Background: Nutrient profiling (NP) aims to identify healthier food options according to the content of selected ‘positive’ nutrients e.g. fibre, protein, and ‘negative’ nutrients e.g. sodium, saturated fat. The British and French food safety agencies developed the WXYfm and SAIN,LIM models, respectively. Their predictive validity in relation to chronic disease has yet to be demonstrated. Aim: To test the hypothesis that ‘healthy’ diets as defined by NP have predictive validity. Methods: Between 1991-93, 7,251 participants of the Whitehall II study completed a 127-item food frequency questionnaire (FFQ). WXYfm and SAIN,LIM scores for each FFQ-item were used to derive energy-weighted aggregate diet scores (AS) for each participant and NP model. Validity was assessed against baseline factors including dietary quality indices. Prospective associations were examined with incident CHD, diabetes and cancer, and all-cause mortality (318, 754, 251, and 524 events, respectively—median follow-up time was approximately 17 years). Results: AS were weakly associated with dietary quality indices. Cox modelling identified U-shaped associations (p quadratic trend <.05) between both AS and all outcomes except diabetes. Participants with middle AS had slightly reduced risk; SAIN,LIM estimates were significant for CHD and all-cause mortality. Dietary misreporting, particularly of energy-dense foods with high ‘negative’ nutrient content, was associated with BMI, hypertension and other risk factors, and explained much of the unexpected U-shaped AS-outcome associations. Alternative AS less sensitive to dietary misreporting confirmed the potential of NP as a public health tool. In particular, the WXYfm ‘positive’ nutrients predicted risk reduction for all outcomes. Conclusions: Predictive validity of the NP approach was partly established. The prospective effects of AS on chronic disease outcomes were confounded by the association between vascular risk and energy misreporting. Further predictive validity studies of NP methods ideally require food-based dietary assessment (e.g. diet diaries, 24h recalls) with less reporting bias.

Background: The relevance of neighbourhoods for inequalities in children’s development, while receiving increasing attention, is still debated. This PhD thesis aimed to examine whether children’s place of residence influenced their social and cognitive development, and to test two specific pathways through which such place effects might operate. It was hypothesised that schools contribute to the variability in children’s socio-emotional and cognitive outcomes across neighbourhoods, and further, that neighbourhood characteristics affect children via maternal psychological distress and parenting. Methods: Participants were 7-year-old children and their mothers from the UK Millennium Cohort Study. Cross-classified multilevel models were run to simultaneously estimate the variability in the child outcomes between neighbourhoods and schools; and to examine potential mediating effects via maternal psychological distress and selected parenting behaviours. Results: Most of the variability in children’s socio-emotional difficulties across neighbourhoods and schools was explained by the clustering of children from similar socioeconomic backgrounds. However, for children’s cognitive test performance, considerable variability between neighbourhoods and schools remained even after allowing for such compositional effects. Structural neighbourhood factors such as median household income were associated with cognitive outcomes, while neighbourhood social processes were related to children’s socio-emotional development. The data did not support a mediating role for maternal psychological distress in relation to teacher-reported socio-emotional difficulties or cognitive test performance. However, maternal psychological distress was on the pathway between social processes in the neighbourhood and socio-emotional difficulties reported by the mother. There was no evidence to suggest mediation via the examined parenting practices. Conclusions: Children’s experiences within their neighbourhoods and schools contribute to their social and cognitive development. Schools appear to be an important area for investment. However, given the fundamental role of families’ individual circumstances, an integrated approach is needed which combines policies directed at schools and neighbourhoods with measures that support children and their families directly.

Previous research strongly suggests that ethnic minorities are more likely to suffer a poorer health profile compared to the overall population. Trends have emerged to suggest that social factors such as socioeconomic status and health behaviours are not fixed across generations and have a role to play in these inequalities in health. This thesis investigated the differences in ethnic inequalities in health between the first and second generations, and determined the extent to which intergenerational changes in socioeconomic status and health behavioural factors might explain any variation that exists. The study used ethnically‐boosted data from the third sweep of the Millennium Cohort Study (n=14,860) and the combined 1999 and 2004 Health Survey for England (n=28,628). Crosssectional analysis investigated generational differences in self rated general health, limiting illness, obesity, hypertension, depression, psychological distress and a range of biomarkers of cardiovascular disease, across the major ethnic minority groups in the UK (Indian, Pakistani, Bangladeshi, Black Caribbean, Black African, Irish, Chinese and Other). Children were additionally assessed for levels of cognitive development using the British Abilities Scales II. The generational change in socioeconomic circumstances (social class, highest educational qualification and household income) and the extent of acculturation (current smoking and drinking status, dietary behaviours and patterns of breastfeeding, immunisations and physical exercise) was examined. Strong upward intergenerational socioeconomic mobility in ethnic minority groups did not lead to improving health profiles. The second generation required greater levels of social advantage than the first generation to achieve the same level of health. Acculturative shifts led to a worsening in health behaviours, although the degree of change was highly ethnic group specific. Findings showed that the social and economic contexts, and the cultural identities and behaviours of ethnic minorities, differ across generations, but ultimately their opposing influences on health result in stable overall patterns of health inequality across generations.

BACKGROUND: As early as the 17th century diabetes was linked to psychosocial stressors, yet this association still remains elusive. The impact of psychosocial work stressors on diabetes is also unknown, as are the pathways involved. AIM: To investigate the prospective effect and impact of psychosocial work stressors on incident T2DM and elucidate direct and indirect pathways in this association. METHODS: Prospective analysis (1991-2004) among 5895 Caucasian middle-aged civil servants free from baseline diabetes in the Whitehall II cohort study. T2DM was ascertained by an oral glucose tolerance test supplemented by self-reports over four consecutive waves of data collection. The demand/control/support and effort-reward imbalance models were used to assess psychosocial work stressors. RESULTS: Job strain (high job demands/low job control), iso-strain (high demands/low control/low work social support) and effort-reward imbalance (high efforts/low rewards at work) are associated with 60% to 2-fold higher risk of T2DM among women but not men. This effect is higher among obese and lower socioeconomic position individuals. An estimated 35-44% of T2DM cases are attributed to psychosocial work stressors among exposed women, while 10-15% of all cases in the study population are estimated to result from exposure to work stressors, assuming a causal association. Biological factors (obesity, HDL-cholesterol and markers of inflammation) explain 1/3 of the effect of psychosocial work stressors on T2DM. CONCLUSIONS: The observed association between psychosocial work stressors and T2DM is internally valid, temporal and biologically plausible, thus most likely causal. The external validity is questionable however as the effect was observed among a very specific sample of white-collar, Caucasian female civil servants. Given the interaction with body weight status, reduction of psychosocial work stressors would offer some benefits in light of the huge impact of the growing obesity epidemic on T2DM. Even though cardiometabolic and inflammatory factors explain a substantial part of the psychosocial effect on T2DM, other novel mediating factors should be identified. These results should be confirmed by other studies, preferably experimental, with special attention on gender differences. Policies for reduction of psychosocial work stressors should be informed by findings from such studies.

Learning about human papillomavirus (HPV) has been identified as a possible source of negative affect in women, but the extent and nature of these emotions is unclear, along with whether they are associated with negative attitudes and behaviours (particularly with respect to HPV vaccination). The goal of this thesis is to examine the psychological impact of human papillomavirus (HPV) information using measures of knowledge, behavioural intentions, mood, attitudes and implicit associations. Existing literature was reviewed to examine a range of methods and outcomes suitable for use. Study 1 examined responses to health information in adolescent-aged women using a randomised between-participants design, and was carried out in a classroom setting. Participants given information about HPV and cervical cancer showed strong interest in future vaccination and did not display any more anxiety (as measured by the short form of State Trait Anxiety Index; STAI) than those participants given alternative control information. Three further studies adapted and refined this method for use with older women of university-going age in one-on-one testing sessions. These studies employed an enhanced range of outcomes, many of which were administered as repeated measures, and although showed positive evaluations of HPV material, strong behavioural effects were more difficult to elicit. Again few effects of anxiety were observed between information conditions. Implicit evaluations of the concept of ‘cancer’ were also examined using a computer-based Implicit Association Test, which showed some evidence of changes in associations following information exposure. Correlates of changes in implicit associations were also examined, with some relationships shown with behaviour and knowledge uptake but not anxiety or attitudes.

Social inequalities in health remain a major social issue globally. One possible explanation of health inequality is health selection: in other words people with poor health move down the social hierarchy. This study started with the conceptual distinction between two types of health selection studies. Type I health selection study (the presence of health selection) examines the impact of poor health on the subsequent social mobility. On the contrary, type II study (the contribution of health selection to social inequalities in health) examines whether health selection changes social inequalities in health. The first 13 waves (1991-2003) of the British Household Panel Survey with 63599 observations from 8819 individuals were used. In accordance with the typology, two different approaches were applied to empirical and theoretical investigation. For type I study, a multilevel multinomial model to fit all possible transition from multiple origins was used to assess social mobility with regard to health status. For type II study, both empirical and hypothetical analyses are applied in order to address the relationship between social mobility, health selection, and social inequalities in health on the population-level framework. Findings from the type I study presented that health selection was negligible in mobility within employment indicated by class and income measures, although it was highly significant in the transition between employment and non-employment. In type II study, changes in social inequalities in health were associated with a set of elements extracted from a social mobility process. Varying levels of health selection and scales of social mobility result in different extents of change in social inequalities in health.

The overall aim of this thesis is to examine the relationship between psychosocial factors and health; the specific aims are: to investigate links between psychosocial factors; to investigate the relationship between psychosocial factors, health behaviour and biology and finally, to investigate these relationships in cross-cultural samples. The recent incorporation of positive aspects, such as happiness and increased social support, into models of health has indicated a protective link between psychosocial factors and health outcomes. Psychosocial factors may impact upon health through behavioural and biological pathways, and there may be interactions between psychosocial factors, including constructs such as psychological and social function and both behavioural and biological pathways, This thesis focuses on the association of three psychosocial constructs, positive well-being, social support and optimism, with health. The first study investigates the relationship between positive well-being and health behaviour in an international sample. The findings showed that life satisfaction was associated with increased healthy behaviours for smoking, exercise, fat intake, sun protection and fruit intake, with no relationship for alcohol consumption or fibre intake. The second study investigated the associations of positive well-being, social support and optimism, and found that social support was strongly related to positive well-being. This study also found a relationship between social support and exercise; between social support and cortisol, and an association between these and positive affect. The third study presents data from a Japanese sample. This study found that social support was related to positive well-being, although effects were different to those found in the UK study. Although effects were small and there were several null findings, overall this thesis concludes that social support and positive well-being may be a part of a protective network of wider psychosocial factors, and that effects on health are exerted by moderation of behavioural and biological pathways.

Coronary heart disease (CHD) is the leading cause of death in the UK. Despite this declines in death rates have been experienced within the UK since the 1970’s but these declines have not been experienced by South Asians. Studies on access have highlighted inconsistencies between South Asians and Whites. The present study was conducted to explore factors which may affect health care seeking behaviour following symptoms of angina amongst South Asians and Whites. 40 face to face interviews were conducted using a purposive sample from a quantitative survey (Chaturvedi et al, 1997). This sub-sample consisted of White and South Asian male and female participants aged between 35-55 years. A further 4 focus groups were conducted with another purposive sample of White and South Asian males and females aged between 35-55 years. This second sample was selected using the same sampling frame (Chaturvedi et al, 1997). Focus groups were only conducted with those participants who had not taken part in the individual face to face interviews. The individual face to face interviews consisted of 10 South Asian male and 10 South Asian female participants and 10 White male and 10 White female participants. The focus groups consisted of 5-6 participants each. The focus groups were homogenous for ethnicity and sex. The epistemological framework used to guide the analysis was social constructionism. Results indicated that trust was an important factor which affected health care seeking behaviour amongst South Asians and Whites. South Asians had high trust of their GP, were over reliant on the GP and had lower expectations following chest pain. White participants had low trust of their GP, were less reliant on the GP and had high expectation following symptoms of chest pain. Issues of trust affected response to symptoms of angina, with Whites stating that they were less likely to delay seeking health care during an acute crisis compared to South Asians who stated they would phone the GP for advice before going to A&E. Conclusion Access to appropriate health care following chest pain is likely to result from patient related factors which include response to chest pain, expectation from the health service following chest pain and most importantly an overly trusting GP patient relationship.

Despite dramatic improvements in the quantity and quality of life for Human Immunodeficiency Virus (HIV) positive people with the introduction of combination antiretroviral therapy (cART), all-cause mortality rates remain higher than the general population. Furthermore, treatment is a lifelong commitment and a substantial burden on patient life. The aim of this thesis was therefore to assess the long term durability of cART through assessing various clinical, virological and immunological outcomes including mortality in HIV-positive patients across Europe. The analyses were based on data from the EuroSIDA cohort, an observational cohort of more than 16000 HIV-positive patients from Europe, Israel and Argentina. Results showed that HIV-positive patients on a well-tolerated and fully suppressive cART regimen have a small risk of treatment failure occurring over the next 6 months and could therefore be monitored less frequently. In contrast patients who have spent a low percentage of time with a suppressed viral load whilst on cART or who have recently rebounded may require more intensive monitoring after making a treatment switch. In patients who have achieved an initial response and tolerated the first three months of treatment, nevirapine efavirenz and lopinavir based cART regimens all have similar durability based on risk of all-cause discontinuation and development of serious clinical events. Starting cART earlier to reduce the proportion of patients with a low CD4 count may decrease the rate of developing many common non-AIDS related malignancies. Individuals in Eastern Europe had an increased risk of mortality from AIDS related causes in part due to differences in use of effective cART. In conclusion results from this thesis provide evidence that could help improve the long term durability of cART for HIV-positive patients through different measures of healthcare capturing the wide aspects of treatment and outcomes.

This thesis examined the bidirectional association between obesity and common mental disorders (CMDs) using alternative longitudinal models. Although previous evidence from different studies suggests that obesity increases the risk of CMDs and CMDs increase the risk of obesity, a more detailed longitudinal analysis is needed in order to better understand the temporal patterns. The participants were from the Whitehall II prospective cohort study with 5 data collection waves between 1985 and 2009 (n=10,265 participants in total contributing 35,880 person-observations over the follow-up), aged 35 to 55 at baseline. Body mass index (BMI) was determined on the basis of height and weight measured in medical examinations, and CMDs were assessed with the self-reported General Health Questionnaire (GHQ). In addition, several covariates (occupational status, sleep duration, bodily pain, alcohol consumption, smoking, physical activity, longstanding illnesses) were included. Associations were examined using multilevel regression. Obesity increased and the level of CMDs decreased with age. The development of both obesity and CMDs were characterized by cumulative developmental patterns, that is, the risk of future obesity (or CMDs) increased progressively with the number of times the person had been obese (or had CMDs) in previous study phases. Standard longitudinal regression models suggested that obesity was prospectively associated with future CMDs, whereas CMDs did not predict future risk of obesity. However, chronic CMDs increased the risk of obesity, so that only individuals with CMDs in several study phases over the follow-up phase had elevated risk of future obesity. Such cumulative pattern was not observed for chronic obesity in predicting future CMDs. Analysis of changes of BMI and CMDs over time indicated that a decrease in BMI was associated with a future decrease in CMDs, and an increase of CMDs was associated with future increase in BMI. An increase in BMI, however, was not associated with future change in CMDs, and a decrease in CMDs was not associated with a decrease in BMI, suggesting that the associations between changes in BMI and CMDs are dependent on the direction of change in the exposure. Except for bodily pain, the covariates had little if any effect on the associations between obesity and CMDs, and only age showed a consistent moderating effect such that the cross-sectional association and the association between CMDs and the future risk of obesity increased in magnitude with age. The results from several alternative longitudinal models suggest that the bidirectional association between obesity and CMDs is likely to represent the effects of multiple mechanisms that exert their influence over different time periods. Standard longitudinal regression models with only two measurement times are not sufficient to capture such complicated temporal patterns.