Thesis (M. Tech.) -- Central University of Technology, Free State, 2008 / Blood transfusions carry a number of risks, one of which is transmitting HIV/AIDS from an infected donor. Since HIV is sexually and parenterally transmitted, the initial HIV risk management of donated blood in the early 1980‟s consisted of screening by visual assessment and completion of a lifestyle questionnaire, followed by deferral of practicing homosexual and bisexual male donors and intravenous drug addicts. The visual assessment was replaced by tests for antibodies directed against HIV, from the middle 1980‟s. In the early 1990‟s HIV was increasingly found in the black population of South Africa, particularly among black women. By 1998 0.26% of the received donations returned a positive test for HIV-1. In 1999 the South African Blood Transfusion Service (SABTS) Blood Safety Policy was introduced, including a donation HIV-risk categorization model which used the donor ethnic group, gender and donation history as indicators of the risk of exposure to HIV. The unacceptable use of the donor ethnic group as an indicator was the motivation to seek a suitable alternative donation risk categorization model which excludes the donor‟s ethnic group. The use of a more acceptable model with a high level of accuracy in predicting the risk of exposure to HIV has the potential of contributing to the reduced risk of HIV transmission through blood transfusion in South Africa. The aim of this study was to compare the suitability of four alternative models based on the information obtained from donors. Donations from new and lapsed donors were categorized in the highest applicable risk category in each model. The study was divided into two phases to achieve the aim. The first phase needed to determine suitable parameters for a model which uses the donor‟s age as an indicator. For this phase the ages of the regular donors returning an HIV-positive test result, were analysed. The second phase was to evaluate the effectiveness of the four suggested alternative blood donation risk categorization models against the model introduced by the SABTS in 1999. During this phase the donor demographic data and donation histories of donors who made donations at the Bloemfontein branch of the South African National Blood Service (SANBS) between October 2004 and September 2005, were analysed statistically. This phase honed in on two aspects to evaluate the effectiveness of the alternative models. Firstly the percentages of HIV-positive donations found in each risk category of each model, were determined as indicators of the residual risk of HIV-positive donations within the window period. Secondly the percentages of the collected blood donations allocated to each risk category within each model, were analysed to give an indication of the availability of “safe” blood associated with each of the models. The first phase of the study highlighted the difference in the age-group prevalence between male and female regular donors who returned an HIVpositive test result. Potentially suitable parameters for an Age-based Model were formulated by comparing this data with the ages of the donors who donated in Bloemfontein during the twelve months covered by this study. The second phase compared a Donation Interval Model, a Combination Model (using donation interval, gender and ethnic group as indicators), the SANBS 2005 Model (using age and gender as indicators) and an Agebased Model (using age and gender as indicators) with the SABTS 1999 Model (using gender and ethnic group as indicators). This study has shown that each of the models analysed has its advantages and disadvantages. The SANBS 2005 Model proved the best model without an ethnic indicator, for SANBS. Several recommendations regarding further investigation emanating from the results of this study were made.