Physical inactivity and adiposity are independent risk factors for several chronic conditions including coronary heart disease. Activity and adiposity also modulate psychophysiological responses to psychosocial stress. Since heightened cardiovascular and inflammatory responses to mental stress predict cardiovascular risk, these two factors may influence cardiovascular risk through modulation of autonomic reactivity to stress. However, experimental evidence to support this hypothesis is scarce. The aim of this project is to investigate the associations between physical activity, adiposity, mental stress and mood and physiological reactivity using naturalistic and controlled laboratory methods. Study one examined the association between self-reported physical activity participation, diurnal cortisol rhythm and mood symptoms in everyday life. Study two used an experimental design to examine the effect of physical activity on mood symptoms and on cardiovascular and inflammatory responses to acute mental stress. Exercise withdrawal was used as a model of physical inactivity to induce mood disturbances in healthy, active participants. Several stress-induced markers relevant in cardiovascular disease were examined including pro-inflammatory factors and cortisol. Study three examined the effect of adiposity on physiological responses to acute mental stress and mood. Weight loss was experimentally induced through caloric restriction in overweight or obese women. Responses to acute stress were compared before and after weight loss. Cardiovascular and inflammatory responses to acute stress were evaluated to establish whether adiposity is associated with a heightened or blunted response. The combination of studies presented in this thesis provides insight into the complex relationships that links behavioural factors such as physical activity with mood and stress. An understanding of the mechanisms involved in the association between adiposity, physical activity and cardiovascular risk is invaluable in informing preventive strategies and health related programmes.

Background: There are gradients in general and oral health by socioeconomic position (SEP). Psychosocial processes play an important role to explain those social gradients. One such process is Antonovsky’s Sense of Coherence (SOC) construct. However, the contribution of SOC in explaining social gradients in oral health has not been fully explored. SOC relates to how individuals make sense of the world, use required resources to respond to it and feel their responses are meaningful and make sense emotionally. Aim and Objectives: This thesis aimed to assess the role of SOC in the relationship between SEP and oral health in Finnish dentate adults. The objectives were to assess the role of SOC as a moderator of the relationship of adulthood SEP with behaviours and oral health; to assess the role of SOC as a mediator of the relationship of childhood SEP with adult behaviours and oral health; and to assess the contribution of SOC to oral health independently of socio-demographic and behavioural factors. Methods: Data were from the nationally representative Finnish Health 2000 Survey of adults aged 30 years and over. Childhood SEP was indicated by parental education and adulthood SEP by education and income. SOC was measured by the Orientation to Life questionnaire. Outcomes were behaviours, clinical oral health measures and self-rated oral health. Regression analysis and structural equation modelling were used. Results: Findings did not support a moderating role of SOC in the relationship of adulthood SEP with behaviours and oral health. SOC was a modest mediator of the relationship of childhood SEP with adult behaviours and oral health, compared to the role of adulthood SEP. Finally, SOC was independently associated with oral health. Conclusion: SOC appears to be an important psychosocial disposition for the development and maintenance of individuals’ oral health. The association between SOC and oral health was not explained away by socio-demographic or behavioural factors.

Background: There are few studies on the patterns of clustering of health-related behaviours among adolescents. Most of the studies have not adequately explored the clustering of a wide range of health-related behaviours. The very few studies examining the determinants and variations in the clustering of health-related behaviours among adolescents merely assessed the variations in the clustering of health-related behaviours by socio-demographic factors. Moreover, they did not use rigorous statistical methods. Objectives: To assess the distribution of six health-related behaviours, namely, intake of fruits, intake of sweets, toothbrushing, physical activity, smoking, and physical fighting, in Saudi Arabian male adolescents. To identify the patterns of clustering of the six health-related behaviours in Saudi Arabian male adolescents. To assess the variations of the clustering count of the six health-related behaviours by demographic and socioeconomic status factors. To identify the role of psychosocial determinants of the clustering count of the six health-related behaviours. The psychosocial determinants assessed were self-confidence, future orientation, life satisfaction, and peer relationships. Methods: A representative stratified cluster random sample of 1335 Saudi Arabian male adolescents living in Riyadh city took part in this study. A modified WHO questionnaire was used to assess health-related behaviours. Hierarchical Agglomerative Cluster Analysis (HACA) was used as a first step to identify broad clusters of the six health-compromising behaviours (HCBs). Then clustering was assessed using an observed to expected ratio (O/E) method for possible combinations of the six HCBs. Poisson regression models were constructed to assess variations between selected explanatory variables and clustering count of the six HCBs. Results: About 85% of adolescents had low fruit consumption, 74% brushed their teeth less frequently, 64% had low physical activity, 51% had been involved in physical fighting, 43% ate more sweets, and 23% smoked tobacco. Hierarchical Agglomerative Cluster Analysis suggested two broad and stable clusters of the six health-compromising behaviours among Saudi male adolescents; 1. The first cluster included low fruit consumption, less frequent toothbrushing and low physical activity. 2. The second cluster included high sweets consumption, smoking and physical fighting. There were more than sixty patterns of health-compromising behaviours among both younger and older adolescent boys. Most patterns with the highest O/E ratio included smoking behaviour. Older adolescents (17-19 years) were more likely to have a greater clustering count of HCBs (RR: 1.06; 95% CI: 1.01-1.13) compared to younger adolescents (13-14 years). Adolescents whose fathers’ education was up to secondary school or lower had a greater rate for clustering count of HCBs (RR: 1.07; 95% CI: 1.01-1.16) compared to those whose fathers completed higher education. There were no significant differences in clustering count of HCBs in relation to adolescents’ perceived self-confidence (RR: 1.05; 95% CI: 0.99-1.12) and future orientation (RR: 1.05; 95% CI: 0.98-1.13). Adolescents who had low levels of life satisfaction (RR: 1.20; 95%CI: 1.10-1.31) and those who had middle levels of life satisfaction (RR: 1.07; 95% CI: 1.00-1.16) had a greater rate of clustering of HCBs compared to those with a high level of life satisfaction. The clustering rate of HCBs was higher for each extra evening meeting with friends (RR: 1.03; 95%CI: 1.01-1.04). Conclusions: The six health-compromising behaviours clustered into two broad clusters. Most of the clustering patterns with the highest O/E ratio included smoking behaviour. Adolescents’ father’s education level and life satisfaction were significantly negatively associated with clustering of health-compromising behaviours. Age of adolescents and evening meetings with friends were significantly positively associated with clustering of health-compromising behaviours. There were no significant variations in the clustering of health-compromising behaviours with adolescents’ self-confidence and future orientation, after adjusting for age and father’s education level.

In England over 2 million young people were tested for Chlamydia trachomatis (chlamydia) in 2010/2011 to control infection and prevent reproductive health problems. Since 2008, the National Chlamydia Screening Programme (NCSP) has delivered a significant proportion of this testing. The main part of my thesis focused on delivery, specifically on socio-economic variations in access to chlamydia testing amongst young people. It is not known whether testing reaches people in disadvantaged socio-economic circumstances (SEC) who often have worse access to preventive healthcare yet poorer outcomes than socially advantaged groups. Firstly, I undertook a systematic review and re-analysis of Natsal-2000 data to select suitable SEC indicators for use in young people. Living in disadvantaged areas and a lack of education were most consistently associated with a higher risk of chlamydia. Secondly, I examined socio-economic variations in young people's access to chlamydia testing at national and local levels. Data from the NCSP's first year of national delivery indicated that chlamydia screening reached more individuals in disadvantaged areas, where positivity was also higher. A cross-sectional study informed by focus groups found that local delivery varied by service model, particularly when SEC was measured by educational participation. It also showed that other social factors need to be considered to understand how SEC may affect young people's risk of chlamydia. The last part of my thesis evaluated the NCSP's rationale. In 2009, the National Audit Office questioned the justification for the Programme, given the lack of evidence surrounding the benefits of screening. My qualitative study revealed there was an implicit driver around improving young people's sexual healthcare underpinning the NCSP's establishment, in combination with the explicit aims of chlamydia control. The implications of my findings with respect to the delivery of chlamydia testing are discussed in light of this implicit rationale for chlamydia screening.

This thesis examined the association between long working hours and health outcomes with high public health relevance; coronary heart disease (CHD), type 2 diabetes, depression and sleep disturbances, in a cohort of middle-aged white-collar British civil servants. Earlier research has shown mixed results on the topic but the evidence relies largely on cross-sectional and case-control studies. I used data from the longitudinal Whitehall II study where self-reported working hours were assessed at phase 3 (1991-1993) when the employed participants were 39 to 61 years of age. The second assessment of working hours was at phase 5 (1997-1999). CHD was assessed at phases 3, 5, and 7 (2003-04); type 2 diabetes at phases 3 and 7; depression at phases 3 and 5, and sleep disturbances at phases 3, 5, and 7. Analyses of each outcome disorder were based on a cohort free from the specific disorder at baseline. Follow-up time ranged between 6 to 11 years depending on the outcome, and the number of participants in each part of the study ranged between 913 and 6014 depending on the baseline and follow-up data available. Incidence of CHD was indicated by CHD death, non-fatal myocardial infarction and angina defined on the basis of clinical examinaton, clinical records, and nitrate medication use. Type 2 diabetes was ascertained from high fasting or postload plasma glucose levels, self-reported information on doctor-diagnosed diabetes and use of medication assessed during clinical examinations. Onset of depression was assessed by University of Michigan version of Composite International Diagnostic Interview (UM-CIDI) and onset of sleep disturbances were requested by survey questions on sleep length and four types of sleep disturbances (the Jenkins Scale). Several known confounding and mediating covariates were assessed and included in the analyses. Working 11-12 hours per day (or >55 hours per week) at baseline, compared with working 7-8 hours per day (or 35-40 per week) was associated with an increased risk of CHD, depression and most types of sleep disturbances at follow-up. Working long hours was not associated with an increased incidence of type 2 diabetes except among prediabetic participants. These findings were robust to adjustments for relevant confounding factors at baseline. The results of this thesis indicate that long working hours could be recognized as a potential risk marker for the development of CHD, depression, and sleep disturbances. However, the results are generalisable to British white-collar workers only, and as this study is based on observational data it is not known whether the associations are causal.

BACKGROUND: With population ageing, prevention of frailty is increasingly important. However, significant gaps in the evidence base exist. Accordingly, the purpose of this thesis was to: (1) identify the current measures of frailty undertaking an overview; (2) validate the ‘phenotype of frailty’ using data from the Whitehall II study; and (3) examine the relation of cardiovascular disease (CVD) and diabetes risk factors with future frailty risk. METHODS AND RESULTS: For objective 1, a literature review identified 27 original articles describing 27 different frailty measurements. Of them, the most tested and frequently used measure was the ‘phenotype of frailty’ which comprises five components: weight loss, exhaustion, physical activity, walking speed, and grip strength. For objectives 2 and 3, I used data from the Whitehall II study, an occupationally-based cohort of 10,308 British men and women aged 35-55 years followed-up since 1985. Of the participants aged 55 to 79 years in 2007-2009 (n=5,169), 2.8% were frail and 38.6% pre-frail. Using survival analyses, in sex- and age-adjusted model, compared with the non-frail group, the frail group was 2.40 (95% confidence interval (CI): 1.83, 3.14) times more likely to be hospitalised for any cause during the mean follow-up of 15.2 months, while for the pre-frail group the risk was 1.20 (95%CI: 1.06, 1.35) greater. Logistic regression models were used to examine the performance of risk algorithms for CVD and diabetes assessed in 1997-1999 in predicting frailty in 2007-2009. CVD and diabetes risk scores were significantly associated with frailty: odds ratios per 1-standard deviation increment (disadvantage) in CVD scores ranged from 1.17 (95%CI: 1.10, 1.25) to 1.20 (95%CI: 1.13, 1.28) and in diabetes scores ranged from 1.05 (95%CI: 0.98, 1.14) to 1.27 (95%CI: 1.17, 1.37) depending on the risk score used. CONCLUSIONS: Both frailty and pre-frailty are associated with increased risk of hospitalisation. Better prevention of cardiovascular and diabetes risk factors in midlife is likely to reduce frailty at older ages.

Congenital heart defects (CHDs) are the most prevalent type of congenital anomaly and the principal cause of death from congenital anomalies during infancy. However reliable information about longer-term survival and the factors predictive of death during later childhood is lacking. In this thesis, I report the findings of a multicentre cohort study established to estimate survival with serious CHDs from birth to 15 years of age and to investigate the independent effects of early life and procedure-related factors on mortality risk. The hospital records of 3,897 children (57% boys) born in the UK between 1992 and 1995 and who required intervention or died in the first year of life were reviewed. Multiple imputation of missing data including time varying covariates was performed using multilevel statistical techniques. Survival was estimated using Kaplan-Meier, Cox proportional hazards and multilevel models and predictive factors evaluated using step-wise logistic regression models using complete case as well as multiple imputed datasets. By 15 years of age, 932 children had died, 144 without any procedure. Overall, survival to one year was 79.8% (95% Confidence Intervals [CI] 78.5, 81.1%) and to 15 years was 71.7% (63.9, 73.4%), with variation by cardiac diagnosis. In the multilevel imputed data model including children who had at least one procedure, higher mortality risk was independently associated with multiple cardiac malformations, younger age at first procedure, lower weight and/or height, longer duration of cardiopulmonary bypass and circulatory arrest, and presence of pre- or post-procedure complications. Although the risk of death from CHDs is highest in the first year, an important continuing risk of death remains throughout childhood. Early life and procedure-related factors predict mortality risk independently of CHD diagnosis. Future research should address social, educational, and health outcomes in later childhood to understand the impact of advances in diagnosis and management of CHDs.

Complex diseases such as coronary heart disease (CHD) are a major and growing burden on public and individual health worldwide. As an increasing number of risk factors for CHD are identified, more questions arise over the pathogenesis of the disease, and potential new therapeutic and preventive strategies. Mendelian randomisation (MR) offers a means of addressing these questions, using common genetic variants as proxies for a putative risk factor. By exploiting the innate properties of the genotype of an individual, MR overcomes many limitations of traditional observational epidemiology. The lifelong and largely immutable influence of genetic variation on pathogenic mechanisms and risk of disease avoid confounding and reverse causation, and allow robust causal inference in the context of observational population studies. The validity of such a MR experiment is dependent on the choice of genetic instrument and considerations of specificity and phenotypic effect size are important in selecting the most appropriate instrument for different applications of MR. The considerations involved in optimising the selection of genetic instruments for MR were explored in the context of CHD. Using variants in a gene encoding a protein that may be a target for pharmacological intervention, the MR paradigm can be applied to the validation of potential therapeutic targets. The practical considerations, limitations, and utility of this approach were demonstrated in a proof-of-principle study of the widely prescribed and well characterised 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) reductase inhibitors (statins) and variants in the gene encoding their target (HMGCR). Associations of variants in HMGCR in a large collaborative sample of population studies were compared to the effects of statin treatment reported in randomised trials in order to investigate the strengths and limitations of the MR approach to drug target validation. Moreover, this investigation allowed the mechanism-based actions of statin therapy to be distinguished from those likely to be off-target. A causal role for inflammatory pathways is suggested in atherosclerosis, and higher circulating concentrations of inflammatory markers such as interleukin-6 (IL-6) are strongly associated in observational studies with higher risk of CHD. Two other inflammatory markers, C-reactive protein (CRP) and fibrinogen, have been shown using MR methods to be unlikely to play a causal role in atherogenesis, but interest in the role of IL-6 and potential for modulation of its actions as a preventive strategy in CHD remains. A iii monoclonal antibody, tocilizumab, licensed for the treatment of rheumatoid arthritis (RA), specifically inhibits the IL-6 receptor (IL-6R). In a large, collaborative general population sample, associations of common variants in the gene encoding the IL-6R (IL6R) with a range of biomarkers relevant to cardiovascular disease (CVD), and with risk of CVD events, were estimated. These estimates were compared with the effects of tocilizumab therapy on similar biomarkers in order to validate the IL-6R as a potential therapeutic target in CHD. Evidence from this large-scale genetic analysis and existing clinical trials of tocilizumab suggest inhibition of IL-6 signalling may be a valuable new therapeutic strategy for prevention of CHD. The work presented here makes a case for the wider use of the MR principle for improving efficiency in drug discovery. By exploiting findings from genome-wide association (GWA) studies, MR may be used to validated and prioritise drug targets earlier in development, and help to reduce rates of late-stage failure.

Objective: Maternal employment has been shown to influence a number of child and adolescent health, cognitive and behavioural outcomes. There are, however, only a limited number of UK studies using health measures as study outcomes. The aim of this thesis is to examine the influence of maternal employment during three periods of childhood on health outcomes among young adults aged 16–21 in the British Household Panel Survey (BHPS). Methods: The BHPS is an annual nationally representative panel study which started in 1991, and data from 18 waves were available for this thesis. There were 3,940 individuals with at least one measurement of self-rated health and/or psychological distress and/or smoking in young adulthood (age 16–21) and maternal employment prior to age 16 years. Other variables, such as gender, maternal age, maternal education and marital status, household income, maternal self-rated health, maternal psychological distress and maternal smoking, were used as covariates. Multilevel logistic regression (using clustering of repeated measurements within individuals) was used for the analysis. Results: 19% of young adults reported poor self-rated health, 23% reported psychological distress and 27% reported being current smoker. 56%, 76% and 79% had mothers who worked during the preschool (age 0–4 years), primary school (age 5–11) and secondary school (age 12–16) periods of their childhood. In general, the results suggest that maternal employment is protective for those from socially advantaged families, while it has no effect or increases the risk of poor self-rated health, psychological distress and smoking among those from less advantaged families. Conclusions: The associations between maternal employment during childhood and young adults’ health and smoking differ in different social groups. Maternal employment might have only a limited role in the health and health behaviours of young adults, and maternal education, household income and marital status seem to be stronger predictors of the study outcomes than maternal employment status.

This thesis contributes to our understanding of the individual and community assets available to ethnic minority people living in areas characterised by high concentrations of co-ethnics. It has been hypothesized that positive attributes found in areas of greater concentration of ethnic minority people, or ethnic density, might provide ethnic minority residents with health promoting, or protective effects. This study explored the effect of ethnic density on the health of ethnic minority people in the UK. It proposed and tested three pathways by which ethnic density is hypothesised to operate: through a change in racism-related social norms; through buffering the detrimental effects of racism on health; and through an increase in civic-political activity. Multilevel methods were applied to three nationally representative cross sectional studies, the 1999 and 2004 Health Survey for England; the Fourth National Survey of Ethnic Minorities; and the 2005 and 2007 Citizenship Survey. Results showed a stronger ethnic density effect on psychological outcomes, as compared to that found for physical health outcomes. Effect sizes were larger when the ethnic density of specific groups was analysed, but more likely to be significant when the density of all minority groups combined was considered. Analyses conducted to test the social norms model reported a significant reduction in experienced racism among ethnic minority people living in areas of high ethnic density, as compared to their counterparts who live in areas of reduced ethnic density. Examinations of the buffering effects of ethnic density indicated a tendency for a weaker association between racism and health as ethnic density increased, although interactions were mostly non-significant. Finally, ethnic minority people were not found to report higher civic engagement as ethnic density increased, but they were found to be more satisfied with local services and to report greater community cohesion.