Background: High prevalence of depression among pregnant women has been observed in low and middle-income countries including Pakistan. Pregnancy is an important stressor for depression and depressed women tend to have poor dietary intake. The present study aimed to determine the prevalence of prenatal depression and its risk factors, and explored the relationship of prenatal depression with dietary intake in a representative sample of women living in a rural setting of Pakistan. Methods: This study was conducted at the baseline of a large cluster randomised controlled trial. Five hundred pregnant women in the second and third trimester of pregnancy, living a in a rural area of district Rawalpindi Pakistan, were recruited for the baseline of the trial. Depression was assessed using “Patient Health Questionnaire” (PHQ9), with a cut-off score of 10, and the dietary intake assessment was carried out by“Food Frequency Questionnaire” and “24 Hour Dietary Recall”.Data on stressful life-events and perceived social support were captured through “Life Events Checklist” and “Multidimensional Scale of Perceived Social Support”. All pregnant women who had PHQ-9 scores of 10 or more were invited to participate. Those women whose PHQ 9 was less than 10 were selected through simple random sampling and invited for baseline measurements after obtaining informed consent. Response rate was around 98%.Appropriate tests of significance were used for bivariate analysis. Final Generalized Linear Model with logit link function was obtained. Results: The prevalence of prenatal depression was found to be 27%.Depressed pregnant women belonged to 23-30 years age group, were less educated had,lived in joint family and had less perceived support from family and friends. Depressed women suffered from at least 3-4 stressful life events .Food variety scores were generated and dichotomized at median for dietary inadequacy. Mean intake of all the energy, macronutrients and micronutrients was significantly less among depressed (p < 0.001).Prenatal depression was significantly associated with dietary inadequacy (P < 0.05). In addition factors like life satisfaction, husband away from home in last six months (P < 0.05), physical IPV (P < 0.05) and stressful life events (P < 0.01) were also independently associated with dietary inadequacy. Conclusion: The current study to the best of my knowledge is the first study to investigate in a large rural community based sample of women the link between prenatal depression and dietary intake. Most of the depressed women did not eat sufficient items from various food groups to meet the recommended dietary allowance. Prenatal depression was independently associated with inadequate dietary intake. Risk factors for prenatal depression as well as dietary inadequacy were psychosocial in nature. There is a need to screen women at antenatal visit and provide nutritional counselling to improve dietary behaviors for better pregnancy outcomes.

Type 2 diabetes (T2D) is a complex multifactorial disorder driven by both genetic and environmental factors. The rapid increase of T2D in Qatar -prevalence of 16.3% in 2014 according to the International Diabetes Federation (IDF) - motivated the introduction of genetic studies among this under-presented population. Major progress to study the genetic basis of T2D came from studying common variants. However, these variants were of mild effect sizes. Studies have been shifted from applying the common variants hypothesis towards the investigation of other genetic variables including rare variants, copy number variants (CNV) and epigenetic mechanisms. This PhD project is focused on identifying genomic risk factors of T2D among the Qatari population. Eight Qatari families were analysed using the advances in genotyping and sequencing technologies. Three analyses were carried out; the aim was to identify known or novel rare variants within candidate T2D genes, identification of potential large CNVs related to T2D and detection of methylation associations with T2D. Three data sets were generated for this PhD project; these were genome-wide genotyping arrays (Illumina HumanOmni2.5M bead chip), whole genome sequencing (WGS) (Illumina hiSeq2500 platform) and genome-wide methylation arrays (Illumina 450K BeadChip capturing more than 485,577 probes). I performed three analytical experiments for the aim of this PhD project. First, linkage analysis and runs of homozygosity in combination with WGS were used to map rare variants. Second, PennCNV prediction algorithm for CNV calling was applied aiming to identify rare T2D-related CNVs. Third, association test using a linear mixed model was used for the identification of methylation associations with T2D. The findings included the identification of three rare variants detected through WGS. One variant identified within known monogenic T2D gene (GCKR) and two variants detected within known T2D genes (IGFBP2 and TGM2). These genes are functionally related to T2D and replication analysis will be required to assess their contribution to T2D among Qataris. A number of large rare CNVs detected from CNV analysis in the second experiment of the project, but potential T2D genes were not found within candidate CNVs. Also, methylation analysis replicated a significant association with T2D at cg19693031 in TXNIP (p=5.28x10¯⁶) among Qataris. The identified candidate genes were proved to be involved in the pathogenesis of T2D by causing insulin resistance and beta-cell dysfunction.

This year saw the announcement of the ambitious aim of ending the HIV epidemic by 2030 from UNAIDS. To meet this ambitious goal, a renewed commitment to the prevention of new HIV infections will be essential to stop further expansion of the epidemic. The growing array of prevention tools available must be utilised efficiently, and the core principles of a targeted and strategic combination prevention response revisited in order to ensure the greatest impact possible is generated. This thesis uses mathematical modelling to answer some of the key questions in HIV prevention programming with a view to increasing the efficiency of the response. The role of different prevention modalities in an optimised combination prevention programme is examined to understand how they can be best applied. Increasing data on the epidemic suggest that it is highly diverse across populations and settings, both in its magnitude and drivers, and also in the intensity of the response, and this provides a blueprint for designing a more strategic and tailored prevention strategy. Epidemiological and programmatic information is used to develop a geographically focused approach to prevention programming, which can generate considerably greater impact through being responsive to local epidemiology and directing funds to populations and locations where they can have greatest impact. How best to set local targets for prevention and monitor the response in light of the local characteristics of the epidemic is also investigated. We also examine how the pattern of funding available to policymakers shapes prevention programmes, and we find greater ‘front-loading’ of investment is needed to maximise impact. In this way, key strategies for improving the impact of investments are identified which can be used to inform the future of HIV prevention.

Metabonomics utilises high resolution analytical platforms to generate spectroscopic profiles that are rich in latent biological information. At present, the two principal analytical platforms used routinely in metabonomic studies are nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS). While these analytical methods in their current state of development in the field of metabonomics give broad coverage across multiple chemical classes, none report sufficiently well, if at all, on the absolute configuration of chiral molecules. As a consequence of the stereospecificity present in biological systems, a very large number of biologically active molecules exhibit chirality. Therefore, an important aspect of the metabolome remains largely unexplored by current metabonomic analytical platforms. Current enantioselective methods exist, but these are largely unsuitable for multi-analyte measurements. The work in this thesis addresses the need for appropriate chiral analytical methods for metabolic profiling. The aim was to develop a practical, fit-for-purpose chiral analytical method that will report simultaneously on numerous endogenous metabolites by untargeted or widely targeted analysis, and importantly, could be aligned with existing analytical workflows. Initial evaluation of an NMR spectroscopic approach using chiral solvating agents concluded that additional spectral complexity and interaction-induced internal reference compound chemical shift instability would prohibit efficient chiral profiling in a metabonomics workflow. An existing targeted MS-based assay for enantioselective separation of amino acids was selected for optimisation and further development. Focused analyses were performed to characterise its performance in the differentiation of individual pairs of enantiomers, and subsequently expanded, first to cover the panel of proteinogenic amino acids, and secondly to other detectable metabolites. Each stage of assay optimisation incorporated an evaluation in a representative set of samples and was able to provide highly relevant biological information that would not be accessible using current achiral metabonomic methods. Further work to expand the number of detectable metabolites and quantitative nature of the assay are discussed.

As international funding for malaria programmes plateaus, it is critical to better understand how to implement interventions such as first-line Artemisinin Combination Therapies (ACTs) most effectively through an existing health system. This thesis presents an expansion of a mathematical model of malaria transmission to provide insight to the role of health systems factors as barriers to the effectiveness of ACTs, and interventions to overcome them; considering dimensions of access to care, different sectors through which care is delivered, and the quality of care provided. Data from the IMPACT 2 study in Tanzania was used to parameterise this approach. Primary-care based interventions had most impact on transmission. In low-prevalence scenarios some single interventions, e.g. ensuring 100% care-seeking, eliminated parasite prevalence. Diagnostic-led therapy with adequate stocks of ACTs was as effective in all settings as a policy of presumptive treatment, reducing parasite prevalence in under-fives in moderate transmission settings by up to 86% depending on the sector of implementation. Modelling combinations of hospital-based interventions shifted the pattern of severe malaria away from a peak at early ages (greater than 70% relative reduction in 0-5 year olds in medium transmission settings) towards a more sustained lower incidence across 0-20 years of age as seen in low prevalence settings. This was not immune-mediated and demonstrates the role of health systems interventions preventing the development of severe malaria in those at risk, and reducing mortality. Weak health systems and a poorly controlled diversity of antimalarial sources act as barriers to deploying ACTs effectively both as a control measure and first-line treatment. Addressing these constraints needs consideration of existing healthcare provision and local priorities, e.g. reducing ACT wastage, but through specific planning may improve progress towards targets set by Roll Back Malaria to decrease clinical disease and mortality, and in low transmission settings, to approach elimination.

Background: The U.K. Government plans to improve access to general practice services in England, partly by extending the opening hours of these services. The Government expects this to enhance patient experience and reduce use of emergency hospital services. The thesis aims to provide empirical evidence relevant to these policy expectations. Methods: Three observational studies of a cross-sectional questionnaire-based survey—the GP Patient Survey—from 2011-12 to 2013-14 (2,912,535 respondents aged ≥18 years old and registered to 8,289 general practices in England). Respondent-level data were deterministically linked at the practice level to Hospital Episode Statistics Accident and Emergency (A&E) and Inpatient data and to routine data on practice characteristics. Multivariable regression estimated associations between: (1) participation in the Extended Hours Access Scheme and patient experience; (2) several patient experience measures; and (3) patient experience and rates of A&E visits and emergency hospital admissions. Results: Most patients were very (40.0%) or fairly (42.3%) satisfied with the opening hours of their practices; results were similar for experience of making an appointment and overall experience. Practices that participated in the Extended Hours Access Scheme had greater adjusted mean values of these measures than non-participants, but the mean differences were small (≤1.25 on 0-100 scales). Overall experience was most strongly associated with doctor interpersonal quality of care (standardised coefficient=0.38; 95% CI: 0.38 to 0.38). A standard deviation increase in the practice-level mean experience of making an appointment predicted a 2.7% decrease (rate ratio=0.973; 95% CI: 0.966 to 0.979) in A&E visit rates and a 1.7% decrease (rate ratio=0.983; 95% CI: 0.978 to 0.988) in emergency admission rates. Conclusions: Changes to opening hours under existing policies are unlikely to improve patient experience of general practice substantially. Realistic short-term improvements in experience would likely have modest effects on use of emergency hospital services.

Oral vaccines have consistently underperformed in the low-income countries where they are needed most. This has formed a potent obstacle to several public health initiatives that rely on such vaccines, including the polio eradication endgame. This thesis examines the hypothesis that the composition of the intestinal microbiota – including both pathogenic and commensal microbes – contributes to the impaired performance of oral vaccines in low-income countries. Based on a systematic review and meta-analysis, we observed a decrease in the likelihood of responding to oral poliovirus vaccine (OPV) among infants with concurrent enterovirus infections or diarrhoea. We subsequently tested for multiple bacterial, viral, and eukaryotic enteropathogens and sequenced the 16S rRNA gene (to characterise the total bacterial microbiota) among infants living in a semi-urban slum in south India who had received Rotarix (an oral rotavirus vaccine) and OPV at 6 and 10 weeks of age. We did not observe significant differences in bacterial microbiota composition according to seroconversion status for either vaccine. However, the presence of bacterial pathogens was positively correlated with Rotarix response and negatively correlated with OPV response, suggesting that distinct mechanisms may impact these vaccines. The same methods were applied to samples from 6–11 month-old infants who had received OPV after a 3-day course of azithromycin or placebo. Once again, the association between bacterial microbiota composition and vaccine outcome was modest, although microbiota diversity was negatively correlated with vaccine poliovirus replication. As expected, viral pathogens were associated with a decrease in OPV immunogenicity. Moreover, recently acquired viral infections appeared to inhibit OPV response to a greater extent than persistent viruses. Together, these findings have substantiated the inhibitory effect of viral pathogens on OPV response, while implicating bacterial pathogens as a potential risk factor for OPV failure in early infancy. Risk factors for rotavirus vaccine failure remain elusive.

Introduction: High quality primary care is considered central to preventing avoidable health system waste such as unplanned short stay admissions (SSA) for minor conditions and alleviating health inequality. Recent policy changes in primary care (2004) may have impacted on provision, access and supply of GPs. This provides an ideal opportunity to study its role on keeping children healthy in the community. I aimed to quantify the impact of policy change; GP timeliness and access; and GP utilisation on potentially preventable admissions and health disparities in children. Methods: My studies included a segmented population based trends study design and a retrospective cohort design. I used national hospital and primary care administrative datasets, focusing on children aged ≤14 years between April 2000 and March 2013, in England. My primary outcome measures were SSA rates (< 2 days stay) for chronic conditions and infectious illness; my secondary outcome emergency department (ED) visits. I investigated: the impact of 1) primary care policy change in 2004; 2) patients’ reported access to their GP; and 3) primary care utilization on unplanned health service use and reducing deprivation gradients. Results: There was a significant increase in the number of children being admitted with chronic conditions for a short stay after primary care policy changes (11% rate increase in year of change), but not for infectious illness. Children were less likely to visit EDs or be admitted for a chronic condition if their GP offered better access. Better preventive care reduced children’s risk of an unplanned admission and deprivation gradients were narrowed in children who regularly consulted their GP. Conclusion: Primary care plays a significant role in limiting use of urgent and unplanned health service use, particularly for deprived children and those with chronic conditions. Investment in primary care is vital in a time of epidemiological transition in children.

HIV transmission has effectively been reduced in many countries by focusing on high transmission foci with interventions tailored to key populations groups and high risk behaviours. However people who inject drugs (PWID) still accounted for significant proportions of new HIV and Hepatitis C infections in several regions in 2014 . Among PWID, as in other populations, risk behaviours are not distributed equally. Polydrug use (using multiple drugs) in particular has increasingly been cited as a source of harm and possibly greater HIV infection risk but remains under-researched. In this thesis, I assessed the epidemiological significance of polydrug use on HIV and HCV risk behaviours among PWID, using several methods, and data sources from different settings. Multiple methodological challenges were faced, related to inherent difficulties in studying PWID, including stigmatised behaviours, comparatively small sample sizes and the absence of sampling frame for PWID. I established that polydrug use was frequent among PWID, in contrast with the common assumption that PWID mainly inject heroin, and considerable proportions of injectors across locations had recently injected or used multiple drugs (chapter 2). I generated new evidence on the biological and behavioural risks associated with injecting certain drugs and drug combinations in three stages. First, in the meta-analysis, HIV incidence rates were consistently higher among PWID injecting cocaine, amphetamine (ATS) or combinations of heroin and stimulants, compared to those not injecting or not injecting the drug(s). Wide confidence intervals and inconsistent reference categories however, limited conclusions (chapter 3). Second, I identified differences in injecting risk and HIV status between ATS and opiate injectors (chapter 5), and greater injecting and sexual risk among polydrug- than single-drug injectors in Russia and Estonia (chapter 6). Finally, using a generic polydrug typology developed in latent variable modelling, I found significant differences in HIV-related injecting and sexual risk behaviours between polydrug classes (chapter 7). The results of this work fill important research gaps by establishing that polydrug use is frequent and associated with substantial heterogeneity in HIV risk among PWID. My thesis examines the implications of these findings and proposes a polydrug use typology to help tailor drug and HIV prevention and treatment interventions to high risk PWID sub-groups to prevent the spread of new HIV and HCV infections.

Background. Quantifying and predicting the effect of control interventions against infectious diseases requires knowledge about prevalence prior to the implementation of such control measures as well as knowledge of key parameters which affect model outcomes. Here is presented the first geostatistical map estimating the prevalence of human onchocerciasis in the former Onchocerciasis Control Programme in West Africa (OCP) before the initiation of vector control operations in 1975. The map is used to quantify the burden of onchocercal skin disease prior to interventions. The results of a socioeconomic survey investigating coverage and compliance to community-directed treatment with ivermectin are then presented. Methods and Findings. The OCP epidemiological database was investigated for spatial autocorrelation. A total of 737 village surveys had parasitological examinations conducted prior to the start of control. Using these 737 pre-control data points plus environmental covariates, a Bayesian model-based geostatistical (B-MBG) approach was used to generate a continuous surface (at a pixel resolution of 25km2) of microfilarial prevalence in West Africa prior to the commencement of the OCP. The mean Pearson’s correlation between observed prevalence and model-estimates of prevalence at hold-out locations was 0.693; mean prediction error was 0.77% and mean absolute prediction error was 12.2%. Within OCP boundaries, 17.8 million people were deemed to be at risk and 7.6 million infected in 1975 which is greater than previous estimates. The mean microfilarial prevalence across all countries was 45% (range: 2–90%). Conclusions and Significance. This is the first map of continuous onchocerciasis prevalence in West Africa, developed using robust geostatistical methods. Important environmental predictors of infection prevalence were identified and used in a model that out-performs those without spatial-random effects or environmental covariates. Our results may be compared with recent epidemiological mapping efforts to find areas of persisting transmission, informing feasibility of elimination with current and novel tools.