This dissertation aims to broaden the understanding of vulnerability in Rwanda by describing, and giving best estimates of sexual risk behaviors and maltreatment among youth heads of household (YHH). YHH vulnerability is examined using both quantitative and qualitative data and multiple datasets. Results pertaining to YHH's sexual risk behavior and experienced abuses are presented. Chapter Three investigates sexual behaviors among YHH and utilizes data from a sample of 692 YHH aged 10--24. The analysis found thirty-seven percent of males and 28.3% of females had a history of penetrative sexual intercourse with 41 percent indicating sexual onset before age 15. Males were more likely to start earlier than females (50.4% versus 26.7%) but females reported more sexual onset with an older partner. Chapter Four documents maltreatment experienced by YHH in the context of high poverty and disrupted community due to war. Thematic analyses of focus group data provide in-depth understanding of major factors potentially related to abuse. A youth survey of 692 YHH indicated that nearly 72 percent of YHH were victims of maltreatment with significantly higher scores among YHH age 18 and above. One out of four YHH reported to have been beaten since being on his/her own, 36.4% of YHH complained of land or property dispossession, and half of YHH reported their properties were damaged. In addition, sexual abuse was reported by 4.5% of YHH and half of the perpetrators were known community members. Chapter Five investigates whether the provision of an adult mentor to youth heads of household through a home visiting program can reduce abuse and improve sexual risk behavior. The findings suggest that mentoring relationship should be supported as a key strategy that may contribute to youth development including the prevention of risk-taking behavior and their consequences, and the protection against maltreatment from community members. Overall, this dissertation draws attention to the high level of vulnerability among YHH and underscores the need for appropriate and comprehensive interventions / acase@tulane.edu

Statement of problem. Socioeconomic status (SES) is considered to be a major determinant of injury and this relationship has been examined in developed countries for a number of years. The relationship between SES and the occurrence of injury is often discussed at either the individual or contextual level. However, few studies have examined the relationship between SES and injuries at both levels at the same time. Injury is an emerging public health problem in low and middle income countries, but little attention has been paid to injury research, and there are very few studies that examine the relationship between socioeconomic status and injury in the developing world. This study aims to examine the association between individual and contextual-level socioeconomic status and the occurrence of injury mortality and morbidity in Vietnam. In this study, individual and contextual-level socioeconomic status were studied together; with special emphasis on evaluating the influence of contextual effects on the occurrence of injury after adjusting for individual factors Methods. The Vietnam Multi-Center Injury Survey was used as the main source of injury and individual SES information. Data on nonfatal and fatal injuries obtained from the Vietnam Multi-Center Injury were used as the outcome variables. Contextual socioeconomic characteristics extracted from the 1999 Vietnam Census of Population were merged with the injury surveillance file The Census file contained information on 128,627 individual records grouped into 72 districts throughout the country of Vietnam. Multilevel logistic regressions were performed to model the relationship of individual and contextual level SES measures to the injury outcomes. The multilevel regression technique allowed examination of both single-level and dual-level relationships between individual and contextual level socioeconomic status and the occurrence of injury in the source population Results. The findings from this study show that there is an emerging injury problem in Vietnam. Both nonfatal and fatal injury rates are very high in comparison with other developed countries. The patterns of injury vary by different socioeconomic status characteristics. Nonfatal and fatal injuries are gender unbalanced with the rates in males always higher than those in females. Injury rates varied by different age groups, living standards of the family, geographical region, and other characteristics. Individual socioeconomic status characteristics were found to have a strong independent relationship with the occurrence of various injury causes, age group, gender and family standard of living. Contextual socioeconomic status measured by a composite index was also associated with the occurrence of nonfatal and fatal injury in Vietnam independently of individual socioeconomic status Conclusions. The findings from this study showed an emerging public health problem of injury in Vietnam with a high rate of nonfatal and fatal injury, specific patterns of injury related to different demographic and socioeconomic variables among the population. The individual and contextual level socioeconomic status measures were related to the injury outcomes with the contextual effect maintained after controlling for the individual level factors. The findings of this study help to improve our understanding about the association between individual and contextual level socioeconomic status and the occurrence of injury in a developing country. The findings also support the theory that contextual socioeconomic status has an independent effect on the health outcomes of population. Understanding this issue could contribute more for the success of injury control program and more general the health programs in Vietnam / acase@tulane.edu

Anopheles gambiae is the principal vector of malaria in Africa. Efforts using aquatic predators to control these mosquitoes as larvae have been hampered by a paucity of research on ground pool trophic dynamics. Loss of structural complexity has been linked to a loss of specialist predators, and an increase in prey survivorship in other ecological systems. There is evidence that structural complexity influences the distribution of mosquito larvae and their predators. Additionally, temporality may diminish the diversity of predators colonizing ground pools, thereby increasing the survivorship of mosquito larvae. Mosquitoes that selectively colonize temporal habitats early in biological succession would be using a novel strategy to escape predation The roles of structural complexity and temporality were investigated in rural ground pools near the Kenyan coast, in brick making pits in the Kenyan highlands, and in laboratory experiments. Habitats in Malindi city were evaluated to describe the temporality and complexity of urban habitats and to associate these ecological variables with predator guild colonization Flood-disturbed ground pools along the Jeribuni River in coastal Kenya revealed a coincidence of vegetation removal and Anopheles larvae appearance. In two coastal villages, all ground pools were temporal, some containing Anopheles larvae. In a highland valley, eight predator taxa were associated with pools undisturbed by brick makers; three were associated with older habitats. In path analysis of laboratory experiments, the magnitude of mortality experienced by backswimmers from loss of structural complexity exceeded that of direct predation pressure. In other experiments, mortality of An. gambiae was found to significantly exceed that of Culex quinquefasciatus in the presence of three different predators. In the port city Malindi, many man-made habitats were found to be temporal and/or structurally simple. Predator colonization of habitats was statistically associated with habitat type. Swimming pools were associated with planned areas of the city while tire track pools were associated with peri-urban areas. Temporality and low structural complexity reduce colonization of aquatic habitats by predators, increasing survivorship of An. gambiae mosquito larvae in Kenya. Further investigations into ground pool ecology will aid in the development of methods that promote predator colonization and mosquito reduction / acase@tulane.edu

acase@tulane.edu

acase@tulane.edu

acase@tulane.edu

Background. Little information is available on tracking of glucose or insulin. The phenomenon known as tracking, defined as the maintenance of relative rank with respect to peers over time, is important in the early identification of individuals who may be at highest risk of metabolic disorders. Previous studies indicate that the tracking phenomenon may be most useful for anthropometric variables, slightly less so for lipids, and lowest for blood pressure Methods. Serial, cross-sectional data collected from 1976 to 2001 as part of the Bogalusa Heart Study were combined to create a longitudinal cohort of individuals with at least one measurement before, and one after, age 18 years. Tracking was assessed by correlations, percentiles, GEE, and mixed models regression analyses for fasting glucose, insulin, and homeostasis model assessment (HOMA) Results. Fasting measurements were available for 2,533 and 1,434 individuals for glucose and insulin/HOMA, respectively. Mean follow-up age was 25--27 (+/-5) years. Average blood pressure and lipid values were in the normal range in adulthood. Mean body mass index (BMI) was 27 +/- 7 kg/m2. Correlations ranged from 0.16--0.22 for glucose, 0.24--0.31 for insulin, and 0.24--0.30 for HOMA (p < 0.0001 for all). The percent of individuals remaining in the top quintile for glucose was 25--34%, and 30--41% remained in the top quintile for insulin during adulthood. Analyses by GEE regression indicated that mean values of outcomes were significantly different for each measurement age group. Mixed models analysis indicated significant random variability in slopes of the lines for glucose and insulin, but not for HOMA. Multiple measurements over time accounted for 27%, 8% and 19% of the variation in glucose, insulin, and HOMA, respectively Conclusions. This study supports the hypothesis that fasting glucose, insulin, and HOMA follow a tracking phenomenon from childhood to adulthood. A strong association was found for all three outcomes by multiple analyses. The tracking relationship was stronger for insulin than glucose, and HOMA offered no advantage with regard to tracking over insulin alone. This information may be useful to help identify, early in life, those at highest risk of future metabolic disorders / acase@tulane.edu

The purpose of this study was to examine the difference in the number of syphilis cases detected by social network analysis (SNA) interviews compared to partner notification interviews and describe the sexual and social networks of syphilis cases Ten primary and secondary syphilis index cases were enrolled in the social network study. The index cases and two more generations of contacts were interviewed using an innovative game board. Social network data collection occurred from 8/1/96-1/31/96 in one urban parish and three rural parishes. Infected syphilis cases agreed to receive the social network interview in place of the partner notification interview. Seventy-two PN cases were collected from 1/1/96-6/30/96 from the same parishes. SNA collects detailed information on sexual and social relationships that PN does not. Statistical comparisons were conducted between the PN and the SNA interviews. Within group analyses were also conducted There were no differences with respect to the mean number of syphilis infected contacts between the PN and SNA interviews (PN mean = 0.24, SNA mean = 0.38, NS). Social contacts from the SNA had a zero yield of syphilis positivity. Sexual and social networks were diffuse and disaggregated. Fifteen percent of the sexual contacts were also members of respondent's social groups It is not practical to conduct a full scale SNA as a routine procedure in place of PN because SNA interviews are time intensive and the analysis is mathematically complex. The PN program may be improved by incorporating elements of the social network interview and analysis. In a network informed approach to STD control, PN interviewers could focus their efforts at a particular neighborhood in order to understand the neighborhood structure This study has identified a diffuse and disaggregated group of individuals that are at 'moderate' risk for syphilis and other STDs. This study revealed a community of many persons with a couple or a few sexual partners, high rates of self reported STDs and similar network placement in their specific community structures. In addition to 'high risk' target groups, groups of persons at 'moderate risk' identified by SNA should also be the target of STD control / acase@tulane.edu

Qualitative methods can be used in public health research to inform the design of a quantitative study, to help interpret the results of a quantitative study, and as a primary method of data collection. This dissertation demonstrates these three uses of qualitative data by presenting three papers prepared for scientific journals and compares and contrasts the methods used for each paper The first paper shows how qualitative methods can be used as part of an epidemiological study to identify regimen, individual, community, and cultural factors that affect the adoption and maintenance of weekly Vitamin A supplementation. This information is used to design an effective information, education and communication campaign for a clinical trial. The second paper uses qualitative data to help interpret and validate the quantitative findings of a study exploring whether the subjects in a Vitamin A supplementation trial understand the consent form that they signed, know that not all the capsules provided by the trial are the same, and identifies factors associated with this knowledge. The final paper uses qualitative methods as an autonomous research tool to explore a new and emerging concept. The paper presents research conducted in three districts in South Africa exploring the range of sexual behavior change among HIV-positive individuals related to the availability of anti-retroviral therapy (ART) and compares this to diagnosis-related behavior change / acase@tulane.edu

acase@tulane.edu