Bibliography: leaves 129-163. / There is little information on the aetiology and outcome of HIV-associated pneumonia in African children and no comprehensive data from South Africa. Studies of HIV-infected adult in Africa reported that the spectrum of pulmonary disease differs from that of developed countries with tuberculosis and pyogenic pneumonia predominating and Pneumocystis carinii pneumonia (PCP) occurring uncommonly. Knowledge of the aetiology and outcome of pneumonia is important for the development of paediatric management guidelines and of policies for allocation of resources especially in South Africa, where the HIV pandemic has resulted in increasing numbers of HIV-positive children requiring admission to hospital or intensive care units for pneumonia. Furthermore in countries with limited resources, development of cost effective diagnostic procedures to investigate the aetiology of pneumonia is necessary.

1. Introduction: Obtaining Informed consent is an ethical prerequisite for enrollment in clinical research. There is a perception that Informed consent documents used in biomedical research are lengthy, overly complex and above the reading capability of typical research participants. In South Africa, ethical committees regulating research on human participants (HRECs) are mandated by the Department of Health’s National Health Research Ethics Council’s (NHREC) guidelines to ensure that researchers have made special considerations for vulnerable groups when conducting research. This includes considerations made for populations with low literacy. For example, the Standard Operating Procedure (SOP) of the University of Cape Town’s Human Research Ethics Committee (UCTHREC), requires that the language used in Informed consent documents should be directed at a reading level of grade 6 to 8 and that common, everyday words should be used rather than complex language syntax. The HREC expects researchers to translate the approved English version documents into local languages such as isiXhosa and Afrikaans. Since ethics committee focus approval on the English language consent documents and only acknowledge translated versions, a potential gap in this process is whether the translated versions meet the same required readability levels. This study aims to investigate whether translated versions of English language informed consent documents used at a single busy clinical research site are readable and meet the readability levels specified by UCTHREC. 2. Methodology: A quantitative descriptive statistical design was used to explore readability levels of informed consent documents used at a single clinical research facility based in a semi-rural community. Informed consent documents approved by UCTHREC over the past thirteen years (2004 to 2017) that met the inclusion criteria were analysed for readability. The LIX readability test tool was used to calculate readability scores and the levels of reading difficulty. These scores were then matched to a grade level conversion chart to determine the equivalent number of education years required to be able to easily understand the information. Readability levels were determined for isiXhosa and Afrikaans translations of the documents and compared to the levels of the English document. 3. Results: The results indicate that informed consent documents used at this single clinical research facility, independent of language type, are difficult to read. A total of 259 sub-sections of informed consent documents from 10 different studies were analysed. The analysis showed that informed consent documents were classified as “very difficult to read” according to the LIX readability tool in a large proportion of English, isiXhosa and Afrikaans languages: 41 (16%), 255 (98%), and 85 (33%) of informed consent sections respectively. Of all the subsections of English, isiXhosa and Afrikaans documents respectively, 98 (38%), 0 (0%) and 126 (49%) were classified as “difficult to read”, while 79 (31%), 3 (1%) and 38 (15%) were found to have an “average” readability level. Twenty eight (11%), 1 (0%) and 10 (4%) were found to be “easy to read” and 13 (5%), 0 (0%) and 0 (0%) had a “very easy” readability level. The mean LIX readability scores across English, isiXhosa, and Afrikaans languages were respectively 42.27 (95% CI 41.20 – 43.34) corresponding to a readability level of “average”, 74.64 (95% CI 73.79-75.49), corresponding to “very difficult to read” and 46.73 (95% CI 45.66-47.8) “difficult to read”. These findings suggest a high level of difficulty in reading of the text in the Informed consent documents. 4. Conclusion: Translations of Informed consent documents used at a single busy clinical research site are difficult to read and are written at high school to tertiary reading level. These reading levels are above the recommended level prescribed by the site’s research ethics committee (UCTHREC). Local ethics committees should employ more stringent guidelines and checks to ensure readability of translated informed consent documents. Researchers and Sponsors should include readability outcomes in the design and with submissions of new protocols.

Includes bibliographical references / In this study we sought to more comprehensively analyse antigen-specific CD4 T cell responses induced by vaccination and to examine the effects of latent M. tb infection on these responses. We had two broad objectives: Firstly, to determine the effects of latent M. tb infection on epitope recognition by mycobacteria-specific CD4 T cells and to design HLA class II tetramers for detection of these cells. Secondly, to characterise antigen-specific CD4 T cells following vaccination with the novel vaccine candidate, H1:IC31, by measuring transcriptomic, phenotypic and functional attributes, and to determine the effects of latent infection on these responses. Firstly, we found that acquisition of M.tb infection did not alter the breadth and/or pattern of Ag85A/ B CD4 T cell epitopes recognised. We determined the HLA allele restriction of identified epitopes, and designed HLA class II tetramers for detection of Ag85-specific CD4 T cells. These results suggest that latent infection does not alter CD4 T cell epitope breadth within Ag85A/ B elicited by BCG vaccination and/or exposure to environmental mycobacteria. The second finding of this work is that underlying infection drives a more effector-like H1-specific CD4 T response after vaccination. Following vaccination M. tb-infected adolescents had higher frequencies of H1-specific CD4 T cells compared with uninfected adolescents. Additionally, H1-specific CD4 T cells from infected adolescents predominantly displayed a CCR7 - CD45RA - effector memory phenotype, had higher proportions of IFN-γ + TNF-α + IL-2 + cells, and expressed higher levels mRNA transcripts encoding effector molecules such as granzyme K and perforin, compared with uninfected adolescents. By contrast, H1-specific CD4 T cells in uninfected adolescents displayed a less differentiated memory phenotype, and had increased expression of central memory genes, compared to cells from infected adolescents. Thirdly, we found that Ag85B and ESAT-6-specific CD4 T cells exhibited markedly distinct transcriptomic profiles, memory phenotypes and cytokine expression patterns in M.tb infected adolescents. The data suggested that ESAT-6-specific cells preferentially drove the effector-like H1-specific response in M.tb infected adolescents. We conclude that while underlying M.tb infection does not affect the epitopes recognized by mycobacteria-specific CD4 T cells, but may promote and maintain effector memory antigen-specific CD4 T cells endowed with immediate effector function and tissue homing.

This work is concerned with the application of scintigraphy in studies of foregut motility in paediatric patients; notably oesophageal transit during swallowing, gastro oesophageal reflux (GOR) and stomach emptying as measured by the gastric contents 30 and 120 minutes after deglutition of liquid. Relevant anatomy, physiology, pathophysiology and non-radioisotopic methods of gastro oesophageal examination are outlined. The extremely important question of dose deposited by ionising radiation in paediatric patients is also considered. Because currently available estimations of absorbed dose after swallowing non- absorbable radionuclides are unsatisfactory and often contradictory, a new model was derived. It is based on more physiological principles than previous models, with a continuously varying exponential passage between four compartments in the gastrointestinal tract. From a review of previous work and our experience, a standard method of examination was developed. There is emphasis on normal physiological manoeuvres. The broad extent of useful information available in the study stems from the positions in which the patient is investigated, the quantity of labelled fluid swallowed, its physiological nature and observations of oesophageal transit, GOR, gastric emptying and pulmonary aspiration after two hours. The wide range of paediatric problems amenable to investigation by this means is described, including those of a life-threatening nature. Also, the importance of using this examination to determine the likelihood of successful surgical intervention in the gastro oesophageal region is explained. By using the observations made during swallows in neonates, the closure of the lower oesophageal sphincter (LOS) was observed during a short series of rapid deglutitions. This extends previous knowledge of the function of the neonatal LOS. These findings were very clearly demonstrated by the condensed image technique. A comparison between the abilities of radiology and scintigraphy to detect GOR in paediatric patients, with and without oesophagitis, gives unequivocal results in favour of scintigraphy. Two independent clear correlations between oesophageal transit time and the severity of GOR and age are demonstrated by variance and correlation analysis. The mean duration of GOR, maximum height of a GOR, gastric contents 30 and 120 min after deglutition, the effect of the nature of liquid swallowed and age are also examined and their relations are described. The application of the method to assess drug response is illustrated by the paediatric use of cisapride, an experimental drug which increases acetylcholinesterase release. The deductions from this work include values of the radiation dose deposited in paediatric patients of various ages, after swallowing non- absorbed Tc-99m compounds. Other contributions are a scintigraphic means to allow recognition of the importance of oesophageal dysmotility in paediatrics, especially in the absence of GOR and its relevance to the efficacy of surgery in cases of apparent GOR. Also, this gastro oesophageal scintigraphic study provided information on the paediatric use of cisapride, the relevance of GOR to repeated respiratory problems and the recognition of a new paediatric syndrome analogous to the diffuse oesophageal spasm syndrome of adulthood. It has been demonstrated that a routine, uncomplicated, paediatric scintigraphic examination, useful for investigating a wide range of problems in both a first and third world setting is possible for a nuclear medicine practitioner with normally available radiopharmaceuticals, equipment and radiographical expertise.

Background Acute kidney injury (AKI) is a frequent complication of children admitted to the paediatric intensive care unit and is associated with significant short term and long -term consequences. The causes vary from conditions that decrease intravascular volume to the use of nephrotoxic medications. One key management modality of AKI is the use of diuretics to reduce fluid overload which is an important indication for initiation of renal replacement therapy (RRT) - dialysis and also an important indicator of morbidity and mortality. Aminophylline, a drug that is well known for its use in the treatment of bronchial asthma is also purported to have diuretic effects on the kidneys. A paucity of data to support the use of aminophylline as diuretic and especially in children has occurred until a recent renewal of interest in this drug as a diuretic. In the Paediatric intensive care unit (PICU) of the Red Cross War Memorial Children's Hospital (RCWMCH), the use of aminophylline to augment urine output started in the early 2000s.This retrospective case-cross over study assesses the effect aminophylline in critically ill children with AKI; specifically, the effect on urine output, improvements in serum creatinine levels or limitations of renal replacement therapy as well as any age-related differences in aminophylline effects. Methods Children admitted to the PICU of RCWMCH with AKI (from 2012 to June 2018) were identified through a search of the PICU database, folders of cases were obtained, and a folder review carried out. Cases that fulfilled the inclusion criteria (that is children admitted to PICU with AKI who received aminophylline) were subsequently recruited. Data captured and analyzed included demographics, underlying disease conditions, medications, vital signs, urine output, renal function, arterial blood gases, RRT and outcomes of AKI. Results Thirty-five children were analyzed. Urine output increased from a median of 0.4mls/kg/hr [IQR: 0.1, 1.1] at six hours prior to aminophylline administration to 0.6mls/kg/hr[IQR: 0.2, 1.9] at six hours post aminophylline therap,1.0mls/kg/hr(IQR:0.2, 2.7) after twelve hours and 1.6 mls/kg/hr(IQR:0.2, 4.2) after twenty-four hours and this was statistically significant based on the Friedman's analysis of variance test (p=0.001).The median change in postaminophylline urine output after six hours was 0.05mls/kg/hr (IQR:0.0, 0.6) and this change did not vary significantly across the age groups (p=0.530). There was no significant change in serum creatinine levels six hours pre and post- aminophylline administration [109(77,227), 125.5(82,200) micromole/l] respectively, p=0.135. Sixteen out of the thirty-five children (45.7%) required renal replacement therapy.

Introduction: In order to better understand the epidemiology and burden of human coronaviruses - NL63, HKU1, OC43 and 229E in South Africa, their role in the aetiology of childhood pneumonia needs to be described. Methods: We used data collected between September 2012 – September 2013 from children aged < 13 years with lower respiratory illness at Red Cross War Memorial Children's Hospital. Respiratory samples including a nasopharyngeal swab (NP) and induced sputum (IS) were taken and tested for the four strains of coronaviruses using FTD33 multiplex realtime PCR. Results: A total of 460 respiratory samples were analysed. Of these, 258 (56.0%) were male and 19 (4.1%) HIV infected. The median age of the children was 8 (IQR 4-18) months. Nasopharyngeal (NP) samples were obtained from 460 children while induced sputum (IS) was not available for six children due to sample loss prior to analysis, leaving 454 available for analysis. A total of 42 (9.1%, 95% CI 6.7- 12.1%) participants tested positive for HCoV in at least one of the two specimens. PCR was able to detect a total of 35 (7.7%) cases from the 454 tested IS specimens compared to 23 (5.0%) detected out of 460 NP samples. The commonest detected HCoVs were coronavirus OC43 with 20 (4.3%) detected from either specimen followed by coronavirus NL63 or coronavirus HKU detected in 14 (3.0%) and 10 (2.2%) of positive test samples, respectively. The least common virus detected HCoV was coronavirus 229E detected in both positive test samples of one participant. Overall HCoVs were detected in 23 (8.9%) of boys compared to 19 (9.1%) of the girls who returned a positive test; p=0.856. The overall age distribution of children with PCR detected HCoVs was similar to that of children with a negative result with median age of 10 (IQR 5- 16) months and median of 8 (IQR 4- 19) months, respectively; p=0.535. Prevalence of HCoV was 11/192 (5.7%), 23/153 (15.0%) and 8/115 (7.0%) in children <6 months old, 6-18 months and over 18 months respectively; p=0.008. Conclusion: Children aged 6 to 18 months had double the risk of other age groups.

Introduction: There are limited data on the chest X-ray (CXR) abnormalities in human immunodeficiency virus (HIV)-infected children in low- and middle-income countries (LMIC's). Aim: To investigate the evolution of CXR abnormalities in HIV-infected children in LMIC's, to correlate this with the severity of HIV-disease, and to assess the impact of anti-retroviral therapy (ART). Method: A prospective longitudinal study evaluating clinical, immunological and radiographic parameters at regular intervals over a minimum of 24 months. CXR abnormalities were stratified by severity and deemed persistent if present for 6 consecutive months or longer. Univariate and multiple logistic regression analyses assessed associations between radiological and clinical/immunological parameters at enrolment. An ordinal multiple logistic regression model assessed the association of enrolment and time-dependent variables with CXR findings over time.

Background The human immunodeficiency virus (HIV) is a leading cause of death among women during pregnancy and the postpartum period, especially in areas of high prevalence. In 2018 there were approximately 1.3 million pregnant women living with HIV globally.Infants born to women living with HIV are at increased risk of contracting HIV as the virus can be transmitted to the foetus/ infant during pregnancy, labour, delivery and breastfeeding, posing a serious risk to their survival and well-being. Viral load (VL) testing of pregnant women living with HIV could contribute to improved care, thereby reducing the risk of vertical transmission of HIV from the mother to her infant. Aim The objective of this study was to describe HIV VL testing amongst pregnant women living with HIV at entry into the prevention-of-mother-to-child-transmission (PMTCT) services at selected health facilities in Mutare district of Manicaland Province, Zimbabwe from January to December 2018. Methods This descriptive cross-sectional mixed methods study evaluated the uptake of HIV VL testing amongst pregnant women living with HIV at entry into the prevention-of-mother-to-child transmission (PMTCT) services at 15 health facilities and explored factors that influence the provision of HIV VL testing services. Results Among 383 pregnant women living with HIV enrolled in antenatal care (ANC) and known to be on antiretroviral therapy (ART), only 121 (32%) had a VL sample collected and 106 (88%) received their results. Among these 106 women, 93 (88%) had a VL< 1, 000 copies/mL and 77 (73%) had a VL< 50 copies/mL. The overall median duration from ANC booking to VL sample collection was 87 (IQR, 7-215) days. The duration was significantly longer among pregnant women newly started on ART [207 (IQR, 99-299) days] compared to those already on ART [50 (IQR, 0-162) days], p< 0.001. The median time interval for the return of VL results from date of sample collection was 14 (IQR, 7-30) days. There was no significant difference when this variable was stratified by time of ART initiation. Viral load samples were significantly less likely to be collected at local authority facilities compared to government facilities [aOR=0.28; 95% CI: 0.16- 0.48]. Barriers for VL testing identified by health care providers included staff shortages, nonavailability of consumables and laboratory forms and weaknesses in sample transportation. Additionally, the turnaround time (TAT) was long as VL testing was centralised at the provincial hospital, and results feedback was not done electronically. High levels of knowledge among health care providers (75%) did not translate into high HIV VL testing coverage amongst pregnant women living with HIV. Conclusions and recommendations The low rate of HIV VL testing among pregnant women living with HIV in Mutare district is a cause of concern and needs to be addressed urgently in the interest of contributing to the eliminating mother to child transmission of HIV. The Ministry of Health should consider disseminating ARV and PMTCT guidelines and other policy documents using electronic platforms as these are more accessible and result in quicker dissemination, which may translate into faster implementation of new policies and policy updates. There is need to conduct regular mentorship and supervision processes and establish quality improvement initiatives for PMTCT services. Interventions like alert systems should be implemented for ease of identifying women who require HIV VL testing. Point of care technology and mHealth could reduce VL result turnaround time. All this should be aimed at ensuring that policies and guidelines are implemented, and targets are reached within agreed timeframes, to ensure that positive outcomes can be experienced by all pregnant women living with HIV.

Introduction/Purpose: Lennox-Gastaut Syndrome (LGS) is one of the most common refractory epilepsies of childhood with significant morbidity and mortality. However there is paucity of data of this syndrome in resource limited settings (RLS). We sought to delineate the phenomenology, diagnosis, aetiologies, management and outcomes of children referred with LGS in the Western Cape Province of South Africa. To further identify early clinical markers differentiating LGS from other types of epilepsy. Methods. This retrospective observational cohort study included all children between 1 to < 18 years of age in the neurology database with a referral label of LGS between 2000-2018. The group were critiqued for those who met the diagnostic criteria of LGS. Then were categorized into those with confirmed LGS and remainder were not- LGS. Data of the social demographics, age of seizure onset, etiology, preceding epileptic spasms, and semiology of epilepsy types, management interventions were reviewed to identify key diagnostic indicators to permit early and targeted interventions for children with this epilepsy syndrome. Results; Of 2551 children managed with epilepsy, 110 were suspected at presentation to have LGS of these 66 records were available for assessment. The median (IQR) age in months at presentation was 35(16.0-54.5) with a slight male (37/66) predominance. 43(65%) met the criteria LGS and 23(35%) were not-LGS. 34(52%) had no identifiable cause for their epilepsy whilst a structural and metabolic cause were identified in 25(38%) and 3(5%) respectively. Moderate or severe cognitive impairment was associated with LGS (OR 2.59, p = 0.02 and OR =3.15, p = 0.01) and so were tonic seizures (OR=4.03, p=0.04). The most common diagnoses in the not-LGS group were other types of DEE not meeting the LGS criteria (15%) and uncategorised epileptic syndromes 11%. Conclusion; Over third of the children in this cohort were erroneously referred with LGS early in their course. This has implications for their management and prognostic counselling. Identification of indicators such as tonic seizures and moderate or severe intellectual/cognitive impairment are useful early markers which support a diagnosis of LGS and could be viable for use in our setting.

Neurodevelopmental and behavioural tools, used to assess children's cognitive and psychological development, play important supportive roles in clinical decision-making processes. It is therefore important that inferences made based on the information generated by these tools are valid and reliable. Because validity and reliability are inextricably tied to the context in which the tool is administered, it is critical that tools, including established “gold standard” tools, undergo ongoing psychometric evaluation. This is particularly important when tools are used in contexts different to that in which the original tool was developed. The use of poorly performing measures in clinical settings may result in a higher risk of misinterpretation of results or misdiagnosis, carrying serious implications for provision of interventions. This dissertation explored the psychometric properties of two frequently used behavioural tools in a South African context. The first aim of the dissertation was to investigate the use, cultural appropriateness, and psychometric soundness of Achenbach System of Empirically Based Assessment (ASEBA) forms in sub-Saharan Africa. The ASEBA forms are used worldwide to screen children and adolescents for behavioural and emotional problems. To achieve this aim, I conducted a systematic review of the psychometric properties of the ASEBA forms in sub-Saharan Africa. The second aim was to evaluate the structural and construct validity of a commonly used measure of AttentionDeficit/Hyperactivity Disorder (ADHD), the Swanson, Nolan, and Pelham ADHD Rating Scale (SNAP-IV), in a sample of South African children with neurodevelopmental disorders (NDDs), including Autism Spectrum Disorder and Intellectual Disability. The systematic review identified 58 studies with sub-Saharan African participants that reported measurement properties of the ASEBA forms. Most studies came from Southern (n = 29, 50%) or East African (n = 25, 43%) countries. Forty-nine studies (84%) used translated versions of the tool, but details regarding the translation process, if available, were often sparse. Most studies (n = 47, 81%) only reported internal consistency (using coefficient alpha) for one or more subscales. The methodological quality of the psychometric evaluations varied considerably across all measurement properties, except for internal consistency. There is limited good quality psychometric evidence available for the ASEBA forms in sub-Saharan Africa. Recommendations include implementing a standardised procedure for conducting and reporting translation processes and conducting more comprehensive psychometric evaluations of the translated versions of the tools. Parents of 109 children with one or more diagnosed NDDs completed the SNAP-IV. A subset of parents (n = 79) also completed the ASEBA Child Behaviour Checklist (CBCL/6-18). We conducted a confirmatory factor analysis to inspect the two-factor structure of the SNAP-IV (Inattention + Hyperactivity-Impulsivity). We also calculated ordinal coefficient alpha (α) to estimate internal consistency. Finally, we correlated scores of SNAP-IV and CBCL/6-18 subscales to estimate concurrent, convergent (ADHD + externalizing behaviour), and discriminant (ADHD + internalizing behaviour) validity respectively. The two-factor model performed acceptably (χ2 (134) = 249.82, p < 0.001, TLI = 0.868, RMSEA = 0.089, p < 0.001). The model fit improved after removing three problematic items, two of which were dependent on the child's level of spoken language (χ2 (89) = 135.17, p < 0.01, TLI = 0.942, RMSEA = 0.069, p = 0.096). The revised SNAP-IV subscales had acceptable internal consistencies (α = 0.85-0.86). Correlation coefficients between the SNAP-IV and ADHD-related CBCL/6-18 subscales were significant (r = 0.53- 0.62, p < 0.001). Correlations between ADHD and externalizing behaviours (r = 0.45, p < 0.001) and internalizing behaviours (r = 0.38, p < 0.001) respectively were not significantly different (z = 0.97, p = 0.165). The findings tentatively support the use of the SNAP-IV in this group of children. However, there are limitations to its performance in this population likely related to the presence of NDDs. Taken together, the findings of these two studies highlight the need for clinicians and researchers to conduct ongoing psychometric testing of behavioural tools for use with linguistically and culturally diverse sub-Saharan African populations. The data also reveal important insights regarding problems associated with using standard behavioural tools in children with complex clinical presentations.