Urbanization and the global malaria recession

The range of malaria has contracted through a century of economic development and disease control[1‐3]. During an era of renewed interest in elimination and eradication there is a need to better understand and quantify the forces behind this recession. A variety of direct control efforts were likely a major factor behind the global contraction in malaria transmission[2, 3], although these gains were often coincident with rapid economic and social development and land use changes[4‐6]. One major aspect of this development that has been shown to significantly impact malaria transmission is urbanization[7‐9].

The process of urbanization includes physical landscape modification and transformation of environs through demand for resources and improved communications. Moreover, urbanization involves significant socio-economic change; generally improved health, housing and increased wealth[10, 11]. These factors, common to urban areas, cause marked entomological, parasitological and behavioural effects that result in reduced malaria transmission both within the urban core and surrounding peri-urban areas[7, 9, 12]. A century of rapid urbanization has therefore likely had an impact on malaria transmission globally, but the size and importance of this impact has never been examined.

The recent construction of 20th Century time series of global urban extent data[13], and a contemporary malaria transmission map[14] that is comparable to historic data[15], enables a more detailed exploration of this relationship. Here, these datasets were utilized to examine for the first time the association at global, national and subnational scales between changes in malaria transmission and urban growth across the last century.

A consistent pattern of greater rates of urbanization in areas that either eliminated malaria or displayed the greatest transmission reductions was found. At the time of elimination certification of those countries that achieved malaria elimination, those areas that were originally malarious had significantly higher proportions (Mann-Whitney test: area, z = -3.207, p < 0.01; populations, z = -5.432, p < 0.01) of their populations living in urban areas and higher percentages of urban land area (Figure 2) than the contemporary situation in those areas of the world that remain endemic today. Moreover, the rate of urbanization from 1900 onwards was significantly larger in those countries that achieved elimination than those that remain endemic today (see Additional file8). Those countries that achieved elimination exhibited significantly greater increases in the proportions of their populations (Mann-Whitney test: z = -1.985, p < 0.05) and land area (Mann-Whitney test: z = -3.232, p < 0.01) that became urban during the 20 years prior to elimination, compared to the recent rates of urbanization in currently endemic countries over the last 20 years. Such findings confirm the presence of a correlation between urbanization and the achievement of malaria elimination at national scales, but this relationship is likely to be intimately tied to wealth. Countries with greater wealth have greater malaria control resources at their disposal, and it is those wealthier countries that have succeeded in eliminating malaria, as confirmed by undertaking similar analyses based on average income (average income data from[30], Mann-Whitney test: z = -6.491, p < 0.01). Moreover, those higher income countries that succeeded in elimination are located in latitudinal regions of lower climatic and environmental receptivity to malaria transmission, making elimination likely an easier task with a fixed set of intervention tools. Therefore, untangling the effects of urbanization upon changes in malaria transmission required analysis independent of these economic or latitudinal biases, and these are described below.

Twenty-nine countries were identified for which at least a fifth of their land area became malaria-free over the past century and for which at least a fifth of their land area remains endemic today. Over 75% of these countries had greater proportions of land area urbanized in the malaria-free areas, and showed a greater percentage increase in urban extent (Figure 3) than the areas that remain endemic today. Both of these differences are significant (Wilcoxen test: urban area: z = -2.505 p < 0.05, urban extent change: z = -2.001, p < 0.05). However, these relationships show regional variations, (Figure 3). Almost all countries analysed in the Americas (Figure 3a) and Asia (Figure 3c) showed greater urbanization in their malaria-free versus still-endemic areas. In the Africa and Arabian Peninsula region (Figure 3b), where both levels of urbanization and transmission reductions were smaller, the results were more mixed. Many of those falling below the one-to-one line (Figure 3a-3c) are countries where extremely arid or mountainous conditions have influenced human settlement to occur principally in areas most suitable for malaria transmission (e.g., Saudi Arabia, Botswana, Swaziland). A full set of national level results can be found in the Additional file9. In addition, the results of a repeat analysis for P. vivax yielding very similar results are shown in Additional file10.

Urban areas have been shown to exhibit lower levels of transmission than surrounding rural ones[8, 9]. If the process of urbanization is a causal factor in malaria declines, a consistent pattern of greater transmission reduction in those areas that have undergone urbanization should be seen. Comparing mean changes in transmission class in areas that are urban today, versus those that have remained rural, for malaria-endemic areas in 1900, was possible in 158 countries. It shows that 82.3% underwent a greater transmission reduction in their urban areas than rural ones (Wilcoxen test: p < 0.001, see Additional file10). Of the 28 countries that displayed a greater malaria reduction in rural areas, half of these were in sub-Saharan Africa, where the smallest levels of urbanization and reductions in transmission occurred. This confirms that greater transmission reductions occurred in areas that are urban today, but does not indicate if the largest declines were coincident with greater rates of urbanization.

Finally, examination was undertaken of whether the rate of urbanization was higher in regions of countries that saw the greatest reductions in Pf R_C. Sixty-three countries were identified that had two or more Pf R_C reduction classes, each covering at least 20% of their land area, and the rates of urbanization 1900-2000 in the areas exhibiting the smallest reduction in transmission were compared to those in the areas displaying the greatest reduction in transmission. Some 84.1% of countries displayed a greater urban increase in the areas that showed the greater Pf R_C reduction than those that showed the smallest transmission reductions (Wilcoxen test: p < 0.001, Figure 4). Further supporting analyses and more detail on these analyses are presented in Additional file11.

The process of urbanization results in a variety of changes that reduce receptivity to malaria transmission[9‐11]. Here, a clear picture of increased urbanization associated with greater malaria transmission reductions across countries and continents is documented for the first time. Whilst it remains challenging to untangle whether this urbanization resulted in decreased transmission, or that malaria reductions promoted development, a close relationship is evident, irrespective of national wealth and latitude.

Other local evidence[7, 12, 31, 32] hints that changes commensurate with urbanization play a substantive role in driving malaria transmission declines. Whether measured by proportion of land area or population urbanized, the majority of nations that remain malaria endemic today exhibit substantially lower levels of urbanization compared to that at the time of elimination for those countries that have achieved it (Figure 2 and Additional file8). Moreover, the magnitude of change in urban extent from 1900 is also correlated with malaria declines, with continental differences, notably sub-Saharan Africa showing lower levels of urbanization today, and smaller changes in urban extent and endemicity over the past century (Figure 3).

Recent malaria declines in sub-Saharan Africa point towards the success of interventions, however, in several cases the decline began before specific interventions were deployed, suggesting the contribution of alternative factors[33]. While malaria declines due to urbanization and its effects are likely to be more gradual than some of the sudden drops seen, it remains a possibility that the rapid urbanization ongoing in sub-Saharan Africa[34] is at least a contributory driver to these changes.

While over three-quarters of countries show decreasing transmission in areas that have undergone urbanization over the past century, a handful of countries go against this trend. A possible reason for this is the likely multi-factorial complexity behind both changes in transmission and human settlement dynamics, and the difficulty of attributing changes to a single cause. Recent analyses have indicated that vector-borne and parasitic diseases have systematically impacted economic growth[35], but more detailed studies of these types of relationships across disease types and ecozones are required to gain a fuller understanding. Almost all of the countries that show greater transmission reductions in rural areas are those where human settlement was constrained to the more malarious areas of the country, due to uninhabitable arid or mountainous areas elsewhere. Equally however, the reverse is true for some countries – i.e. uninhabitable areas with intense transmission forced human settlement in the lower transmission regions.

It is clear that various sources of uncertainty exist in the inputs and methodologies used in this study. Uncertainties are inherent in the urbanization datasets through the lack of data for many regions and time periods, and the assumptions made to fill these gaps[13]. The proposed levels of historical endemicity[15, 23] are plausible when triangulated against other values reported from the pre-intervention era (for example,[26, 36]), but the relatively crude categorization of all-cause malaria endemicity strata and the cartographic approach used preclude a more formal quantification of the global P. falciparum endemicity declines and their link to urbanization beyond the broad relationships presented here. Nevertheless, recent mapping of the limits of P. vivax transmission[28] and analyses of contemporary impacts of urbanization[8], indicate very similar effects and contractions as that seen for P. falciparum, with results repeated, where possible, for P. vivax showing almost identical results (Additional file10), and transmission rarely more intense than meso-endemic[37, 38].

Much of the low-income world, where malaria burden is highest and levels of urbanization are lowest, is set to undergo an urban and demographic transition in the coming decades[17, 34], ultimately likely arriving at levels of urbanization similar to those exhibited by countries that eliminated malaria (Additional file8). Significant efforts towards modelling future malaria scenarios have been completed or are underway, focused principally on the effects of a variety of interventions[39‐41] or on climate change scenarios[42‐45], but the impacts of urbanization are rarely considered[46]. Yet, if the past century of malaria declines are indicative, the study of its impacts should receive more attention as nations start to monitor their progress toward elimination[47]. There exist multiple data gaps and uncertainties in obtaining suitable data to build urbanization into scenario models, however. Firstly, simple consistent definitions of what constitutes an urban area in general are difficult to outline. Beyond this, multiple approaches to mapping urban extents in a consistent fashion have been attempted (e.g.,[48‐50]), but the spatial modelling of their future growth is lacking. Secondly, definitions and measures of urbanization that are relevant to understanding transmission patterns are poorly quantified, with only occasional attempts at empirical definitions made[8, 9], and little consideration of the urban preferences of the dominant Anopheles species[51‐53], or adaptation of them to urban environments[54, 55]. Finally, the treatment of urban areas as single homogenous entities ignores the great variations in demographic, socio-economic and land uses within cities with, for example, urban agricultural practices often maintaining transmission within urban areas[56‐58].

The quantification of a global recession in the range and intensity of malaria over the 20th century has allowed us to review the impact that urbanization has had on these declines, and gauge its importance as a driver of future changes in malaria epidemiology. Results highlight for the first time the consistent relationship between urbanization and malaria declines over the past century globally, and point towards continuing declines as urbanization permanently alters the receptivity of many areas to malaria transmission. The findings presented here suggest that these trends will likely continue to catalyze malaria declines on the path to the goal of a malaria-free future.