When wildfires threaten human populations, untimely or late evacuation can mean the difference between life and death. In the fire prone regions of the world, last minute evacuations have been responsible for hundreds of deaths and thousands of injuries. In Southern California, where fire season persists for the majority of the year and where human populations continue to expand into lands conducive to wildfire, emergency management officials commonly issue evacuation orders to initiate protective action behavior for those threatened. While early evacuation generally provides the safest course of action for people in the path of wildfire, past research has shown that for a variety of environmental hazards, humans do not always immediately heed official warnings, and delay evacuating. This cross-sectional study uses a mixed-mode survey design and survival analysis to investigate the evacuation timing behavior of residents drawn from a simple random sample of Southern California households (n=221) during the 2016 Blue Cut fire in San Bernardino County, California. Findings from these analyses indicate that only half of those receiving a mandatory evacuation order evacuated because of the Blue Cut fire. Residents with past evacuation experience were almost twice as likely to evacuate compared to residents without prior experience. Gender and length of residence were found to be associated with shorter evacuation times and higher evacuation rates, with women and short-term residents more likely to evacuate compared to men and longer-term residents. The findings from this study suggest that if emergency managers and local public agencies can get wildland residents to evacuate when an evacuation order is issued, they may have future success getting them to evacuate during subsequent wildfire evacuation events. However, this study also identifies a significant number of residents who are resistant to evacuation. Addressing the concerns of this group may help increase the speed and rate of future evacuation events.

Indian adolescents are particularly vulnerable to HIV due to their well-documented lack of information about the disease. While there is ample evidence pertaining to the role of structural factors in the transmission of HIV-related information, much less is understood about the transmission of accurate and inaccurate information among information sources that Indian adolescents have at their disposal.

This dissertation investigated the role of information sources in the transmission of HIV-related knowledge in a nationally representative sample of Indian adolescents who are aware of the disease. A combination of multivariable regression and multiple mediation models examined secondary data from the National Family Health Survey-3 (2005-2006) (N=35,603).

Results revealed a disproportional reliance on non-reciprocal information sources coupled with an under utilization of reciprocal sources that served as conduits for accurate information. Television was the most common source of information and the most effective in terms of increasing comprehensive knowledge. In contrast, health workers were the least utilized source but the most efficacious at increasing accurate information pertaining to condom availability, mother to child transmission, and the availability of health services. Multiple mediation models also determined that sources of information fully or partially mediated nearly three-quarters of the established relationships between demographic characteristics and HIV-related knowledge.

Globally, social networks are rapidly increasing in size, composition, and speed. These findings underscore the importance of understanding, mitigating, and leveraging the social network properties that both constrain and promote HIV prevention efforts in our increasingly interconnected world.

Statement of the Problem: Men who have sex with men (MSM) make up only 2% of the population in the United States, but represent 60% of all HIV infections each year. Pre-exposure prophylaxis (PrEP), a once daily medication to prevent HIV infection, has emerged as a prevention tool for populations most heavily affected by HIV. However, the criteria for PrEP have only been broadly defined and lead many individuals and their medical providers to question if PrEP use is appropriate, given individual risk profile and potential drug side effects. The primary goal of this dissertation is to use risk assessment and HIV testing data to create an HIV risk score for HIV-negative MSM. The secondary goal is to use this HIV risk score to create more targeted criteria for PrEP use and subsequently inform individualized PrEP candidacy.

Methods: Behavioral risk assessment and HIV testing data were collected at the Los Angeles LGBT Center from January 2009 to June 2014. Individuals were included in the analysis if they were MSM, had an HIV-negative test result at baseline, and returned for at least one follow-up HIV testing visit (n = 9,981). Bivariate and Multivariate Cox Proportional Hazards Models were used to determine the biological, behavioral, and substance use variables significant in HIV contraction over the 5.5 year follow-up period.

Summary of Findings: Self-reported history of gonorrhea (p = 0.03), chlamydia (p < 0.0001), and syphilis (p = 0.01); having receptive anal sex (p < 0.0001), race/ethnicity of partners (p = 0.005), and number of sexual partners in the last three months (p = 0.0003); methamphetamine (p = 0.0008), and nitrate use (p = 0.002) were all significant predictors of HIV infection during follow-up. Age of partners, intimate partner violence, and ecstasy use were not significant. Following risk score creation, approximately 45% of all individuals were above a chosen risk score, which consisted of 76% of all HIV-positives. The use of this targeted strategy is beneficial in that it more accurately outlines PrEP candidacy criteria, subsequently allowing individuals and their medical providers to make a more informed decision before use.

Background: Community organizations active in disasters play a vital role in community disaster response and recovery, but academic understanding of this organizational population is limited by untested and imprecise typological differentiations. An organizational taxonomy would better quantify and define this population and subgroups within it. This would allow for contextualizing research and findings against a validated framework that relates organizational groups and subgroups within the broader population. Taxonomies also serve a role similar to theory by enabling the development of new research questions and hypotheses.

Objectives: This dissertation proposes a taxonomy to classify the organizations of interest, and the taxonomy uses typological benchmarks that ensure coherent classificatory groups to provide meaning-in-context and salience to the needs of its users. The taxonomy evaluated the utility of structural/operational, functional, and financial traits for classifying the organizations. The taxonomy can guide research and policy development, and it can also provide utility to the community organizations themselves and their collaborative networks.

Methods: A novel study population of 660 organizations was created from a stratified non-probability quota sample of 28 Voluntary Organizations Active in Disasters (VOAD) networks. The 660 organizations cover the full range of organizational subtypes/subgroups of interest, and the results are not meant to be generalizable to the VOADs themselves without additional and planned validation. Two sets of hierarchical clustering results were produced and compared using both polythetic and parsimonious trait selection. Iterative and heuristic modeling procedures assessed and compared the results of several important permutations and methodological choices.

Findings: The best set of results classified the population based on a parsimonious set of structural/operational traits: charitable/religious and faith-based/not faith-based. The results from both approaches were robust and congruent with the typological understanding of these organizations, but up to one-fourth of the organizations in the study population exhibited noteworthy deviation from common typological distinctions. The results were expanded into a taxonomy with three branches and three tiers to create a combined total of 26 clusters and subclusters. The best set of results also used a modified unit of analysis that classified 47 coherent organizational subgroups, akin to “species”, rather than 660 individual organizations.

Background

Methods

Results

Conclusions

Trial registration

Background

Methods

Results

Conclusions

Background

Methods

Results

Conclusions

Background

Methods

Results

Conclusions