UC Berkeley

Abstract

Statistical models for longitudinal analysis of single and mixed species infections

By

Kathryn Louise Colborn

Doctor of Philosophy in Biostatistics

University of California, Berkeley

Professor Terence P. Speed, Chair

There are numerous examples of infectious diseases that are caused by various species of

the same pathogen. Some examples include Lyme disease, malaria, Leishmaniasis, Dengue

fever, and Ehrlichiosis. The advancement of laboratory methods has facilitated more sensitive

detection of mixed species infections in humans, which has resulted in a surge of research focussing

on the eects of mixed infections on clinical outcomes. Cross-sectional blood samples

compared with clinical outcome measures provide a limited scope of the interactions between

species. It is important to study these infections in humans longitudinally, and within their natural

environments, in order to develop an understanding of the complex relationships between

hosts, pathogens and vectors of transmission.

Papua New Guinea is a country with high prevalence of both Plasmodium falciparum and P.

vivax, two species of parasites that can cause malaria. It is well known that these two parasites

can cause severe morbidity and mortality independently, but there has not been conclusive

evidence of the eect of mixed P. falciparum and P. vivax infections on clinical symptoms.

Children under age five are at highest risk of experiencing adverse outcomes from Plasmodium

infections. In 2006, a cohort study was implemented to conduct an investigation of the eects

of mixed P. falciparum and P. vivax infections on clinical episodes of malaria in children living

in a rural area of Papua New Guinea. The data collected from this study are used throughout

this dissertation to address both the epidemiological questions of the study investigators and to

present statistical models for analyzing longitudinal malaria data and mixed species infections.