Globally, over the last decade primary school access has improved significantly. Yet challenges remain: 67 million primary school-aged children are not in school. Poor nutrition and health among schoolchildren are important barriers in achieving education-for-all goals. School feeding is a popular intervention supporting the education, health and nutrition of children in food-insecure settings. However, school feeding programmes are complex, involving a broad range of stakeholders across different sectors and implementation levels. This thesis is aimed at providing evidence to support policy-makers in managing trade-offs among alternative targeting approaches, feeding modalities, and costs. This work is also aimed at building an evidence-based framework to guide Governments in managing the inherent complexity of school feeding interventions. The thesis includes an analysis of a natural experiment involving survey data from 32 countries across sub-Saharan Africa that suggested that school feeding increased enrolment by 10 percent. Enrolment changes varied by modality and gender, with onsite meals having stronger effects in the first year of treatment in lower grades, and onsite combined with take-home rations being effective post-year 1, particularly for girls. Expenditures across 62 countries indicated considerable differences in costs across modalities, ranging from $23 USD for fortified biscuits to $75 USD for take-home rations. This raises important questions of cost-effectiveness and sustainability, also in terms of school-level costs not normally captured in programme expenditures. Findings also suggest that school level costs are substantive, and are a considerable overhead, considering that these costs are generally borne by food-insecure communities. The thesis also highlights that scaling-up school feeding requires significant financing, on average equal to 40 percent of primary education costs. Despite these opportunity costs there is strong buy-in on school feeding from governments in sub-Saharan Africa. The implications of this thesis also suggest that the complexity of school feeding as an intervention has perhaps been underestimated by policymakers. Strengthening the evidence linking outcomes to the design of school feeding and to the quality of the service delivery, including the trade-offs between implementation modalities, remains a critical area of future research. This thesis provides both a foundation and a step towards answering these complex questions in a comparable and meaningful way.

Background: Healthcare associated infections (HCAI) pose a significant burden to health systems both within the UK and internationally. Surveillance is an essential component to any infection control programme, however traditional surveillance systems are time consuming and costly. Large amounts of electronic routine data are collected within the English NHS, yet these are not currently exploited for HCAI surveillance. Aim: To investigate whether routinely collected electronic hospital data can be exploited for HCAI surveillance within the NHS. Methods: This thesis made use of local linked electronic health data from Imperial College Healthcare NHS Trust, including information on patient admissions, discharges, diagnoses, procedures, laboratory tests, diagnostic imaging requests and traditional infection surveillance data. To establish the evidence base on surveillance and risks of HCAI, two literature reviews were carried out. Based on these, three types of innovative surveillance tools were generated and assessed for their utility and applicability. Results: The key findings were firstly the emerging importance of automated and syndromic surveillance in infection surveillance, but the lack of investigation and application of these tools within the NHS. Syndromic surveillance of surgical site infections was successful in coronary artery bypass graft patients; however it was an inappropriate methodology for caesarean section patients. Automated case detection of healthcare associated urinary tract infections, based on electronic microbiology data, demonstrated similar rates of infection to those recorded during a point prevalence survey. Routine administrative data demonstrated mixed utility in the creation of simplified risk scores or infection, with poorly performing risk models of surgical site infections but reasonable model fit for HCA UTI. Conclusion: Whilst in principle routine administrative data can be used to generate novel surveillance tools for healthcare associated infections; in reality it is not yet practical within the IT infrastructure of the NHS.

Dengue is a vector-borne disease found across much of the world, with an increasing number of cases annually. This thesis explores the dynamics of dengue infection within an individual, and the possible impact of this at a population level. I use mathematical modelling and statistical analysis, tightly coupled with data, as a way of tying together the important components and processes during infection. I model the virus and immune dynamics, capturing the differences between individuals, disease severity and primary/secondary disease (with a focus on hypothesised secondary mechanisms). Within the immune dynamics I concentrate on antibody, looking at the role of antibody in limiting infection. Within this framework I also consider the impact on these dynamics of an antiviral. The final section of this thesis brings together this closer consideration of virus dynamics and considers their impact at a population level. Using data from biting experiments I am able to characterise the “infectivity” of an individual over time, how this varies between individuals and groups (as above), and how this compares to previous transmission modelling assumptions. In terms of control I look at how this “infectivity” is altered by antivirals and by wolbachia infected mosquitoes.

There has been considerable interest recently in quantifying uncertainty beyond that due to random error in meta-analyses. This is particularly relevant to meta-analyses of observational studies, since error in estimates from these studies cannot be attributed to a randomization mechanism. Typically, observational studies are also subject to error due to measurement error, non-participation, and incomplete adjustment for confounding. Errors due to these sources are often referred to as bias. To quantify uncertainty due to bias, researchers have proposed using "bias models" and giving subjectively elicited probability distributions to parameters that are not identifiable in the models. In a typical meta-analysis, probability distributions involving tens of parameters will have to be elicited. At the same time, the resulting estimate and uncertainty interval of the overall (meta-analytic) effect measure will generally be very sensitive to this multi-dimensional subjectively-elicited distribution. To overcome some of the problems associated with the use of such a distribution, I propose an alternative method for eliciting and quantifying uncertainty due to bias. In the method of this thesis, the lower and upper bounds of bias parameters are elicited instead of probability distributions. The most extreme Bayesian posterior inference for the target parameter of interest within the specified bounds is sought through an algorithm. The resulting lower and upper bounds for the target parameter of interest have interpretation of a Robust Bayes analysis. In this thesis, the method is applied to a meta-analysis of childhood leukaemia and exposure to electromagnetic fields. The method of this thesis was found to produce uncertainty intervals that are generally more conservative in comparison with the standard approach. It is also proposed that the method be used as a tool for sensitivity analysis, and some interesting insight is gained from the childhood leukaemia data. [For supplementary files please contact author].

Amphibians are the most threatened taxon assessed by the IUCN Red List, with over 42% of all species in decline. The emerging infectious disease chytridiomycosis, caused by the fungus Batrachochytrium dendrobatidis, has been shown to a driver of many of these declines. The broad aim of this thesis is to develop mechanistic models that realistically capture the observed disease dynamics of Bd within Europe, and use these to help understand its ecological and environmental drivers. Mathematical modelling, field work and experimental work are used in order to obtain an understanding of a Bd-host system on Mallorca, and in combination, to show how the host population response to Bd is highly context-dependent. The understanding gained is used to help predict the consequences of an attempt to mitigate (reduce or avoid the negative effects of) Bd in this system, and the mitigation attempt’s short-comings are then analyzed in order to better inform future efforts. The context-dependence of a host-population response to Bd will depend of two components: a component dependent on the host and a component dependent on Bd. Modelling of infection experiment results in the frog species Silurana tropicalis is used to show a temperature-dependent host response which is separate to the temperature-dependent response of Bd. Multi-host models of Bd are created and used to show how increasing species diversity can increase disease risk, and a method of estimating unknown epidemiological parameters for Bd based on known facts about the host’s biology is presented.

The use of routinely collected, or administrative, data for measuring and monitoring patient safety in primary care is a relatively new phenomenon. With increasing availability of data from different sources and care settings, their application for adverse event surveillance needs evaluation. In this thesis, I demonstrated that data routinely collected from primary care and secondary care can be applied for internal monitoring of adverse events at the general practice-level in England, but these data currently have limited use for safety benchmarking in primary care. To support this statement, multiple approaches were adopted. In the first part of the thesis, the nature and scope of patient safety issues in general practice were defined by evidence from a literature review and informal consultations with general practitioners (GPs). Secondly, using these two methods, measures of adverse events based on routinely collected healthcare data were identified. Thirdly, clinical consensus guided the selection of three candidate patient safety indicators for investigation; the safety issues explored in this thesis were recorded incidents with designated adverse event diagnostic codes and complications associated with two common diseases, emergency admissions for diabetic hyperglycaemic emergencies (diabetic ketoacidosis, DKA and hyperglycaemic hyperosmolar state, HHS) and cancer. In the second part of the thesis, the contributions of routinely collected data to new knowledge about potentially preventable adverse events in England were considered. Data from a primary care trust (NHS Brent), national primary care data (from the General Practice Research Database, GPRD) and secondary care data (Hospital Episode Statistics, HES) were used to explore the epidemiology of, and patient characteristics associated with, coded adverse events and emergency admissions for diabetic hyperglycaemic emergencies and cancer. Low rates of adverse events were found, with variation by individual patient factors. Finally, recommendations were made on extending the uses of routinely collected data for patient safety monitoring in general practice.

Bloodstain Pattern Analysis (BPA) is a forensic application of the interpretation of distinct patterns which blood exhibits during a bloodletting incident, providing key evidence with its ability to potentially map the sequence of events. The nature of BPA has given the illusion that its evidentiary significance is less than that of fingerprints or DNA, relying solely on the interpretation of the analyst and focusing very little on any scientific evaluation. Recent preliminary literature studies have involved a more quantitative approach, developing directly crime scene applicable equations and methodology, which have established new ways of predicting the angle of impact, impact velocity, point of origin of blood and blood pattern type. Using these new equations and further improving on them to include a variation of impact surfaces, surface properties (i.e. porosity, roughness, manufacturing process etc.) and changes in blood properties is the principal focus of this work. The primary objective of this research is to expand the knowledge of blood and surface interactions and generate general equation/s or quantitative approaches that encompasses some of the possible conditions, in relation to Bloodstain Pattern Analysis (BPA), which may be encountered at a crime scene. Overall validating BPA and supporting a more reputable / respected scientific field giving credence to its usage within criminal trials. This thesis is presented in three parts: The first part explores blood, its characteristics and how manipulating the components of blood (i.e. packed cell volume, PCV), can alter the way a bloodstain forms and dries. Since packed cell volume is instrumental in the overall viscosity of blood, which ultimately determines the final bloodstain diameter via the natural fluctuation exhibited throughout the body and by the individual human characteristics, it was deemed necessary to investigate its effect on the interpretation of bloodstains. Packed cell volume was found to alter the size of bloodstains significantly, where increments in their diameter were experienced when PCV% was decreased; angled impacts were unaffected. The mechanism of drying blood was also analysed, the current understanding being that blood dries primarily by the Marangoni Effect. However this is found to be altered when PCV% is considered; low PCV% exhibits a strong Coffee Ring Effect where higher PCV% levels dry by the Marangoni Effect. Other drying characteristics considered were volume analysis, skeletonisation and the halo effect where PCV% was manipulated. Volume analysis methods were significantly affected by PCV%, where new drying constants were established and several established scientific methods were shown to be unreliable at determining the volume. The second part of this thesis investigates surface interaction, exploring the fundamentals of various common surface types, and how individual features (i.e. surface roughness) affect the interpretation of bloodstains; four common surfaces were considered (wood, metal, stone/tile and fabric). Blood drop tests were performed at different heights and angles where recently formulated equations were applied to the results to create new constants, which could be used to distinguish between surface types. Wood and fabric were found to alter the spread of blood most significantly, constants increased or decreased substantially, compared to the original value. The last part of this thesis expands the groundwork set forth in part two. Surfaces were manipulated, either by heat or cleaning. Since it is possible that blood may interact with a surface which may have been cleaned (to remove blood, or simply to clean surface prior to any blood impaction) or heated (i.e. radiators), it is important to fully explore surface alterations which commonly occur in an everyday environment and therefore are highly probable to be encountered at a crime scene. Surface manipulation is investigated in the form of a heated surface, where a blood boiling curve reminiscent of the water boiling curve was created establishing four visibly recognizable boiling regimes. Heat was found to decrease the resultant bloodstain diameter, separate blood into its components and create reduction rings as the temperature increased. An equation accounting for these changes was deduced, further showing how simple alterations to the surface, which have previously been overlooked, can interfere with the results. Further surface manipulation was implemented in the form of cleaning, since cleaning can be performed before blood impacts, therefore causing a surfactant layer, of after blood has impacted the surface, indicating crime evasion. Secondary analysis of blood on a heated surface in conjunction with cleaning was implemented, establishing the effectiveness of presumptive testing and the ability to extract valuable DNA. Initial presumptive testing and DNA extraction was found to be successful for all temperatures, however when various cleaning methods were applied (a common occurrence at crime scenes) DNA testing produced negative results at temperatures of 50oC onwards. Fabric washing, using various household detergents and methods of washing/drying were also evaluated. Detergents significantly increased the resultant diameters of bloodstains, secondary rings were experienced on all polyester and silk fabrics, establishing constants relating to the secondary ring produced. Repeated cycles of washing were found to produce a stable fabric after 6 cycles, for most fabric types.

Introduction Coronary heart disease (CHD) is one of the leading global causes of morbidity and mortality. The underlying biological mechanisms are well understood, and a host of causal risk factors for the disease have been identified, mainly related to diet, smoking and physical activity. Evidence-based treatments for the disease are also available, reducing mortality and improving quality of life. The decline in CHD mortality rates observed in most developed countries since the 1960s represents a most remarkable epidemiological phenomenon. However, this decline is not universal, and may now be in jeopardy. Thus, the mortality decline has recently plateaued in young adults in the United States. Furthermore, the absolute burden of disease is set to increase mainly because of an increasingly ageing population, and will represent a heavy burden to high, middle and low income countries alike. Furthermore, CHD incidence may rise in future because of recent adverse trends in major CHD risk factors, namely the worldwide increases in obesity and diabetes prevalence observed since the 1980s. Moreover, new technology and improved treatments are decreasing case fatality in CHD patients, increasing life expectancy and thus expanding the pool of patients surviving with clinically apparent disease. Finally, and crucially, important socioeconomic inequalities persist, perhaps reflecting disease determinants. The complex interplay of these factors and potential changes over time together suggest that the CHD epidemic may still be evolving. Further attention is therefore essential. The analysis of time trends in disease specific mortality can thus potentially help us to understand the population dynamic of diseases such as CHD, warn about key changes and perhaps offer some novel insights for better prevention and control. However, most previous analyses have been focused on age-adjusted rates that might conceal important differences by age or by socioeconomic status, which might provide further understanding of trend drivers. Aims and objectives: My aim is to study recent coronary heart disease mortality time trends in different countries, in order to better understand the current state of the CHD epidemic. Furthermore, I will analyze the relative importance of CHD treatments and risk factors as drivers of the mortality trends. Finally, I will consider the Public Health implications of my findings. My objectives therefore are: 1. To summarize our current understanding of Coronary Heart Disease (CHD) causation 2. To describe recent CHD mortality time trends focusing on age and gender specific trends by identifying periods with similar rate of change in diverse populations (England & Wales, the Netherlands, Poland and Australia). 3. To describe recent CHD mortality time trends by Socio-Economic Status in England and Scotland. 4. To quantify the role of risk factors and evidence-based treatments as drivers of the CHD mortality trends, first using a modelling approach in Poland, and then in England while also considering socioeconomic factors. 5. To consider the public health policy implications of dynamic trends in coronary heart disease mortality. Methods CHD mortality trends were analysed using the joinpoint regression approach. Widely used in cancer epidemiology, but rarely in CHD, this method explores trend data to find points in time (“joinpoints”) that define segments where the trend has a constant pace of change. The key strength of this technique is objectivity- (it avoids the detection of potentially biased patterns when trends are described using time intervals defined subjectively by the researcher). Joinpoint avoids this potential bias by essentially removing the observer from the selection process, instead using a formal and objective exploration of the time-series data. My analysis therefore focused on age-adjusted rates, then age and gender specific rates. The analysis for Scotland and England also considered socio-economic status (using area-based measures of material deprivation). The contributions of risk factors and treatments to the observed CHJD mortality trends in Poland were studied using the IMPACT model, a comprehensive, population-based model of CHD epidemiology. The model goal is to quantify the decline in coronary heart disease deaths in the Polish population between 1991 and 2005 which might be explained by risk factor changes and by treatments. The model is comprehensive, incorporating all usual treatments for coronary heart disease and heart failure plus all major cardiovascular risk factors, including smoking, blood pressure, cholesterol, diabetes, obesity and physical activity. Similar analyses but also exploring the socio-economic differences were conducted in England, using a modified IMPACT model (IMPACTsec). That was used to estimate the contribution of risk factors and evidence based treatments to the observed decline in mortality in England between 2000 and 2007, for each quintile of the index of multiple deprivation. Results Age-adjusted trends in England and Wales, Scotland, Australia and the Netherlands conceal important recent age specific patterns. In these countries, the age-adjusted rates show continuing declines; however, among young adults a recent period of slowing down of the rate of decline in CHD mortality has been observed. Furthermore, trends are very dynamic, and the patterns can change surprisingly quickly. In the Netherlands, the sustained period of minimal change in young adults was followed by a period of further decline. Poland offers a strikingly different example of trend dynamism. After a period of constant increase, Poland showed a sudden, sharp decline in CHD mortality rates within a period of a very few years. This decline occurred in all age and gender groups, and still continues. The recent mortality trends are probably attributable more to changes in risk factors rather than medical treatments. For example, using the IMPACT model to study the decline phase of the Polish CHD epidemic, approximately 55% of the observed fall in mortality might be attributed to changes in risk factors, and only about a third to evidence based therapies. Because of the social patterning of risk factors levels, further insights on the role of risk factors as major contributors to trend changes can be obtained by studying trends in levels stratified by socioeconomic circumstances. Scotland and England offer particular opportunities for detailed studies of trends in CHD mortality using high quality data including socioeconomic status. The resulting picture is complex. The recent flattening in CHD mortality trends observed in young adults was confined to the most deprived groups in Scotland, but was more uniform in England. A marked deterioration of medical care is implausible, meaning that the most likely explanation for this recent flattening of CHD mortality must be adverse trends in major cardiovascular risk factors. The CHD mortality modelling in England produced intriguing results. As expected, socio-economic patterning of risk factor changes were observed. For example, decline in smoking levels contributed more to the observed decline amongst the more deprived groups. Social patterning was less clear among young adults in England. Moreover, the IMPACT SEC model analysis suggested that approximately half the CHD mortality fall was attributable to improved treatment uptake, with benefits occurring surprisingly equitably across all social groups. A similar analysis of the Scottish trends is therefore urgently needed to gain better insights on the drivers of the socioeconomic patterning underlying the observed trends. Conclusions The recent flattening in CHD mortality in young adults seen in many countries experiencing an overall decline in deaths strongly suggests that favourable trends can reverse.

Background The year on year increase in the incidence of malignant melanoma skin cancer is a global health problem. In 2009 the International Agency on Research on Cancer raised exposure to artificial UV to the highest category, “carcinogenic to humans”. Surveys in England in 2006-10 indicated that young people were using sunbeds regularly, with the North West, especially Merseyside, having some of the highest rates. There has however, been little research conducted that attempts to understand the attitudes and motivations of young sunbed users. Inadequate regulation coupled with their desire to use sunbeds has meant that young people have been using sunbeds hazardously. In 2010 the English government introduced legislation to ban under-18 sunbed use. Aims The study aimed to explore the attitudes, motivations and experiences of young people aged 14 to 16 years living in Merseyside, in relation to sunbed use in the context of the imminent legislation. The study also aimed to use the findings to promote the development of an evidence-based local skin cancer prevention strategy for Liverpool, in an attempt to begin to tackle hazardous sunbed use by young people in the locality. Methods Qualitative research was conducted in five schools in the North West of England between September 2009 and March 2010, prior to the implementation of the sunbed legislation banning under-18 year olds using them. Girls and boys aged between 14-16 years were recruited to the study. Eight focus groups were conducted, involving a mixture of sunbed users and non-users, sunbed users only, boys and girl only focus group and separate year groups. Twenty-two in-depth one-to-one interviews were also conducted. An interpretive approach was taken to the analysis of the qualitative data, drawing on the approach taken by Edwards and Tichen (2003). For the development of the skin cancer prevention strategy, a stakeholder workshop was organised in March 2013, bringing together approximately sixty key stakeholders from a variety of organisations in Liverpool and including representatives from the public, public health, primary care and voluntary sectors and the local authority. Findings from the qualitative research with young people were fed into the strategy development, with the author also acting as facilitator and coordinator. Findings Key motivations for sunbed use among these young people in Merseyside were to improve self-esteem and confidence and to conform to social norms and peer-expectation. Mothers and older siblings were reported to influence sunbed initiation or continued use. Poor salon practice emerged as a risk to over-exposure to UV rays. Young people were quick to play down the risks associated with sunbed use; however some young people also reported being addicted to sunbeds. Young people feared the introduction of the legislation banning sunbed use because some had become dependent on using sunbeds as a way of expressing their identity, to socialise and as a strategy to improve their self-esteem and confidence. The study demonstrated how the qualitative findings could be used to influence the development of national and local health policy. Following the developmental process, the strategy was endorsed and supported by the Mayor of Liverpool, the head of policy at Cancer Research UK and the head of policy at the Chartered Institute of Environmental Health. Conclusions Whilst some young people reported the physical and psycho-social benefits of sunbed use, young sunbed users were also exposed to increased risks associated with artificial UV damage. Poor salon practice, inadequate regulation and policy and continued pressure to conform to social ideals left sunbed users vulnerable to the effects of UV damage. Moreover, the research also highlighted the need for health providers to develop psychological support pathways for young people who may be addicted to sunbeds. Prevention initiatives should take into consideration young people’s ideals concerning appearance when aiming to reduce or prevent sunbed use among young people.

Introduction: Reducing inequalities in health is a public health imperative. In the UK and internationally policies are being implemented to try to reduce health inequalities, with limited success. This thesis examines the effect of socio-economic status (SES) on clinical outcomes, healthcare use and employment opportunities in people with cystic fibrosis (CF). Poorer socio-economic circumstances have been linked with worse outcomes in cystic fibrosis. Because CF is genetically determined, this offers an opportunity to investigate the impact of SES on health and social outcomes, in a chronic condition without a socio-economic gradient in incidence. This provides a useful case for understanding how health inequalities are generated, in order to develop more effective interventions, for people with CF and more generally. Methods: I analyse, for the first time, the national CF registers from the UK and Denmark, using longitudinal modeling techniques. Mixed-effects models are used to assess the association between measures of SES and longitudinal outcomes, adjusting for clinically important covariates. Study 1 explores longitudinal weight, height, BMI, %FEV1, risk of Pseudomonas colonisation, and the use of major CF treatment modalities, and their association with small-area deprivation (8055 people with 49,337 observations between 1996 and 2010). Study 2 explores longitudinal employment status in adults with CF in the UK, and its association with small-area deprivation, disease severity, and time in hospital. Study 3 presents a novel longitudinal analysis of the Danish CF registry (70,448 %FEV1 measures on 479 patients seen monthly between 1969 and 2010), to understand the way %FEV1 changes over time. Results: Compared with the least deprived areas in the UK, children with CF from the most deprived areas weighed less, were shorter, had a lower body-mass index, were more likely to have chronic P. aeruginosa infection, and have a lower %FEV1. These inequalities were apparent very early in life and did not widen thereafter. On a population level, after adjustment for disease severity, children in the most deprived quintile were more likely to receive intravenous antibiotics and nutritional treatments compared with individuals in the least deprived quintile. Patients from the most disadvantaged areas were less likely to receive DNase or inhaled antibiotic treatment. In adults deprivation, disease severity, and time in hospital all influence employment chances in CF. Furthermore, deprivation amplifies the harmful effects of disease severity on employment: the employment chances of people with CF with poor lung function from disadvantaged areas are damaged to a greater extent than for their counterparts living in the least disadvantaged circumstances. The Danish analysis quantifies the short-term variability in %FEV1 (SD 6.3%) and shows that lung function measures are correlated for over 15 years. Conclusions: In the UK, children with CF from more disadvantaged areas have worse growth and lung function compared with children from more affluent areas, but these inequalities do not widen with advancing age. Policies to reduce inequalities should thus focus on the early years. Clinicians consider deprivation status, as well as disease status, when making decisions about treatments, and this might mitigate some effects of social disadvantage. The differential social consequence of having CF in terms of employment is likely to be an important pathway for the amplification of health inequalities.