Of four million annual neonatal deaths, 80% occur in Asia and sub-Saharan Africa. To improve survival, cheap, equitable strategies are needed that are effective within community settings. The overall aim of the thesis was to explore early infant feeding practices and neonatal mortality across two trial surveillance systems in rural sites in Nepal, with particular interest in the discarding of first breastmilk as a new dimension in early infant feeding. The primary objectives were to examine the associations between maternally reported discarding of first breastmilk and neonatal mortality in (i) Makwanpur and (ii) Dhanusha districts, Nepal, and (iii) to describe qualitatively the practices and reasons behind discarding of first breastmilk in Makwanpur. Secondary objectives were to describe associations of first milk discarding with suspected acute respiratory infection and infection-specific mortality, replicate previous analyses of established breastfeeding patterns, breastfeeding initiation time and neonatal mortality, clarify any relationship between prelacteal feeding and neonatal mortality, and to determine characteristics predicting discarding. A qualitative secondary objective was to explore the behaviours represented by mothers categorised as having discarded or not discarded their first milk. Data on a range of variables were collected from mothers aged 15-49 who had given birth at least one month previously. Discarding of first milk was significantly associated with neonatal mortality in Makwanpur (n=15 919, aOR 2.21 [1.47 - 3.32], p<0.0001) and Dhanusha (n=14 991, aOR 1.28 [1.00 - 1.62], p<0.043) after adjustment for several covariates including other infant feeding dimensions, and exclusion of deaths within two days of birth. There was dichotomy in perceptions of first milk. Practices of discarding included squeezing out small volumes of breastmilk before initial feeding. The findings suggest that discarding of first milk represents an important dimension of early infant feeding in addition to breastfeeding initiation time, with implications for policy and research.

Background: Interest in and debates around health literacy have grown over the last two decades and key to the discussions has been the distinction made between functional health literacy, communicative/interactive health literacy and critical health literacy. Of these, critical health literacy is the least well developed and differing interpretations of its constituents and relevance exist. Rigorous analysis of the concept of critical health literacy is necessary in order to offer some clarity of definition upon which appropriate theory, well grounded practice and potential measurement tools can be based. Research Aim: The aim of the study is to clarify the concept of critical health literacy and to examine its usefulness in enabling individuals and groups to take control over their health. Method: The study develops an Applied Evolutionary Concept Analysis method to systematically identify the features associated with this concept. A unique characteristic of this method is that it practically combines an analysis of the literature with in depth interviews undertaken with practitioners and policy makers who have an interest in the field. Finally, it explores the concept within a practice context using a case study method to examine two projects. Findings: Findings revealed that critical health literacy is a unique concept that has a different set of characteristics from similar but related concepts such as empowerment. Its framework of attributes is complex and multi faceted, consisting of advanced personal skills, health knowledge, advanced information and analysis skills, effective interaction between service providers and users, informed decision making and empowerment including political action. Critical health literacy is a transactional concept in that it exists at an individual and community level while being dependent upon the skills, commitment and political will of the health professionals, services and decision makers with whom they interact. Conclusion: This study has systematically delineated the concept of critical health literacy. It concludes that it is an important and unifying concept for health promotion and that there is value in developing it as a way of enabling individuals and communities to take control of their health at both a personal and structural level. The conceptual clarity that this study provides enables critical health literacy to be recognised and operationalised and offers problem solving opportunities for the goals of health promotion in addressing the social determinants of health, improving health outcomes and reducing health inequalities. The Applied Evolutionary Concept Analysis method developed in this study has been shown to be rigorous, with a clear philosophical basis and is useful and relevant for the field of health promotion.

Forming habits – broadly defined as learned automatic responses to contextual cues – is proposed as a means to health behaviour maintenance. Habits form through context dependent repetition, but additional variables may affect this process. This work aimed to investigate additional predictors of habit formation. The potential predictors investigated were: placement in a routine; attraction versus health information; perceived reward and its impact; and individual differences. The study investigated habit formation for dental flossing and taking a vitamin C tablet daily, in a sample of 118 participants from the general public. Eighty participants received an online intervention for vitamin C tablets at the start of the study. All participants received an intervention for flossing after four weeks, randomised to be based on attraction versus health, and to advise flossing before or after brushing. Self-reported behaviour, habit, and motivational variables were measured every four weeks, for sixteen weeks. Habit strength and behaviour increased for both target behaviours. Neither behaviour nor automaticity were affected by the attraction and health interventions, or placement of flossing before or after toothbrushing. Experiencing the behaviour to be rewarding led to greater automaticity, but was not mediated by behaviour frequency. Reward strengthened the behaviour-automaticity relationship, as predicted by theories of reinforcement. Pleasure and intrinsic motivation also acted in this way, but positive attitudes about the behavioural outcomes did not predict behaviour or automaticity. Rational, rather than experiential, thinking style was associated with weaker habits, and higher prospective memory ability predicted greater initial gains in behaviour, followed by stronger habits. The results suggest that there are experiential and trait factors which affect the habit formation process and which can reinforce habits, beyond their impact upon repetition. These findings advance habit theory and lead to suggestions for intervention development, such as recommending encouraging intrinsic motivation, and tailoring interventions according to individual dispositions.

What role, if any, the concept of personal responsibility for health should play in health policy is a burning question. Human beings have long known that their behaviour can cause ill-health, but it is only in the last century that we have come to understand that our behaviour is causally responsible for a substantial amount of the global burden of disease. Partly as a result of these new findings politicians have started to advocate "cost bearing" policies which require people who are responsible for their ill-health to bear some of the costs of their risky health affecting choices. In this respect, policy makers and other stakeholders have stolen a march on philosophers because relatively little work has hitherto been carried out by these academics to determine whether cost bearing health policies are normatively justifiable and practically feasible. As such, the primary goals of my thesis will be to develop a framework for analysing these questions and to provide some substantive answers of my own. The main conclusions that I will draw in thesis is that some forms of cost bearing policies - namely, "risk tax" and "risk insurance" policies - are both practically feasible and normatively reasonable and that these policies can be defended by at least five different normative arguments - including a theory of justice known as "luckist" luck egalitarianism. During the course of my thesis I will also defend the supplemental claims that ill-health is, causally speaking, very much in our own hands and that most people are substantively morally responsible for their risky health affecting choices.

Onset of Type 1 diabetes leads to a “biographical disruption” where the individual’s life completely changes. International standards for clinical management of diabetes exist, but fail to adequately consider the needs of individuals. The aim of this research was to: investigate the needs of people with Type 1 diabetes; use Maslow’s hierarchy of needs as a framework to present the results in a hierarchical form; and propose that as Type 1 diabetes can be viewed as a tracer condition the results of this work may be applicable to other Chronic Noncommunicable Diseases. A qualitative approach was taken using semi-structured interviews with 101 people from 13 countries. Grounded Theory was used for data collection and analysis. Content analysis was used to ascertain whether the needs defined by professionals were met or not, and Thematic Analysis to identify the “needs” as defined by the interviewees. The results of the Thematic analysis were developed into a pyramid with “Policies”, “Organisation of Health System”, “Insulin”, “Delivery of insulin”, “Control e.g. blood or urine glucose”, “Healthcare workers” and “Information and education” at the base as they were needed for survival. The next level included “community, family and peers”, and also a changing role for “Healthcare workers” in their approach to care and delivering “Information and education”. This enabled people to start learning how to use “Insulin”, “Delivery of insulin” and “Control e.g. blood or urine glucose” in a flexible way. People’s “Experience” and “Personality” then helped them “Adapt” and “Be Open” about their diabetes and fully participate in society. In turn this allowed for “Acceptance” and ultimately viewing “diabetes as something positive”. This research developed a hierarchy of needs for people with Type 1 diabetes and highlighted the importance of both health and social needs in order to ensure ideal management of diabetes.

Background: Maternal and neonatal mortality are high in Malawi, and cost-effective and sustainable interventions are needed in order to reduce mortality rates and make progress to achieve Millennium Development Goals 4 and 5 for mother and child health. Where health systems are weak and many women deliver at home, community-based interventions have the potential to make an important contribution to health improvements. Methods: A cluster-randomised study with a factorial design was used to evaluate the impact of two community-based interventions on maternal and child health outcomes. A prospective pregnancy and birth monitoring system was developed to collect information on pre-specified pregnancy, birth and infant outcomes. The research presented here focuses on the women’s group intervention, which uses participatory methods to mobilise communities to take actions for maternal and child health problems they identify. Results: 18,562 pregnancies were followed up, resulting in 18,340 live births, 362 stillbirths, 434 neonatal deaths and 73 maternal deaths. 11,450 live births were identified retrospectively, resulting in 484 infant deaths. Statistically significant reductions in maternal and neonatal mortality as a result of the women’s group intervention were not seen (adjusted odds ratio 0.94 (95% CI 0.56-1.61) and 0.95 (95% CI 0.71-1.28) respectively). There were significant improvements in antenatal care and immunisation, and reductions in births attended by traditional birth attendants, and there were non-significant reductions in mortality and increases in health-care seeking. Discussion: Although women’s groups showed promising signs of community-level action for mother and child health, methodological factors, such as low power and baseline imbalance after randomisation, may have limited the ability of this study to detect an impact of the intervention on mother and child health outcomes. Design and implementation factors may also have caused delays and limited the measurable impact of the intervention at this time. Follow-up over a longer period may show greater impact.

Background: Interventions targeting female sex workers (FSWs) are pivotal to HIV prevention in India. Societal factors and legislation around sex-work are potential barriers to achieving this. In recent years several high profile closures of red-light areas and dance bars in India have occurred. In this thesis I describe the effects of the demolition of Goa’s red-light area on the organsiation of sex-work, HIV risk environment, and implications for evidence-based HIV prevention. Methods: The pre-demolition phase was a detailed ethnographic study. The early post-demolition phase included rapid ethnographic mapping of sex-work in the immediate aftermath. The late post-demolition phase was a cross-sectional survey supplemented by an in-depth qualitative study. 326 FSWs were recruited throughout Goa using respondent-driven-sampling, and completed interviewer-administered questionnaires. They were tested for sexually transmitted infections (STIs) and HIV. Results: The homogeneous brothel-based sex-work in Goa evolved into heterogeneous, clandestine and dispersed types of sex-work. The working environment was higher risk and less conducive to HIV prevention. Infections were common with 25.7% prevalence of HIV and 22.5% prevalence of curable STIs. Women who had never worked in Baina, young women, and those who had recently started sex-work were particularly likely to have curable STIs, a marker of recent sexual risk. STIs were independently associated with young age, lack of schooling, no financial autonomy, deliberate-self-harm, sexual-abuse, regular customers, streetbased sex-work, Goan ethnicity, and being asymptomatic. Having knowledge about HIV, access to free STI services, and having an intimate partner were associated with a lower likelihood of STIs. HIV was independently associated with being Hindu, recent migration to Goa, lodge or brothel-based sex work, and dysuria. Conclusions: Tackling structural and gender-based determinants of HIV are integral to HIV prevention strategies. Prohibition and any form of criminalisation of sex-work reduce the sex workers’ agency and create barriers to effective HIV prevention.

Disability is an issue of human rights and equal opportunity and is no longer focused on impairment and medical intervention. Adolescents with disabilities (AWD) are marginalised throughout the world (UNICEF, 2005), particularly in low- and middle-income countries where they are affected by poor access to resources, and by stigma and local taboos. Poverty has been shown to accentuate the impact of disability. The cycle of poverty and disability is marked by poor access to education, vocational training, and employment. The need of AWD for inclusion and participation in education, health and social life are the same as their non-disabled peers. For many AWD, these needs continue to go unmet (Groce, 2004). The research question was: ‘How do adolescents with physical disabilities perceive the factors that determine their participation and inclusion in their communities in South India?’ Quantitative and qualitative techniques were used. Thirty-seven AWD, 25 parents of AWD, and 24 non-disabled adolescent peers participated in the study in Bangalore, India. A background information questionnaire and two measures of quality of life (QOL) were conducted with all participants. Qualitative data was gathered with each participant group using interviews and focus group discussions (FGD). Photography was used with the AWD to facilitate discussions. Analysis of QOL data revealed no differences between AWD and parent-proxy scores. Differences exist in QOL scores between AWD and their peers. Analysis of interview and FGD data revealed three key themes that influence the perception of Participation and Inclusion among AWD. These are: 1) Personal Factors; 2) Interpersonal Relationships; and 3) External Factors. Belief in themselves and parental support were two key facilitators of participation. Negative attitudes of others, physical barriers, and poorly implemented legislation were significant barriers to participation. Recommendations are suggested to enhance the participation and inclusion of AWD and to facilitate success in adulthood.

Background: Of the world’s four million neonatal deaths, 99% occur in low/middleincome countries, but most information relates to the 1% dying in high-income countries. Reliable cause-of-death data are lacking. The aim of this thesis is to develop programmatically-relevant, national estimates for neonatal cause-of-death, focusing on “birth asphyxia” to illustrate specific challenges in the available data and for systematic national estimates. Objectives: 1. Review estimation methods, giving implications for neonatal cause-of-death estimation. 2. Propose programmatic categories for neonatal cause-of-death, reviewing measurement options for intrapartum-related outcomes (“birth asphyxia”). 3. Identify and analyse existing neonatal cause-of-death data. 4. Estimate intrapartum-related neonatal deaths for all countries, comparing single-cause and multi-cause models. 5. Summarise priorities for improving neonatal cause-of-death estimates and input data. Data inputs: Case definitions were reviewed for neonatal cause-of-death and intrapartumrelated outcomes. Six programmatically relevant cause-of-death categories were defined, plus a residual “other neonatal” category. Two sources of neonatal cause-of-death data were examined: Vital Registration (VR) datasets for countries with high coverage (>90%) based on a new analysis from 83 countries; and published/unpublished studies identified through systematic searches. Inclusion criteria for representativeness and comparability were applied. Data from 44 countries with VR (96,797 neonatal deaths) and from 56 studies (29 countries, 13,685 neonatal deaths) met inclusion criteria, despite screening almost 7,000 abstracts. These data represent <3% of the world’s neonatal deaths. Thus estimation is necessary for global level information. No useable data were identified from Central and North-West Africa, or Central Asia. Modelling: Methods were developed to estimate intrapartum-related neonatal deaths (single-cause), and then simultaneously estimate seven causes of neonatal death (multi-cause). Applying these proportions to the numbers of neonatal deaths in 192 countries gives a global estimate of intrapartum-related neonatal deaths of 0.90 (0.65-1.17) million using single-cause and 0.91 (0.60-1.08) million using multi-cause methods. Discussion: The multi-cause model has become WHO’s standard method for neonatal cause-of-death estimates. However, complex statistical models are not a panacea. More representative data are required. Simplified case definitions and consistent hierarchical cause-of- death attribution would improve comparability, especially for intrapartum-related deaths.

Background: Low birth weight (LBW) remains a major public health problem in developing countries, but is only one measure of size at birth. Others include small-for-gestational-age (SGA) and low ponderal index (PI). The objectives of the thesis were: to estimate the prevalence of LBW, SGA and low PI in a cohort of Nepalese infants; to identify risk factors for small size, and to investigate whether prediction models were useful for screening; and to assess the effects of size at birth on subsequent outcomes. Methods: Mothers enrolled in a prospective trial were followed through pregnancy and delivery. Child anthropometry was collected at birth and at two years of age. A range of indices of size at birth were described. Multivariable regression models were developed to predict them, and their associations with subsequent outcomes. Results: There was a high prevalence of LBW (25%), SGA (55%) and low PI (70%) at birth. None of the prediction models for size at birth was particularly good, the strongest being for birth weight (R2=33%). Common predictors were parity, pre-pregnancy weight, gestational weight gain, gestational duration and infant sex. LBW was associated with neonatal (OR 3.5, 95% CI 1.4-8.9), infant (3.6, 1.6-7.9) and young child (3.7, 1.7-7.8) mortality, and stunting (3.4, 2.2-5.3), wasting (2.9, 1.5-5.6) and underweight (3.7, 2.5-5.5) at two years of age. Discussion: In southern Nepal, many newborn infants were classified as small, and most were disproportionate. The modifiable risks for small size at birth were few, even though it was associated with mortality and size in childhood. Conclusion: The previously undescribed disproportionate majority of Nepalese infants is worrying for public health. However, measurement of birth weight is not yet routine and it seems better to recommend LBW as a single risk measure than to add new and more complicated activities.