An application of Bayesian Network modelling to the HIV/AIDS epidemic in the Philippines.

Abstract

Historically, the Philippines was one of the few countries that had not faced a large HIV epidemic. This thesis evaluated the current epidemiology, trends in behaviour and public health response in the Philippines to identify factors that could account for the HIV epidemic just prior to a recent epidemiological outbreak, and found that the likely reasons for the epidemic's slow development include: the country's geography is complicated; injecting drug use is relatively uncommon; a culture of sexual conservatism exists; sex workers tend to have few clients; anal sex is relatively uncommon; and relatively high circumcision rates. As well as to review conditions that may be of concern for facilitating an emerging epidemic were examined and found that there are numerous factors suggesting that HIV is increasing and ready to emerge at high rates, including: the lowest documented rates of condom use in Asia; increasing casual sexual activity; returning overseas Filipino workers from high-prevalence settings; widespread misconceptions about HIV/AIDS; and high needle-sharing rates among injecting drug users.

This thesis has also proposed an application of Bayesian Networks to investigate the impact of funding changes on HIV incidence in four populations of interest (men who have sex with men (MSM), intravenous drug users (IDUs), female sex workers (FSWs) and those in the non most at risk population (Non-MARPs)) in the Philippines. The model, which is the first of its kind in this field, focused on the probabilistic relationships and the factors that contribute to new infections in the population. It investigated the impact of funding changes on the probability of increases in HIV incidence in the four populations of interest and found that changes in primary prevention program funding have a greater impact on HIV incidence changes, than changes in testing and treatment programs.