| This dissertation explores the use of capture-recapture methods, a sampling and estimation method used by ecologists to study wildlife population characteristics. In epidemiology, capture-recapture usually relies on independent records as "samples"; thus, the approach is often referred to as dual- or multiple-record system estimation. In this document, we assess how well dual- and multiple-record system estimation methods perform in producing valid and reliable estimates of population parameters of (1) the frequency of a human genetic polymorphism (called Duffy) that confers resistance to infection of erythrocytes by Plasmodium vivax, a cause of malaria in some parts of the world, and (2) the prevalence of P. vivax blood-stage infections. In each of these analyses, we use data from cross-sectional malaria prevalence surveys performed in the Wosera region of the East Sepik Province in Papua New Guinea between 1998 and 2003 as well as primary data from blood samples collected from study participants. As you will see, although both estimates are counts of population members---the number of individuals who are carriers of the Duffy-negative allele and the number of individuals who have a blood-stage P. vivax infection---the outcomes can be quite different because of the object of measurement. In addition, we describe the pattern of association between the Duffy-negative heterozygous genotype and P. vivax blood-stage infections and assess the effect of the Duffy-negative allele on the prevalence of P. vivax infections and P. vivax parasitemia within a cohort of children who participated in a short 7-month longitudinal study. |
