Estimating the effectiveness and efficiency of US HIV prevention efforts using scenario and cost-effectiveness analysis

Scenario and cost-effectiveness analyses were used to estimate the effectiveness and efficiency of HIV prevention activities in the USA (1978-2000). Under four conservative scenarios on the course of the HIV epidemic that might have occurred had there been no prevention efforts, we estimate that prevention activities averted 204 000-1 585 000 infections at a cost of between US$49 700 and US$6400 per infection prevented (less than the medical costs of treating a case of HIV disease).

Incident HIV infections in the United States reached a peak in the mid-1980s, and then declined to approximately 40 000 infections per year by 1990 (Fig. 1) [^1,2]. This can be viewed as a sign of HIV prevention success, in that the number of incident infections dropped dramatically from the peak level. It might also be viewed as a failure, in that there have been approximately the same number of new infections per year for over a decade. However, both views are partly flawed. To estimate the effectiveness of US HIV prevention efforts, one would have to compare the HIV incidence curve, as experienced, with the HIV incidence curve that would have occurred had there been no HIV prevention activities; the area between these two curves represents the number of HIV infections prevented. By combining this information with an estimate of the cost of national HIV prevention efforts to date, a crude overall cost-effectiveness ratio can be calculated. This ratio can provide a general estimate of the efficiency of the HIV prevention efforts.

Fig. 1

The challenge in developing even general estimates of effectiveness and efficiency of HIV prevention efforts is that the HIV incidence curve that would have occurred had the prevention activities not been in place is impossible to measure. However, we can use scenario analysis to estimate the number of infections prevented under a wide range of uncertain yet plausible possibilities. Here, we utilize scenario and cost-effectiveness analysis [^3,4] to assess: (i) the order of magnitude of the number of HIV infections prevented in the USA from 1978 to 2000; and (ii) whether the national HIV prevention effort could be considered to be generally cost-effective.

Four scenarios were constructed (Table 1). Without an HIV prevention program in place, the course of the epidemic would be determined by (at least) the natural history of the disease, mortality rates, patterns of risk behavior, disease co-factors, and any spontaneously occurring changes that individuals undertook simply as a function of seeing others become ill and, usually, die (note that without any HIV prevention program in place, individuals would not necessarily know why others were falling ill or what to do to prevent the illness). Therefore, the US HIV incidence curve without a program in place might have followed a pattern seen more often in developing regions with less intensive HIV prevention programs, in which HIV incidence continues to rise for years, before levelling off as the pool of susceptible individuals becomes smaller (following an S-shaped curve) [^5,6]. Our four scenarios are conservative in that none assumes an incidence rate greater than the peak year 'observed', and most incorporate an incidence drop.

Table 1

The first scenario considered a possible epidemic that reached a peak of 161 000 HIV infections per year and stayed flat at that level. The second scenario posited a decline after the peak but then a flattening at 123 000 HIV infections per year. This figure of 123 000 incident infections is based on: (i) estimates of the sizes of US populations of injection drug users, men who have sex with men, and high-risk heterosexuals (according to the 1996 National Household Survey of Drug Abuse) [⁷]; (ii) the size of the US population (many of whom are at very low risk of infection) [⁸]; and (iii) estimates of the HIV incidence rates for these high- and low-risk communities [⁹]. For each population we took the estimate of the size of that population, multiplied it by the incidence rate estimate, and then summed the products across all four populations (taking care to avoid double counting).

In the HIV epidemic, as observed in the USA, the first year past the peak of HIV infections had 12.4% fewer infections. The third scenario describes an annual decline of HIV infections at one-half of the observed 12.4% rate of decline for the first year past the incidence peak. The fourth scenario considers an annual decline at exactly the 12.4% rate observed for the first year past the peak.

Fig. 1 shows the estimated HIV incidence curve experienced in the USA [^1,2] and the estimated incidence curve from scenario 3. The cross-hatched area between the curves is the estimate of HIV infections prevented. Similar figures could be drawn for the other scenarios. Table 1 displays the number of HIV infections prevented under each scenario. To gain a perspective on these effectiveness estimates, they range from approximately the size of the population of Baton Rouge, Louisiana, to the size of the population of Philadelphia, Pennsylvania [¹⁰].

Up to and including fiscal year 2000, the federal Centers for Disease Control and Prevention (CDC) had a total budget for HIV prevention of approximately US$7.2 billion (summed over all years as raw dollar figures unadjusted for inflation) [^2,11]. However, this is not the entire national contribution to HIV prevention; other federal, state and private contributions must be added. We estimated that an additional 8.1% was spent on other federal programs [¹²], another 27.0% on state-funded efforts [¹³], and a further 5.6% from private sources [¹⁴]. (There is uncertainty as these additional percentages were based on limited data available for a short time period and were extrapolated across the time-frame of the epidemic; furthermore, these figures are budgetary and may or may not reflect the true costs of the programs). Including all of these sources of funding, the estimated expenditure on HIV prevention programs in the USA to fiscal year 2000 is US$10.1 billion.

Table 1 shows the results of dividing this gross budget estimate by the number of HIV infections prevented under each scenario. In all cases, the gross cost per HIV infection averted is well below the discounted lifetime treatment costs of one case of HIV disease (approximately US$154 000-195 000 currently; approximately US$56 000 before the era of highly active antiretroviral therapies) [^15,16]. Therefore, the findings indicate a strong likelihood that the HIV prevention efforts to date in the USA have resulted in an overall cost saving to society.

Clearly, this analysis is limited by multiple sources of uncertainty; and therefore, sensitivity analysis is important to demonstrate the robustness of the findings to changes in the values of input parameters. However, because the formulae used in this analysis are straightforward, the results of the sensitivity analyses are easy to anticipate and we do not report them here in detail (for example, increasing the cost by 100% increases the cost-effectiveness ratio by 100%). Furthermore, it is not the specific numerical results but rather the general finding that is of interest. The analysis indicates that HIV prevention efforts in the USA may well have prevented a number of infections roughly equal to the size of a moderate to large US city, and probably resulted in overall cost savings to society. These estimates and this analytical framework will hopefully foster further research and refinement. Additional interesting analyses would compare the epidemic as observed with what may have occurred had there been improved resource allocation, policymaking and program implementation.

Acknowledgements

The author would like to thank Drs Carol Scotton and Steven Pinkerton, as well as an anonymous reviewer, for their very helpful comments on a previous draft of this manuscript. The content of this paper was presented in part by the author as a presentation at the Second National HIV Prevention Conference in Atlanta, GA, 12-15 August 2001.