Multilevel models have long been used by health geographers working on questions of space, place, and health. Similarly, health geographers have pursued interests in determining whether or not the effect of an exposure on a health outcome varies spatially. However, relatively little work has sought to use multilevel models to explore spatial variability in the effects of a contextual exposure on a health outcome. Methodologically, extending multilevel models to allow intercepts and slopes to vary spatially is straightforward. The purpose of this paper, therefore, is to show how multilevel spatial models can be extended to include spatially varying covariate effects. We provide an empirical example on the effect of agriculture on malaria risk in children under 5 years of age in the Democratic Republic of Congo.
BACKGROUND: Long-lasting insecticidal nets (LLINs) remain a cornerstone of malaria control, but strategies to sustain universal coverage and high rates of use are not well-defined. A more complete understanding of context-specific factors, including transmission intensity and access to health facilities, may inform sub-district distribution approaches and tailored messaging campaigns. METHODS: A cross-sectional survey of 2190 households was conducted in a single sub-county of western Uganda that experiences highly variable malaria transmission intensity. The survey was carried out approximately 3 years after the most recent mass distribution campaign. At each household, study staff documented reported LLIN use and source among children 2 to 10 years of age and performed a malaria rapid diagnostic test. Elevation and distance to the nearest health facility was estimated for each household. Associations between parasite prevalence and LLIN use were estimated from log binomial regression models with elevation and distance to clinic being the primary variables of interest. RESULTS: Overall, 6.8% (148 of 2170) of children age 2-10 years of age had a positive RDT result, yielding a weighted estimate of 5.8% (95% confidence interval [CI] 5.4-6.2%). There was substantial variability in the positivity rates among villages, with the highest elevation villages having lower prevalence than lowest-elevation villages (p < .001). Only 64.7% (95% CI 64.0-65.5%) of children were reported to have slept under a LLIN the previous night. Compared to those living < 1 km from a health centre, households at ≥ 2 km were less likely to report the child sleeping under a LLIN (RR 0.86, 95% CI 0.83-0.89, p < .001). Households located farther from a health centre received a higher proportion of LLINs from government distributions compared to households living closer to health centres. CONCLUSIONS: LLIN use and sourcing was correlated with household elevation and estimated distance to the nearest health facility. The findings suggest that current facility-based distribution strategies are limited in their reach. More frequent mass distribution campaigns and complementary approaches are likely required to maintain universal LLIN coverage and high rates of use among children in rural Uganda.
BACKGROUND: Long-lasting insecticidal nets (LLINs) remain a cornerstone of malaria control, but strategies to sustain universal coverage and high rates of use are not well-defined. A more complete understanding of context-specific factors, including transmission intensity and access to health facilities, may inform sub-district distribution approaches and tailored messaging campaigns. METHODS: A cross-sectional survey of 2190 households was conducted in a single sub-county of western Uganda that experiences highly variable malaria transmission intensity. The survey was carried out approximately 3 years after the most recent mass distribution campaign. At each household, study staff documented reported LLIN use and source among children 2 to 10 years of age and performed a malaria rapid diagnostic test. Elevation and distance to the nearest health facility was estimated for each household. Associations between parasite prevalence and LLIN use were estimated from log binomial regression models with elevation and distance to clinic being the primary variables of interest. RESULTS: Overall, 6.8% (148 of 2170) of children age 2–10 years of age had a positive RDT result, yielding a weighted estimate of 5.8% (95% confidence interval [CI] 5.4–6.2%). There was substantial variability in the positivity rates among villages, with the highest elevation villages having lower prevalence than lowest-elevation villages (p < .001). Only 64.7% (95% CI 64.0–65.5%) of children were reported to have slept under a LLIN the previous night. Compared to those living < 1 km from a health centre, households at ≥ 2 km were less likely to report the child sleeping under a LLIN (RR 0.86, 95% CI 0.83–0.89, p < .001). Households located farther from a health centre received a higher proportion of LLINs from government distributions compared to households living closer to health centres. CONCLUSIONS: LLIN use and sourcing was correlated with household elevation and estimated distance to the nearest health facility. The findings suggest ...
BACKGROUND: The Democratic Republic of the Congo (DRC) remains one of the countries most impacted by malaria despite decades of control efforts, including multiple mass insecticide treated net (ITN) distribution campaigns. The multi-scalar and complex nature of malaria necessitates an understanding of malaria risk factors over time and at multiple levels (e.g., individual, household, community). Surveillance of households in both rural and urban settings over time, coupled with detailed behavioral and geographic data, enables the detection of seasonal trends in malaria prevalence and malaria-associated behaviors as well as the assessment of how the local environments within and surrounding an individual's household impact malaria outcomes. METHODS: Participants from seven sites in Kinshasa Province, DRC were followed for over two years. Demographic, behavioral, and spatial information was gathered from enrolled households. Malaria was assessed using both rapid diagnostic tests (RDT) and polymerase chain reaction (PCR) and seasonal trends were assessed. Hierarchical regression modeling tested associations between behavioral and environmental factors and positive RDT and PCR outcomes at individual, household and neighborhood scales. RESULTS: Among 1591 enrolled participants, malaria prevalence did not consistently vary seasonally across the sites but did vary by age and ITN usage. Malaria was highest and ITN usage lowest in children ages 6–15 years across study visits and seasons. Having another member of the household test positive for malaria significantly increased the risk of an individual having malaria [RDT: OR= 4.158 (2.86–6.05); PCR: OR= 3.37 (2.41–4.71)], as did higher malaria prevalence in the 250m neighborhood around the household [RDT: OR= 2.711 (1.42–5.17); PCR: OR= 4.056 (2.3–7.16)]. Presence of water within close proximity to the household was also associated with malaria outcomes. CONCLUSIONS: Taken together, these findings suggest that targeting non-traditional age groups, children >5 years old ...
