If you are not counted, you don't count: Estimating the size of hidden populations
- Author(s): Wesson, Paul Douglas
- Advisor(s): Reingold, Arthur;
- McFarland, William
- et al.
Background: Despite advances in treatment and prevention services, HIV infection remains a leading cause of morbidity and mortality worldwide, identified by the 2010 Global Burden of Disease report as the fifth leading cause of global disability adjusted live years. While the epidemiologic features of HIV infection vary globally, marginalized populations, such as men who have sex with men (MSM), female sex workers (FSW), and injection drug users (IDUs) are consistently at increased risk for HIV infection relative to the general population. Targeting such marginalized, or hidden, populations has become a global priority to maximize the effectiveness of the public health response to the HIV pandemic. Members of these populations are often difficult to find, and the size of these populations is largely unknown, making it difficult to calculate epidemiologic measures of disease and to evaluate the reach and coverage of public health programs.
Methods: Through three separate analyses, this dissertation will investigate the reliability and the plausibility of population size estimation methods when applied to hidden populations. Chapter 1 systematically reviews the literature on population size estimation methods and assesses the degree to which different methods, applied to the same population, calculate similar estimates of the target population. Chapter 2 evaluates a novel size estimation method, the SS-PSE, by applying it to a Respondent-Driven Sampling study of African-American MSM in San Francisco, and comparing results to other methods. Chapter 3 applies capture-recapture models to evaluate the completeness of the Alameda County HIV surveillance system, and examines the role of sampling bias in this application.
Discussion and Significance: The results of the analyses featured in this dissertation demonstrate that variability in population size estimates from different size estimation methods is common, though often unaddressed. Population size estimation is fundamental to public health surveillance, serving as the basis for policy decisions and quantifying the magnitude of disease. To produce reliable population size estimates, which have implications for the allocation of limited public health resources to marginalized populations, investigators should consistently apply multiple size estimation methods and carefully consider the influence of sampling bias.