The OptAIDS project: towards global halting of HIV/AIDS

Introduction We face a unique, transitory opportunity in the history of the HIV/AIDS epidemic, because we have collectively pooled money faster than the epidemic has grown [1]. Can we then seize the moment and halt this epidemic now? Most scenarios for the future of HIV/AIDS project modest reductions spread out over decades [2]. The very timescale of such projections, beyond the persistence time of all models, makes them unreliable [3]. Can we do better, quicker? The OptAIDS project was conceived as a means to address this issue. Its implementation thus far has been twofold: a workshop held in July 2008 and this supplement on the eradication of AIDS. The aims of the project are to address two questions: 1. Can we optimally spend our way out of the HIV/AIDS epidemic? 2. Can we work together to build a World Halting AIDS Model (WHAM) that would permit us to estimate the quickest way to halt HIV/AIDS, monitor our success, and adjust our strategy as we go? The OptAIDS project grew out of a frustration with existing attempts to tackle the disease. AIDS exceptionalism means that HIV/AIDS is handled differently from other public-health epidemics, which has likely been detrimental [4,5]. Consequently, much of the funding of HIV/AIDS efforts has been for qualitative observations of the expanding epidemic rather than quantitatively effective intervention. Although fund accumulation has recently outpaced the epidemic, we argue that plans to spend donor money are too long range in the face of a growing epidemic [6]. Long-range scenarios have no reality to them, so that only short-term solutions - those that fall within the persistence time of their models - have any possibility of being realistic [3]. Furthermore, disease is a global problem that is only tackled locally [7]; epidemics cross borders, whereas we fund mostly local or national "solutions". The OptAIDS project was an outgrowth of the Stop Afghan AIDS project [8]. This project was led by mathematical modellers planning to continuously adapt their models to new data and predicting what data should be collected. The Stop Afghan AIDS project showed how it should be possible to intervene quantitatively in an epidemic. The usefulness of modelling in complex systems is not new. Mathematical models of the economy tell us whether a decrease in income tax will result in an increase in investment or an increase in imported consumer goods. Mathematical models of the atmosphere tell us what the effects of carbon dioxide emissions or of nuclear wars may be. Mathematical modelling is used routinely in such things as aircraft design and the design of traffic systems [9]. So too, epidemics are quantitative creatures with predictable thresholds. Models that can be adapted to new results and to changes in control policy have been identified as an integral part of disease-control programs [10]. Modelling-led interventions were instrumental in halting the 2001 Foot and Mouth outbreak in the UK [11]. A mathematical model of the dynamics of measles in New Zealand developed in 1996 successfully predicted an epidemic in 1997 and was instrumental in the decision to carry out an intensive immunisation campaign in that year. While the epidemic began some months earlier than anticipated, it was rapidly brought under control, and its impact on the population was much reduced [12]. The West African Onchocerciasis (river blindness) Control Program successfully used modeling to supplement intervention programs [13]. By using clearly delineated endpoints, these models helped convince donors and the scientific community that the aims of the program were achievable [14]. As a result, mathematical models have retained a role in subsequent policy discussions [15]. Insights from mathematical models during the SARS epidemic helped determine how serious the epidemic might become, as well as the impact of proposed control measures. These models provided important guidance to public-health authorities at a critical time when little other information was available. Insights from the models showed that, if unchecked, the virus could cause a significant epidemic, but that basic epidemiological control measures - patient isolation, contact tracing, etc - could have a substantial impact on the extent of the epidemic. Subsequently, these control measures played a major role in limiting the spread of the 2003 epidemic [16]. Weather prediction models provide a workable analogy. Such models consist of continually updatable inputs, that must adapt to an enormous array of incoming data [17]. Short-term predictions, especially those associated with discrete, extreme weather events such as floods and hurricanes, have proven useful in supporting emergency management strategies, unlike events such as earthquakes or acid rain, which have longer lead times [18]. Complex mediating models which themselves have explanatory power and which embody techniques of modeling can be refined and passed down to successor models [9]. The virtue of mathematics in such a context is that it forces clarity and precision upon the conjecture, thus enabling meaningful comparison between the consequences of basic assumptions and the empirical facts [19]. Existing scenarios for HIV control have typically been spread out over two or more decades [20], which means that the reliability of their predictions is low. The basic concept of OptAIDS is to spend more money up front, effectively, based on the best models and their parameters we can formulate, with the goal being a rapid halt to the epidemic with the fewest additional cases. This means that models can be shorter term and therefore more reliable, because we stay within the models' persistence time. OptAIDS emphasises continuous monitoring to check the accuracy and adjust the parameters of the global model. Mathematically, this is an optimal halting problem.