This dissertation assesses neurodevelopment in children affected by HIV in Botswana, by examining associations between a range of factors and neurodevelopmental outcomes at 24 months of age and adjusting for potential confounders through linear regression. We followed children from birth to approximately 24 months to assess their neurodevelopment with an adapted version of Bayley Scales of Infant Development third edition (BSIDIII: cognitive, gross motor, fine motor, expressive and receptive language domains) and with the Development Milestones Checklist (DMC: locomotor, fine motor, language and personal-social domains), a parent-completed questionnaire. Chapter I compares neurodevelopment between HIV-exposed uninfected (HEU) children whose mothers took antenatal 3-drug combination antiretroviral therapy (ART) vs. zidovudine (ZDV) within a prospective study, nested within two cohorts of HIV-infected mothers and their children in Botswana (one observational, one interventional). We observed that neurodevelopmental outcomes at 24 months of age were generally at least as good among HEU children exposed in utero to ART when compared to those exposed in utero to ZDV. Chapter II compares neurodevelopment between HEU vs. HIV-unexposed uninfected (HUU) children in Botswana, within a prospective observational study. We observed that neurodevelopmental outcomes at 24 months among HEU children were generally as good as those among unexposed children. Results from Chapter I and II provide reassurance, easing concerns that HIV or ARV-exposure may detrimentally affect neurodevelopment in young children. Chapter III examines the potential of a family-based intervention for use in reducing harmful alcohol use and intimate partner violence, to protect and promote child mental health, within families affected by HIV in Rwanda. Quantitative data from a randomized controlled trial (RCT) were analyzed to demonstrate significant reductions in alcohol-use and intimate partner violence within HIV-affected families receiving the intervention, when compared to control families. Quantitative and qualitative data from the RCT, were integrated using a mixed-method approach, and support the potential of family-based interventions to reduce adverse caregiver behaviors as a major mechanism for improving child well-being, for families affected by HIV in Sub-Saharan Africa.

Background: The association between meat consumption and breast cancer risk and the relationship between vitamin D and colorectal cancer risk have been examined in numerous studies, but questions still remain. For colorectal cancer survival, research into diet has been sparse. Methods: Chapter one is a pooled analysis of 22 prospective cohort studies examining meat and egg consumption and breast cancer risk overall and by estrogen and progesterone receptor (ER/PR) status using Cox proportional hazards models to estimate study-specific relative risks which were then pooled using random effects models. Chapter two is a pooled analysis of 17 prospective cohort studies in which the relationship between prediagnostic circulating 25-hydroxyvitamin D [25(OH)D] and colorectal cancer risk was explored using conditional logistic regression to calculate study-specific relative risks which were then pooled using random effects models. In Chapter three, the association between pre- and post-diagnosis sugar-sweetened beverage (SSB) and sugar intake and mortality among colorectal cancer patients was assessed using Cox proportional hazards models to estimate relative risks in two prospective cohort studies. Results: There were no significant associations between red and processed meat, seafood, and egg consumption and breast cancer risk, regardless of hormone receptor status. For poultry intake, there was a marginally statistically significant increased risk of ER+ and ER+PR- breast cancers only. We found that 25(OH)D concentrations were significantly inversely associated with colorectal cancer risk, with significant decreased risks for levels beyond current recommendations. These inverse associations were stronger in women than men. Pre-diagnosis total fructose, glucose, sucrose, and added sugar intakes were positively associated with mortality, but no association was observed for SSBs. In post-diagnosis analyses, there were marginally significant increases in all-cause mortality with higher intakes of fructose and glucose only. Conclusion: The results of the pooled analysis of meat and egg consumption and breast cancer risk provide some clarification on the inconsistent findings for these associations in individual studies, while the findings of the pooled analysis of 25(OH)D and colorectal cancer risk should be considered in developing public health guidelines on vitamin D. For colorectal cancer survival, future studies should explore pre- and post-diagnosis sugar intake and mortality.

Conditional cash transfer programs have diffused rapidly across the world since the mid- 1990s and have been adopted as a poverty-reduction tool in the majority of Latin American countries. This dissertation focuses specially on the Colombian conditional cash transfer program, Familias en Acción, and examines the relationship between social welfare policy and health in the Colombian context. The work is divided into three separate papers. Paper 1 assesses the impact of Familias en Acción on adult health outcomes that capture disruptions in daily life due to health conditions. The analysis utilizes data collected by the Colombian National Department of Planning for program evaluation purposes and employs a differences-in-differences approach specifying mixed effects logistic models to examine the impact of the program on impairment, bedridden status, and hospitalization among individuals 18 years of age and older. Paper 2 explores the impact of Familias en Acción on mortality assessing the possibility of differential impacts by age group and cause of death. The work pools program evaluation data with vital statistics and census data to create a dataset of municipal population, municipal death, and municipal exposure to the anti-poverty program. The analysis uses a differences-in- differences approach specifying mixed effects negative binomial regression models. Paper 3 is a qualitative political economy project and draws from both primary as well as secondary data. Familias en Acción was modeled after the Mexican conditional cash transfer program, and the two programs are nearly identical in terms of programmatic components with one major exception which is in the area of health programming. The Mexican program is more comprehensive in the area of health, targeting all ages and offering a more comprehensive package of health services. The work examines the political and economic reasons that led to the adaptation of the Mexican conditional cash transfer program in the Colombian context.

Increases in the rates of sexually transmitted infections (STIs) suggest that control programs may not be targeting the population responsible for the spread of STIs - core groups. The objective was to examine STI transmission within these groups using both traditional epidemiology and social network analysis. Routine partner notification data, supplemented with more detailed voluntary information, was collected from individuals diagnosed with, or exposed to a STI in Manitoba. Groups of individuals were identified (n=2,508), and their profiles described. Larger groups (size > 15) had more repeat cases, and contacts that were repeatedly named. Three different groups were identified, containing individuals which differed demographically and clinically. This study uniquely identified same-sex partnerships within larger groups. A greater understanding of disease transmission patterns within these groups will clearly aid in the development of targeted education and prevention programs for all STIs.

Statement of problem. Bronchiolitis is a common childhood lower respiratory tract infection. Determining risk of developing severe disease is important. Methods. 312 children, aged ≤ 12 months, with bronchiolitis were prospectively enrolled over 5 months at 2 emergency departments. We assessed 22 standardized variables and clinical outcome. A decision rule predicting severe bronchiolitis (apnea, intubation, pediatric ICU admission, death) was derived using chi2 recursive partitioning techniques. Results. 8 children (2.6 %) developed severe bronchiolitis. Age, length of breastfeeding, heart rate (HR) and systolic blood pressure outside normal, respiratory exam score, oxygen saturation, and lung atelectasis were associated (p < 0.05) with severe bronchiolitis. Preliminary decision rule included (1) heart rate outside normal range and (2) oxygen saturation < 88%. Rule sensitivity was 100% (95% CI 68,100) and specificity was 93% (95% CI 90,96). Conclusion. Small sample size resulted in the rule's large CIs. Successful model development suggests a decision rule is feasible.

There are very few established causes of primary brain tumours in adults. Associated with short survival times, increasing effort is being put forth in an attempt to better understand the risk factors of these neoplasms, including investigating the possible relationship with a family history of cancer and germline genetic polymorphisms. This thesis was conducted to evaluate both of these potential associations. Using an international population-based case-control study, the self-reported family histories of cancer were compared between 1089 glioma cases and 1922 matched controls and between 307 meningioma cases and 1095 controls. Significantly lowered odds of glioma were associated with the reporting of any type of cancer in a first degree relative (OR = 0.8, 95% CI = 0.7-0.99) and with any type of cancer excluding brain tumours (OR = 0.8, 95% CI = 0.7-0.9). No significant associations were found amongst the meningioma cases and controls, though elevated point estimates were found for those reporting parental lung and genitourinary cancers, while the presence of breast, lip, oral, pharyngeal and unspecified cancers all produced great reductions in meningioma odds, suggesting that further study is required. In order to evaluate the association between adult brain tumours and genetic polymorphisms, a systematic review of the literature was completed. A total of 41 case-control studies were included, covering 46 separate genes and more than 100 different single nucleotide polymorphisms. When possible, quantitative data synthesis was performed to establish a more refined point estimate and confidence intervals. Heterogeneity across the studies and variability in the subject matter often prevented any possible data synthesis so establishing associations that were statistically significant was difficult. All told, there were 41 significant associations found amongst the included studies and each varied by the particular polymorphism or histology studied. None of the estimates produced in the quantitative data syntheses suggested a statistically significant association. The results of this thesis suggest that a family history of cancer is not a risk factor for primary brain tumours in adults and that further work is necessary to better establish the possible association between various genetic polymorphisms and adult brain tumours.

Inflammatory factors, both centrally produced and released from adipocytes (fat cells), have been shown to affect metabolism and to contribute to the pathogenesis and development of many chronic diseases such as heart disease, type 2 diabetes and cancer. Inflammatory response has been shown to be affected by environmental stresses such as heat and cold exposure. The aim of this thesis was to study the effects of heat and cold stresses on inflammatory response over the course of 3 experiments which exposed healthy subjects to one of two heat intensities or to cold stress. This thesis is presented in the classical format and is composed of the following: Chapter 1 introduces the main concepts of inflammation and summarizes the existing literature on temperature stress and inflammation; Chapter 2 describes the methodologies used; Chapter 3 presents the results from the three studies; Chapter 4 discusses the relevance of our results in terms of the pre-existing research in this area and Chapter 5 provides some speculation into the underlying causes of the observed changes, suggests further studies to investigate the speculations and ends with some concluding remarks. Subjects were exposed to either moderate intensity heating (Experiment 1: three hours in chamber set at 42°C, 1.1+/-0.01°C core temperature increase), low intensity heating (Experiment 2: two hours in liquid perfusion suit with circulating water temperature set at 48°C, 0.4+/-0.02°C core temperature increase) or cold exposure (Experiment 3: two hours in liquid perfusion suit with circulating water temperature set at 5°C, no change in core temperature). Fasting blood samples were taken before and during the temperature stress. Plasma levels of pro-inflammatory factors, interleukin-6 (IL-6) and tumor necrosis factor alpha (TNFalpha) were measured along with the adipocyte-derived anti-inflammatory factor, adiponectin. Results showed that, during the higher intensity heat exposure, plasma IL-6 increased by 440% while no statistical change were observed for adiponectin and TNFa levels. No changes were noted in inflammatory markers during the lesser intensity heating. With cold exposure, plasma IL-6 and adiponectin increased over time, falling short of statistical significance. No change in TNFalpha was noted at any point. Increases in plasma IL-6 and adiponectin are positively correlated with increases in core temperature. This thesis supports the notion that temperature stress influences inflammatory response. The novelty of this work lies in the use of two different heat exposure intensities, both lower than the majority of heating intensities used in previous literature, the study of cold stress in human subjects without the confounding effect of other co-existing stresses and the quantification of adiponectin levels in human subjects during temperature stress.

The prevalence of obesity in young children is increasing. Sleep duration is suggested to affect the regulation of body weight through changes in metabolic hormones that regulate appetite, food selection, and energy expenditure. Using data obtained from the Quebec Longitudinal Study of Child Development, this study prospectively examined: 1) the association between longitudinal sleep duration pattern (LSDP) (age 2.5 to 6 years) and overweight/obesity at age 6 and 7 years; and, 2) whether such an association may be mediated by dietary intake and/or eating behaviour at the population level. Associations were examined using multivariate logistic regression methods, with adjustments for potential confounders. Boys with a 'short persistent or short increasing ' LSDP had significantly greater odds of being overweight/obese at age 6 and 7 years, in comparison to boys with an '11-hour persistent ' LSDP. This association was partially mediated by problematic eating behaviours. LSDP did not associate with overweight/obesity in girls.

This thesis describes the epidemiology, quality and methods of systematic reviews of health professional behaviour change interventions. The epidemiology was explored using descriptive methods and the quality was assessed with the AMSTAR tool. 'Lumping' and 'splitting', which refers to how broad or narrow a systematic review question is framed, was explored by assessing a subset of reviews to see how review authors framed their systematic review question and justified this decision. The results indicate that there has been an increase in the number of systematic reviews published on professional behaviour change interventions, they are dispersed across various literature sources and the reviews were generally of poor quality. Furthermore, many reviews are 'split', with little justification for the authors' choice of research question. The overlapping subject areas in addition to the low methodological quality raise concern about the organization of the field, including redundancy of publications and potential duplication of efforts.

Gene-environment and gene-gene interactions lie at the root of many human diseases. This thesis evaluated the risk of bias in the case-only design applied to studies of genetic interaction by way of a systematic review and meta-regression analysis. The case-only design was then used to investigate interactions between BRCA gene mutations and alcohol consumption among breast cancer patients, and the results were compared to a case-control analysis of BRCA mutation carriers with and without breast cancer. The systematic review suggests that the case-only design is unbiased when applied to studies of genetic interaction. The case-control analysis found that increasing wine consumption may reduce the risk of breast cancer among BRCA1 mutation carriers, and the results of case-only analysis may be compatible with this. Among BRCA2 mutation carriers, consumption of alcohol other than wine increased breast cancer risk in the case-only analysis while no association was observed in the case-control analysis.