Systemic Lupus Erythematosus (SLE), a prototypic multi-organ autoimmune disease, is associated with a dramatically increased risk of coronary artery disease (CAD) manifesting as angina, myocardial infarction and sudden cardiac death. Traditional cardiac risk factors such as hypertension and hypercholesterolemia, measured at baseline in accordance with the Framingham model, only partially account for the increased risk of CAD in SLE. In this thesis, I have shown that blood pressure (BP), lipids and novel risk factors such as the inflammatory marker high-sensitivity C-reactive protein (hsCRP), take a dynamic course in SLE, with more than half of the variance in serial measurements over time occurring within rather than between individuals. This variability is due to changes in disease activity, treatment, accrual of other cardiac risk factors, and complications such as infection. I have demonstrated that by capturing cumulative exposure over time, ‘summary measures’ such as arithmetic mean and time-adjusted mean (AM) are better able to quantify CAD risk in patients with SLE than single-point-in-time measurements of risk factors. By incorporating ‘summary measures’ such as mean and AM into time-dependent covariate survival analysis models, I was able to quantify the magnitude of increase in CAD risk associated with increments in systolic and diastolic BP, and to demonstrate and quantify the association between several lipids / lipoproteins and CAD risk in SLE. Using this methodology, I was also able to demonstrate that despite marked variability over time, ‘summary measures’ of hsCRP are independently predictive of CAD risk among patients with SLE, highlighting the pivotal role of inflammation in atherosclerosis. Furthermore, I was able to determine lipid and hsCRP ‘cut-points’ that will aid clinicians in identifying a subgroup of patients with SLE who are at significantly increased cardiac risk.

Background: Alzheimer disease (AD) is a progressive, late-life neurodegenerative disorder. Given the aging population, AD is a significant health concern. According to the Alzheimer Society of Canada (Smetanin et al., 2009), in 25 years 2.8% of the Canadian population will have AD or a related dementia. Presently, there is no cure for AD; therefore, efforts to either delay AD onset or prevent AD altogether are a primary focus. The ability to proficiently speak many languages has been associated with certain cognitive advantages. Based on these findings, multilinguals are hypothesized to be more resistant to cognitive decline than monolinguals. More research is warranted in order to further this theory and to contribute to strategies to prevent or delay AD. Objectives: The first study objective was to evaluate whether multilingualism was associated with the development of AD. The second study objective was to assess whether multilingualism was associated with later dementia onset. Methods: Analyses were based on data from the Nun Study, a longitudinal study of aging in 678 participants 75+ years living in the United States. In order to address the first study objective, the association between multilingualism and AD was assessed in 157 participants using logistic regression models adjusted for age, education, apolipoprotein E-E4 (ApoE-E4) status, immigrant status, and occupation. Additional subgroup analyses also included covariates associated with career length and linguistic ability (grammatical complexity and idea density). AD was diagnosed based on criteria for both clinical dementia and AD neuropathology. Dementia was diagnosed using the Diagnostic and Statistical Manual of Mental Disorders, 4th edition criterion (American Psychiatric Association, 1994) (based on the Consortium to Establish a Registry for Alzheimer’s Disease battery of tests (Morris, Heyman, Mohs, & Hughes, 1989) and performance on activities of daily living), while AD neuropathology was based on the National Institute on Aging and Reagan Institute criterion (The National Institute on Aging - Reagan Institute (NIA-RI) Working Group on Diagnostic Criteria for the Neuropathological Assessment of Alzheimer's Disease, 1997). In order to address the second study objective, dementia likelihood was assessed in 325 participants using discrete-time survival analyses adjusted for age, ApoE-E4 status, education, and linguistic ability. Results: When adjusted for age, education, ApoE-E4 status, occupation, and immigrant status, participants speaking two or more languages had similar AD risks compared to monolinguals (OR = 1.05; 95% CI = 0.45-2.50). However, when grammatical complexity was held constant across participants, speaking two or more languages was associated with a four-fold decrease in AD risk compared to speaking one language (OR = 0.25; 95% CI = 0.04-1.23), although this did not reach statistical significance. When the association between multilingualism and time of dementia onset was assessed, the dementia hazard function estimates for all participants were constant and persisted throughout the follow-up period of the study. When ApoE-E4 status and baseline age were held constant, participants speaking four or more languages were significantly less likely to develop dementia than monolingual participants (OR = 0.14; 95% CI = 0.01-0.66). An interaction between multilingualism and the other two covariates (ApoE-E4 status and baseline age) was observed: the oldest participants with an ApoE-E4 allele who spoke four or more languages had smaller dementia risks than younger participants without an ApoE-E4 allele who spoke one, two, or three languages. Participants speaking two or three languages were no less likely than monolinguals to develop dementia across the study duration. When idea density was held constant across participants, multilingualism was associated with a nonsignificant decreased risk of dementia for individuals speaking three (OR = 0.62; 95% CI = 0.16-2.41) or four or more languages (OR = 0.53; 95% CI = 0.06-4.91) while participants speaking two languages were no more at risk for dementia than monolinguals (OR = 1.08; 95% CI = 0.43-2.69). Discussion: Initially, multilingualism did not appear to confer protection against AD. After holding grammatical complexity constant across all participants, however, multilingualism was found to be associated with AD risk. Therefore, linguistic ability confounded the initial relationship measured by this study. When the association between multilingualism and time of dementia onset was evaluated, participants were no more likely to develop dementia in one time period than another, and monolingual participants were no more likely to develop dementia in earlier time periods than multilinguals. While a trend of decreasing dementia risk with ascending number of languages spoken was not observed, speaking four or more languages was consistently associated with decreased dementia risk compared to speaking one language. The presence of an ApoE-E4 allele and low linguistic ability had a strong and consistent significant association with increased AD and dementia risk. Therefore, the influence of these variables on the association of multilingualism with AD and dementia is worthy of further exploration. Overall, this study provided some support for a protective effect of multilingualism on AD and dementia. Some of the present investigation’s results differ, however, from those of previous studies. This is not surprising, considering the present study utilized different methodologies than other studies in this research area. For instance, our study employed a definition of multilingualism based on self-report data – participants were classified as multilingual based on the number of languages they reported proficiency with. Therefore, our definition of multilingualism was less strict than definitions used in previous studies. However, our study employed much stricter outcome criteria than those used in previous studies, as our study is the first in this area to confirm AD cases with AD neuropathology evaluations. Our study is also the first io utilize prospective data and to include participants who remained dementia-free in addition to participants developing AD and dementia. In addition, this is the only study in this research area to evaluate the relationship of multilingualism with AD and dementia in the context of important covariates such as ApoE-E4 status and linguistic ability. Therefore, while some of our results contrast with other findings in this area, this is understandable given our novel methodologies. A broad range of study methods must be used in the future if we are to generate the depth of evidence needed for a full understanding of the relationship of multilingualism with AD and dementia. A better understanding of this relationship may also provide insight into both cognitive and brain reserve mechanisms, which could help more individuals maintain cognitive function into late life.

Background: Premenopausal black women have a 2-3 fold greater rate of coronary heart disease (CHD) than premenopausal white women. The purpose of this study was to provide insight into the reasons for this difference. Methods and Results: We compared CHD risk factors in 100 black and 100 white, healthy premenopausal women age 18-45 years and of relatively advantaged socioeconomic status. Black women consumed diets higher in saturated fat and cholesterol (12% of kcal as saturated fat and 360 mg of cholesterol per day) than did white women (10% of kcal and 290 mg/day) (p=0.008). Black women also had a higher body mass index (BMI) (32.0±9.2 vs. 29.0±9.4 kg/m², p=0.021), and higher systolic (124±17 vs. 115±14 mmHg, p<0.0001), and diastolic (79±14 vs. 75±11 mmHg, p=0.048) blood pressures. The mean plasma Lp(a) concentration was higher in the black women (40.2±31.3 mg/dl) than in the white women (19.2±23.7 mg/dl)(p<0.0001). The black women, however, had lower plasma triglyceride levels (0.91±0.46 vs. 1.22±0.60 mmol/L, p<0.0001), and a trend toward higher HDL cholesterol levels (1.37±0.34 vs. 1.29±0.31 mmol/L, p=0.064) than the white women. Plasma total and LDL cholesterol levels were similar. Rates of cigarette smoking and alcohol intake were low and similar between the races. Black women additionally had higher levels of plasma total homocysteine (8.80 vs. 7.81 μmol/L, p=0.013), lower plasma folates (3.52 vs. 5.23 ng/ml, p<0.0001), and higher vitamin B₁₂ levels (522 vs. 417 pg/ml, p<0.0001) than white women. More white women than black women took a multivitamin supplement (42.4% vs. 24.7%, p=0.019). When adjusted for multivitamin use, homocysteine levels did not differ, but plasma folate remained significantly lower in black women. Sixty-eight percent of black women carried the wild-type methylenetetrahydrofolate reductase genotype, 32.0% were heterozygotes, and none were homozygotes. Of the white women, 47.4% were wild-type, 40.3% heterozygotes, and 12.3% homozygotes (p=0.013). Conclusions: Premenopausal black women consumed more saturated fat and cholesterol and had a higher mean body mass index, blood pressure, Lp(a), and plasma total homocysteine levels than white women. These differences in coronary risk factors may explain the higher incidence of CHD in premenopausal black compared to white women. / Graduation date: 1998

Thesis (Ph. D.)--University of Hawaii at Manoa, 1994. / Includes bibliographical references (leaves 126-143). / Microfiche. / xi, 143 leaves (1 folded), bound ill. 29 cm

<p>The aim of this study was to assess the factors associated with HIV risk behaviours in women in Livingstone, Zambia. A cross-sectional analytical survey using a structured questionnaire was carried out in two sites in Livingstone, which were selected on the basis of differences in socio-economic status. Married women and women in the urban community are at high risk of contracting HIV and every effort should be made to ensure that HIV/AIDS programmes help to reduce their vulnerability to HIV infection.</p>

<p>Physical inactivity is one of the leading risk factors for major non-communicable diseases, which contribute substantially to the global burden of chronic diseases, disability and death. The burden of disability, morbidity and mortality, attributable to non-communicable diseases, is currently enormous in the developed countries and is increasingly growing in the developing countries. The purpose of this study was to investigate the relationship between participation in physical activity and health risk behaviours among high school students in the Mtwara region, Tanzania. The objectives of this study were to identify the physical activity levels among high school students in Mtwara region, Tanzania, to identify health risk behaviours among the above mentioned high school students, to identify the factors that influenced them to engage in health risk behaviours and to establish the relationship between physical activity and health risk behaviours.</p>

Aim: - This research project was a community based intervention study of the impact of increased access to exercise opportunities and kangaroo meat on diabetes and cardiovascular disease risk factors in the Aboriginal community of Woorabinda. The exercise and dietary interventions were based on principles of community development including: training of community members to conduct exercise programmes; supply of exercise equipment; training of community members to harvest kangaroos; establishment of a butcher’s apprenticeship; and establishment of the kangaroo meat processing in the community. Methodology: - The impact of the kangaroo meat and exercise interventions on diabetes and cardiovascular disease risk factors was determined by pre- and post-intervention assessment. Pre-intervention baseline data was obtained by community clinical assessment, household meat surveys,and community exercise surveys. The post-intervention assessment was a repeat of modified meat and exercise surveys. A post-intervention community clinical assessment was not conducted due to unresolved difficulties encountered in conducting the research project. The Study Population: - The study population for the community clinical assessment and exercise surveys were volunteer adults B 15 years of age. Approximately one third of the adult population participated in the community clinical screening, 29% in the pre-intervention exercise survey, and 20.2% in the post-intervention exercise survey. The meat surveys were conducted on a household basis. There was 84% household participation in the pre-intervention meat survey and 44.3% postintervention. Findings and Discussion: - It was found that the exercise and kangaroo meat interventions did not follow the planned linear trajectory but rather proceeded in an episodic and incremental manner. The community clinical assessment found a high prevalence of diabetes (18.6%; 95% CI, 13.04 – 24.36%) and impaired glucose tolerance (13.2%; 95% CI, 8.3 – 18.1%) comparable to that found in other Indigenous communities. There was a low prevalence of hypercholesterolaemia (30.9%; 95% CI, 24.2 – 37.6%) and hypertension (19.7%; 95% CI, 13.9 – 25.5%). High prevalence for other diabetes and cardiovascular disease risk factors were found including: current smoking (48.3%;95% CI, 40.95 – 55.64%); and obesity as measured by body mass index (35%; 95% CI, 27.9 – 42%), waist circumference (83.7%; 95% CI, 76.5 - 90.86% in women; and 55.5%; 95% CI, 43.9 – 61.1% in men), and waist / hip ratio (75.5%; 95% CI, 66.9 – 84% in women; and 57.1%; 95% CI,46 – 68% in men). A high prevalence of abnormal ACR was found: the prevalence ACR 3.4 – 33.9g/mol was 11.7% (95% CI, -ve2.39 – 25.7%), and the prevalence ACR B 34g/mol was 7.6% (95% CI, -ve6.8 – 22%). The prevalence of proteinuria was 67.8% (95% CI, 59.3 – 76.3%). It was found that this high prevalence of renal disease indicators coincided with an escalating incidence of end-stage renal disease in the community. Analysis of the kangaroo meat surveys found evidence of positive dietary change including reduction in the amount of fat used to cook non-kangaroo meats, and a positive shift in stage of change for cooking for family health. In regard to kangaroo meat however it was found that the most common cooking method was unfavourable with the nutritional value of the meat being compromised by a significantly higher prevalence of frying than other meats. It was further found that this method for cooking kangaroo meat was unchanged by the research intervention. Evidence of positive change was also seen in analysis of the exercise surveys. It was found that at baseline, and post-intervention, that the majority of adults in the community met the recommended duration for exercise per week through activities of daily living. There was a positive shift in stage of change for exercise behaviour with a significant movement beyond the ‘precontemplation' towards the ‘action’ and ‘maintenance’ stages of change. This positive shift in thinking about exercise corresponded with a significant increase in the proportion of people exercising specifically for health and fitness. Conclusion: - Though not all the proposed intervention objectives were accomplished the research project contributed to furthering community aspirations and capacity. The community clinical assessment provided a useful overview of the health status of Woorabinda, and an opportunity of a thorough health check for community members. The community clinical assessment drew attention to future projections of disease in Woorabinda and prompted a concentrated health system response. Evidence of positive change in regard to meat consumption and exercise behaviours were found, changes however were slow and uneven. Improved infrastructure was important to increasing community capacity for kangaroo meat supply and exercise, but essential to the sustainability of community initiatives is skilled people, and on-going maintenance and support. The findings of this study indicate that simplistic assumptions around the health benefits of ‘traditional’ diet need to be reconsidered cognisant that communities such as Woorabinda are cultures in transition. Whilst limitations in the methodology require the findings to be considered with caution, this study provides useful evidence for planning future health education and health promotion initiatives for Woorabinda.

ABSTRACT Background Oesophageal cancer has a high mortality; it is the 6th most common cause of death due to cancer worldwide. Of the common subtypes of oesophageal cancer, it is the adenocarcinomas that have been rising rapidly in incidence throughout the western world. The incidence of adenocarcinomas now exceeds the previously common squamous cell carcinoma. These recent changes in the incidence patterns of oesophageal cancer suggests that the environmental risk factors associated with these subtypes differ, and that changes in the prevalence of these exposures over time are the most likely explanation for the observed shifts in the incidence. However, due to its low incidence until a few decades ago, the adenocarcinoma subtype has been less studied compared to squamous cell carcinoma, and the environmental factors associated with this cancer have not been so clearly defined. Smoking and alcohol have been the strongest environmental risk factors reported for oesophageal squamous cell carcinoma (OSCC) whereas for oesophageal adenocarcinoma (OAC), the effect of smoking appears to be weaker, and the evidence for an effect of alcohol is scant and inconsistent. However, epidemiologic studies consistently identify people with frequent symptoms of gastro-oesophageal reflux (GOR) as having the highest risk of OAC, but the effect of GOR on OSCC has been negligible. Furthermore, it has been argued that adenocarcinoma occurring at the gastro-oesophageal junction (GOJAC) may have different aetiology again. Together, these reports suggest the three subtypes of oesophageal cancers (OAC, GOJAC and OSCC) may arise through different mechanisms with different strengths in the impact of risk factors. This thesis investigated the independent associations of smoking, alcohol and gastro-oesophageal reflux on cancers of the oesophagus by considering the possibility of variation in the risks due to differences in the dose effect patterns of various measures such as smoking, alcohol and GOR. Method Data from a population-based case-control study of oesophageal and ovarian cancers in Australia were used. Study participants comprised histologically confirmed cases of OSCC (n=308), OAC (n=367) and GOJAC (n=426) who were frequency matched to 1580 controls from the general population. Exposure history for both cases and controls were derived from health and lifestyle questionnaires. Unconditional multivariate logistic regression was used to calculate the odds ratios and 95% confidence intervals for the risk factors analysed. In addition, generalised additive model with a logit link was also used to explore and present the non-linearity in the dose effect pattern for continuous exposures adjusting for other confounding factors. The effects of two exposures combined on these cancers were assessed by obtaining synergy index. Results Smokers were at significantly higher risk of all three subtypes of oesophageal cancer with the risk greatest for OSCC. The effect of smoking was greater for adenocarcinoma occurring at the gastro-oesophageal junction compared to that of the oesophagus. Of the various measures of smoking, duration was significantly associated with all three subtypes of cancer whereas intensity was associated with only OSCC and GOJAC and the dose effect was non-linear. Time since quitting was associated with a steady decline in risk of all three cancers emphasising the health benefits of quitting among smokers. Alcohol was not associated with OAC or GOJAC but was significantly associated with OSCC among those drinking in excess of 170g/week. The association between alcohol and OSCC was modified by smoking; the association with alcohol was significantly greater among current smokers with effect. Low to moderate wine consumption was associated with significant risk reduction for all three cancers compared to non-drinkers. Increased frequency of GOR symptoms was associated with increased risks of OAC and GOJAC, although the risk of OSCC was constrained to frequent GOR symptoms only. The effect of GOR symptoms were exacerbated by smoking whereas it was weakened by regular NSAID use. Lastly, the sensitivity analysis that assessed the effect of non-participation among controls in the estimated effect of smoking and BMI (the two risk factors most likely to be affected by non-participation) showed a slight overestimation of effect of smoking assuming higher exposure rate among non-participants but not BMI while the effect remained strong and statistically significant. Conclusion Smoking, alcohol and GOR symptoms were the environmental factors strongly associated with all subtypes of oesophageal cancers. However, the dose effect patterns of these exposures varied by cancer subtypes. Smoking and alcohol were the larger contributing factors for OSCC whereas smoking and GOR symptoms had greater impact on OAC and GOJAC. Low to moderate wine consumption and regular NSAID use reduced the risk of all three subtypes significantly. While selection bias may have led to mildly inflated risks for smoking, the effects persisted even when modelled under extreme scenarios of biased participation amongst controls, and there was no evidence that selection bias materially affected the other associations.

ABSTRACT Background Oesophageal cancer has a high mortality; it is the 6th most common cause of death due to cancer worldwide. Of the common subtypes of oesophageal cancer, it is the adenocarcinomas that have been rising rapidly in incidence throughout the western world. The incidence of adenocarcinomas now exceeds the previously common squamous cell carcinoma. These recent changes in the incidence patterns of oesophageal cancer suggests that the environmental risk factors associated with these subtypes differ, and that changes in the prevalence of these exposures over time are the most likely explanation for the observed shifts in the incidence. However, due to its low incidence until a few decades ago, the adenocarcinoma subtype has been less studied compared to squamous cell carcinoma, and the environmental factors associated with this cancer have not been so clearly defined. Smoking and alcohol have been the strongest environmental risk factors reported for oesophageal squamous cell carcinoma (OSCC) whereas for oesophageal adenocarcinoma (OAC), the effect of smoking appears to be weaker, and the evidence for an effect of alcohol is scant and inconsistent. However, epidemiologic studies consistently identify people with frequent symptoms of gastro-oesophageal reflux (GOR) as having the highest risk of OAC, but the effect of GOR on OSCC has been negligible. Furthermore, it has been argued that adenocarcinoma occurring at the gastro-oesophageal junction (GOJAC) may have different aetiology again. Together, these reports suggest the three subtypes of oesophageal cancers (OAC, GOJAC and OSCC) may arise through different mechanisms with different strengths in the impact of risk factors. This thesis investigated the independent associations of smoking, alcohol and gastro-oesophageal reflux on cancers of the oesophagus by considering the possibility of variation in the risks due to differences in the dose effect patterns of various measures such as smoking, alcohol and GOR. Method Data from a population-based case-control study of oesophageal and ovarian cancers in Australia were used. Study participants comprised histologically confirmed cases of OSCC (n=308), OAC (n=367) and GOJAC (n=426) who were frequency matched to 1580 controls from the general population. Exposure history for both cases and controls were derived from health and lifestyle questionnaires. Unconditional multivariate logistic regression was used to calculate the odds ratios and 95% confidence intervals for the risk factors analysed. In addition, generalised additive model with a logit link was also used to explore and present the non-linearity in the dose effect pattern for continuous exposures adjusting for other confounding factors. The effects of two exposures combined on these cancers were assessed by obtaining synergy index. Results Smokers were at significantly higher risk of all three subtypes of oesophageal cancer with the risk greatest for OSCC. The effect of smoking was greater for adenocarcinoma occurring at the gastro-oesophageal junction compared to that of the oesophagus. Of the various measures of smoking, duration was significantly associated with all three subtypes of cancer whereas intensity was associated with only OSCC and GOJAC and the dose effect was non-linear. Time since quitting was associated with a steady decline in risk of all three cancers emphasising the health benefits of quitting among smokers. Alcohol was not associated with OAC or GOJAC but was significantly associated with OSCC among those drinking in excess of 170g/week. The association between alcohol and OSCC was modified by smoking; the association with alcohol was significantly greater among current smokers with effect. Low to moderate wine consumption was associated with significant risk reduction for all three cancers compared to non-drinkers. Increased frequency of GOR symptoms was associated with increased risks of OAC and GOJAC, although the risk of OSCC was constrained to frequent GOR symptoms only. The effect of GOR symptoms were exacerbated by smoking whereas it was weakened by regular NSAID use. Lastly, the sensitivity analysis that assessed the effect of non-participation among controls in the estimated effect of smoking and BMI (the two risk factors most likely to be affected by non-participation) showed a slight overestimation of effect of smoking assuming higher exposure rate among non-participants but not BMI while the effect remained strong and statistically significant. Conclusion Smoking, alcohol and GOR symptoms were the environmental factors strongly associated with all subtypes of oesophageal cancers. However, the dose effect patterns of these exposures varied by cancer subtypes. Smoking and alcohol were the larger contributing factors for OSCC whereas smoking and GOR symptoms had greater impact on OAC and GOJAC. Low to moderate wine consumption and regular NSAID use reduced the risk of all three subtypes significantly. While selection bias may have led to mildly inflated risks for smoking, the effects persisted even when modelled under extreme scenarios of biased participation amongst controls, and there was no evidence that selection bias materially affected the other associations.

ABSTRACT Background Oesophageal cancer has a high mortality; it is the 6th most common cause of death due to cancer worldwide. Of the common subtypes of oesophageal cancer, it is the adenocarcinomas that have been rising rapidly in incidence throughout the western world. The incidence of adenocarcinomas now exceeds the previously common squamous cell carcinoma. These recent changes in the incidence patterns of oesophageal cancer suggests that the environmental risk factors associated with these subtypes differ, and that changes in the prevalence of these exposures over time are the most likely explanation for the observed shifts in the incidence. However, due to its low incidence until a few decades ago, the adenocarcinoma subtype has been less studied compared to squamous cell carcinoma, and the environmental factors associated with this cancer have not been so clearly defined. Smoking and alcohol have been the strongest environmental risk factors reported for oesophageal squamous cell carcinoma (OSCC) whereas for oesophageal adenocarcinoma (OAC), the effect of smoking appears to be weaker, and the evidence for an effect of alcohol is scant and inconsistent. However, epidemiologic studies consistently identify people with frequent symptoms of gastro-oesophageal reflux (GOR) as having the highest risk of OAC, but the effect of GOR on OSCC has been negligible. Furthermore, it has been argued that adenocarcinoma occurring at the gastro-oesophageal junction (GOJAC) may have different aetiology again. Together, these reports suggest the three subtypes of oesophageal cancers (OAC, GOJAC and OSCC) may arise through different mechanisms with different strengths in the impact of risk factors. This thesis investigated the independent associations of smoking, alcohol and gastro-oesophageal reflux on cancers of the oesophagus by considering the possibility of variation in the risks due to differences in the dose effect patterns of various measures such as smoking, alcohol and GOR. Method Data from a population-based case-control study of oesophageal and ovarian cancers in Australia were used. Study participants comprised histologically confirmed cases of OSCC (n=308), OAC (n=367) and GOJAC (n=426) who were frequency matched to 1580 controls from the general population. Exposure history for both cases and controls were derived from health and lifestyle questionnaires. Unconditional multivariate logistic regression was used to calculate the odds ratios and 95% confidence intervals for the risk factors analysed. In addition, generalised additive model with a logit link was also used to explore and present the non-linearity in the dose effect pattern for continuous exposures adjusting for other confounding factors. The effects of two exposures combined on these cancers were assessed by obtaining synergy index. Results Smokers were at significantly higher risk of all three subtypes of oesophageal cancer with the risk greatest for OSCC. The effect of smoking was greater for adenocarcinoma occurring at the gastro-oesophageal junction compared to that of the oesophagus. Of the various measures of smoking, duration was significantly associated with all three subtypes of cancer whereas intensity was associated with only OSCC and GOJAC and the dose effect was non-linear. Time since quitting was associated with a steady decline in risk of all three cancers emphasising the health benefits of quitting among smokers. Alcohol was not associated with OAC or GOJAC but was significantly associated with OSCC among those drinking in excess of 170g/week. The association between alcohol and OSCC was modified by smoking; the association with alcohol was significantly greater among current smokers with effect. Low to moderate wine consumption was associated with significant risk reduction for all three cancers compared to non-drinkers. Increased frequency of GOR symptoms was associated with increased risks of OAC and GOJAC, although the risk of OSCC was constrained to frequent GOR symptoms only. The effect of GOR symptoms were exacerbated by smoking whereas it was weakened by regular NSAID use. Lastly, the sensitivity analysis that assessed the effect of non-participation among controls in the estimated effect of smoking and BMI (the two risk factors most likely to be affected by non-participation) showed a slight overestimation of effect of smoking assuming higher exposure rate among non-participants but not BMI while the effect remained strong and statistically significant. Conclusion Smoking, alcohol and GOR symptoms were the environmental factors strongly associated with all subtypes of oesophageal cancers. However, the dose effect patterns of these exposures varied by cancer subtypes. Smoking and alcohol were the larger contributing factors for OSCC whereas smoking and GOR symptoms had greater impact on OAC and GOJAC. Low to moderate wine consumption and regular NSAID use reduced the risk of all three subtypes significantly. While selection bias may have led to mildly inflated risks for smoking, the effects persisted even when modelled under extreme scenarios of biased participation amongst controls, and there was no evidence that selection bias materially affected the other associations.