Background Studies which have sought to explore the associations between diabetes and cancer have produced heterogeneous results and there is a paucity of evidence related to glycated haemoglobin (HbA1c) and cancer risk. Methods Initial analyses utilised data from the Health Survey for England (HSE) and SHeS combined linked to mortality and Cancer Registry data (n=204,533, including 7,199 with diabetes) to explore the associations between diabetes, HbA1c and cancer incidence and mortality. Additional analyses used linked Whitehall I data (n= 19,019, including 237 with diabetes). Odds Ratios (ORs), Hazard Ratios and 95% Confidence Intervals (CI) were estimated adjusted for a range of confounding factors using logistic, multinomial and Cox regression. Results 18,310 deaths occurred within the HSE/SHeS follow-up period (4,997 from cancer). The adjusted OR for cancer among those with diabetes was 1.27, CI 1.12-1.43. Raised HbA1c was associated with an excess risk of dying/developing cancer; diabetes and HbA1c were associated with a number of site-specific cancers. When analyses were stratified by cardiovascular disease (CVD) baseline status, only those with diabetes who did not report CVD had a statistically significant excess in cancer mortality (adjusted OR: 1.27, 1.08-1.48). There were also sex differences in cancer incidence and mortality risk. 81% of Whitehall I participants died during follow-up (including 4,076 from cancer). These results did not replicate the initial analyses in finding no association between diabetes and cancer mortality -this is likely to relate to the age of the two cohorts and the differences in CVD mortality and incidence. Conclusion The association of diabetes and HbA1c with increased cancer incidence and mortality was not consistent across studies or population groups. Differences in risk by sex and CVD status suggest the need for health professionals to tailor services to take account of the individual circumstances of their diabetic patients.

Background Mental stress is a recognized risk factor and trigger for heart disease. Socioeconomic status (SES) is associated with morbidity and mortality, with low SES people having poorer health compared to their counterparts. I hypothesised that those two factors may interact with each other, so that when they are present simultaneously the total harmful effect is more than the sum of the two risk factors alone. Research aim My aim was to test whether lower SES interacts with mental stress and amplifies its effect on heart disease, so that the effect of mental stress on heart disease would be more pronounced in people from low SES backgrounds. Studies I carried out three studies. For my first study, I analysed data from a cross-sectional study involving about 500 disease-free middle/old-aged men and women drawn from the Whitehall II epidemiological cohort. I evaluated their salivary cortisol responses to standardized mental stress tests (exposure variable) and their cardiac troponin T plasma concentration (a marker of heart damage) using a high-sensitivity assay (HS-CTnT, outcome variable). I also used measures of coronary calcification levels using electron-beam dual-source computed tomography and Agatston scores. After adjustment for demographic and clinical variables associated with heart disease as well as for inflammatory factors, I found a robust association between cortisol response to mental stress and detectable troponin T (odds ratio [OR] =3.8, 95% confidence interval [CI] =1.5-9.4, P =0.005). The association remained when I restricted the analysis to participants without coronary calcification (n =222, OR =4.8; 95% CI =1.2-18.3; P =0.023) or when I further adjusted for coronary calcification in participants with positive Agatston scores (n =286, OR =6.2, 95% CI =1.9-20.6; P =0.003). In analyses stratified by SES, there was a trend showing that the lower the SES was, the higher the OR, although this trend was not significant (P >0.05). In my second study, I selected about 67,000 male and female participants from the Health Survey for England who were 35 years or older, free of cancer and cardiovascular disease at baseline, and living in private households in England from 1994 to 2004. Selection used stratified random sampling (hence representative of the nation), and participants were linked prospectively to mortality records from the Office of National Statistics (mean follow-up, 8.2 years). Mental stress was measured using the 12-item General Health Questionnaire, and SES was indexed by occupational class. The crude incidence rates for heart disease and all-cause mortality of the cohort were 1.9 (95%CI =1.7-2.0) and 14.5 (95%CI =14.2-14.8) per 1,000 person-years. After adjustment for age and sex, mental stress was associated with increased mortality rates. In a stratified analysis, the association of mental stress with the outcomes differed with SES, with the strongest associations being observed in the lowest SES categories (the adjusted P values for interaction were 0.012 for all-cause mortality and 0.047 for heart disease mortality). My third study involved about 80,000 post-menopausal women selected from the United Kingdom Collaborative Trial of Ovarian Cancer Screening study, who were followed up for about three years on average. Mental stress was measured using the hopelessness/helplessness index and incident heart disease was assessed using hospital electronic records. The overall incidence rate of hospitalisation for acute heart disease event was 2.7 per 1,000 person years (95% CI=2.5-3.0). The augmented incidence for people who experienced mental stress was higher in people of low SES, medium in people of medium SES, and lower in people of high SES (adjusted P value for interaction =0.013). Conclusion These studies suggest that the interaction between socioeconomic status and mental stress is associated with ischemic heart disease, in such a way that people in low socioeconomic circumstances are more vulnerable to the negative effects of mental stress. In other words, the harmful effect of mental stress for human cardiac health may be modified by socioeconomic position and rendered more deleterious for people from disadvantaged backgrounds. Further research is needed to disentangle the dynamics of this effect amplification.

Background: Treatment for coronary heart disease with coronary artery bypass graft (CABG) surgery provides benefits for physical and psychological health. Poor recovery and adjustment is experienced by some patients and their partners. Aspects of social relationships may be important psychosocial determinants of physical and psychological outcomes for both partners. Methods: A longitudinal study of CABG patients and their partners was conducted with the aim of determining the role of social relationships for short-term recovery and adjustment from surgery. Participants completed measures of emotional adjustment, physical health status, support and caregiving (partners only), 4 weeks before and 8 weeks after surgery, and clinical data was obtained from medical notes. The trajectories of variables were analysed, and support variables were examined as predictors for emotional and physical outcomes. The provision of support (caregiver burden) was also assessed as a predictor for partner outcomes, as was its relationship with support. Results: Patients experienced improvements to emotional variables after surgery but only anxiety improved for partners. Both spouses suffered reductions to physical health. After controlling for covariates social support predicted length of hospital stay in patients and marital functioning predicted depression symptoms and anxiety. Social support predicted mood disruption in partners and caregiver burden predicted emotional distress. Caregiver burden predicted decreasing social support, but support was not protective of distress in those with greater burden. Partners reported less favourable levels of emotional and support variables than patients. Conclusions: Support influences the post-surgery adjustment of patients and their partners. The provision of care impacts the partner’s emotional outcomes and their perceptions of support. Particular types of support and the provision of support are risk factors for worse psychological and physical outcomes in CABG patients and their partners, with implications for the development of interventions.

Non-drinkers have been consistently found to have worse health outcomes than moderate drinkers in later life. Explanations for this include a protective effect of moderate alcohol consumption on health, or alternatively that some non-drinkers are ex-drinkers who may have had to stop drinking because of poor health hence suffering from a pre-existing poor health bias. Another factor, which has been unexplored in the literature, is the early life health and social circumstances of non-drinkers; this is the subject of investigation in this thesis. The Health Survey for England was used to explore the early life social, health and health behaviours of non-drinkers aged 18 to 34 years. The National Child Development Study and the 1970 British Cohort Study were used to investigate the childhood health characteristics of non-drinkers in early adulthood. Binary logistic regression was carried out to assess whether poor health from an early age and persistent poor health was associated with being a persistent non-drinker across time at different ages, adjusting for sex, highest qualification, mental health and marital and parental status. Poor health from an early age and persistent poor health were associated with being a lifetime abstainer, consistently between two cohorts, which is an original contribution to knowledge. Non-drinkers from an early age had higher rates of emotional and behaviour problems than drinkers; this may contribute to greater risk of cognitive decline. Furthermore non-drinkers in early adulthood had higher rates of health conditions in adolescence, and had lower educational levels from early adulthood. This might increase the risk of mortality among non-drinkers in later life through persistent multiple disadvantage from an early age. The health and social characteristics of non-drinkers in early life need to be considered when comparing health outcomes of non-drinkers with drinkers in later life. The worse health and lower social circumstances of non-drinkers from an early age may be why non-drinkers consistently have worse health outcomes than drinkers across a broad range of conditions.

Background: Macro-level factors (related to the economic and political context) have been considered as determinants of health inequalities. In particular, the role of political factors (such as welfare state regimes) has recently received increasing attention. However, very little is known in that respect for oral health inequalities. Aim: To examine the relationship between oral health inequalities and political factors (welfare state regimes) in Europe and the US. Methods: The project involved three stages. First, oral health inequalities were compared across 21 European countries grouped into different welfare state regimes (Scandinavian, Anglo-Saxon, Bismarckian, Southern, and Eastern). Second, a multilevel approach was employed to assess the influence of welfare regimes on the variation in oral health between European countries. Third, inequalities were compared between two countries classified in the same welfare regime, but with different health care systems: England and the US. In stages one and three, relative and absolute socioeconomic inequalities were examined using the relative and slope indices of inequality (RII and SII, respectively). Results: The Scandinavian welfare regime showed consistently lower prevalence rates of edentulousness, no functional dentition and oral impacts than the other regimes. Significant educational and occupational inequalities on edentulousness and no functional dentition were observed in all welfare regimes. The comparison on the magnitude of inequalities across regimes showed a complex picture with different findings according to the outcome, socioeconomic indicator and nature of the inequalities (absolute and relative). Overall, results of this comparison did not support the hypothesis of lower inequalities in the Scandinavian regime. When using a multilevel approach, results revealed that grouping countries into welfare regimes contributed to explaining the variation in oral health among European countries. In the England-US comparison, significant relative (RII) and absolute (SII) inequalities were found in the two countries in all oral health measures. These inequalities were consistently higher in the US compared to England. Conclusions: Oral health inequalities exist in all European welfare state regimes. The Scandinavian regime exhibited better oral health, but not lower inequalities compared to the other regimes. The US showed consistently larger inequalities than England. Overall, results suggest that political factors influence socioeconomic inequalities in oral health.

Background: Obesity prevalence in Mexico has risen substantially over the last 25 years. Its social patterning has not been systematically studied. Aim: To test the nutrition transition proposition of a crossover from lower to higher rates of obesity among the more disadvantaged groups, leading to emerging and increasing obesity inequalities as Mexico develops economically. Methods: Data came from four nationally representative surveys (1988, 1999, 2006, 2012); N=51,387 non-pregnant 20-49 year old women and N=18,988 20-49 year old men. Level of education and a household wealth index were used to calculate the relative and slope indexes of inequality (RII and SII respectively). Trends in RII and SII were examined in the period 1988-2012 for women. Change from 2006 to 2012 was examined for men. The contribution of mediating factors to obesity inequality was investigated. Results: There was support for the nutrition transition proposition among Mexican women. As the country developed economically, obesity became more prevalent among more disadvantaged women. Among men, there was no evidence of a reversal of the social gradient. Higher education and wealth were associated with higher obesity prevalence. Unexpectedly, educational inequalities in obesity among urban women declined over the study period. This was due to faster increases in obesity prevalence among women with more years in education compared to those with less. Psychosocial factors (food insecurity and aspired body size) explained a proportion of educational inequalities in obesity among women. Gender differences in educational inequalities in obesity were partially explained by differences in aspired body size. Conclusion: This detailed analysis of obesity inequalities in Mexico, and their recent trends, significantly develops existing literature. By using both education and household wealth as markers of SEP, the nutrition transition proposition was investigated in depth. The nutrition transition proposition fits the educational inequality pattern among Mexican women but not men.

The prevalence of obesity has reached dramatic proportions over recent years, and obesity among older adults is becoming an increasingly important concern in developed countries with ageing populations. Weight loss is recommended for all obese individuals, regardless of age, yet while there has been a vast amount of research into factors surrounding weight loss across younger and middle-aged populations, the evidence base on weight loss in older adults is lacking. This thesis uses data from a cross-sectional survey of UK adults, and two large epidemiological studies of ageing, the English Longitudinal Study of Ageing and the Health and Retirement Study in the US, to address this gap in the literature. Study 1 highlights the high prevalence of desire to weigh less and attempts at weight loss among older adults, and Study 2 reveals that a surprisingly high proportion of those who are overweight or obese are achieving clinically meaningful (≥5%) weight loss. Studies 3 and 4 show that despite reductions in cardio-metabolic risk, weight loss in overweight/obese older adults is associated with increased risk of depressed mood, and this association grows stronger with each decade of age. However, Study 5 finds some evidence to suggest that rates of depressed mood might only be increased during the process of weight loss, and that if weight loss is maintained there might be benefits for mood relative to baseline. This research contributes to the understanding of weight loss in older adults – particularly that which occurs in the general population, outside of the trial context – and emphasises the need for health professionals to take into consideration patients’ psychological wellbeing when recommending or responding to weight loss at older ages. Limitations of this work and directions for future research are discussed.

Dramatic increases in the prevalence of overweight and obesity have prompted a focus on prevention. Weight is known to have a strong genetic basis, but the speed of change in rates of overweight and obesity against a relatively stable gene pool suggests that exposure to an ‘obesogenic’ environment is important. The home environment is thought to play a key role in early weight trajectories, providing an avenue for long-term obesity prevention. There is evidence for associations between various aspects of the home environment and energy-balance behaviours; however, evidence for associations with weight is limited, particularly in early childhood. Few studies have used comprehensive, psychometrically-tested measures of the home environment, and no studies have tested for gene-environment interaction in the home context. This thesis uses data from the Gemini twin cohort to further examine the role of the home environment. Study one describes the development of a comprehensive measure of the home environment in early childhood, including the quantification of the extent that the home is likely to be obesogenic. Study two explores the utility of a novel tool called SenseCam to examine and validate aspects of the home environment measure. Study three identifies a number of maternal characteristics associated with the obesogenic quality of the home environment. Study four shows associations between the obesogenic quality of the home environment and energy-balance behaviours; while study five finds no association with weight. Findings from study six highlight the role of gene-environment interaction, showing that the heritability of weight is higher among children living in home environments with greater obesogenic potential. Overall, the findings of this thesis further understanding on how the home environment contributes to the development of overweight and obesity. Implications, limitations, and avenues for future research are discussed.

Identifying risk factors associated with normal cognitive ageing is a prerequisite for understanding dementia. Potential modifiable risk factors include socioeconomic factors and health behaviours. This thesis investigated the importance of life course socioeconomic position (SEP) and two core health behaviours, alcohol consumption and smoking, for mid-late life cognitive function in four previously unstudied Central and Eastern European populations with historically smaller income inequalities and significant contributions of alcohol and smoking to the high premature mortality in these populations. The thesis used data from over 29,000 men and women aged 45-78 from random population samples in Novosibirsk (Russia), Krakow (Poland), Kaunas (Lithuania) and six Czech towns participating in the HAPIEE study. Cognitive function was measured using four tests of fluid cognition. SEP measures, alcohol consumption and smoking were self-reported using structured interviews. Structural equation analyses revealed significant associations between SEP measures from across the life course and cognition. Education consistently showed the strongest association with cognition and some accumulation of disadvantage across the life course was observed, similar to studies in Western countries. However, variation in magnitude of these associations across centres may partly reflect the influence of contextual factors. Regression analyses showed modest associations of cognitive function with alcohol and smoking, and neither of these behaviours appeared to significantly mediate the associations between life course SEP and cognition. An inverted U-shaped association indicated slightly worse cognitive performance among male heavy drinkers and lower scores in non-drinkers, compared to light drinkers. Binge drinking and alcohol type were not associated with cognitive performance. Smoking was associated with poorer mental speed in both genders but not with any other cognitive test. The findings suggest a pattern of associations between life course SEP and cognition similar to Western populations and modest associations of alcohol and smoking with mid-late life cognitive performance in these Central and Eastern European populations.

Proposed changes to hospital services, such as the closure of Accident & Emergency departments (A&E), often create high profile, contentious debates. Whilst clinicians tend to focus on potential health gains for patients, public concerns may include non-clinical factors such as ease of access to services. Previous research has largely concentrated on policy issues and little is known about the priorities and preferences for emergency hospital care amongst the public at large. The first study in this thesis explored a process carried out to engage the local community with proposals to reconfigure hospital services in North London. This included mapping key areas of concern for the public and critically reviewing the impact of the methods used in the public engagement process. Study 1 demonstrated that the techniques currently used draw on traditional approaches aimed at improving the public understanding of science. This includes an apparent assumption of ‘public ignorance’ in matters of science and technology, and the belief that science offers a uniquely privileged view of the world. The second study involved a series of in-depth interviews examining priorities for emergency care in more detail. Interviews were conducted with four groups of participants: patients with a chronic condition; parents of young children; older people; patient representatives and community groups campaigning against service closures. Interviews were carried out in an area where a reconfiguration was being discussed and an area where it was not. The analysis drew on theories relating to risk perception and risk communication. It demonstrated the importance of the widespread belief that timely access is associated with better outcomes, as well as the way in which the public’s assessment of service quality influences their response to reconfiguration proposals.