Modeling current and future influences of climate on infectious disease transmission

Various model-based estimates have been made of how climate change scenarios would affect the future transmissibility (both geographic range and seasonal incidence) of malaria (Lindsay and Birley, 1996; Rogers and Randolph, 2000; Tanser et al., 2003; Kovats et al., 2004; Thomas et al., 2004; van Lieshout et al., 2004). Two contrasting types of models have been used. One is based on known climate-disease relationships from laboratory and local studies. Such models comprise an integrated set of equations that express those relationships mathematically. They are referred to as "biological models". The other type, the "statistical" model, uses an empirical-statistical approach. A statistical equation is derived that expresses the currently observed relationship between the geographic distribution of the disease and local climatic conditions, and then applies that same equation to the specified future climate scenario.

The controversy about the relative roles of climatic conditions, which set spatial-seasonal limits for transmission, and the many non-climatic variables, which (separately or together) may even preclude occurrence of the infectious disease entirely within some locations, has regrettably bred some confusion within the scientific literature. To model how a given scenario of climate change would alter the receptive zone and season limits for some infectious disease is not to model where and when the disease will occur; it is to model where and when it could occur. We cannot, of course, know what the future pattern of infectious disease transmission will be, because we cannot know the future of vaccine technologies, vector controls, public health surveillance strategies, and antimicrobial resistance - nor, more generally, the future impacts of changes in levels of wealth, mobility, social organization, and public literacy about infectious diseases.

Nevertheless, in accord with classic experimental scientific practice, we can sensibly ask the following question: if all non-climatic factors were held constant over coming decades, how would a change in climate alter the potential geographic range and seasonality of infectious disease transmission? Indeed, with the increasing sophistication of both our knowledge base and our modeling capacities, we can incorporate plausible scenarios of how at least some of those non-climatic factors will change in future, and thus estimate the net impact of climate change on potential infectious disease transmission. There has been a second, cruder, confusion: some critics appear to presume that those who publish modeled estimates of future climate-induced changes in the potential transmission of a specified infectious disease therefore also assume that the current pattern of transmission reflects recent climate changes. This criticism has its basis more in politics than in logic, and will not be explored further.

Several modeling studies have projected limited geographical expansion of malaria transmissibility over the next few decades (Rogers and Randolph, 2000; Thomas etal., 2004), while some estimate more extensive changes later this century (Martens et al., 1999; Thomas et al., 2004; van Lieshout et al., 2004). Those studies that have modeled how climate change would affect seasonal changes in transmission project a substantial increase. One study, based on thorough documentation of current malaria occurrence in sub Saharan Africa, estimates that climate change by 2100 would cause a 16-28 percent increase in person-months of exposure to malaria (Tanser et al., 2003).

Dengue fever, the world's most common mosquito-borne viral infectious disease, is also well known to be sensitive to climatic conditions. Various research groups have developed ways of modeling how future changes in climate would be likely to affect the geographic range and seasonality of this disease. As with malaria, it is well understood that many other non-climate factors influence, indeed can preclude, the occurrence of dengue - as is well illustrated by the huge differential in rates between Texas (very low rates) and adjoining Mexico (very high rates). Public health programs of monitoring, mosquito control, and rapid case detection and treatment are important. Holding constant such non-climate factors around the world as at present, statistical modeling indicates a substantial potential for increased geographic spread of dengue in warmer and wetter conditions over the coming century (Hales et al., 2002). In a recent further development, the non-stationary temporal-spatial relationship between El Niño and the spread of dengue in Thailand has been modeled (Cazelles et al., 2005). This study suggests that the El Niño event acts as a "pacemaker," resulting in a point-source "surface ripple" spread of the infectious outbreak.

A recent study in Canada has modeled the impact of projected climate change on the potential geographic extension of Lyme disease in that country, to 2080 (Ogden et al., 2006). The disease, currently confined to the southern extremity of the country, would, approximately, become transmissible throughout much of the southern half of the country by the middle of this century.

There is, in all of this recent modeling of how climate change would affect infectious disease risks within a specified single country, a strong tendency towards an "inverse law." Those countries that have the professional and economic resources to carry out such research are generally countries at relatively low risk, and vice versa. However, a welcome recent development has been the advent of such studies for countries such as India and Zimbabwe. In the Zimbabwe study (Ebi et al., 2005), plausible country-level climate change scenarios were generated and then applied to a mathematical model of how climatic parameters affect malaria transmissibility. The study showed that, with rising average and minimum daily temperatures accompanied by minimum necessary monthly rainfall, the future risk of malaria would progressively extend to higher altitudes. An important corollary here is that even if Zimbabwe were to become very wealthy and socially modernized, it would still cost much more than today to prevent the population's risk (exposure) from rising temperatures each morning.

Legionella pneumophila lives in the water of (evaporative) air-conditioning cooling towers, and is spread by aerosolized droplets. There is therefore the possibility of increased outbreaks of legionellosis with climate change, especially in developed countries that are becoming increasingly dependent on air-conditioning to cool both private and public buildings.

As noted earlier, weather disasters may also affect outbreaks of infectious diseases. One important manifestation of climate change is a change in climatic variability. Hence, regional patterns of extreme weather events are expected to alter as climate change proceeds. Following Hurricane Mitch in 1998, which directly killed 11,000 people in Central America, dramatic increases occurred in rates of cholera, malaria, dengue leptospirosis, and dengue fever - especially in Honduras, with estimates of 30,000 cholera cases, 30,000 malaria cases, and 1000 dengue cases (Epstein, 1999). In similar fashion, extreme flooding in Mozambique in early 2000 caused a surge in malaria cases three months later (see Figure 14.6).

This genre of modeling has, so far, usually not included various non-climate characteristics of the future world that would also affect infectious disease transmission probabilities, since many of those characteristics are not easily foreseen. If the pathogen were not locally present (e.g. because of efficient case surveillance and treatment) or if the vector species had been eliminated (e.g. by mosquito control programs), then the disease could not be transmitted. Future modeling will become more versatile if it can incorporate plausible scenarios (or, better, probabilistic projections) of these non-climatic contextual changes. Nevertheless, estimating how the intrinsic probability of infectious disease transmission would alter in response to climate change alone is itself informative - and, indeed, accords with classical experimental science. It serves to alert us to the range of future potential risks, and it focuses attention on areas that need more attention and research (see Box 14.5, page 398).

-Malaria cases

■Maputo precip

1000 900

es 700 ase 600 S 500

lari 400

Mal 300 M200 100 0

-Malaria cases

■Maputo precip

1 cip re

13 25 37 49 61 73 85 97 109 121 133 145 Weeks (Jan 99-Dec 01)

Figure 14.6 Weekly cases of malaria (dark gray) and association with floods (pale gray) in Maputo, Mozambique, 2000. Reprinted from Milne (2005), with permission.

1 cip re

13 25 37 49 61 73 85 97 109 121 133 145 Weeks (Jan 99-Dec 01)

Figure 14.6 Weekly cases of malaria (dark gray) and association with floods (pale gray) in Maputo, Mozambique, 2000. Reprinted from Milne (2005), with permission.

0 0

Post a comment