Philosophiae Doctor - PhD / The current study was, in part, prompted by the high case fatality rates for severe acute malnutrition in two district hospitals in the Eastern Cape Province in South Africa. These case fatality rates were being attributed to Human Immunodeficiency Virus infection rather than to mismanagement by nurses involved in the hospital management of SAM cases. There were also some anecdotes from clinicians in the same hospitals that, depending on the clinical stage of HIV infection, the World Health Organisation's ten-step protocol may show no effect. This left some uncertainties as to whether these guidelines are suitably designed for use during the management of HIV positive children who are severely malnourished and at different HIV clinical stages. This study sought to reinforce the design of a longstanding facility-based intervention originally developed to improve the management of severe acute malnutrition in two district hospitals in South Africa. The aim was to design an improved intervention which was implemented and evaluated to determine its potential effect on treatment outcomes, specifically in the context of high HIV comorbidity. The study also sought to provide the context for the effectiveness of this intervention, in terms of its implementation fidelity and associated moderating factors. Lastly, the study evaluated the sustainability of the intervention after it was discontinued. Methods The current study reports on the development, implementation and evaluation of an intervention to improve the management of severe acute malnutrition in two district hospitals in the Eastern Cape Province. A Sequential Explanatory Mixed Method Design was used. During the study, the effect of HIV infection, disease stage and other clinical characteristics on the survival of children with severe acute malnutrition was assessed. The relationship between the rate of weight gain and duration of hospitalisation based on HIV status and disease stage were also examined. The data were collected prospectively during the study using retrospective record review of a total of 450 severely malnourished children who were admitted and treated at the two facilities from 2009 to 2013.A pre-tested 76- item patient evaluation form was used to collect data on patient characteristics on admission, treatment processes and outcomes. Data analysis was performed using STATA13.0 and involved simple descriptive computation of quantitative variables as well as non-parametric tests to compare groups between and within hospitals. Kaplan-Meier curves and Cox proportional hazard modelling were used to analyse time to event data. The study also assessed the impact of the intervention at time intervals on outcomes of interest. The analysis focused on modelling and plotting monthly mortality statistics collected over a period of 69 months. This was done to detect related trend and level changes before, immediately (after the first two months) and after (following the two months) the removal of the intervention. Lastly ethnographic and focus group enquiries were used to explain the quantitative results. Two focus group discussions were held in each hospital with clinicians and the management staff. This was done at the end of phase three. The focus group data were analysed using the framework analysis approach.