A dissertation submitted to the Faculty of Medicine, Universlty of the Witwatersrand, in partial fulfilment of the requirements for the degree of Master of Medicine in the branch of Haematology. / Despite the efforts, for more than twenty years, to control malaria, the incidence of this disease still appears to be escalating globally. At the Chris Hani/Baragwanath Hospital, data analysis of malaria admissions between January 1994 to December 1996 showed an increasing trend from year to year. The main objective of the study was to try and provide some insight into this increasing rate of malaria at the ChrisHani/Baragwanath Hospital. The study was structured into two main parts: a retrospective analysis which concentrated on malaria admissions between and including January 1994 and March 1994, and a prospective analysis interviewing and examining all the malaria cases that were diagnosed between and including January 1995 and March 1995. Both aspects of the study assessed the patients socioeconomically, haematologically and immunologically. A detailed travel,medical and drug history was taken through the aid of a questionnaire. Two hundred and sixty-three patients (175 male and 88 female), of which 35% (91/263) were children (< 13 years old), were diagnosed with malaria. The clinical and laboratory presentations were consistent with other studies. The prevalence of complicated disease however was less than what has been described in the literature; cerebral malaria (as defined by Warrell, 1982) was documented in 1% of patients, hypoglycaemia (glucose < 2.2 mmol/l) and renal failure (creatinine > 265 umol/l) accounted for 5% and 3% of the cases respectively. In contrast to this 32% had features of liver dysfunction, however it appeared that haemolysis was the main contributing factor to the liver derangement. The most common infecting species was Plasmodium falciparum, alone (91.3% of the patients) or part of a mixed infection with either P. vivax or P.ovale (3.8% of patients). More than 88% of the infections were contracted in other African, mainly southern African countries, the most important of which was neighbouring Mozambique (58%). About 11% were contracted in endemic areas of South Africa i.e Northern Province, Mpumalanga and Kwazulu Natal. Twelve percent of cases (24/203) gave a history of previous malaria, The underlying immune status of these patients was analyzed using the Indirect Fluorescence Antibody Test (IFAT) and compared with the total study group. The test did not reveal any striking differences between the two groups. The previously exposed patients however did demonstrate a much lower parasitaemia, with 71% (17/24) of cases presenting with a parasite density <_;1%. These results may indicate the ability of these patients to clear their parasitaemias earlier due to previous sensitization, with the subsequent establishment of a low grade chronic infection. Seven of 88 women admitted had a documented pregnancy at the time of diagnosis. Foetal death was recorded in 517 cases which confirms the poor prognosis in pregnancy associated malaria infection, reported by other authors. Fifteen patients (13 adults and 2 children) required admission into the intensive care unit. Indications included high parasite loads, > 5% (67%) and renal failure, creatinine> 265 umol/l (33%). Standard chemotherapy was administered to all the patients with the most frequently used being quinine (94% of cases), alone or in combination with other drugs. The use of prophylactic agents for the prevention ofmalaria was restricted to twelve patients (8%), with the majority of individuals being ignorant about( malaria and therefore being unaware that any medication along with other preventative measures, were necessary prior to entering , and while staying in an endemic region. It was also apparent that the correct dosage was not adhered to as none of the patients completed their antimalarial course after returning from the malaria area. The most commonly used prophylactic drugs were chloroquine, alone or in combination with proguanil and pyrimethamine plus dapsone (Maloprim), The latter is no longer recommended routinely as a prophylactic agent. Following univariate analysis using Fisher's exact test and the student t-test, and a multivariate: nalysis (using a logistic regression model), hyperparasitaemia (p=O.0070) and renal failure (p=O.0016) were identified as significant predictors of poor outcome. Significant differences were also demonstrated in the mean WCC and the mean HB levels between the survivors versus the patients that died, indicating that a significantly elevated WCC and an anaemia at presentation, may be important risk factors towards the establishment of severe/complicated infection. The overall mortality rate was 3%. Climatic data (which was limited to the Johannesburg area), together with evidence that the malaria bearing Anopheles vector does not exist in Gauteng suggests that the conditions in the city may not be suitable for local transmission of malaria during the summer. It therefore appears that all efforts need to be channeled into the education of our travelers who visit malaria endemic regions and upon returning succumb to 'imported' malaria. Perhaps the education of the traveler alone is not sufficient, and medical personnel who diagnose the condition and who prescribe prophylactic agents need to revise their knowledge of this life threatening infection, so that their advice and drug therapies are optimal and effective. Lastly the pharmaceutical companies that manufacture these drugs might be persuaded to make their products more affordable for those individuals who are most at risk. We need to utilize our limited options to the fullest until such a time as the ultimate challenge is realised - an effective malaria vaccine. / Andrew Chakane 2018

A dissertation submitted to the Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, in fulfillment of the requirements for the degree of Master of Medicine in Ophthalmology Johannesburg, 2014 / Retinoblastoma is the most frequent malignant intraocular tumour of childhood, and in Africa it is the commonest and most important life threatening ocular neoplasm. In resource-rich countries such as the United States and Canada, survival rates for patients with retinoblastoma approach 100%. This is in stark contrast to developing countries where survival rates may be less that 20%. Despite the vast majority of affected patients living in less developed countries, there is little attention and published literature on disease characteristics in these populations. OBJECTIVES: To characterise retinoblastoma in the South African population through a 20-year retrospective analysis of patient records at two tertiary academic hospitals in Johannesburg. DESIGN AND METHOD: Retrospective clinical case series analysis of medical records of patients with retinoblastoma presenting to Charlotte Maxeke Johannesburg Academic Hospital and Chris Hani Baragwanath Academic Hospital between 01 January 1992 and 31 December 2011. RESULTS: The total number of retinoblastoma cases identified was 282, with 245 of these meeting the study inclusion criteria. Retinoblastoma comprised 6.9% of total paediatric oncology presentations. 65.3% were unilateral, 34.3% bilateral and 0.4% trilateral. The overall male to female ratio was 1.08. Mean age at presentation overall was 32.6 months (median 28.0 months), for unilateral 39.4 months (median 33.0 months) and for bilateral 19.7 months (median 17.0 months). The mean delay to presentation overall was 7.0 months (median 4.0 months), for unilateral 8.5 months (median 5.0 months) and for bilateral 4.4 months (median 3.0 months). The most frequent presenting symptoms were leukocoria (37.1%) and proptosis (34.7%). Distribution of disease stage at presentation (using the International Retinoblastoma Staging System) was 1.6% with Stage 0, 24.1% with Stage I, 27.8% Stage II, 16.3% Stage III and 25.3% Stage IV (data not available in 4.9%). 26.5% of patients defaulted care. The five-year survival rate was 57.7% in the overall study population, and according to disease stage at presentation: 95.3% - Stage I, 84.8% - Stage II, 49.7% - Stage III and 5.7% - Stage IV. CONCLUSION: The five-year survival rate for Stage I disease is similar to the overall five-year survival rates reported in the developed world. This suggests that treatment standards at the study hospitals are comparable with those of developed countries and that similar overall survival rates may be achieved if patients were to present as early. Delay to presentation, disease stage at presentation and defaulting care were identified as key factors contributing to the poor overall survival rate. This study provides information regarding patient demographics, social challenges in management, and patient outcomes in comparison to developed countries and to reports from other African populations. It also highlights the need for educational campaigns and screening initiatives to address poor survival outcomes as a result of late presentation and high rates of patients defaulting care. Finally, this report serves as a platform for comparison with future research in this area.

Despite the efforts, for more than twenty years, to control malaria, the incidence of this disease still appeal's to be escalating globally. At the Chris Hani/Baragwanath Hospital, data analysis of malaria admissions between January 1994 to December 1996 showed an increasing trend from year to year. The main objective of the study was to try and provide some insight into this increasing rate of malaria at the Chris Hani/Baragwanath Hospital. The study was structured into two main parts: a retrospective analysis which concentrated on malaria admissions between and including January 1994 and March 15)94, and a prospective analysis interviewing and examining all the malaria cases that were diagnosed between and including January 1995 and March 1995. Both aspects of the study assessed the patients socioeconomically, haematologicaliy and immunologicailly. A detailed travel,medicai and drug history was taken through the aid of a questionnaire. Two hundred and sixty-three patients (175 male and 88 female), of which 35% (91/263) were children (< 13 years old), were diagnosed with malaria. The clinical and laboratory presentations were consistent with other studies. The prevalence of complicated disease however was less than what has been described in the literature; cerebral malaria (as defined by Warrell, 1982) was documented in 1% of patients, hypoglycaemia (glucose < 2.2 mmol/1) and renal failure (creatinine > 265 pnol/1) accounted for 5% and 3% of the cases respectively. In contrast to this 32% had features of liver dysfunction, however it appeared that haemolysis was the main contributing factor to the liver derangement. The most common infecting species was Plasmodium falciparum, alone (91.3% of the patients) or part of a mixed infection with either P.vivax or P.ovale (3.8% of patients). More than 88% of the infections were contracted in other African, mainly southern African countries, the most important of which was neighbouring Mozambique (58%). About 11% were contracted in endemic areas of South Africa i.e Northern Province, Mpumalanga and Kwazulu Natal. Twelve percent of cases (24/203) gave a history of previous malaria. The underlying immune status of these patients was analyzed using the Indirect Fluorescence Antibody Test (IFAT) and compared with the total study group. The test did not reveal any striking differences between the two groups. The previously exposed patients however did demonstrate a much lower parasitaemia, with 71% (17/24) of cases presenting with a parasite density <1% . These results may indicate the ability of these patients to clear their parasitaemias earlier due to previous sensitization, with the subsequent establishment of a low grade chronic infection. Seven of 88 women admitted had a documented pregnancy at the time of diagnosis. Foetal death was recorded in 5/7 cases which confirms the poor prognosis in pregnancy associated malaria infection, reported b / other authors. Fifteen patients (13 adults and 2 children) required admission into the intensive care unit. Indications included high parasite loads, > 5% (67%) and renal failure, creatinine > 265 p.mol/1 (33%). Standard chemotherapy was administered to all the patients with the most frequently used being quinine (94% of cases), alone or in combination with other drugs. The use of prophylactic agents for the prevention of malaria was restricted to twelve patients (8%), with the majority of individuals being ignorant about malaria and therefore being unaware that any medication along with other preventative measures, were necessary prior to entering , and while staying in an endemic region. It was also apparent that the correct dosage was not adhered to as none of the patients completed their antimalarial course after returning from the malaria area. The most commonly used prophylactic drugs were chloroquine, alone or in combination with proguanil and pyrimethamine plus dapsone (Maloprim). The latter is no longer recommended routinely as a prophylactic agent. Following univariate analysis using Fisher’s exact test and the student t-test, and a multivariate analysis (using a logistic regression model), hyperparasitaemia (p=0.0070) and renal failure (p=0.0016) were identified as significant predictors of poor outcome. Significant differences were also demonstrated in the mean WCC and the mean HB levels between the survivors versus the patients that died, indicating that a significantly elevated WCC and an anaemia at presentation, may be important risk factors towards the establishment of severe/complicated infection. The overall mortality rate was 3%. Climatic data (which was limited to the Johannesburg area), together with evidence that the malaria bearing Anopheles vector does not exist in Gauteng suggests that the conditions in the city may not be suitable for local transmission of malaria during the summer. It therefore appears that all efforts need to be channeled into the education of our travelers who visit malaria endemic regions and upon returning succumb to ‘imported’ malaria. Perhaps the education of the traveler alone is not sufficient, and medical personnel who diagnose the condition and who prescribe prophylactic agents need to revise their knowledge o f this life threatening infection, so that their advice and drug therapies are optimal and effective. Lastly the pharmaceutical companies that manufacture these drugs might be persuaded to make their products more affordable for those individuals who are most at risk. We need to utilize our limited options to the fullest until such a time as the ultimate challenge is realised - an effective malaria vaccine.

A research report submitted to the Faculty of Health Sciences, University of the Witwatersrand, in fulfillment for the requirements for the degree of Master of Medicine in Microbiology. Johannesburg, 2017 / Background Invasive Candida infections (ICI) have emerged as an important cause of increased morbidity and mortality in specific patient populations in recent years. Multiple risk factors coupled with changes in epidemiology have made clinical management of these patients challenging. A laboratory-based surveillance project, Tracking Resistance to Antifungal drugs for Candida species in South Africa (TRAC-SA) was conducted at Chris Hani Baragwanath Hospital (CHBH) from 2009 to 2010 and allowed for collection of laboratory information related to episodes of candidaemia, delineation of the situation at the hospital and distribution of information to relevant stakeholders to help make informed clinical decisions. Objective Determine the clinical epidemiology and risk factors for bloodstream Candida infection at CHBH over an 18-month period Methods A retrospective, cross-sectional analysis was carried out on cases of blood culture-confirmed candidaemia from inpatients from 1 February 2009 until 31 August 2010. These cases were identified from the TRAC-SA database, inpatient files were traced and clinical data recorded on a standard case report form. Additional laboratory data of selected tests done within 72 hours of the initial blood culture were obtained from the National Laboratory Health Service Corporate Data Warehouse (CDW). Results A total of 167 episodes of candidaemia were identified during the study period with an incidence of 2.09 per 1000 admissions. The distribution of episodes occurred among 55 children (33%), 41 adults (25%) and 71 neonates (43%). The overall species distribution was Candida species other than C. albicans (98/167, 58.7%) and C. albicans (69/167, 41.3%). Candida species other than C. albicans comprised mainly of C. parapsilosis (73/167, 43.7%), C. glabrata (10.2 %, 17/167) and other species combined including C. tropicalis and C. krusei (8/167, 4.7%). Factors associated with C. albicans (versus Candida species other than C. albicans) infection included older age, use of 2 or more antibiotics, use of broad spectrum antibiotics specifically meropenem, aminoglycosides, vancomycin, co-trimoxazole and mechanical ventilation (p < 0.001). The overall case-fatality was 59/163 (35.3%). The highest case fatality was noted among adults with C. albicans infection, i.e. 15/22 (68.18%). Significant risk factors associated with in-hospital mortality were use of central lines, urinary catheters, total parenteral nutrition, 2 or more antibiotics, beta lactam - beta lactamase inhibitors, proton pump inhibitors, aminoglycoside and abdominal surgery (p < 0.01). Of the C. parapsilosis isolates tested, 40 (57. 9%) tested non-susceptible to fluconazole. Risk factors associated with fluconazole resistance included neonatal age, involvement of the respiratory system, mechanical ventilation, chemotherapy, use of a prior antifungal agent and use of 2 or broader spectrum antibiotics (p<0.01). Of 71 neonates, 16 (22.5%) received empiric antifungals, in comparison to children (5/55, 9.0%) and adults (4/41, 9.7%) (p = 0.272). Conclusion CHBH had a high incidence of candidaemia with a predominance of Candida species other than C. albicans especially in the neonate age group. Risk stratification of in-patients is of paramount importance in choice of empiric antifungal drug due to the differing azole resistance patterns observed. / MT2017

This dissertation was motivated by the growth in the number of elderly with dementia in the United States and worldwide, and the consequent need to assist caregivers who face the chronic stress and consequent negative outcomes of caring for dementia patients for increasingly longer periods of time. While behavioral and psychological symptoms of dementia (BPSD) are routinely cited as important predictors of negative caregiving outcomes, they consist of a wide variety of patient behaviors (e.g., depression, physical aggression and paranoid delusions). This dissertation aimed to further elucidate the BPSD-caregiver relationship by determining if and why individual components of BPSD impacted caregiver depression, how these relationships change over time, and how the timing of behaviors across the disease course affect the caregiver. First, via a systematic literature review we found that despite the significant amount of research conducted examining BPSD and caregiving, the literature focused on the aggregate effect of symptoms rather than on the effect of individual symptoms. Existing studies categorized BPSD based on how they clinically manifest in the patient, rather than from the perspective of how they may differentially impact the caregiver. Second, we tested the differential impact of four symptom clusters derived based on their likely impact on the caregiver (accusatory/aggressive symptoms, non-threatening psychotic symptoms, depressive symptoms, and difficult behaviors) on caregiver depression. All clusters exerted similar effect sizes on caregiver depression, with patient depressive symptoms exerting the strongest and most consistently stable relationship with caregiver depression. The patient depression-caregiver depression relationship was mediated by both perceived burden to caregivers and impact of symptoms on the patient. Finally, we did not find evidence that the presence of individual BPSD early in the disease course impacted subsequent caregiver depressive symptoms. When examining change patterns over time, we found that most caregivers had stable trajectories of depressive symptoms, with a smaller subset showing more evidence for wear-and-tear over time. Given that there may be a differential effect of individual symptoms on caregivers, research in this field should continue to study the effects of individual BPSD symptoms rather than the cumulative effects of a range of behaviors and should also consider how symptoms may negatively impact caregivers by evoking empathy in addition to increasing burden for the patient. Future work should continue to utilize a longitudinal perspective on caregiving to better characterize the relationship between individual BPSD and caregiver outcomes and to understand varied outcome trajectories. The continued focus on understanding how, why and when BPSD are most detrimental to caregivers can help target and improve patient treatment and caregiver intervention efforts.

In this thesis, I explore the epidemiological features of CNS infections using a population-based dataset and analyze CSF samples from patients with meningitis or encephalitis of unknown cause using molecular methods. In the second chapter, I review the literature on the epidemiology of meningitis and encephalitis, and current diagnostic approaches, focusing on the strengths and limitations of various diagnostic methods. In the third chapter, I investigate the causes of meningitis and encephalitis using a population-based dataset from the National Health Insurance Research Dataset (NHIRD) from Taiwan, representing the interval from 1996 to 2008. The analysis assesses differences in the disease between known and unknown cause groups in terms of incidence, demographic features, seasonal and geographic distribution. In the fourth chapter, I analyze CSF samples from patients with meningitis or encephalitis of unknown cause using a tiered molecular approach to discover the undetected or novel pathogens; techniques used include MassTag PCR, DNA microarrays and high-throughput pyrosequencing. I conclude the thesis in the fifth chapter, summarizing and highlighting the main results of these studies.

Type 1 diabetes (T1D) is an autoimmune disease characterized by destruction of beta cells in the pancreas resulting in insulin deficiency, which leads to hyperglycemia and organ damage. T1D patients experience an increased risk of morbidity and mortality due to long-term complications, specifically retinopathy, nephropathy, and neuropathy. Studies demonstrating familial aggregation support the claim that a genetic contribution may influence the development of complications. This dissertation aims to identify genes/chromosomal regions that predispose T1D patients to, or protect them from, the expression of the chronic microvascular complications: retinopathy, neuropathy and nephropathy. In my first chapter, I introduce the history of type 1 diabetes and microvascular complications and their importance as a public health concern. Data show that the prevalence for T1D is increasing, and thus it is likely that the prevalence for the associated complications will also increase. Further, a large proportion of T1D patients develop at least one microvascular complication within 15 years of T1D diagnosis. By identifying risk alleles for the microvascular complications of T1D, the findings of this study could allow physicians to determine which patients are at greater/lesser risk for developing complications, help to develop interventions to delay or protect against the development of complications and thus reduce medical expenditure and suffering due to diabetes. In the second chapter, I provide the background for, and review the literature on, the genetics of T1D and microvascular complications. The genetic risk factors for T1D are well-established, but there is conflicting research on the question of whether T1D-predisposing HLA alleles may also be in part responsible for the occurrence of microvascular complications seen in T1D patients. We also investigate whether T1D HLA risk alleles are associated with all forms of microvascular complications or whether there are HLA alleles specifically associated with a given complication. In the work described, I address these questions by examining the relationship of HLA alleles to the risk for any complication and to the risk for some specific complication. In the third chapter, I perform case-control analysis and evaluate known type 1 diabetes HLA susceptibility alleles and their association with microvascular complications. I used data from the Human Biological Data Interchange (HBDI), which includes 425 Caucasian families (2,506 family members) with cases diagnosed with type 1 diabetes. Using a case-control study design nested on the cohort of the HBDI type 1 diabetes patients and their families, probands with at least one microvascular complication were considered cases, and the probands with T1D without microvascular complications (T1D only) were considered controls. Our findings suggest that the HLA class II DRB1*03:01 allele is a protective factor for complications, specifically for retinopathy, as is the DQA1*05:01-DQB1*02:01 haplotype. The DRB1*04:01 allele showed no evidence of association, except when the carriers of the protective DRB1*03:01 were removed from the analysis. Findings also showed a strong positive association between the HLA class I allele B*39:06 and complications. In the fourth chapter, using the same sample of HBDI type 1 diabetes families that I used in Chapter 3, I perform linkage analysis and test markers along chromosome 6 for co-segregation with microvascular complications. Using SNP data that were genotyped by the Center for Inherited Disease Research (CIDR), I performed linkage analysis examining 1) the phenotype of T1D itself, 2) the presence of any microvascular complication, 3) retinopathy alone, 4) nephropathy alone, and 5) neuropathy alone. Initially, we confirmed the linkage of the HLA locus to T1D. In subsequent analyses, using all complications as well as retinopathy alone as the phenotypes, we identified two linkage peaks; a linkage peak located at the HLA locus and another novel locus was telomeric to HLA. We did not find evidence for linkage for nephropathy alone or neuropathy alone. Findings from this dissertation show that both HLA and non-HLA regions are involved in the expression of complications with strong evidence of genetic influences specifically for retinopathy.

As the proportion of adults aged 65 years and older continues to grow across the globe and edentulism rates decline, there is widespread concern about a rise in the prevalence of periodontal disease, characterized by chronic inflammation of tooth-supporting tissues induced by persistent infection. Compared to their younger counterparts, older adults experience a higher burden of periodontal disease, which can result in tooth loss, poor nutritional intake, higher prevalence of other chronic diseases, and a decrease in overall quality of life. While cross-sectional studies have underscored the role of chronic stress on periodontal disease progression in older adults, longitudinal evidence is currently lacking. This dissertation draws on prospective data from a birth cohort of older Danish adults (1914 Glostrup Aging Study) with 25 years of follow-up. Using physiological markers than span the metabolic, inflammatory, and cardiovascular systems, I developed a composite measure of allostatic load (AL) at age 80, defined as the cumulative biological damage that results from a whole-body adaptation to chronic stress. First, I identified social and behavioral predictors of high scores on AL. In men, those with no vocational training, unskilled occupation, low income, and a sedentary lifestyle were more likely to have high AL, consistent with a "weathering" pattern of biological systems resulting from chronic adversity over the life course. To test the hypothesis that high AL is longitudinally associated with periodontal disease, I evaluated bidirectional longitudinal associations using multiple measures of AL and periodontal disease. Results showed a positive nonlinear association of AL at age 70 with periodontal disease at age 85, but no association between periodontal disease at age 70 with AL at age 80. This finding confirms previous cross-sectional data, and supports the role of chronic stress on infection-induced inflammation. To test the hypothesis that high AL is associated with mortality risk, I examined this association longitudinally from ages 70 - 95. Compared to low AL, high AL was positively associated with all-cause mortality, and even stronger when cardiovascular disease mortality was considered. AL-mortality associations were higher among those who were dentate as compared to edentate, suggesting that dentate status modifies the relationship. Findings from this dissertation contribute to our understanding of the consequences of stress on periodontal disease in relation to aging and offer potential avenues for intervention.

Alopecia areata (AA) is the most prevalent autoimmune disease in the US. With An estimated lifetime risk of 1.7%, it affects both genders with similar frequencies and people of all ages. AA affects more individuals than most other autoimmune diseases combined, and yet despite its prevalence, there is an enormous unmet medical need, in part due to the dearth of information about the underlying pathogenesis. In AA, autoimmunity arises against the hair follicles in the skin, which causes hair loss associated with an aberrant accumulation of immune-response cells around the affected hair follicles. Evidence supporting a genetic basis for AA stems from multiple lines of research, including increased risk of disease in first degree relatives, twin studies, and more recently, our initial family-based linkage study and genome wide association study (GWAS) in a cohort of unrelated individuals. Importantly, our GWAS identified a set of 16 statistically independent risk haplotypes across 8 loci, implicating specific genes that increase risk of AA, all of which have been validated. Genome wide genetic studies can provide critical insight into disease mechanisms, especially when little is known about the underlying causes of disease. In this study, I use complementary gene mapping methods, performing one study in a cohort of families and a second study in a cohort of unrelated cases and controls. Using these two approaches, I obtain new evidence about the genetic influences on AA. Our family cohort contains statistically significant evidence for linkage at a new locus, on chromosome 2q36.1-q37.3 (LOD=4.17) and family-based tests of association implicate 47 genes. I then conducted a GWAS that expanded our initial cohort with the addition of 800 cases and obtained statistically significant evidence for a new locus at chromosome 16p13.13 (p=4.6x10-7). This region has been implicated in several other autoimmune diseases and contains several genes that are known to be involved with immune processes. Taken together, these two studies demonstrate the presence of both rare and common variants are contributing to AA etiology and support emerging evidence that suggests the genetic architecture of common complex diseases involves both rare and common variants.

Purpose and Background: The aim of this paper is to investigate the causal effect of the timing of Kasai procedure, a surgical intervention, on Biliary Atresia (BA) patient outcomes using quasi-experimental techniques developed in economics and education. BA is a rare pediatric disease and is the most common cause of liver disease-related death in children. The causes of BA are not well established and are probably multifactorial. The Kasai procedure is usually performed at the early stage of disease, within the first few months of a patient’s life. Although performance of the Kasai procedure before 60 days of age suggests improved prognosis, no available research to date has investigated the causal effect of the timing of the Kasai procedure on patient outcomes. Methods: This study uses data drawn from the Japan Biliary Atresia Registry. The dataset contains a total of 2743 patients; 1743 girls, and 1000 boys registered since 1989 to 2012. Subjects were eligible for study inclusion provided that they were born between 1989 and 2012 and underwent the Kasai procedure. In addition, eligible subjects had to have had an observable primary predictor variable and outcome. The primary predictor variable was the age at Kasai procedure and the outcome was 1-year native liver survival without jaundice. Other variables included in the analyses were sociodemographics (i.e. gender, birth order, birth weight, gestational age, parental age), clinical characteristics (i.e. types of obstruction, associated anomalies, type of choleretic agents), and hospital characteristics defined as hospital caseload. Bivariate analysis, multivariate logit/probit regressions, stratified multivariate logit regression, hospital fixed effects, and instrumental variable (IV) approaches were conducted. Findings: The IV approach was used to examine whether the timing of the Kasai procedure was endogenous in the model. Although using the length of stay at the last hospital prior to Kasai procedure as a proxy for accessibility to the nearest high-level hospital had its strengths as the instrument as confirmed by the first-stage diagnostics, the Wu-Hausman test indicated that the estimates of the model with and without IV were consistent. The results from the multivariate covariate-adjusted logit regression suggested that for patients with obstruction types I and I with cyst, the timing of Kasai procedure was not associated with 1-year native liver survival. However, in patients with type III obstruction (the majority of BA patients [86%]), when comparing to the timing of Kasai procedure at 61 to 75 days, the timing of Kasai procedure at less than 30 days was suggestively associated with the primary outcome of 1-year native liver survival without jaundice (log-odds ratio: 0.43 [t=1.79]). This was 1.5 times as likely to achieve 1-year native liver survival without jaundice while holding other variables constant. The timing of Kasai procedure after 76 days had a statistically negative association (log-odds ratio: -0.36 [p<0.05]) with the outcome. Moreover, the Kasai procedure at 106 to 121 days of age decreased the log odds of 1-year native liver survival without jaundice by 0.83, which equated to an increase in likelihood of achieving the primary outcome of 0.43 times. The Kasai procedure after 121 days of age also decreased the log odds of the outcome by 1.17 that was 0.3 times as likely to achieve 1-year native liver survival without jaundice. These differences are quite significant; therefore, the effort to increase the patients who could have the Kasai procedure at less than 30 days and to prevent the Kasai procedure after 76 days should be made to improve 1-year native liver survival without jaundice for the type III patients. As previously suggested, cholangitis and the use of corticosteroid have a negative association with the outcome after adjusting for covariates. Ursodeoxycholic acid indicates a positive association with the outcome, as does the hospital caseload, even when both are adjusted for covariates.