Risk Profiling of Malaria Epidemiology in Rural Bangladesh
SCHAUGHENCY-DISSERTATION-2016.pdf (58.73Mb) (embargoed until: 2020-12-01)
Schaughency, Katherine Chunmin Lin
MetadataShow full item record
Background. Of 214 million estimated new malaria cases worldwide in 2015, 10% were from south-east Asia. Bangladesh is one of the 95 countries with ongoing malaria transmission. Of 13 malaria endemic districts in Bangladesh, Bandarban District---located in Chittagong Hill Tracts---has one of the highest malaria prevalence. This dissertation aimed to study risk factors associated with malaria endemicity and provide risk profiles of malaria epidemiology of the area. Methods. This dissertation was conducted under Mapping Malaria Epidemiology Project, a prospectively surveillance project, in Bandarban, Bangladesh from 2009 to 2013. There were 5,006 households and 22,325 individuals resided in Bandarban Study Area, which included Kuhalong Union and Rajbila Union. We used logistic regressions to model field performance of FalciVax Rapid Diagnostic Test (RDT) against Giemsa-stained microscopy (Chapter 3). How levels of Plasmodium falciparum density were associated with malaria symptoms was analyzed by logistic regressions (Chapter 4). Linear regressions were used to examine the relationship between household building materials and average number of Anopheles mosquitoes found at households at night (Chapter 5). Finally, we conducted Generalized Estimating Equation (GEE) Poisson regression models to study how living standards (e.g. 33 durable assets and household building materials) would be associated with malaria incidence (Chapter 6). Results. Among 616 Plasmodium falciparum tested individuals, 529 of them were malaria positive. Overall, sensitivity, specificity, positive and negative predictive values of FalciVax RDT were 99.6% (527/529), 33.3% (29/87), 90.1% (527/585) and 93.5% (29/31), respectively. Being 30 years old or above, not having measured fever during malaria diagnosis, having symptom duration for more than 6 days, having self-reported fever at night, and not having self-reported fever with sweating were related to having lower level of Plasmodium falciparum parasite density (i.e. parasite density below median (5400 parasites/μl)). Approximately 5 Anopheles mosquitoes were found per night per household. Using mud as a building material (wall: N = 123 households (HHs), 95% CI: [0.31, 1.74]; partition: N = 119 HHs, 95% CI: [0.24, 1.70]; floor: N = 420 HHs, 95% CI: [0.15, 1.15]), comparing to the use of "bamboo", was associated with a higher number of Anopheles mosquitoes found at households at night. Having "bamboo" as a wall, partition and flooring material, having "corrugated tin or iron sheet" as a roofing material, as well as having "elevated ground floor at home" were related to elevated malaria incidence comparing to other types of building materials. Discussion. With smaller sample size in stratum specific category, logistic regression could provide smoother estimates of sensitivity, specificity and predictive values. Sensitivity and specificity of rapid diagnostic devices were previously assumed to be independent from prevalence of malaria. Our field performance study may had shown otherwise. Future studies are needed to examine the assumption. How to utilize identified risk factors to integrate case awareness, reactive case search and hot spot analysis is essential to reduce malaria transmission in the study area. Although having some "mud" or "Bamboo" as part of building materials did not significantly change the overall Anopheles population, species specific preferences among Anopheles mosquitoes should be further studied. Although many factors were identified as risk factors for malaria incidence, the link among building materials, mosquitoes and malaria incidence, should be further explored.