This study documents an apparent decline in the prevalence of infection with P. falciparum in children less than five years old and an increase in the use of bed nets over a six-year period in Greater Mbarara district, a mesoendemic area of Uganda. Similar results have been reported in other settings but were often based on routine surveillance data [3–10]. In this study, the same area was observed before the implementation of the Roll Back Malaria strategy for malaria control and six years after, with data systematically collected at individual, household, and village levels, and malaria infections assessed using blood smear or RDT. These results are thus less likely to be confounded by ecological bias or by changes over time in the reporting system than reports from routine surveillance. Moreover, the high participation rate observed makes non-response bias very unlikely.
Despite these encouraging findings, the malaria situation remains of concern in the rural areas of this region, contrasting with the urban situation; some rural villages still have medium to high prevalence of P. falciparum infection, and almost 10% of the rural villages had a mean prevalence exceeding 50%. A parallel could be drawn with the global Ugandan malaria epidemiological situation that is characterized by areas of high malaria transmission contrasted to low malaria transmission in urban areas [14, 16–18, 30].
Also, despite improvements since 2004, the coverage of households with bed nets remained low in rural areas and below the target proposed by WHO . Estimations of bed net coverage are commonly limited by the fact that they are based on reported information. In this study, actual presence of bed net was controlled in all households, thus the estimated bed net coverage is believed to be reliable and confirms the result from a recent national survey . As reported in other studies, it was found that prevention strategies are different between rural and urban areas, mainly because of differences in the household wealth and education levels [31, 32]. Further research is needed to better understand these differences and more generally the reasons for bed net use .
To identify factors associated with increased malaria risk, individual-, household-, and cluster-level characteristics were considered. At cluster/village level, an altitude above 1,500 m was a strong predictor of very low malaria prevalence. However, below 1,500 m, the association between altitude and risk of malaria infection vanished. Also at cluster level, a strong association between the malaria risk and latitude was observed that indicates the presence of unexplained heterogeneity with a spatial structure that should be further investigated.
At household level, higher wealth as measured by higher housing and socioeconomic scores was strongly associated with decreased risk of malaria infection in both rural and urban areas, consistent with other studies [34, 35]. Wealth was an important factor associated with bed-net ownership. A possible explanation for the lower risk of malaria infection with increased wealth is that more expenses may be dedicated to prevention in wealthier households. A complex association between household characteristics (wealth, education level, and size) on one hand, and knowledge, ownership, and actual use of bed nets on the other was found, which could explain the weak association between bed-net use and malaria infection. The difference in the origin of the nets between rural and urban settings suggest that net donation remains an important way to improve the net coverage. However, as indicated by the moderate use of net utilization by children, bed-net donations should be coupled with health education.
At individual level, the risk of malaria increased with the age of the children in both rural and urban areas. A possible explanation of this is that younger children may be more likely to stay under bed nets as they usually sleep with their mothers and play less in the evenings. Although the association did not reach statistical significance in the separate analyses, in the pooled analysis malaria risk was lower for children reported to have slept under a bed net the night preceding the survey interview.
This study has several limitations. First, accurate information on malaria interventions in the study area over the study period (especially on ACT availability) was not collected. In 2004, the information collected was less comprehensive and the diagnosis method was less accurate, compared with 2010. Although the same area was investigated, the cross sectional design of this study leads to limitations that include the inability to account for short term parasitemia fluctuations. Also, it should be noticed that malaria risk is known to vary on very small scale, thus only a rough estimation of the spatial variation of malaria could be done with the sampling design used. This study focused on children less than five years of age, since they constitute the most vulnerable population for malaria morbidity and mortality. However, as a consequence of changes in malaria transmission, the peak in malaria morbidity and mortality might shift toward older age groups . Future studies including older age groups are, therefore, needed. Also, entomological and environmental data would be useful to monitor changes in malaria endemicity.