Salud Pública de México

Mesa redonda XXXIOptimizing HIV/AIDS prevention programs: towards multidimensional allocative efficiency

Mesa redonda XXXIOptimizing HIV/AIDS prevention programs: towards multidimensional allocative efficiency


Sergio Bautista-Arredondo, Paola Gadsden, Stefano M. Bertozzi



The total number of people living with HIV worldwide reached its highest level in history at the end of 2005: 40.3 million, double the figure in 1995. In addition, the incidence of new HIV infections has grown every year, with close to 5 million people newly infected in 2005. In that same year less than one fifth as many people were able to start antiretroviral therapy, despite the focus in recent years on expanding treatment. Faced with these statistics it is not surprising that the recent International AIDS Conference called for a redoubling of global prevention efforts.1

1 The central message of the Toronto International AIDS Conference was “we can’t treat our way out of this epidemic”.

In the past decade however, we have witnessed a significant increase in the resources allocated to HIV prevention, a trend that will likely continue in the future. Unfortunately, there remains a large gap between the resources needed and those available. UNAIDS estimated recently that 27 billion dollars will be needed for HIV/AIDS prevention globally in 2005-2007, but only 18 billion (66% of the amount needed) are expected to be available in that period. Further, the estimated need for 2005-2015 is US$122 billion which it is estimated could avert 28 million infections. At an estimated cost per infection averted of US$3,900 and savings of US$4,770 in treatment and care costs per infection averted, such an investment would not just be cost-effective but cost-saving.

It is clear that the majority of experts in the field are convinced that more resources should be invested on prevention. What is more ambiguous Optimizing HIV/AIDS prevention programs: towards multidimensional allocative efficiency for almost everyone, UNAIDS to National AIDS programs in most countries included, is how available resources should be allocated in order to maximize their returns in terms of new infections averted. What particular interventions should be purchased, in what order, for whom and at what scale?

Scholar debate regarding the optimization of resource allocation for HIV prevention has focused primarily on the relative cost-effectiveness of different interventions. Targeting of interventions to different sub-populations has been included implicitly in some of these analyses, to the extent that many interventions are partly defined by their target populations (for example, peer education of sex workers is typically treated as a different intervention than workplace peer education). A third dimension, technical efficiency, has hitherto been ignored when considering allocation of resources, with most analyses assuming that each producer of a prevention intervention is equally efficient. We propose here that optimal policy implementation should explicitly and simultaneously consider the allocation of resources among different sub-populations and among different interventions. Additionally, we argue that explicitly acknowledging the third dimension of program implementation, technical efficiency, is required in order to optimize use of resources for prevention.

In this paper, we provide evidence suggesting that current allocation of resources for prevention efforts is very likely, highly inefficient in all three of the above dimensions. Our perspective is strategic (national – regional) and assumes that policy makers seek to maximize health benefits from an investment in prevention. In any given context, improving performance on any of the three dimensions requires different levels of investment of time and/or money, with different (and not necessarily  corresponding) returns in health benefit. Decisionmakers must therefore balance investment in improving each of the dimensions. We argue that the marginal cost of achieving the same improvement in health can vary substantially among the dimensions, suggesting the need for information and guidance on how to prioritize efforts. Furthermore, investments are needed at different levels. Some are most efficiently done globally, such as trials to evaluate intervention effectiveness; some are national, such as strengthened surveillance systems; and others must be done at the level of the individual provider, such as management training and intervention monitoring.

Conceptual Model

In this section we develop a conceptual framework for optimal resource allocation for HIV prevention. In our model, a level of optimization is determined simultaneously by three dimensions of allocation: a) allocation among interventions, b) allocation among subpopulations and c) allocation among inputs to produce a specific intervention.

The optimization of prevention resources can be viewed as a problem of simultaneous optimal allocation at different levels or dimensions. Our model considers three criteria that must be optimized at the same time. The first dimension, which has been previously discussed in the literature, is allocation among different types of interventions, with special emphasis on cost-effectiveness analyses as criterion for decisions between interventions. The second dimension, allocation among subpopulations, proposes that scarce resources are invested on targeted interventions towards populations at greatest risk. Finally, the third criterion that has not been deeply analyzed before is the allocation among inputs or technical efficiency to produce a specific intervention. More specifically, technical efficiency is the ability of a firm to optimize its production, given the level of inputs available. The production function is the relationship that describes the way in which a firm transforms inputs into products, and for any given level of inputs available there is an optimal amount of product that can be produced. Technical inefficiencies exist whenever less than optimal output is produced. Figure 1 presents graphically the model’s rationality.


Graphically, the structure of the model is made up by three dimensions that together create a cube;

each axis corresponds to one dimension that should be considered for optimal allocation of resources. Every point inside the cube represents a specific allocation of resources given the level reached in the three axes at the same time.

The origin of the cube represents absolute allocative inefficiency in all 3 dimensions and as one go from zero to one hundred percent of allocative efficiency in each axis, the optimal allocation point is achieved. For example, at point A we may have chosen the most cost-effective intervention (e.g. condom distribution), but it is delivered to the wrong subpopulation (e.g. Mother Mary and Mother Cristina) and implemented inefficiently (e.g. condoms are bought but not delivered). Therefore, in this extreme situation there are no HIV infections averted at point A. Point B can be reached, if the right intervention is delivered to the correct subpopulation (e.g. condom distribution targeted to sex workers), but the implementation is completely inefficient (condoms are not delivered), so again, there are no infections averted.


In this paper we argue that optimal prevention programs must simultaneously consider three dimensions for the allocation of resources: (a) allocation among different types of interventions; (b) allocation geographically and among subpopulations; and (c) allocation among inputs to produce a specific intervention. In dimension (a), the aim is the optimization of resources by choosing the most cost-effective prevention interventions for a specific setting. In dimension (b), the epidemiology of the epidemic in a given context is considered in order to target the prevention efforts to those people where the returns in cases averted will be greatest. Finally, for dimension (c), which has previously received little attention, the goal is to allocate the resources among inputs (labor, equipment, supervision, monitoring, incentives, etc) so as to produce the interventions in the most efficient manner.

Cost-effectiveness and cost-utility analyses compare alternative interventions in terms of their relative costs and returns in health. These methodologies, consider dimension (b) and (c) only implicitly. Dimension (b) targeting, is implicit in the measure of effectiveness and in the measure of costs, since those parameters must be estimated from implementing the intervention in a specific setting and targeted to a specific population, even if it is the general population. Similarly, dimension (c) technical efficiency, is implicit in the parameters of the analysis, however cost per unit of service delivered it is almost always assumed to be constant across providers, despite evidence of very large (multiple orders of magnitude) heterogeneity that in many cases would swamp the calculated differences in cost effectiveness among interventions.

Optimizing dimension (a) requires generating data on intervention effectiveness and costs in different contexts. This is important and warrants additional investment, especially of prospective evaluation of large-scale interventions. However, enormous gains could be obtained in the short-term by improving the optimization of dimensions (b) and (c). We argue that targeting effectively and producing services efficiently, neither of which is currently achieved, could result in significant gains.

Enlaces refback

Salud Pública de México es una publicación periódica electrónica, bimestral, publicada por el Instituto Nacional de Salud Pública (con domicilio en Avenida Universidad núm. 655, col. Santa María Ahuacatitlán, Cuernavaca, Morelos, C.P. 62100, teléfono 329-3000, página web,, con ISSN: 1606-7916 y Reserva de Derechos al Uso Exclusivo con número: 04-2012-071614550600-203, ambos otorgados por el Instituto Nacional del Derecho de Autor. Editor responsable: Carlos Oropeza Abúndez. Responsable de la versión electrónica: Subdirección de Comunicación Científica y Publicaciones, Avenida Universidad núm. 655, planta baja, col. Santa María Ahuacatitlán, Cuernavaca, Morelos, C.P. 62100, teléfono 329 3000. Fecha de última modificación: 16 de julio de 2020. D.R. © por el sitio: Instituto Nacional de Salud Pública.