needed before the causal pathways from ENSO and climate variability to disease prevalence can be understood.
Click to enlarge image
Figure 4. Floods in Peru caused by heavy rains attributed to El Ni?o, 1998
In areas where El Ni?o can reliably be associated with regional or local climate variations such as droughts and floods, El Ni?o forecasts can provide decision makers with the earliest possible warning of an increased risk of such adverse climate conditions. Seasonal forecasts have already been included into many local and regional famine and drought early warning systems. The need to define predictive factors on which to base forecasts of epidemic risk94 is an enormous challenge for public health agencies working at the central and local level. Preparedness for epidemics is one of the focal points of WHO's Roll Back Malaria programme.14 Health planners are used to dealing with spatial risk concepts, but there is a lack of experience with temporal risk management. Furthermore, many developing countries lack the infrastructure for a prospective operational disease-forecasting system. Improved surveillance of diseases and health outcomes that seem to be influenced by the ENSO phenomenon will provide better quality data for research and enhance attempts to prevent adverse effects. The science of climate forecasting is developing rapidly. Health planners should be aware of similar advances in epidemic forecasting and use the best available evidence to apply scarce resources to reduce disease risk.
Back to top
Search strategy and selection criteria
We searched Medline, EMBASE, BIDS, and Web of Science using the search terms ENSO, El Ni?o, teleconnection, La Ni?a, and SOI. We searched for articles published between 1980 and April, 2002 in all languages, except with the terms El Ni?o and La Ni?a, to exclude Spanish language studies, because these terms resulted in a large number of articles about child health. We also used reference lists to identify additional articles, and we contacted authors who had published work on El Ni?o and health. We included articles that
?were published in peer reviewed journals
?were original research articles using epidemiological data.
?quantified any association with an ENSO parameter (eg, El Ni?o year, sea surface temperature, SOI or other index).
?had an outcome that was an infectious disease in human beings.
?had time series data that included more than one El Ni?o event.
We included 21 articles that met these criteria. Further information about the search strategy can be obtained from the authors.
Back to top
Conflict of interest statement
None declared.
Acknowledgments
Menno Bouma is partly funded by NOAA Office of Global Programs and DFID malaria programme. Eve Worrall was previously funded by the Liverpool Malaria Knowledge Programme. The authors would also like to thank the following for helpful comments: David Bradley, Eleanor Riley, Simon Hales.
Back to top
<!--start tail=-->References
1. Glantz MH. Currents of change: El Ni?o's impact on climate and society. Cambridge: Cambridge University Press, 1996:.
2. In: , Glantz MH, ed. Once burned, twice shy? Lessons learned from the 1997/98 El Ni?o. Tokyo: United Nations University Press, 2001:.
3. Walker GT, Bliss EW. World weather V. Mem R Met Soc 1932; 4: 53-84.
4. Ropelewski CF, Halpert MS. Global and regional scale precipitation patterns associated with the El Ni?o/Southern Oscillation. Monthly Weather Review 1987; 115: 1606-1626.
5. IPCC. Climate change 2001: the scientific basis. Contribution of Working Group I to the Second Assessment Report of the Intergovernmental Panel on Climate Change. New York: Cambridge University Press, 2001:.
6. Vinnikov KY, Robock A. Trends in moments of climatic indices.. Geophys Res Lett 2002; 29 (2): 10.
7. Timmerman A, Oberhuber J, Bacher A, Esch M, Latif M, Roeckner E. Increased El Ni?o frequency in a climate model forced by future greenhouse warming. Nature 1999; 398: 694. CrossRef
8. Workshop on applications of seasonal forecasting, 14?15 June 1999. Reading: European Centre of Medium-Range Weather Forecasts (ECMWF), 1999:.
9. Nicholls N. El Ni?o-Southern Oscillation and vector-borne disease. Lancet 1993; 342: 1284-1285. MEDLINE | CrossRef
10. Bouma MJ, Sondorp HE, van der Kaay HJ. Health and climate change. Lancet 1994; 343: 303. MEDLINE
11. Hales S, Weinstein P, Woodward A. Dengue fever epidemics in the South Pacific region: driven by El Ni?o Southern Oscillation?. Lancet 1996; 348: 1664-1665. Full Text | Full-Text PDF (19 KB) | MEDLINE | CrossRef
12. Bouma MJ, Kovats RS, Goubet SA, Cox JS, Haines A. Global assessment of El Ni?o's disaster burden. Lancet 1997; 350: 1435-1438. Abstract | Full Text | Full-Text PDF (70 KB) | MEDLINE | CrossRef
13. Checkley W, Epstein LD, Gilman RH, et al. Effects of El Ni?o and ambient temperature on hospital admissions for diarrhoeal diseases in Peruvian children. Lancet 2000; 355: 442-450. Abstract | Full Text | Full-Text PDF (443 KB) | MEDLINE | CrossRef
14. WHO. Malaria early warning systems, concepts, indicators and partners: a framework for field research in Africa.Roll back malaria/technical support network for prevention and control of malaria.. Geneva: World Health Organization, 2001:.
15. Noji EN. The public health consequences of disasters. New York: Oxford University Press, 1997:.
16. Zurbrigg S. Re-thinking the ?human factor? in malaria mortality. Parassitologia 1994; 36: 121-135. MEDLINE
17. Prothero RM. Forced movements of population and health hazards in tropical Africa. Int J Epidemiol 1994; 23: 657-664. MEDLINE
18. Dilley M, Heyman B. ENSO and disaster: droughts, floods, and El Ni?o/Southern Oscillation warm events. Disasters 1995; 19: 181-193. MEDLINE
19. WMO/UNESCO/UNEP/ICSU. The 1997?98 El Ni?o: a scientific and technical retrospective. Geneva: WMO, 1999:WMO No 905.
20. Yip R. Famine In: , Noji EN, ed. The public health consequences of disasters.. New York: Oxford University Press, 1997: 305-335.
21. Davis M. Late Victorian holocausts: El Ni?o famines and the making of the third world.. London: Verso Books, 2000:.
22. Nicholls N. ENSO, drought and flooding rain in south east Asia. In: Brookfield H, Byron Y, eds. South-east Asia's environment future: the search for sustainability.. Tokyo: United Nations University Press/OUP, 1993: 154-175.
23. WHO. Health guidelines for vegetation fire events: background papers. Singapore: Institute of Environmental Epidemiology, 1999:.
24. Cronica de Desastres: Fenomeno El Ni?o 1997?1998. Washington DC: PAHO, 2000:.
25. Kuhnel I, Coates L. El Ni?o southern oscillation: related probabilities of fatalities for natural perils in Australia. Natural Hazards 2000; 22: 117-138. CrossRef
26. Spennemann DH, Marschner IC. The association between El Ni?o/Southern Oscillation events and typhoons in the Marshall Islands. Disasters 1995; 19: 194-247. MEDLINE
27. Murphy SJ, Washington R, Downing TE, et al. Seasonal forecasting for climate hazards: prospects and responses. Natural Hazards 2000; 23: 171-196. CrossRef
28. Buchanon-Smith M, Davies S. Famine early warning and response?the missing link.. London: Intermediate Technology, 1996:.
29. Kovats RS, Bouma MJ, Haines A. El Ni?o and health.. Geneva: World Health Organization, 1999:.
30. Lindblade KA, Walker ED, Onapa AW, Katungu J, Wilson ML. Highland malaria in Uganda: prospective analysis of an epidemic associated with El Ni?o. Trans R Soc Trop Med Hyg 1999; 93: 480-487. MEDLINE | CrossRef
31. Corwin AL, Larasati RP, Bangs MJ, et al. Epidemic dengue transmission in southern Sumatra, Indonesia. Trans R Soc Trop Med Hyg 2001; 95: 257-265. MEDLINE | CrossRef
32. Gagnon AS, Smoyer-Tomic KE, Bush A. The El Ni?o southern oscillation and malaria epidemics in South America. Int J Biometeorol 2002; 46: 81-89.
33. Poveda G, Rojas W, Quinones M, et al. Coupling between annual and ENSO timescales in the malaria climate association on Colombia. Env Hlth Pers 2001; 109: 307-324.
34. Bouma MJ, Poveda G, Rojas W, et al. Predicting high-risk years for malaria in Colombia using parameters of El Ni?o southern oscillation. Trop Med Int Hlth 1997; 2: 1122-1127.
35. Bouma MJ, van der Kaay HJ. The El Ni?o Southern Oscillation and the historic malaria epidemics on the Indian subcontinent and Sri Lanka: an early warning system. Trop Med Int Hlth 1996; 1: 86-96.
36. Hay SI, Myers MF, Burke DS, et al. Etiology of interepidemic periods of mosquito-borne disease. Proc Natl Acad Sci 2000; 97: 9335-9339. MEDLINE | CrossRef
37. Bouma MJ. Epidemiology and control of malaria in northern Pakistan.. Dordrecht: ICG Printing, 1995:.
38. Bouma MJ, Dye C. Cycles of malaria associated with El Ni?o in Venezuela. JAMA 1997; 278: 1772-1774. MEDLINE
39. Gagnon AS, Bush ABG, Smoyer-Tomic KE. Dengue epidemics and the El Ni?o Southern Oscillation. Clim Res 2001; 19: 35-43.
40. Hales S, Wienstein P, Souares Y, Woodward A. El Ni?o and the dynamics of vectorborne disease transmission. Env Hlth Pers 1999; 107: 99-102.
41. Anyamba A, Linthicum KJ, Tucker CJ. Climate-disease connections: rift valley fever in Kenya. Cad Saude Publica 2001; 17: 133-140.
42. Tong S, Bi P, Parton K, Hobbs J, McMichael AJ. Climate variability and transmission of epidemic polyarthritis. Lancet 1998; 351: 1100. Full Text | Full-Text PDF (50 KB) | MEDLINE | CrossRef
43. Harley DO, Weinstein P. The southern oscillation index and Ross River virus outbreaks. Med J Aust 1996; 165: 531-532. MEDLINE
44. Bi P, Tong S, Donald K, Parton K, Hobbs J. Southern oscillation index and transmission of the Barmah Forest virus infection in Queensland, Australia. J Epidemiol Community Health 2000; 54: 69-70. MEDLINE | CrossRef
45. Maelzer D, Hales S, Weinstein P, Zalucki M, Woodward A. El Ni?o and arboviral disease prediction. Env Hlth Pers 1999; 107: 817-818.
46. Franke CR, Ziller M, Staubach C, Latif M. Impact of the El Ni?o southern oscillation on visceral leishmaniasis, Brazil. Emerg Infect Dis 2002; 8: 914-917. MEDLINE
47. Bouma MJ, Pascual M. Seasonal and interannual cycles of endemic cholera in Bengal 1891?1940 in relation to climate and geography. Hydrobiologia 2001; 460: 147-156. CrossRef
48. Pascual M, Rodo X, Ellner SP, Colwell R, Bouma MJ. Cholera dynamics and El Ni?o southern oscillation. Science 2000; 289: 1766-1767. MEDLINE | CrossRef
49. Rodo X, Pascual M, Fuchs G, Faruque ASG. ENSO and cholera: a nonstationary link related to climate change?. Proc Natl Acad Sci 2002; 99: 12901-12906. MEDLINE | CrossRef
50. Ebi KL, Exuzides KA, Lau E, Kelsh M, Barsnton A. Association of normal weather conditions and El Ni?o events with viral pneumonia hospitalizations in females, California, 1983?1998. Am J Public Hlth 2001; 91: 1200-1208.
51. Parmenter RR, Yadav EP, Parmenter CA, Ettestad P, Gage KL. Incidence of plague associated with increased winter-spring precipitation in New Mexico. Am J Trop Med Hyg 1999; 61: 814-821. MEDLINE
52. Pascual M, Bouma MJ, Dobson AP. Cholera and climate: revisiting the quantitative evidence. Microbes Infect 2002; 4: 237-246. MEDLINE | CrossRef
53. Schwartz J, Spix C, Touloumi G, et al. Methodological issues in studies of air pollution and daily counts of death or hospital admissions. J Epidemiol Community Health 1996; 50 (suppl 1): S3-S11.
54. Chatfield C. The analysis of time series: an introduction. London: Chapman and Hall, 1989:.
55. Myers MF, Rogers DJ, Cox J, Flahault A, Hay SI. Forecasting disease risk for increased epidemic preparedness in public health. Adv Parasitol 2000; 47: 330.
56. Anderson RM, May RM. Population biology of infectious diseases part I. Nature 1979; 280: 361-367. MEDLINE | CrossRef
57. Swaroop S. Forecasting of epidemic malaria in the Punjab. Am J Trop Med 1946; 29: 1-17.
58. Poveda G, Graham NE, Epstein PR, et al. Climate and ENSO variability associated to malaria and dengue fever in Colombia In: Diaz HF, Markgraf F, eds. El Ni?o and the southern oscillation, multiscale variability and global and regional impacts.. Cambridge: Cambridge University Press, 2000: 183-204.
59. Kovats RS. El Ni?o and human health. Bull World Health Organ 2000; 78: 1127-1135. MEDLINE
60. Anon. Sistema Nacional de Vigilancia en Salud Publica: sivgila number 39.. Colombia: Colombian Ministry of Health, 1998:.
61. Reiter P. Global warming and vector-borne disease in temperate regions and at high altitude. Lancet 1998; 351: 839-840. Full Text | Full-Text PDF (46 KB) | MEDLINE | CrossRef
62. Bouma MJ, Dye C, van der Kaay HJ. Falciparum malaria and climate change in the Northwest Frontier Province of Pakistan.. Am J Trop Med Hyg 1996; 55 (2): 131-137. MEDLINE
63. Loevinsohn ME. Climatic warming and increased malaria incidence in Rwanda. Lancet 1994; 343: 714-718. MEDLINE | CrossRef
64. Lindsay SW, Bodker R, Malima R, Msangeni HA, Kisinza W. The effect of 1997?98 El Ni?o on highland malaria in Tanzania. Lancet 2000; 355: 989-990. Abstract | Full Text | Full-Text PDF (61 KB) | MEDLINE | CrossRef
65. Yacob KBM, Swaroop S. Investigations of the long term periodicitiy in the incidence of epidemic malaria in the Punjab. J Malaria Inst India 1946; 6: 273-284.
66. Akhtar R, McMichael AJ. Rainfall and malaria outbreaks in western Rajasthan. Lancet 1996; 348: 1457-1458. Full Text | Full-Text PDF (24 KB) | MEDLINE | CrossRef
67. Bouma MJ, van der Kaay HJ. Epidemic malaria in India and the El Ni?o Southern Oscillation. Lancet 1994; 344: 1638-1639. MEDLINE | CrossRef
68. Bangs M, Subianto DB. El Ni?o and associated outbreaks of severe malaria in highland populations in Irian Jaya, Indonesia: a review and epidemiological perspective. Southeast Asian J Trop Med Public Hlth 1999; 30: 608-619.
69. In: Aron J, Patz JA, eds. Ecosystem change and public health.. Baltimore: Johns Hopkins University Press, 2002:.
70. Gabaldon A. The nation-wide campaign against malaria in Venezuela. Trans R Soc Trop Med Hyg 1949; 43: 113-160. CrossRef
71. Bouma MJ, van der Kaay HJ. Epidemic malaria in India's Thar desert. Lancet 1995; 373: 132-133.
72. Thomson MC, Connor SJ, Milligan PJ, Flasse SP. The ecology of malaria - as seen from Earth-observation satellites. Ann Trop Med Parasitol 1996; 90: 243-264. MEDLINE
73. Hay SI, Snow RW, Rogers DJ. From predicting mosquito habitat to malaria seasons using remotely sensed data: practice, problems, and perspectives.. Parasitology Today 1998; 14: 306-313. MEDLINE | CrossRef
74. Kovats RS, Campbell-Lendrum D, Reid C, Martens P. In: Climate and vector borne disease: an assessment of the role of climate in changing disease patterns.. Maastricht: University of Maastricht, 2000: 50.
75. Hopp M, Foley J. Global-scale relationships between climate and the dengue fever vector, Aedes Aegypti. Climatic change 2001; 48: 441-463. CrossRef
76. Hales S, deWet N, Maindonald J, Woodward A. Potential effect of population and climate changes on golbal distribution of dengue fever: an empircal model.. Lancet 2002; 360: 830-834. Abstract | Full Text | Full-Text PDF (242 KB) | MEDLINE | CrossRef
77. Broom AK, Lindsay MD, Johansen CA, Wright AE, Mackenzie JS. Two possible mechanisms for survival and initiation of Murray Valley encephalitis virus activity in the Kimberley region of Western Australia. Am J Trop Med Hyg 1995; 53: 95-99. MEDLINE
78. Nicholls N. El Ni?o Southern Oscillation impact prediction. Bull Am Meteorol Soc 1988; 69: 173-176.
79. Woodruff RE, Guest CS, Garner MG, et al. Predicting Ross River virus epidemics from regional weather data. Epidemiology 2002; 13: 384-393. MEDLINE | CrossRef
80. Linthicum KJ, Anyamba A, Tucker CJ, Kelley PW, Myers MF, Peters CJ. Climate and satellite indicators to forecast Rift Valley fever epidemics in Kenya. Science 1999; 285: 397-400. MEDLINE | CrossRef
81. Davies FG, Linthicum KJ, James AD. Rainfall and epizootic Rift Valley fever. Bull World Health Organ 1985; 63: 941-943. MEDLINE
82. Wenzel RP. A new hantavirus infection in North America. N Eng J Med 1994; 330: 1004-1005. MEDLINE | CrossRef
83. Engelthaler DM, Mosley DG, Cheek JE, et al. Climatic and environmental patterns associated with hantavirus pulmonary syndrome, four corners region, United States. Emerg Infect Dis 1999; 5: 87-94. MEDLINE
84. Hjelle B, Glass GE. Outbreak of hantavirus infection in the four corners region of the United States in the wake of the 1997?1998 El Ni?o-southern oscillation. J Infect Dis 2000; 181: 1569-1573. MEDLINE | CrossRef
85. Mills JN, Yates TL, Ksiazek TG, Peters CJ, Childs JE. Long-term studies of hantavirus reservoir populations in the southwestern United States: rationale, potential and methods. Emerg Infect Dis 1999; 5: 95-101. MEDLINE
86. Cairncross AM, Feachem RG. Environmental health engineering in the tropics: an introductory text.. Chichester: J Wiley, 1993:.
87. Hirsch A. Handbook of geographical and historical pathology.. London: New Sydenham Society, 1883:.
88. Colwell RR. Global climate and infectious disease: the cholera paradigm. Science 1996; 274: 2025-2031. MEDLINE | CrossRef
89. Lobitz B, Beck L, Huq A, et al. Climate and infectious disease: use of remote sensing for detection of Vibrio cholerae by indirect measurement. Proc Natl Acad Sci 2000; 97: 1438-1443. MEDLINE | CrossRef
90. Najera JA, Kouznetsov RL, Delacollette C. Malaria epidemics; detection and control, forecasting, and prevention.. Geneva: World Health Organization, 1998:.
91. Thomson MC, Palmer T, Morse AP, Cresswell M, Connor SJ. Forecasting disease risk with seasonal climate predictions. Lancet 2000; 355: 1559-1560. Full Text | Full-Text PDF (59 KB) | MEDLINE | CrossRef
92. WHO/AFRO/ICP. Presentation by WHO/AFRO/ICP for Southern Africa. Malaria epidemics in Southern Africa: a review of the progress in malaria epidemic forecasting, detection and response. In: At RBM/TSN for malaria epidemic prevention and control. Geneva: WHO, Dec 2001: 10-11.
93. Worrall E. An economic evaluation of malaria early warning systems in Africa: a population dynamic modelling approach (thesis).. Liverpool: Liverpool School of Tropical Medicine, 2001:.
94. Onori E, Grab B. Indicators for the forecasting of malaria epidemics.. Bull World Health Organ 1980; 58: 91-98. MEDLINE
Back to top
<!--end tail-->Affiliations
a. Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, UK
b. Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, London, UK
c. Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, London, UK
d. Department of Dean's Office, London School of Hygiene and Tropical Medicine, London, UK
Correspondence to: Sari Kovats, Centre on Global Change and Health, Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London WC1E 7HT, UK</TD></TR></TBODY></TABLE><!--start eln:enhanced-links=--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--start eln:ref=--><!--start eln:link=--><!--end eln:link--><!--end eln:ref--><!--end eln:enhanced-links--><!--start hsp:referedArticles=--><!--start hsp:summaryCollection=--><!--end hsp:summaryCollection--><!--end hsp:referedArticles-->
Comment