Adverse drug reactions (ADRs) are common in children. They contribute significantly to patient morbidity, mortality and hospitalisation costs. There is limited data on the avoidability of ADRs in children and wide variation in avoidability rates has been reported. There is currently no standardised method for determining avoidability and many of the established tools are not suitable or designed for use in paediatrics. The aim of this thesis was to develop and test a new avoidability assessment tool that is suitable for use in paediatrics. The stimulus for this work was difficulty using other tools including the one developed by Hallas et al. (1990). Ideally the new tool should also be applicable and generalisable to a variety of other settings. A secondary objective was to identify potential strategies for clinical practice that might reduce the incidence of ADRs. Three key themes for avoidability have been established through a review of existing literature these are: inappropriate or suboptimal prescribing, inadequate monitoring and inadequate patient or parent education. The development of the LAAT was a multistep process which involved a multidisciplinary team (MDT). Individual and group assessments were conducted and qualitative and quantitative analyses of the assessments were carried out. The LAAT has undergone validity and reliability testing for groups and individuals. The newly validated LAAT was used to assess 249 ADR case reports from a prospective paediatric admissions study by one individual and compared to existing avoidability assessments conducted using the Hallas scale. Assessment of these ADR case reports using the LAAT found that 19.3% were either possibly or definitely avoidable. This was similar to results using the Hallas scale where 22% of the reactions were either possibly or definitely avoidable. Overall percentage exact agreement (%EA) between LAAT and the Hallas scale was 90%; when subcategorised into oncology and non-oncology cases the %EA was found to be 94.2 and 86% respectively. The kappa score between LAAT and Hallas scale assessments was 0.71 (95% CI 0.60 - 0.82) for all cases, 0.54 (95% CI 0.40 - 0.68) for the oncology cases and 0.73 (95% CI 0.58 - 0.88) for the non-oncology cases. The most common avoidability theme detected in this study was inappropriate or suboptimal prescribing. Assessing the avoidability of ADRs is a complex process which requires taking into account a number of factors. Strategies to avoid ADRs can be applied at different levels including: patient, ward, departmental institutional, professional, and national. A common theme that emerged from this work was the lack of available guidelines that could be used to assess whether ADRs were avoidable. Where guidelines were available few contained information about ADRs or their prevention. The majority of clinicians relied on their experience and tacit knowledge rather than on guidelines. Some of the ADRs categorised as either possibly or definitely avoidable may have been avoidable with improved prescribing, more frequent monitoring or improved education of patients and/or parents. Other possible prevention strategies include creating an awareness of ADRs in general and their prevention throughout a clinician’s training. Improved communication and documentation in patient records is a simple but effective method of ADR reduction. In summary, we have designed a novel avoidability assessment tool, developed by a multidisciplinary team, and have shown that the new tool is comparable to an existing avoidability tool, can be used by individuals and most importantly is suitable for use in paediatrics or other areas where clinical conditions extend beyond the expertise of individuals. The LAAT refers to guidelines and patient history rather than to abstract concepts such as ‘present-day knowledge of good medical practice’ and ‘effort exceeding the obligatory demands’ as per Hallas. Further work to identify potentially avoidable ADRs and strategies to prevent them is needed.

In Malawi there has been significant progress in reducing post-neonatal and under-5 deaths over the past decade but very little progress in reducing neonatal deaths. The major causes of neonatal deaths in Malawi are prematurity, infections and birth asphyxia. Neonatal sepsis has been shown to have long term complications ranging from motor deficits to cognitive impairment, epilepsy and behavioural disorders in preterm very low birth weight infants in the developed setting. Contrary to the epidemiology in the developed setting where neonatal sepsis is predominantly seen in preterm low birth weight infants, in the developing setting neonatal sepsis is also common among term babies. However, very little is known on the long term outcome of neonatal sepsis in the resource restrained setting. In this thesis the aetiology and outcome of neonatal sepsis and meningitis is investigated. METHODOLOGY This was study had 2 components; a cross sectional arm and a prospective cohort arm. The cross sectional study was looking at the aetiology, resistance pattern and in hospital outcome of severe neonatal infection cases presenting at QECH in Blantyre. The prospective cohort arm involved participants who were recruited in the cross sectional arm at QECH and were residing within Blantyre urban and infants that never had an episode of severe neonatal infection were recruited from Zingwangwa health Centre. The infants from Zingwangwa acted as controls. The participants in the prospective cohort arm were followed up to the age of 1 year where neurodevelopmental outcomes were assessed using the Bayley’s assessment tool. These participants also had detailed neurologic examination during the follow up visits at 6 and 12 months of age. A comparison between the cases and controls was made to ascertain the impact of neonatal infection outcome. RESULTS During the study a total of 412 cases were enrolled in the cross sectional arm. 75% of the cases had late onset disease. GBS was the commonest organism grown in blood culture 17/42(40%) and CSF culture 16/33(48%). 44% had abnormal serum sodium levels on admission and hypernatraemia was independently associated with an increased risk of dying in hospital (8.34[95% CI 1.95-35.7]). 51% of the gram negative organisms were multidrug resistant. In the long term outcome neonatal sepsis without overt meningitis was associated with an up to 6.6 –fold {95% CI (2.38-18.4) increased risk of developmental delay at 1 year of age. Meningitis was associated with a 17-fold {95% CI 4.89- 61.7} increased risk of developmental delay at 1 year of age. Positive blood or CSF culture and being HIV exposed were independent predictors of delay at I year of age. CONCLUSION GBS is a significant cause of neonatal infections in Malawi. The magnitude of developmental delay observed in infants who had neonatal sepsis without meningitis is worrying up to 35% of these infants were delayed. It is therefore important to employ measures that can prevent neonatal infections. Follow up is recommended in infants who had an episode of severe neonatal infection.

Shaken Baby Syndrome is a form of child abuse estimated to occur to one in 20,000 babies and presumed to occur when a carer cannot cope with a constantly crying child and so gives it a sharp shake. This causes the brain to move within the skull, stretching and possibly tearing the veins that bridge the fluid filled gap. To better understand this condition, experiments were performed to measure the mechanical properties of bridging veins followed by detailed mathematical modelling of the motion of a baby’s brain in response to shaking. Few finite element models of shaken baby syndrome exist, and those either ignore the fluid surrounding the brain or model it as a soft solid. The importance of modelling the fluid properly is demonstrated, and the reliability of MSC.Dytran’s fluid-solid interaction modelling is confirmed. The first three-dimensional finite element model of shaken baby syndrome which accurately includes the cerebrospinal fluid is created and used to estimate tolerance criteria for causing subdural haematoma by shaking. This research concludes that shaking of a baby could produce bridging vein strain close to the tolerance for failure and hence should be considered a possible cause of subdural haematoma.

Arthrogryposis, Renal dysfunction, and Cholestasis (ARC) syndrome is a multisystem disorder associated with abnormal localisation of some polarised membrane transporter proteins. Distinct apical and basolateral poles are essential for epithelial function and organ development but the molecular pathways determining the biogenesis of polarised membranes are not fully characterised. Mutations in VPS33B, a Sec1-Munc18 protein, account for 75% of ARC patients. Reduced expression of VPS33B at both the RNA and protein level was demonstrated in all ARC syndrome patients, even if mutations were not identified in VPS33B. A novel protein POLARIN (PLRN) was identified that interacts with VPS33B, and is crucial for VPS33B function. Pathogenic mutations in PLRN occur in ARC patients without VPS33B mutations. Decreased Polarin and Vps33b expression in mouse renal collecting duct cells led to abnormal localisation of specific apical membrane proteins and to disordered apical junction complex formation. In an in vivo model, knockdown of polarin in zebrafish resulted in defects in biliary tract development. These findings establish that a VPS33B-POLARINRab11a intracellular trafficking pathway is functionally distinct from another VPS33-related pathway (VPS33A/VPS16) and is required for (a) normal epithelial polarisation and apical junction complex formation, and (b) normal liver and kidney development and function.

The work presented in the present thesis investigated the neural, behavioural and genetic markers that may be associated with the manifestation of behavioural problems during the early years of life. Across four different empirical studies, and by incorporating, behavioural, neurophysiological and genetic investigations, it was demonstrated that: (1) there are neurophysiological signatures that may be associated with the manifestation of behavioural problems early in life; (2) common genetic variations that determine serotonin variability are strongly associated with affectivity-related patterns of frontal brain activation; and that (3) normal genetic variations that modulate serotonin availability and neuroplasticity are each associated with affectivity-related patterns of visual scanning behaviours in response to faces and aversive scenes. Taken together, the results illustrate the existence of robust neural, genetic and behavioural markers that may be associated with the manifestation of behavioural problems in early childhood and prompt further investigation of the area by generating novel hypotheses. Together, the empirical findings of the thesis provide a first stage contribution to the complex mechanisms that may yield risk and resilience for behavioural problems during the early years of life by generating a more comprehensive insight on the field of affectivity.

INTRODUCTION Brain tumours are the most common solid tumour in childhood, while Magnetic Resonance Spectroscopy (MRS) studies have been performed on brain tumours previously, the majority have been performed on the adult patient population and at long echo times. The work presented in this thesis outlines work performed in the usage of short echo time MRS in the characterisation of childhood brain tumours. METHODS Short echo time MRS was performed on children with brain tumours at the time of diagnostic imaging. In addition, follow up data was accrued for some patients. Resulting spectra were assessed for characteristics either of diagnosis, prognosis or treatment response. RESULTS Spectra collected were used in assessing characteristics of an array of childhood brain tumours. Initially the technique was tested on a well understood dataset of cerebellar tumours, and was later expanded to provide diagnostic aids for both brain stem tumours and pineal region tumours. A group of pilocytic astrocytomas were assessed for differences by location within the brain, for prognosis and for response to treatment. CONCLUSIONS The additional information given by short echo time MRS was useful in the characterisation of childhood brain tumours

Adolescents are especially vulnerable to mental health difficulties (Collip et al, 2008; Bhardwa, 2013; Costello et al, 2003). The Child and Adolescent Mental Health Service (CAMHS) Review (DCSF & DoH, 2008) highlighted barriers to support including capacity difficulties, stigma, young people’s lacking knowledge and a lack of tailored services. Research, demonstrates weaknesses within young people’s knowledge and beliefs regarding mental health and mental health support and historically accumulated stigmatising attitudes. A paucity of research also exists on young people’s wishes for support (Dogra, 2005; Worall-Davies & Marino-Francis, 2008). The aim of this study is to use a mental health literacy framework (Jorm, 2000) to explore the knowledge, beliefs, attitudes and wishes of a community sample of young people regarding mental health and mental health support. This is to inform prevention and intervention within Hightown Local Authority and Highfields school(1). Adopting critical realism, focus groups and thematic analysis (Guest et al, 2012; Boyatzis, 1998) were used. Main findings illustrate the young people’s broad knowledge-base, largely appropriate beliefs and attitudes. Opportunities were revealed for educationally-based, preventative support which enhances detail and security of knowledge, and addresses attitudes in gender sensitive ways. At the intervention-level, preferences were indicated for tiered support, with professional support delivered externally to school. In terms of delivery of support, feeling well-supported was important, along with confidentiality, accessibility, empathy and tailored support to their needs. In carrying forward these findings, the role of the EP, alongside the school and other stakeholders, is exemplified as integral.

Background: Caring for a child with a chronic health condition is often associated with high levels of anxiety, depression and stress (Burton et al., 2003). Demanding caring duties, a lack of time and social isolation create barriers for carers accessing mental health support. The internet may be useful way for carers to access psychological supports but to date no research has explored this. Aims: To identify common difficulties which carers face and adapt a pre-existing computerised CBT (cCBT) based course to reduce carer stress. To determine the feasibility of providing mental health support to carers through an online course and assess its potential effectiveness and accessibility. Methods: Focus groups and questionnaires completed by hospice staff provided information regarding common stressors encountered by carers. Thematic analysis (Braun and Clarke, 2006) was used to identify topics which were then used to adapt a pre-existing cCBT course which consisted of 6 primary modules and additional online materials/modules plus the offer of weekly email support for 6 weeks from a support worker to encourage use. A single arm feasibility trial of the tailored cCBT was conducted using carers of children with chronic conditions recruited through advertisements and local hospice staff. Participants were provided with access to the cCBT course for 6 weeks and received weekly personalised support emails. Self-report measures of anxiety, depression and daily functioning were measures at baseline and at post-treatment. Questionnaires administered online explored participant satisfaction and course use which were then analysed. Results: 19 hospice workers were recruited to complete questionnaires regarding common difficulties faced by carers. Several themes emerged from thematic analysis of the data including: physical and interpersonal difficulties, external pressures, time constraints, limited support, responding to the child’s illness and difficulties accessing appropriate care for their child. 29 carers were recruited to trial the online course, with 55.6% recruited through social media. 12 (30.8%) carers logged onto the course and 6 (15.4%) completed follow-up measures at 8 weeks post intervention. Of the participants that started the course only 2 (18.2%) completed all 6 course modules. Exploratory analyses showed no statistically significant differences in depression, anxiety or functional impairment. However, results indicated a general reduction in clinically significant anxiety symptoms and functional impairment. Satisfaction questionnaires completed by carers post-intervention indicated a high level of satisfaction with the course. Qualitative results showed that participants valued the email support they received and the flexibility that an online approach provided. Some carers still struggled to find time to complete the course and suggestions were made to amend the course to enable it to be accessed offline and on portable devices. Conclusion: Computerised courses may be an acceptable way of offering mental health support to some carers. Difficulties in retention and recruiting suggest that further adaptations to the course are needed to improve engagement (i.e. delivering the course via apps, or moving to stand alone course topics rather than a recommended course). Alternative methods of delivering psychological support should be explored to avoid excluding those who do not have internet access.

Two-thirds of UK mothers begin breastfeeding, but many soon stop. Although breastfeeding benefits health, infant feeding is influenced by social and attitudinal factors. Study one prospectively investigated the attitudes and experiences of 514 women. Past experience predicted which multiparae would stop by six weeks. Manual social class and considering bottle feeding did so for primiparae. Perceived insufficient milk was the commonest reason for stopping. Study two, a randomised trial of support from breastfeeding counsellors, recruited 720 women. At four months, 46.1% (143/310) intervention and 42.3% (131/310) control women breastfed (Chi\(^2\)=0.942, P=0.33); 73.9% (229/310) vs 79.4% (246/310) gave bottle feeds (Chi\(^2\)=2.60, P=0.11). Survival analysis confirmed that differences between intervention and control women's partial and full breastfeeding duration were not significant (P=0.45 and 0.15 respectively.) Significantly fewer intervention women felt they had insufficient milk. Qualitative analysis of women’s comments revealed they wanted better information, practical help with positioning, effective advice, encouragement and their feelings acknowledged. Women valued counselling, but their feeding behaviour changed little, which may reflect the strength of social influences and that not all mothers contacted the counsellors postnatally. Practical support in the early postnatal period is important. Counselling may increase women's confidence in breastfeeding and producing enough milk.

Background: This thesis sits within the arena of weight status during and after childhood acute lymphoblastic leukaemia (ALL), with a particular focus on the prevalence of unhealthy weight status amongst (ALL), Saudi and UK populations. Each chapter in the thesis explores different aspects of unhealthy weight status in ALL which had been highlighted as gaps in the literature at a conference in Puebla, Mexico, at the end of 2006. A summary of each study is given below. Study 1: Background: This study estimated prevalence of unhealthy weight status and metabolic syndrome (MS) amongst Saudi survivors of standard risk ALL. Methods: We recruited 56 survivors, mean age 13.4 years (SD 4.1), a mean of 9.1 years (SD 4.1) post-diagnosis. The BMI for age was used to define weight status relative to national (Saudi) and international (Cole et al., International Obesity Task Force (IOTF), World Health Organisation (WHO), and Centre for Disease Control and Prevention (CDC)) reference data. We measured body composition by dual energy X-ray absorptiometry (DXA), waist circumference, blood pressure, lipid profile (HDL-C, Triglycerides), fasting glucose and insulin. Results: According to international definitions based on BMI for age, around half of the sample had unhealthy weight status. All of the approaches based on BMI for age underestimated over-fatness, present in 27/51 (53%) of the sample according to DXA. Prevalence of MS was 7.1% (3/42 of those over 9-years old) and 5.4% (3/56) by applying the International Diabetes Federation (IDF) definition and National Cholesterol Education Program Third Adult Treatment panel Guidelines (NCEP III), respectively. However, MS by the NCEP III definition was present in 19% of the overweight and obese survivors and 7.1% of the sample had at least two of the components of MS. Conclusions: Unhealthy body weight and over-fatness may be common amongst adolescent Saudi survivors of standard risk ALL, though overweight and obesity may be no more common than in the general Saudi adolescent population. Defining weight status using BMI underestimates over-fatness in this population, as in other populations. Study 2: Background: Underweight, overweight, and obesity at diagnosis may all worsen prognosis in childhood ALL, but no studies have estimated prevalence of unhealthy weight status at diagnosis in large representative samples using contemporary definitions of weight status based on BMI for age. Methods: Retrospective study which aimed to estimate prevalence of underweight, overweight, and obesity at diagnosis for patients with childhood ALL on three successive UK treatment trials: UKALL X (1985-1990, n 1033), UKALL XI (1990- 1997, n 2031), UKALL 97/97-99 (1997-2002, n 898) .The BMI for age was used to define weight status with both UK 1990 BMI for age reference data and the IOTF definitions. Results: Prevalence of underweight was 6% in the most recent trial for which data were available. Prevalence of overweight and obesity was 35% in the most recent trial when expressed using IOTF definitions; 41% when expressed relative to UK 1990 reference data. Conclusions: Even with highly conservative estimates >40% of all UK patients with ALL were underweight, overweight, or obese at diagnosis in the most recent trial for which UK data are available (UKALL 97/99, 1997-2002). Study 3: Background: This study tested the hypothesis that overweight/obesity at diagnosis of childhood ALL was related to risk of relapse. Methods and results: In a national cohort of 1033 patients from the UK there was no evidence that weight status at diagnosis was related significantly to risk of relapse: log ranks test (p value= 0.90) with overweight and obesity as the exposure (n 917); individual (p value= 0.42) and stepwise (p value= 0.96) proportional hazards models, with BMI z score as the exposure (n 1033). Conclusion: The study does not support the hypothesis that overweight/obesity at diagnosis impairs prognosis in childhood ALL in the UK. Study 4: Background: In the sample of Saudi patients recruited to study 1 we compared DXA whole body and lumbar spine bone mineral density (BMD) using manufacturers software with a body size correction which derived bone mineral content (BMC) for bone area and Apparent bone mineral density of lumbar spine (BMADLS). Methods and results: The survivors of ALL were from Saudi Arabia (n 51, mean age 13.5 years). With no corrections, 29 patients (57%) had lumbar spine BMD z score < -1.0 and 21 (41%) had whole body BMD z score < -2. After correction, by using BMC for bone area method only 6 (12%) had lumbar spine BMC z score <-1.0 and 4 (8%) had whole body BMC z score <-2. By using BMADLS method, 18 (35%) had BMC <-1.0 and 6 (11%) had BMC Z score <-2. Conclusions: Correction for body size seems essential to accurate interpretation of DXA bone health data in adolescent survivors of ALL. The three correction methods provided different conclusions, but bone health remains a concern after treatment for ALL.