Do Figures Lie? AIDS Prevalence, Cost-Effectiveness and Donor Priorities

November 20, 2007

Yesterday in anticipation of World AIDS Day on December 1, UNAIDS released a dramatic update of the numbers they provide every year: the estimated number of people who are HIV infected in each country and in the world at large. As the Washington Post announced this morning, the surprise this time is that the numbers went down from last year, not up.

OK, so instead of 39 million people infected with HIV there are "only" 33 million. What does it mean? Does this matter for resource allocation? Does this mean we should cut back our spending on AIDS?

First of all, the downward adjustments of the total number of people infected has been going on for years. Last year too, UNAIDS decreased its estimates of the number of people infected, but they accompanied that change with the introduction of range estimates. Since the upper end of the range estimate was above previous estimates, the fact that their best guess had been revised downwards went relatively unremarked. This time the size of the downward estimate is larger than ever before. But half of the reduction is explained by the availability for the first time in history of an accurate estimate of HIV prevalence for India. I was part of a team of authors who criticized India's lackadaisical approach to HIV surveillance in 2004. Now that India has finally done a competent nationwide, population-based survey of HIV prevalence, UNAIDS must adjust its estimates. Years of support of national Demographic and Health Surveys by USAID have paid off with population-based estimates of HIV prevalence in the African countries, including Nigeria, Mozambique, Zimbabwe, Kenya and Angola, all of which have been discovered to have less severe problems than was previously thought.

In any case HIV prevention and treatment interventions should be funded because of the benefits from those expenditures in relation to their costs. That is, governments should allocate resources where their use will be cost-effective. But the total number of people living with HIV has only a tenuous connection with the benefits or the cost-effectiveness from most HIV prevention and treatment interventions. For example, the comprehensive review of cost-effectiveness estimates for HIV/AIDS by Bertozzi, Padian and Wegbreit for the Disease Control Priorities Project highlights that peer education of sex workers in Chad was estimated to cost between US$6 and US$30 per HIV infection averted or from 40 cents to US$2 per healthy life year saved. This estimate will not change due to the UNAIDS’ revision of the total number of people living with HIV.

Which interventions will be less cost-effective in situations where the total number of infected people is reduced? There are two types of interventions that must be down-graded. First are those that strive to eradicate the disease. The benefits of eradication are directly related to the total number of cases. Second are programs which are population based or untargeted. For example, screening donated blood at blood banks and hospitals is a crucial public health protection. The lower the percentage of people infected, the more it will cost to find each one. This will not dissuade any health system from a policy of blood screening simply because no system, not even the poorest, is willing to place its transfusion recipients at risk of HIV infection. Systems will simply attempt to fold the cost of the screening test into the cost of hospital procedures which require blood transfusions. However treatment interventions will continue to be as cost-effective as ever, as will targeted interventions, to the extent they are focused on population groups which are indeed at high risk of HIV infection.

So why have advocates for AIDS interventions, including UNAIDS, talked so much about the total number of HIV infected people and so little about the cost-effectiveness of HIV prevention and treatment programs? First, treatment is not cost-effective relative to other underutilized public health interventions in poor countries. Estimates of the cost per life year saved with antiretroviral treatment range from US$200 to US$500 per DALY or higher if second-line treatment is used. Second, the information available to support the claims for the cost-effectiveness of HIV prevention is surprisingly weak. Bertozzi and his co-authors were able to find no research at all on the cost-effectiveness of information campaigns or school based education programs, which form the foundation of any HIV prevention program and absorb most expenditure in countries with low-level epidemics. They found only a handful of studies of peer based programs and none of condom promotion, which is by far the most widely used intervention.

So the scandal here is not about exaggeration of the total number of HIV infections. It's about the failure of international donors over a period of 20 years to conduct enough research on HIV prevention. The world seems to be split between two types of people. On the one hand are those who think the benefits and cost-effectiveness of HIV interventions are so obvious that it would be unethical to divert money from implementation to research. In this well populated category we can place the $250+ million Gates-funded Avahan prevention program in India, which favored rapid start-up of implementation over building in data collection that would have permitted a rigorous evaluation so that those in India and elsewhere could learn how well it was working. On the other hand are people who think HIV prevention is hopeless because "people can’t change their sexual behavior" and who are only willing to consider medical interventions like treatment or a vaccine while opposing HIV prevention research. Where are the donors in the middle, who believe that HIV prevention may be cost-effective, but recognize the gaps in knowledge and are willing to invest to find out exactly how cost-effective they can make it?

Thus, the short answer is "no," the reduction does not matter for resource allocation to either AIDS treatment or to targeted, cost-effective HIV prevention. The good news for treatment is that we are somewhat closer to universal coverage in some countries than we thought we were - but we still have a long way to go in most of them. If the effect of the lower aggregate numbers on HIV prevention is to re-focus attention on the need for cost-effectiveness estimates - of all interventions in all epidemic settings - then this announcement will be a good thing for the battle against AIDS.

See also: The Economist weighs in.


CGD blog posts reflect the views of the authors, drawing on prior research and experience in their areas of expertise. CGD is a nonpartisan, independent organization and does not take institutional positions.