Background People with disabilities in Australia experience poorer mental health than people without disability. However, the mechanisms by which disability leads to poor mental health are inadequately understood. This PhD thesis aims to form a better understanding of how people’s socio-economic circumstances influence the effect of disability acquisition on mental health. Elucidating the causal mechanisms underpinning this relationship will inform the development of effective public health and social policies to improve the mental health of people with disabilities. Methods I used data from the Household, Income and Labour Dynamics in Australia Survey – a large nationally representative longitudinal dataset – to quantify the effect of disability on mental health and to examine the socio-economic mechanisms leading from disability acquisition to poor mental health. I identified adults who acquired a disability during their participation in the survey and used data both before and after they acquired the disability to estimate the causal effect of disability on mental health. Firstly, I conducted analyses of effect modification, examining a wide range of different socio-economic factors to determine whether they influenced the magnitude of the effect of disability acquisition on mental health, using inverse probability weighting and fixed effects models to better control for confounding. Secondly, I conducted causal mediation analyses to further examine the socio-economic mechanisms leading from disability to poor mental health, using sequential mediation analysis to examine a broad range of socio-economic characteristics and an interventional mediation approach to quantify indirect effects operating through two distinct socio-economic characteristics: employment and income. Finally, I examined the indirect effect mediated by employment in greater detail, further decomposing the natural indirect effect through employment to estimate the proportion attributable to interaction, mediation and their joint effects. Results There is a clinically significant and large effect of disability acquisition on mental health. The analyses of effect modification provided evidence that the magnitude of the effect differed according to people’s socio-economic characteristics, with greater effects observed for more disadvantaged groups. The mediation analyses provided additional evidence that socio-economic characteristics contribute to the effect of disability acquisition on mental health. A third of the effect was found to be mediated by material socio-economic factors such as employment, income, wealth, financial hardship and housing characteristics, and a further investigation of the indirect effect through employment and income highlighted employment (but not income) as an important contributing factor, explaining 11% of the effect alone. Finally, further decomposition of the indirect effect through employment suggested that the mediated effect was due to interaction between disability and employment, rather than pure mediation. Conclusion The findings of this thesis highlight the importance of the social determinants of health in generating mental health inequalities. Interventions should prioritise addressing the social determinants of health to improve the mental health of people with disabilities and reduce disability-related mental health inequalities. Furthermore, the evidence that employment is a key mediator of the effect of disability acquisition on mental health indicates that policy strategies are needed to target the causes of low employment rates for people with disabilities.

Background The number of people with type 2 diabetes is increasing rapidly, particularly in low- and middle-income countries (LMICs). High and increasing prevalence of type 2 diabetes contributes to increased morbidity and mortality due to its complications, most importantly, cardiovascular diseases (CVD). Therefore, preventing type 2 diabetes is a high priority. This thesis focuses on the effectiveness of two types of lifestyle interventions that aim to prevent type 2 diabetes and reduce CVD risk by reducing underlying modifiable lifestyle risk factors. The interventions include 1) community-wide lifestyle interventions that target the whole population living in a particular geographic area and 2) structured lifestyle interventions among individuals at high risk of developing type 2 diabetes based on a screening tool. Aims The overarching aim of the current thesis is to assess the effectiveness of lifestyle intervention programs in the prevention of type 2 diabetes and the reduction of cardiovascular diseases risk in resource-constrained settings. This thesis has three main objectives to address its overarching aim: i) To evaluate the effectiveness of a community-wide lifestyle intervention on the prevention of type 2 diabetes in a trial conducted in Iran (Chapter 4), ii) To evaluate the performance of different screening methods in identifying those at high risk of type 2 diabetes (Chapters 5 and 6), iii) To evaluate the effects of a structured lifestyle intervention on the reduction of CVD risk among individuals at high risk of developing type 2 diabetes in a cluster-randomized trial conducted in India (Chapter 7). Methods To meet the study objectives, I conducted a secondary analysis to datasets of two lifestyle intervention trails entitled “Tehran Lipid and Glucose Study (TLGS)” and “Kerala Diabetes Prevention Program (KDPP)”: In TLGS, three areas in District 13 of Tehran, the capital city of Iran, were selected for the study. People living in one area received a multi-component community-wide intervention including an educational session in a medical health centre, newsletters and health education materials four times a year and public education in community gatherings from 1999 until 2015. People in the other two areas were assigned to the control status. People in all three areas were followed up every three years, from 1999 until 2015 (Wave 1 to Wave 5). In KDPP, 60 randomly-selected polling areas in the Kerala state (India) were screened by a non-invasive risk score entitled “Indian Diabetes Risk Score (IDRS)”. Those with IDRS score of >=60 were recruited in the KDPP study if they did not have type 2 diabetes in an oral glucose tolerance test. Identified high-risk individuals in 30 randomly-selected polling areas were assigned to the lifestyle intervention and their counterparts in the other 30 polling areas were assigned to the control. People in the intervention group received 15 group sessions delivered over 12 months. High-risk individuals in both intervention and control groups were followed at baseline, 12 months, and 24 months. I used generalized estimating equations (GEEs) to evaluate the effectiveness of lifestyle intervention programs in the prevention of type 2 diabetes in TLGS. I also used the Framingham risk prediction model for CVD to compare the change in predicted 10-year CVD risk between intervention and control groups of KDPP study, using mixed-effect models. Moreover, I validated several non-invasive and step-wise screening methods in the control group of TLGS, and then I assessed their performance for identifying those at high risk of type 2 diabetes. Results I found that lifestyle intervention in TLGS was associated with a 28% reduction in the incidence of type 2 diabetes in short-time (i.e., 3.4 years after starting the intervention). However, there was no statistically significant and clinically meaningful reduction in the incidence of type 2 diabetes in the following waves (Chapter 4). I also showed that favourable effects of the lifestyle intervention on the reduction of smoking prevalence and improvement of diet quality maintained even in the latest follow-up wave (Chapter 4). In addition, I found that non-invasive risk prediction models have limited utility in identifying those at high risk of developing type 2 diabetes and cannot be used as the only measurement for screening in Iran (Chapter 5). However, stepwise screening methods that combine a non-invasive measurement (e.g., anthropometrics) with a lab-based measurement (e.g., fasting plasma glucose level) have acceptable performance in identifying those at high risk of type 2 diabetes. Moreover, I found that the KDPP lifestyle intervention was associated with a one percentage point reduction in the 10-year CVD risk based on the Framingham risk prediction model after 24 months. My further investigations showed that favourable effects of the intervention on CVD risk are mainly due to a reduction in the incidence of type 2 diabetes and the reduction of smoking prevalence in the intervention group as compared to the control group. Conclusions This thesis provides much-needed evidence on the effects of lifestyle intervention programs on type 2 diabetes prevention and CVD risk reduction in LMICs. Findings from this thesis also highlight the importance of selecting a valid, reliable, and accurate screening method for identifying those at high risk of type 2 diabetes in designing lifestyle intervention programs with a high-risk strategy.

Co-production is where citizens and clients actively contribute to the work of delivering public services and the production of publicly valuable outcomes. This thesis examines co-production in the care of ageing populations, as one of the greatest challenges facing 21st century governments. It comprises both theoretical and applied research at the point where two literatures, in co-production and in health and social care intersect, with important implications for policy and practice. A narrative review of the literature was conducted revealing gaps in the current analysis. From a top- down perspective, a systematic policy document analysis and key stakeholder interviews were performed, that illustrates the impact of emerging Australian health reforms on co-production. The main study of the thesis adopted a bottom-up perspective, that involved qualitative semi-structured interviews to capture the lived experiences of older people with complex needs and their carers as ‘co- producers’ of complex care in the community. This thesis provides frameworks for facilitating and sustaining the co-production of complex care that consider what influences people’s propensity to co- produce in this context and how care tasks are distributed between clients, their informal carers and the service system.

Background Common mental health problems are a rising cause of disability in the workplace. Help-seeking rates are low and there are considerable individual, social and economic impacts. One way to address the impact of mental health problems in the workplace as part of an integrated approach is to improve mental health literacy through training. An existing evidence-informed mental health literacy training program is Mental Health First Aid (MHFA). The MHFA program has a large body of evidence for improving participant mental health first aid knowledge and confidence and reducing stigmatising attitudes. Follow-up studies also indicate that participants go on to use their skills to offer assistance to others they are concerned about. The Standard MHFA course is focused on adults helping other adults and the content is informed by a series of guidelines developed through the Delphi consensus method. Given the MHFA program is widely disseminated internationally, it has good potential for uptake by workplaces wishing to train their workers in how to offer mental health first aid to co-workers. Further research is required to ensure this work is guided by evidence. This thesis aims to build knowledge around mental health first aid in the workplace setting to inform future interventions by: 1. Better understanding how the existing MHFA course is delivered in Australian workplaces and inviting feedback on how the course could be tailored for the workplace setting 2. Developing expert consensus guidelines on how to provide mental health first aid to a co-worker 3. Developing evidence-based recommendations for workplaces wishing to establish formal Mental Health First Aid Officer (MHFAO) roles. Three studies were carried out to fulfil the thesis aims. Study 1 Method The first study focused on exploring the experiences of instructors who had delivered MHFA training to workplaces. One hundred and twenty Australian MHFA instructors completed a purpose designed survey retrospectively reporting on MHFA course delivery in workplaces in the previous 12 months, including the industries, reasons for delivering training and any observed impacts after training. The survey also asked for feedback on how the MHFA course content and mode of delivery could be tailored for workplaces. Finally, instructors reported on the perceived benefits and challenges in delivering workplace training and further support they may require. Results The first study found that MHFA training in workplaces was largely concentrated within three industry areas, more often with a focus on developing skills to help clients rather than co-workers and prompted by an internal policy recommending mental health training. Instructors indicated they most wanted workplace specific content and scenarios added to the course and most endorsed face-to-face delivery to an audience of mixed role types, rather than separate training for employees and managers/leaders. A number of benefits, challenges and support needs were also identified. Study 2 Method The second study used Delphi consensus methodology drawing on the expertise of three groups of panellists from English speaking developed countries; individuals with lived experience of mental illness whilst working, managers and workplace mental health professionals. The study aimed to build consensus amongst the expert panellists on the importance of a series of mental health first aid actions in the workplace drawn from the literature to be included in a guideline document. Results A literature search yielded 246 possible helping statements. Of these, 201 items were agreed to be important or very important to be included in a guideline document by at least 80% of the panellists. These items included actions on how to approach and offer support to a co-worker and additional considerations when the person being offered first aid is a supervisor or manager, or if the person is in a crisis situation such as acute distress or intoxication. The study found that panellists found difficulty in reaching consensus on a number of items related to role delineation, privacy and assisting a person in a more senior position. Study 3 Method The final study explored the experiences of workplaces that had appointed employees trained in MHFA into formal MHFAO roles using qualitative case study methodology. Employees from five diverse workplaces with Australian sites completed semi-structured interviews on their implementation approach and the perceived benefits and challenges experienced. Results The results suggested that organisations tended to differ in their implementation approach based on the amount of experience they had with the MHFAO role. There was strong agreement across organisations on the benefits and challenges experienced with the role. Respondents across all organisations agreed it was important to offer MHFA training to all staff if feasible and identified important differences between the MHFAO role and a traditional first aid officer role. A number of recommendations for future workplaces appointing MHFAOs were developed based on the insights shared by organisational representatives. Discussion The results of these three studies provide new knowledge on mental health first aid in the workplace in a number of ways. The results from the first study highlighted that the dissemination across industries has been relatively narrow to date, suggesting that the current MHFA course may need further tailoring to enable broader reach across industries. The guidelines resulting from the second study provide clear consensus-based strategies on how to offer mental health first aid to a co-worker. There was contention on certain topics, suggesting that providing mental health first aid in a workplace context can be complex possibly due to the constraints of workplace roles and responsibilities. In collaboration with MHFA Australia, the insights provided by instructors and the guidelines have now informed the content and delivery of a tailored MHFA course for the workplace. An important finding from the third study was that organisations appointing MHFAOs found it challenging to provide adequate support to those trained in MHFA, in addition to this being the most cited area of suggested improvement by participants. This suggests that organisations may benefit from being offered guidance when planning their approach to MHFA training in the workplace. An MHFA implementation guide for workplaces has now been developed as a result of the third study. The findings from all three studies can be used to develop further relevant resources and services in order to enhance the well-established benefits of MHFA training. Future work could evaluate the guidelines as a standalone minimal intervention and refine the process of revising the guidelines so that they are more targeted for their intended use. Since this thesis research commenced, the field of workplace mental health has significantly progressed and a number of evaluations of mental health literacy interventions in the workplace have been published. These recent findings are discussed in the context of how a possible workplace specific MHFA intervention informed by the three research studies would be conducted in future. Suggestions for future evaluations include providing implementation support to organisations before and after training, including broader workplace measures to assess impact such as help-seeking behaviours and productivity within the workplace and finally, including MHFA training in an integrated workplace mental health intervention.

In 2018, 3,046 people in Australia died by suicide of whom the vast majority, 2,320 (76.2%), were men. A range of reasons have been put forward to explain the higher rate of suicide among men compared with women but perhaps underpinning all these are conformity to masculine norms. If the rate among men is to be reduced, then suicide prevention interventions are needed that adopt a gendered approach and attend to the influence of masculine norms. In Australia, the ‘dominant masculinity’ is one that endorses the norms of stoicism, independence, invulnerability and avoidance of negative emotions. Conformity to these masculine norms has been associated with suicidal thinking, poor mental health and reduced and delayed help-seeking. Evidence has been found for the mediating role of harmful masculine norms in men’s help-seeking for mental health and suicide concerns. Men demonstrate lower rates of help-seeking for mental health problems and suicidal thoughts compared to women, and help-seeking behaviour has been cited as a key target for improving men’s health. If conformity to masculine norms both confers increased risk of suicidal thinking and influences reduced help-seeking, then interventions that address masculine norms may be a fruitful avenue from a suicide prevention perspective. Little is known about how masculine norms might be targeted to improve these outcomes. The Man Up project was funded by the Movember Foundation in Australia with the intention of addressing the high suicide rate among men by promoting help-seeking for personal or emotional problems via an exploration of Australian masculinity. Man Up was a media-based population-level suicide prevention intervention, the centrepiece of which was the Man Up documentary. The documentary examined how society shapes the way men and boys see themselves and the way in which they act and explored how this might affect mental health and, potentially, lead to thoughts of suicide. It was aired by the Australian Broadcasting Corporation (ABC), Australia’s national public free-to-air broadcaster, in October 2016 and was accompanied by a digital strategy across social media and a website. The overall aim of this thesis is to examine the ways in which Man Up contributed to help-seeking by men with a view to informing future suicide prevention interventions. The thesis aims to determine whether, and how, Man Up brought about changes in men’s help-seeking via a focus on masculine norms. This aim is addressed through four papers that are incorporated into the thesis. A randomised controlled study demonstrated quantifiable changes in help-seeking intentions and conformity to masculine norms in those who viewed Man Up. Qualitative feedback demonstrated impacts of Man Up on men’s awareness of suicide and mental health problem, others’ emotional lives, willingness to help others, and openness about emotions and problems, as well as behavioural changes related to helping others and being more emotionally expressive. In an in-depth analysis of feedback provided through the trial and the website survey regarding the impacts of Man Up on men’s views of masculinity, participants commented they had rethought stereotypes of masculinity and considered the inclusion of a wider diversity of attitudes and behaviours in relation to ‘being a man’. An evaluation of the website, website survey and emails to the website demonstrated that the website provided an important opportunity for people to engage with Man Up and seek help. These findings indicate that Man Up had positive impacts on men’s help-seeking and conformity to harmful masculine norms. It is likely that Man Up had significant population-wide positive impacts when it was delivered as a multimedia intervention in late 2016. The thesis demonstrates the importance of focusing on masculine norms to bring about improvements in men’s help-seeking and the media is highlighted as a valuable tool to be harnessed to this end. Strategies and associated actions for future help-seeking interventions for men are made based on the Theory of Planned Behaviour, and avenues for further research are recommended. Key among the strategies, and perhaps underpinning them all, is attention to changing men’s conformity to masculine norms that discourage help-seeking and emotional expressiveness, and the need for a community that is supportive of men’s help-seeking. Further research regarding the sustainability of change in these domains is also needed. These suggestions for future interventions and research have the potential to further improve men’s help-seeking and ultimately may contribute to reduced suicides by men.

Mammography screening is deemed cost-effective for women aged 50-70. Yet the utilities informing breast screening policy are limited in their ability to adequately capture the benefits and risks. The evaluation of many cancer screening programmes present results in terms of cost per QALY but fail to include any disutility for the patients who have been over-diagnosed and may receive unnecessary treatment. This thesis presents an overview of the challenges associated with valuing breast cancer states, using the results of an empirical study deriving utilities from 172 women in Melbourne, Australia as an example of potential methods to capture the disutility of overtreatment. Ductal carcinoma in situ was used as a proxy to quantify the benefits and harms associated with the sequelae of screening. Utilities derived from 172 women for health states explicitly describing overdiagnosis were lower than those from the literature, where it is unlikely that women were informed that their treatment may have been unnecessary. Qualitative interviews of 26 patients validated that this risk was important in the valuation of treatment. The findings were used to inform an economic model which demonstrated that the explicit inclusion of the harms of screening may change the decision on breast screening strategy.

Statistical variable selection, also known as feature selection, has become an indispensable tool in many research areas involving machine learning and data mining. The object of statistical variable selection is to select the best subset of predictors for fitting or predicting the response variable from a potentially large collection of candidate predictors. It is particularly important in high-dimensional problems such as cancer genetics, where there are potentially thousands of predictors and only a few are associated with the outcome. Moreover, predictors may have underlying group structures in them, so it is desirable to take the underlying group effects into consideration when performing variable selection and handle the grouping structure present in the model when selecting important variables. This thesis presents a Bayesian grouped regression model with continuous global-local shrinkage priors to tackle the high-dimensional variable selection problem by introducing shrinkage parameters at the group level as well as within each group. This model is able to handle complex group hierarchies that include overlapping and multilevel group structures, and also enjoys the advantages of handling sparsity as well as strong signals. The proposed method is also applied to a real high-dimensional GWAS breast cancer dataset, Haiman-Hopper dataset, to analyse the breast cancer genome-wide association study.

As of 2019, Highly Pathogenic Avian Influenza (HPAI) H5N1 is under control in Indonesia as a result of improved biosecurity and vaccination in large poultry enterprises. Nevertheless, the virus still circulates in less specialized village chicken production systems posing economic burden and public health risk. Past field research established that the trading activity associated with village chicken production is the primary pathway for HPAI-H5N1 release and spread in Indonesian villages; lamentably, the scarcity of data on village trading limits the exploration of risk frameworks to identify high-risk HPAI-H5N1 villages. This thesis investigates the village chicken productive landscape (that is, the collective of village chicken systems in a given village) as a determinant of the risk of HPAI-H5N1 events in Indonesian villages. Three core dimensions of these systems: type of village chicken system, village chicken productive dynamics, and risk mitigation strategies, are explored in relation to HPAI-H5N1 in four studies. First, a systematic review and meta-analysis that evaluates the efficacy of commercial vaccines against HPAI-H5N1 in Indonesia. Second, a Leslie matrix model to determine the underlying village chicken population dynamics and their effect on the maintenance of vaccine coverage. Third, an exploration and characterization of the Indonesian village chicken systems, along with the introduction of a deterministic age-structured model that simulates population dynamics and allows estimation of the presumed frequency of trading events to approximate the risk of HPAI-H5N1 release into villages. Fourth, a stochastic susceptible-exposed-infected age-structured model that allows exploration of transmission dynamics in different village chicken productive landscapes and evaluation of plausible risk mitigation strategies. The first study showed that viral drift reduces vaccine efficacy and that seed-homologous vaccine immunogenicity is not a good proxy for efficacy against wild isolates. The meta-analysis demonstrated that extra-label vaccination and the use of alternative seed homologous formulations are emerging sources of heterogeneous vaccine efficacy. The second study demonstrated that a mix of village chicken systems and not a "traditional scavenging" landscape, as suggested in the literature, drives the underlying village chicken population dynamics in Indonesia and these may quickly undermine perfect vaccine coverage. The third study demonstrated the limitations of the current classification of village chicken systems based on farming practice. The hierarchical clustering analysis in this study suggests that village chicken systems transitioned from traditional to semi-commercial modes of operation expressed as produce specialty (non-specialist, bird-specialist, or egg-specialists) and alternative trading strategies. The presumed frequency of trading events based on simulated productive dynamics suggests that the egg-specialists are heavily engaged in trading activities that increase the risk of HPAI-H5N1 release into villages. The fourth study shows that the type of village chicken landscape might not determine the probability of an epidemic event of HPAI-H5N1; however, outbreaks in landscapes that are abundant in bird-specialist and egg-specialist premises may persist longer. The simulations performed suggest that the extinction of the infected population seems invariably required to control these outbreaks. This thesis provides evidence that the Indonesian village chicken productive landscapes may determine the risk of HPAI-H5N1 events in Indonesian villages. This research suggests that a transition from traditional modes towards semi-commercial modes of operation that leads to a more significant presence of egg-specialists in the village results in a higher risk of viral release and likely more persistent epidemic events. The exploration of mitigation strategies which are plausible at the local level revealed challenges of controlling outbreaks in villages. These results emphasize the importance of identifying high-risk villages to enable adequate surveillance that reduces the need for mitigation strategies based on culling. In conclusion, this thesis suggests that an adequate characterization of the village chicken productive landscape may help, or even suffice, to identify villages at high risk of HPAI-H5N1 events, guiding surveillance, education, and future control efforts.

Background: Asthma in early and mid-life, and chronic obstructive pulmonary disease (COPD) in later life, are two very common respiratory disorders that contribute enormously to the burden of disease in developed countries such as Australia. Researchers have investigated the risk factors associated with the onset of these diseases and have conducted many studies exploring treatments that have the potential to provide effective control of disease symptoms. The analysis of family data with appropriate statistical techniques has the potential to ensure the magnitude of the effects of both genetic and environmental risk factors are estimated free from confounding by shared influences. My systematic review on asthma phenotypes on twins and sibship (CHAPTER 3) targeted only those studies that examined the atopic march diseases from infantile eczema to hay fever and asthma. My study was the first review to assess systematically these relationships using studies that generated family data. Most previous studies reported only estimates of prevalence, incidence, heritability, concordance rate and intraclass correlation between asthma and its phenotypes without synthesizing the evidence on associations between asthma phenotypes. Also, although there exist clear sequential and plausible biological mechanisms to explain such associations, previous evidence has been generated from studies featuring only young children. Such studies were not in a position to explore whether these allergic conditions continue to adult life was. My review confirmed the need for more sibship and or twin studies to investigate potential relationships between these phenotypes from childhood to adulthood. For childhood asthma and COPD association, there are, to my knowledge, no previously published analyses that feature sibship data of the type analysed in this thesis. Methods: In CHAPTER 4, I used data from the Tasmanian Longitudinal Health Study (TAHS) on which the research findings presented in this thesis are based. This study provides prospective data on respiratory and lung health and associated factors from the first to the fifth decade of life for sibling pairs with 29,615 participants. Comprehensively documented data collected within the TAHS cohort provide a unique opportunity to investigate the role of early-life factors such as childhood eczema, childhood hay fever in determining the risk of adult asthma, and the relationship between childhood asthma and Chronic Obstructive Pulmonary Disease (COPD). Both the traditional Conditional Logistic Regression (CLR) and B-W regression model (BWR) have been used to analyze the sibling pair data adjusting for possible confounders. These analyses generate estimates of the magnitude of the association between exposure and outcome within-pair free of confounding by factors shared within families. Results: In CHAPTER 6, I examined the associations between childhood eczema and asthma in adult life. My findings suggest that the effect of childhood eczema on asthma risk continues well past childhood. Although most children with eczema will not develop childhood asthma, childhood eczema in children who are asthma-free may progress to incident asthma in adult life. Also, childhood eczema predicted asthma that persisted to middle age, but the association was weak. In CHAPTER 7, I examined the association between childhood hay fever and asthma incidence in adult life. From the sibling pair analysis, it is evident that childhood hay fever and childhood asthma are strongly associated, and that this association is apparently independent of childhood eczema, parental history of asthma and socioeconomic status. Childhood hay fever was associated with almost a two-fold increase in the risk of the onset of asthma in adult life, independent of those same covariates. The risk of Incident asthma risk in adult life is also apparently independent of active and passive smoking. In CHAPTER 8, I investigated the association between childhood asthma and COPD by mean age 45 years. From the sib-study, childhood asthma was found to be associated with an increased risk of COPD by middle age, but evidence against the null hypothesis of no association was weak and the effect of childhood asthma on COPD is substantially reduced when adjusted for current asthma, current smoking status, and childhood eczema and other childhood environmental factors, although the primary motivation for this was the study of effect modification. LLN classified more COPD cases as compared to GOLD for some sibling pairs with age range between 42 and 49. Both statistical models gave similar estimates of the magnitude of the association between childhood asthma and the risk of satisfying criteria for COPD in later life. Conclusions: Overall, my findings in the three results chapters add to the existing literature that has reported associations between atopic march features and childhood asthma towards COPD in sibling studies by providing a more precise and less biased estimation of the respective exposure-outcome association free of confounding. Previously published estimates of the magnitude of these associations may have been larger, but these estimates of risk are potentially confounded by many unmeasured early-life factors and are therefore overestimated. My finding through the analysis of data from a sibling study design has attempted to correct for these confounding effects and, at least theoretically, should produce a more precise estimate of these associations which still support the continued importance of asthma and COPD management plans.

Moral distress is experienced when clinicians feel constrained to provide care they believe is not in a patient’s interests. It is recognised as a significant issue within acute care settings. No studies have longitudinally examined moral distress in relation to the clinical course of the patient. Objective: To explore the relationship between moral distress of healthcare professionals in neonatal intensive care units, the clinical trajectory of preterm babies, and clinician predictions of death, disability and perceptions of best interest. Methodology: A novel mixed methods staged design was used. Clinician baseline understanding and experiences of moral distress were sought through clinician questionnaires. A multi-centre mixed-methodology longitudinal cohort study (MoDaLiTy) was conducted that followed 99 neonates born <28 weeks post menstrual age from birth until discharge in May 2016 until December 2017. Clinicians were surveyed at regular intervals on their experience and intensity of moral distress, their predictions of death and disability, and whether ongoing intensive care was in the patient’s interests. A Neonatal Intervention Score (NIS) was designed and validated to reflect the relative illness severity of the patient throughout their admission. Associations between moral distress and clinical judgements were examined using clustered linear regression, Wilcoxon rank-sum test and Kruskal-Wallis H test. Clinician groups were compared using a two-sample t-test. A follow-up survey was conducted to explore clinician experiences of participating in a longitudinal study on moral distress, identifying changes both at an individual and institutional level. Results: 525 clinicians consented to participation in one or more components of this study. A total of 4593/5332 surveys (86% response rate) were collected during MoDaLiTy. Moral distress was reported on 687 (15%) occasions. Moral distress was only weakly explained by the clinical trajectories of patients, and clinician predictions of death were inaccurate. Nurses and trainees reported moral distress more frequently than neonatologists (16.9%, 18.5%, and 6.3% respectively, p<0.05) despite making similar predictions about patient outcomes. Neonatologists preferred levels of care for the patients more closely aligned with family wishes than did other provider groups. MoDaLiTy was reported to improve clinician awareness about ethical issues, encouraged self-reflection and was perceived to improve communication around end-of life care. Despite these improvements, levels of moral distress were not generally thought to have changed. However, a concerning shift in focus from the patient to clinicians was expressed by a minority of clinicians. Conclusions: Moral distress remains a significant issue within NICUs, particularly for nursing and medical trainees. However, this dissertation challenges current literature on moral distress by highlighting that whilst moral distress is strongly associated with predictions of death and disability, moral distress is only weakly explained by the clinical trajectory of the patient. The presence of moral distress provides an opportunity to question underlying assumptions of a moral judgment and review the goals of care for a patient but does not necessarily dictate a change in clinical management. The interactions between clinicians, families and the wider institution require further consideration in order to promote clinician well-being whilst maintaining the interests of the patient and their families.

The emergence of Plasmodium falciparum strains resistant to first-line antimalarial treatments the artemisinins in the Greater Mekong Sub-region (GMS) instigated the goal of total malaria elimination for the entire region. Pivotal to achieving this goal will be a thorough understanding of the epidemiology of Plasmodium spp. and how to monitor and evaluate elimination progress. In this thesis, I examined naturally acquired immunity to malaria, which underpins the epidemiology of malaria, in three naturally exposed cohorts participating in trials of malaria interventions in the GMS. Specifically, I sought to determine the role of acquired antimalarial immunity in parasite clearance, subclinical infection and surveillance of artemisinin resistance and malaria rebound. Firstly, I sought to determine the contribution of functional antimalarial antibodies to the surveillance of emerging artemisinin resistance in the GMS and South-East Asia. I measured cytophilic IgG subclasses, antibody mediated opsonic phagocytosis, and fixation of early complement proteins in response to merozoite targets in participants of a multi-centre artemisinin therapeutic efficacy study. I found that IgG3, complement fixation and phagocytosis of merozoite targets were associated with faster parasite clearance half-life (hours) during artemisinin treatment, the main metric currently used to assess the emergence of artemisinin resistance. These findings expand our current understanding of the role of functional antibody mechanisms in the surveillance of artemisinin resistance phenotypes. Secondly, I sought to determine the role of acquired immunity in subclinical Plasmodium spp. infection, which is yet to be characterised in the GMS. I utilised samples collected by village health volunteers in South-East Myanmar to determine the temporal trends in subclinical P. falciparum and P. vivax infection and antimalarial IgG responses. The prevalence of Plasmodium spp. by RDT was extremely low throughout the entire study period (0.14%); however, by qPCR the prevalence of P. falciparum and P. vivax infections was approximately 22-fold higher. Considering the relatively low levels of infection, levels and seroprevalence of antimalarial IgG in response to sporozoite and merozoite parasite targets were generally high and fluctuated over the study period. In adjusted analyses IgG seroprevalence was associated with a reduction in the odds of subclinical P. falciparum, indicating that at this point in the elimination curve, antimalarial IgG may act as a marker of protective immunity in this region. This seminal finding of a protective role of antibodies in subclinical malaria has implications for malaria rebound following decreases in transmission. Finally, in order to determine the effects of mass drug administration aimed at eliminating subclinical P. falciparum reservoirs on acquired antimalarial antibody responses and subsequent malaria rebound, I measured antimalarial IgG in response to P. falciparum merozoite targets in participants randomised to receive mass drug administration. Throughout the study, IgG levels and seroprevalence fluctuated slightly but were mostly maintained over the follow-up period. In adjusted analyses IgG seroprevalence was associated with a reduction in the odds of P. falciparum infection; however, due to the effectiveness of the intervention and significant declines in P. falciparum incidence there was considerable uncertainty in these estimates. Findings from this chapter indicate that up to two years post-intervention, antibody responses to conserved merozoite antigens do not significantly decline despite significant reduction to subclinical P. falciparum reservoirs. As such, longer follow-up studies will be required to understand the role of immunity in malaria rebound. Collectively, the findings presented in this thesis provide a comprehensive analysis of acquired antimalarial antibody responses in the GMS. These findings provide a knowledge base for the role of acquired antimalarial immunity in the epidemiology of Plasmodium spp. in the GMS, and their role in the surveillance of Plasmodium spp. infections, including strains of artemisinin resistant P. falciparum.

Background: Hospital admission rates due to cardiovascular disease are rising. Consequently, evidence-based, secondary prevention programs that enable people to self-manage their condition(s) within the community are becoming critical. For people with heart disease, cardiac rehabilitation (CR) is recommended within clinical guidelines to enable risk factor reduction and prevent new cardiac events. However, utilisation rates of CR programs remain low and the quality of program delivery is variable. In Victoria, Australia, there is no mechanism to systematically monitor attendance of CR programs, the quality of the service provided or associated patient outcomes. Registries are effective instruments for capturing data that reflect daily clinical practice, providing insights into patient characteristics and patterns of care to help address variability in service referral, uptake and quality. The purpose of this thesis was to use a learning healthcare system model to determine how the monitoring of CR could be improved in Victoria. Methods: This thesis consists of several complementary projects. First, to quantify the benefits of attending CR, a cohort of Victorian patients with acute coronary syndrome (ACS) was analysed to determine the impact of CR on all-cause hospital readmission rates up to 24-months post-ACS event. Second, a systematic review of the international literature on CR registries was undertaken to identify how other countries evaluate CR, and the associated challenges. Subsequently, the feasibility of collecting CR minimum variables via an automated data capture tool and manual-entry, web-based tool was assessed. Finally, a national think tank with researchers, policy-makers and clinicians representing every Australian state and territory was held to develop recommendations for improving the measurement of CR quality and effectiveness across Australia. Results: In a Victorian ACS population (n=416, 79% male, mean age 58.4 years) CR attendance was associated with lower frequency of hospital admissions (OR: 0.53; 95%CI: 0.31, 0.91; p-value:0.022) and length of stay (Coef. -1.21 days; 95%CI: -2.46, 0.26; p-value: 0.055) in fully adjusted models. Within the systematic review, it was identified that the use of registries for characterising CR and providing a standard for quality improvement is in its early stages but shows promise for national and global benchmarking. Data are commonly entered manually into web-based tools; a process that potentially limits sustainability. However, the feasibility study determined that an automated data capture tool is currently unfeasible in Victoria due to under-developed electronic health infrastructure. Last, the national think tank determined that to improve the monitoring of CR at a national level, enhanced governance structures, leadership and collaboration across jurisdictions are required. Conclusion: This thesis substantiates that attending CR is associated with lower frequency of hospital admissions in an ACS population. However, the ability to quantify healthcare quality relies on the implementation of appropriate systems that can accurately capture how care is being delivered. Clinical registries provide a vehicle for measuring the delivery of care. Nevertheless, challenges associated with the establishment and cost of registries often impact implementation and sustainability. Innovative solutions are required to reduce the burden of data entry on time-poor clinicians and these solutions need to be supported by enhanced infrastructure, governance and national collaborative efforts. This thesis has contributed towards the important discussion on how CR can be improved in Victoria and more broadly across Australia, as evidenced by the collective development of national CR quality indicators that is currently underway.