Limits...
Associations between urbanicity and malaria at local scales in Uganda.

Kigozi SP, Pindolia DK, Smith DL, Arinaitwe E, Katureebe A, Kilama M, Nankabirwa J, Lindsay SW, Staedke SG, Dorsey G, Kamya MR, Tatem AJ - Malar. J. (2015)

Bottom Line: One site (Walukuba) had significantly higher urbanicity measures compared to the two rural sites.In Walukuba, all individual measures of higher urbanicity were significantly associated with a lower household density of mosquitoes.At finer scales, individual household measures of higher urbanicity were associated with lower mosquito densities and parasite prevalence only in the site that was generally characterized as being urban.

View Article: PubMed Central - PubMed

Affiliation: Infectious Diseases Research Collaboration, Kampala, Uganda. skigozi@yahoo.com.

ABSTRACT

Background: Sub-Saharan Africa is expected to show the greatest rates of urbanization over the next 50 years. Urbanization has shown a substantial impact in reducing malaria transmission due to multiple factors, including unfavourable habitats for Anopheles mosquitoes, generally healthier human populations, better access to healthcare, and higher housing standards. Statistical relationships have been explored at global and local scales, but generally only examining the effects of urbanization on single malaria metrics. In this study, associations between multiple measures of urbanization and a variety of malaria metrics were estimated at local scales.

Methods: Cohorts of children and adults from 100 households across each of three contrasting sub-counties of Uganda (Walukuba, Nagongera and Kihihi) were followed for 24 months. Measures of urbanicity included density of surrounding households, vegetation index, satellite-derived night-time lights, land cover, and a composite urbanicity score. Malaria metrics included the household density of mosquitoes (number of female Anopheles mosquitoes captured), parasite prevalence and malaria incidence. Associations between measures of urbanicity and malaria metrics were made using negative binomial and logistic regression models.

Results: One site (Walukuba) had significantly higher urbanicity measures compared to the two rural sites. In Walukuba, all individual measures of higher urbanicity were significantly associated with a lower household density of mosquitoes. The higher composite urbanicity score in Walukuba was also associated with a lower household density of mosquitoes (incidence rate ratio = 0.28, 95 % CI 0.17-0.48, p < 0.001) and a lower parasite prevalence (odds ratio, OR = 0.44, CI 0.20-0.97, p = 0.04). In one rural site (Kihihi), only a higher density of surrounding households was associated with a lower parasite prevalence (OR = 0.15, CI 0.07-0.34, p < 0.001). And, in only one rural site (Nagongera) was living where NDVI ≤0.45 associated with higher incidence of malaria (IRR = 1.35, CI 1.35-1.70, p = 0.01).

Conclusions: Urbanicity has been shown previously to lead to a reduction in malaria transmission at large spatial scales. At finer scales, individual household measures of higher urbanicity were associated with lower mosquito densities and parasite prevalence only in the site that was generally characterized as being urban. The approaches outlined here can help better characterize urbanicity at the household level and improve targeting of control interventions.

No MeSH data available.


Related in: MedlinePlus

Distribution of three urbanicity metrics across the three sites, based on the frequency distribution of individual metric values within a 100 m buffer around households participating in the cohort and entomology study in each site. a Land cover classification in Walukuba at 30 m spatial resolution overlaid with the participating study households. b NDVI in Walukuba at 6 m spatial resolution. c Satellite-derived night-time light brightness across Walukuba overlaid with the participating study households
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4587721&req=5

Fig2: Distribution of three urbanicity metrics across the three sites, based on the frequency distribution of individual metric values within a 100 m buffer around households participating in the cohort and entomology study in each site. a Land cover classification in Walukuba at 30 m spatial resolution overlaid with the participating study households. b NDVI in Walukuba at 6 m spatial resolution. c Satellite-derived night-time light brightness across Walukuba overlaid with the participating study households

Mentions: A supervised classification of Landsat imagery was undertaken, using Google earth [18] to define training samples of known land cover type. Landsat Enhanced Thematic Mapped (ETM) images matching the period of data collection were obtained from the US geological survey repository [19]. Following previously defined approaches [20], the training samples were used within a maximum likelihood supervised classifier in the satellite image processing software, Erdas Imagine (Geosystems, L. 2004; ERDAS imagine. Atlanta, GA, USA), to produce a land cover map that contained classes representing different gradations of urbanicity (Fig. 2). Among the three sites, only Walukuba displayed distinctly observable urban classes. The others, Nagongera and Kihihi, did not display similar classes, and thus, classification results were not used here for those sites. Within Walukuba, three distinct ‘urban’ related classes existed: dense urban, medium dense urban and residential low dense urban (Fig. 2). The percentage of 30 × 30-m grid cells in the observational buffer in each of these classes was used as a household level metric of urbanicity to test against the malaria metrics. Only the residential low dense urban classification of land cover showed significant associations with malaria metrics and hence was maintained while the other two were dropped from further analysis.Fig. 2


Associations between urbanicity and malaria at local scales in Uganda.

Kigozi SP, Pindolia DK, Smith DL, Arinaitwe E, Katureebe A, Kilama M, Nankabirwa J, Lindsay SW, Staedke SG, Dorsey G, Kamya MR, Tatem AJ - Malar. J. (2015)

Distribution of three urbanicity metrics across the three sites, based on the frequency distribution of individual metric values within a 100 m buffer around households participating in the cohort and entomology study in each site. a Land cover classification in Walukuba at 30 m spatial resolution overlaid with the participating study households. b NDVI in Walukuba at 6 m spatial resolution. c Satellite-derived night-time light brightness across Walukuba overlaid with the participating study households
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4587721&req=5

Fig2: Distribution of three urbanicity metrics across the three sites, based on the frequency distribution of individual metric values within a 100 m buffer around households participating in the cohort and entomology study in each site. a Land cover classification in Walukuba at 30 m spatial resolution overlaid with the participating study households. b NDVI in Walukuba at 6 m spatial resolution. c Satellite-derived night-time light brightness across Walukuba overlaid with the participating study households
Mentions: A supervised classification of Landsat imagery was undertaken, using Google earth [18] to define training samples of known land cover type. Landsat Enhanced Thematic Mapped (ETM) images matching the period of data collection were obtained from the US geological survey repository [19]. Following previously defined approaches [20], the training samples were used within a maximum likelihood supervised classifier in the satellite image processing software, Erdas Imagine (Geosystems, L. 2004; ERDAS imagine. Atlanta, GA, USA), to produce a land cover map that contained classes representing different gradations of urbanicity (Fig. 2). Among the three sites, only Walukuba displayed distinctly observable urban classes. The others, Nagongera and Kihihi, did not display similar classes, and thus, classification results were not used here for those sites. Within Walukuba, three distinct ‘urban’ related classes existed: dense urban, medium dense urban and residential low dense urban (Fig. 2). The percentage of 30 × 30-m grid cells in the observational buffer in each of these classes was used as a household level metric of urbanicity to test against the malaria metrics. Only the residential low dense urban classification of land cover showed significant associations with malaria metrics and hence was maintained while the other two were dropped from further analysis.Fig. 2

Bottom Line: One site (Walukuba) had significantly higher urbanicity measures compared to the two rural sites.In Walukuba, all individual measures of higher urbanicity were significantly associated with a lower household density of mosquitoes.At finer scales, individual household measures of higher urbanicity were associated with lower mosquito densities and parasite prevalence only in the site that was generally characterized as being urban.

View Article: PubMed Central - PubMed

Affiliation: Infectious Diseases Research Collaboration, Kampala, Uganda. skigozi@yahoo.com.

ABSTRACT

Background: Sub-Saharan Africa is expected to show the greatest rates of urbanization over the next 50 years. Urbanization has shown a substantial impact in reducing malaria transmission due to multiple factors, including unfavourable habitats for Anopheles mosquitoes, generally healthier human populations, better access to healthcare, and higher housing standards. Statistical relationships have been explored at global and local scales, but generally only examining the effects of urbanization on single malaria metrics. In this study, associations between multiple measures of urbanization and a variety of malaria metrics were estimated at local scales.

Methods: Cohorts of children and adults from 100 households across each of three contrasting sub-counties of Uganda (Walukuba, Nagongera and Kihihi) were followed for 24 months. Measures of urbanicity included density of surrounding households, vegetation index, satellite-derived night-time lights, land cover, and a composite urbanicity score. Malaria metrics included the household density of mosquitoes (number of female Anopheles mosquitoes captured), parasite prevalence and malaria incidence. Associations between measures of urbanicity and malaria metrics were made using negative binomial and logistic regression models.

Results: One site (Walukuba) had significantly higher urbanicity measures compared to the two rural sites. In Walukuba, all individual measures of higher urbanicity were significantly associated with a lower household density of mosquitoes. The higher composite urbanicity score in Walukuba was also associated with a lower household density of mosquitoes (incidence rate ratio = 0.28, 95 % CI 0.17-0.48, p < 0.001) and a lower parasite prevalence (odds ratio, OR = 0.44, CI 0.20-0.97, p = 0.04). In one rural site (Kihihi), only a higher density of surrounding households was associated with a lower parasite prevalence (OR = 0.15, CI 0.07-0.34, p < 0.001). And, in only one rural site (Nagongera) was living where NDVI ≤0.45 associated with higher incidence of malaria (IRR = 1.35, CI 1.35-1.70, p = 0.01).

Conclusions: Urbanicity has been shown previously to lead to a reduction in malaria transmission at large spatial scales. At finer scales, individual household measures of higher urbanicity were associated with lower mosquito densities and parasite prevalence only in the site that was generally characterized as being urban. The approaches outlined here can help better characterize urbanicity at the household level and improve targeting of control interventions.

No MeSH data available.


Related in: MedlinePlus