Drivers of autochthonous malaria cases over time: could the Central European present the African future?

Abstract

Background: Results of spatial and temporal comparison of malaria hotspots and coldspots could improve the health measures of malaria control and eradication strategies. The study aimed to reveal the spatially and temporally independent correlations between the potentially most effective background variables and the number of autochthonous malaria cases.

Methods: Relationships between malaria cases and background variables were studied in 2 km × 2 km sized quadrates (10 Central European and 10 African). In addition to the current habitat structure of the African sites, annual precipitation, and annual mean temperature, data of the above parameters detected in the nineteenth and twentieth centuries and currently in the Central European sites were included in the analyses (n = 40). Mann-Whitney tests, Principal Component Analysis, and Generalized Linear Models were used for the examinations.

Results: In addition to the apparent significant positive correlation of malaria cases with annual rainfall and mean temperature, several correlations were found for habitat parameters. The cover of marshlands in the 19th-century habitat structure of Central European quadrates was considerably the same as in the recent African ones. The extent of rural residential areas was significantly smaller in the 19th-century habitat structure of Central European quadrats than in present-day African ones. According to the revealed correlations, the surface cover of rural residential areas is the main driver of the number of autochthonous malaria cases that we can directly impact.

Conclusions: The study confirmed with historical comparison that not only the annual rainfall and mean temperature, the cover of marshlands and other habitats with breeding sites, but also the elements of the rural human environment play a significant role in the high number of autochthonous malaria cases, probably through the concentration and enhancing sites for vector mosquitoes. The latter confirms that a rapid urbanization process could reduce malaria cases in the most infected areas of Africa. Until the latter happens, extensive biological control of Anopheles larvae and chemical control (both outdoor and indoor) of their imagoes, further mosquito nets, repellents, and carbon dioxide traps will need to be applied more widely in the most heavily infested areas.

Keywords: Aerial photographs; Drainage; Marshland; Nineteenth century; QGIS; Rural; Urban.

Fig. 1

The study areas and the location of the 20 quadrates in Central Europe and Africa

Fig. 2

Examples for habitat maps of the study sites (CEU01 and AF01 sites, all maps see in Suppl. Mat. Figure S2a–S2m)

Fig. 3

Differences of relative surface cover of the most characteristic habitat types in the studied quadrates: mean values (± SE) in CEU19th (n = 10), CEU20th (n = 10), CEUrec (n = 10), and AFrec (n = 10) sites. Statistically significant (p < 0.05) differences detected using Mann–Whitney tests are indicated by different letters

Fig. 4

PCA analysis detected differences in the habitat structure of the studied quadrats. African study sites (AF) showed isolation mainly in the presence of forest (Fo), lake (La), wet arable land (Wa), and rural residential area (Ru) habitat types. CEU quadrats from the nineteenth century differed from other sites in the high relative surface cover of arable fields (Ar) and canals (Ca) (n = 40). Abbreviations of the habitat types: Ar: arable land, Ca: canal, Fo: forest, Gr: grassland, La: lake, Ma: marshland, Ri: river, Ro: road, Ru: rural residential area, Sc: scrub, Ur: urban residential area, Vy: vineyard, Wa: wet arable land

Fig. 5

Differences of annual mean temperature (a) and annual precipitation (b) in the studied quadrates: mean values (± SE) in CEU19th (n = 10), CEU20th (n = 10), CEUrec (n = 10), and AFrec (n = 10) sites. Statistically significant (p < 0.05) differences detected using Mann–Whitney tests are indicated by different letters

Fig. 6

GLM analyses detected robust significant correlations between a the annual mean temperature, b the annual precipitation, c the surface cover of rural residential areas (Ru) and the number of autochthonous malaria cases (n = 30) (all data were log-transformed)