The HIV epidemic continues to be a major public health problem in the United States. Nearly 1.2 million persons are living with the disease [1], and about 50,000 new infections occur annually [2]. In 2010, the White House issued the National HIV/AIDS Strategy, setting goals for decreasing the annual number of new infections by 25% by 2015, increasing access to care and improving the health of persons living with HIV, and reducing HIV-related disparities [3]. Achieving those goals will require sufficient funding for HIV prevention and treatment, and more strategic use of existing funding.

In the United States, HIV prevention programs are primarily funded by the federal, state and local governments and are administered by state and local health departments. In fiscal year 2007, 58% of prevention funding ($337 million) was provided by the federal government, 35% ($205 million) by state and local governments, and 7% ($39 million) by non-governmental entities, such as foundations and pharmaceutical and diagnostic companies [4]. Although there have been modest increases in federal HIV prevention funding since 2007 [5], state governments have experienced deficits during that period that have resulted in reductions in or eliminations of a wide range of programs, including HIV prevention [6]. As of mid-2010, of the 33 states funding HIV prevention [5], eight had reported cuts to HIV testing programs, nine to behavioral interventions, such as those aimed at risk reduction, and seven to partner services [6].

In 2009, an estimated 107,138 persons were living with HIV in California, and 4,981 were newly diagnosed with HIV. Men who have sex with men (MSM) accounted for 73% of HIV prevalence and almost 80% of new diagnoses. Historically, the state of California has allocated a substantial amount of state funds to HIV prevention [4], [7], [8]. In 2005–2007, the state provided 75% ($41 million) of the combined state and federal funding for HIV prevention, while the federal government provided the remaining 25% ($14 million) [7], [8]. However, since fiscal year 2009–2010, the state health department has relied solely on federal funding to support HIV prevention programs [7], [8]. Arnold et al. used a qualitative approach to examine the impact of the California state budget cuts on the provision and access of HIV-related services in Alameda, Fresno, and Los Angeles Counties [8]. In this paper, we quantify the effect of the reduced prevention budget on new HIV infections and their associated treatment costs, and explore the effects of different allocation strategies.

During the pre-cut timeframe, the HIV prevention budget for the 59 jurisdictions was on average $21.8 million, 91% provided by the state and 9% provided by the federal government (Table 2). In those 59 jurisdictions, 51,745 persons, on average, were living with HIV/AIDS and 2,874 new HIV cases were diagnosed annually. MSM accounted for 73% of the prevalence, while IDU and HET each accounted for 13%. Among the new diagnoses, MSM, HET, and IDU accounted for 72%, 17%, and 12%, respectively. During the post-cut year, the budget was funded entirely by the federal government at $5.9 million. The $5.9 million was allocated to 15 jurisdictions that contained 87% of the HIV prevalence in all of the 59 jurisdictions [27] and an annual number of 2,470 new diagnoses (86% of the new diagnoses in all 59 jurisdictions). Among those infected in the 15 post-cut jurisdictions, MSM accounted for 74% of the prevalence, while HET and IDU each accounted for 13%. Among the new diagnoses, MSM, HET, and IDU accounted for 72%, 17%, and 11%, respectively.

During the pre-cut years, 143 agencies received HIV prevention funds; afterwards, 36 agencies received funds. During the pre-cut years, more than 75% of the budget was allocated to risk reduction programs. Afterwards, about 50% went to risk reduction and 50% to testing.

During the pre-cut years, 83,968 tests were performed and an estimated 813 persons (sero-positive rate of 0.97%) were notified of a positive HIV diagnosis annually. Afterwards, the number of tests performed dropped to 53,001, and 465 persons (sero-positive rate of 0.88%) were notified of a positive HIV diagnosis. Seventy-one percent of the tests were provided to HET, 9% to IDU, and 20% to MSM in the pre-cut timeframe; post-cut, 68% of tests were provided to HET, 8% to IDU, and 24% to MSM. Among the positives notified of test results, 63% were MSM, 31% were HET and 6% were IDU in the pre-cut timeframe; post-cut, 71% were MSM, 25% were HET and 4% were IDU.

An average of 11,784 unique clients was served by risk reduction programs annually in pre-cut timeframe, including 2,884 (24%) positive clients and 8,900 (76%) negative clients. Post-cut, the number of unique risk reduction clients decreased to 3,386, including 1,100 (32%) positive clients and 2,286 (68%) negative clients. During the pre-cut years, 47% of HIV-positive risk reduction clients were HET, 5% were IDU, and 48% were MSM; while 65% of HIV-negative clients were HET, 12% were IDU, and 23% were MSM. Those proportions remained the about same following the cuts.

Based on Bernoulli models, we were able to estimate the effect of HIV prevention interventions on annual transmission and acquisition rates among MSM, HET and IDU (Table 3). We estimated that HIV-infected MSM experienced the greatest decrease (0.17) in their transmission rate following a new diagnosis of HIV, and that other transmission categories experienced an annual decrease ranging from 0.014 among HET females to 0.058 among IDU males. The annual transmission-rate decrease following risk reduction for HIV-infected persons ranged from 0.002 among IDU and HET females, to 0.025 among MSM. The annual infection-rate decrease following risk reduction for HIV negative persons ranged from 0.00002 for HET males to 0.005 for MSM.

Based on these calculations of annual transmission and incidence rates, we estimated that 55 additional HIV infections would occur in connection with the first year of the state’s budget cut (Table 4). This represented a 1. 91% increase over the 2,874 infections otherwise expected to incur in the 59 jurisdictions at a societal cost of $20.2 million, compared to the $15.9 million reduction in funding.

For scenario 1, we estimated that, based on the relative cost-effectiveness of testing compared with risk reduction programs, the redirection of a greater proportion of prevention funding to testing over risk reduction, compared with programmatic allocations in pre-cut timeframe, averted 15 infections that otherwise would have occurred (Table 4). In other words, had these funds not been reallocated to focus more heavily on testing, the budget cut would have resulted in a 2.45% increase over those otherwise expected. Under scenario 2, if testing and risk reduction were additionally allocated to transmission categories proportionate to each group’s contribution to HIV prevalence in the 15 funded jurisdictions, 46 additional infections could have been averted, for a 1.66% decrease in the total number of new diagnoses in all 59 jurisdictions, compared to the pre-cut timeframe. Under scenario 3, if funding for HIV prevention were restored to the pre-cut budget of $21.8 million and all funds were allocated among testing and risk reduction in the same proportion as in fiscal year 2009–2010, and services were allocated to transmission categories proportionate to each group’s contribution to HIV prevalence in all 59 jurisdictions, 466 new cases (16%) could be averted each year compared to the pre-cut timeframe.

We presented the results of one-way sensitivity analysis in a tornado graph (Figure 1). In the one-way sensitivity analysis, the most influential parameter was the annual number of sex acts for MSM. If MSM were assumed to have anal sex every day (or 5 times as frequently as the baseline value of 70), the expected number of new infections associated with the budget cut would increase 2.5 times. If MSM were assumed to have anal sex every two weeks (or a third of the baseline value), the expected number of new infections associated with the budget cut would have decreased by almost 50%. Other influential parameters included the per-act transmission probabilities for anal sex, the proportion of sex acts protected by condoms for undiagnosed positive MSM, the effect size of risk reduction for HIV-infected and uninfected individuals, the annual number of sex acts for HET, the proportion of individuals receiving a first-time positive test result out of all of those who received a positive test result, and the annual number of partners of MSM. Other variables tested changed the number of new infections associated with the budget cut by less than 10%. In the probabilistic sensitivity analysis, when we varied all the parameters together in the base case scenario, the number of infections associated with the budget cut ranged from 19.1 to 108.8, and the associated lifetime treatment costs ranged from $6.1 to $42.4 million.

HIV prevention funding for California’s 59 local health departments outside of Los Angeles and San Francisco declined 70%, by almost $16 million in fiscal year 2009–2010. As a result, an estimated 348 fewer persons with HIV were diagnosed and 8,000 fewer clients were served by risk reduction programs. We estimated that 55 more HIV infections occurred because of the first year of budget cuts, generating $20 million in lifetime treatment costs to the health care system, and indicating that California’s pre-cut HIV prevention funding generated more in medical care savings than the cost of prevention programs.

This study systematically analyzed the effects of HIV prevention programs related to testing and risk reduction on annual transmission and acquisition rates among MSM, HET and IDU. It found that MSM, whose transmission and acquisition rates are highest, also achieve the largest reductions in new HIV cases when they are served by prevention programs. For all transmission categories, the diagnosis of a new infection led to the greatest reduction in annual transmission rates, followed by risk reduction for persons living with HIV. This is because among HIV-infected but undiagnosed individuals, awareness of HIV infection has been shown to increase condom use to a greater extent and for a more sustained period than receiving risk reduction services alone [11]–[13]. Individuals diagnosed with HIV also then have the option to seek treatment to reduce their HIV viral loads, protecting their own health and substantially reducing their ability to transmit the infection to others. Generally, risk reduction for HIV-uninfected persons had modest effects, although it was higher for MSM than for HIV-positive HET and IDU of both genders.

Our findings have been generally supported by other studies [28]–[30]. Holtgrave et al. concluded that promoting knowledge of HIV serostatus for undiagnosed infected persons is critical and that prevention services should focus on HIV-infected individuals [28], [29]. Lasry et al. further underscored the importance of greater focus on HIV-infected MSM and IDU, who are at the highest risk of transmitting the disease [30]. Overall, our analyses and those reported by other studies highlight the benefits of allocating resources to testing, prioritizing risk reduction programs to those who already are infected, and directing HIV prevention programs to MSM.

Findings of the sensitivity analysis indicate that the main outcome, the expected number of new infections associated with the cuts, was sensitive to biological and behavioral parameters for MSM, such as their annual number of sex acts, per-act transmission probabilities of anal sex, proportion of all sex acts protected by condom, and number of partners. Our sensitivity analysis underscores the important role of MSM in the HIV epidemic in California (and many other parts of the United States), and the need for more accurate data on their sexual behaviors. HIV prevalence did not play a key role in this analysis, because when resources were focused on 15 jurisdictions with 87% of the new diagnoses reported in all 59 jurisdictions, the positivity rate among those tested declined slightly. In general, however, HIV prevalence is likely to be an important factor in the cost-effective targeting of prevention resources both because of a theoretically higher positivity rate among those tested and because of a greater likelihood of exposure to HIV among uninfected individuals.

Our analysis is subject to several limitations. Our estimate of the effect of the budget cut does not provide a complete picture for the State of California because the analysis does not consider San Francisco and Los Angeles counties, where half of all Californians living with HIV/AIDS reside. However, the budget and programmatic data for San Francisco and Los Angeles were not sufficiently complete to analyze the impact of the budget cuts in those jurisdictions. We believe our overall findings – that the most efficient allocation of HIV prevention funds in the 59 studied jurisdictions would focus on MSM, with testing prioritized over risk reduction, is likely to hold true for Los Angeles and San Francisco, given that 87% of the living cases of HIV and 81% of new diagnoses in these two counties are among MSM. Our analysis only captures the first generation of transmission by those assumed to become infected with HIV as a consequence of the service reductions, making our estimates conservative. We assumed HIV prevention services provided in California achieve the same level of efficacy reported in published studies. In reality, the delivery and effectiveness of programs likely varies across jurisdictions. Reductions in HIV prevention services, particularly those designed to decrease risky sexual behaviors, could have resulted in increases in other sexually transmitted diseases or unintended pregnancy. The data available to us did not allow us to examine those potential effects.

Most of our analytic scenarios did not take into account possible barriers, including cost, to expanding testing programs or reaching greater numbers of MSM (36% of the MSM at risk for HIV infection). To the extent that these barriers exist, our estimates of new HIV infections averted from budget reallocation are too high. The estimates of annual transmission rates, and reductions in transmission rates associated with prevention services, based on Bernoulli process models, rely on self-reported behavioral data, which are subject to recall and social desirability bias, and other uncertain inputs. Parameter uncertainty is reflected in the results of the probabilistic sensitivity analysis, which provides wide intervals around the base case estimates of new cases associated with the budget cuts and corresponding lifetime HIV treatment costs. The probabilistic sensitivity analysis and the scenario analyses do, however, point to relatively more efficient allocation decisions. Although we did not perform a formal budget optimization analysis, our exploration of various budget scenarios suggest where the California state and local health departments avoided additional increases in HIV infections following the severe budget cut, as well as how additional infections might have been prevented. Our analysis did not examine the impact of the simultaneous state cuts to HIV care and treatment programs, which may as well have resulted in additional infections.

Changes in federal allocation of HIV prevention funding [31], and the lack of data from San Francisco and Los Angeles counties and on state budget cuts to HIV treatment programs, will likely make validation of our impact estimates challenging. However, we recommend that the state Office of AIDS continue to closely monitor trends in new HIV diagnoses and incidence and use these data to better understand the effects of the budget cut and to further refine future funding allocations.

One estimate suggests that achieving National HIV/AIDS Strategy prevention goals by 2015 will require an additional annual investment of $420 million [32]. However, overall funding for HIV prevention has decreased due to state budget cuts in spite of modest increases in federal funding [5], [6]. While the National HIV/AIDS Strategy calls for a 25% decrease in the annual number of new HIV infections by 2015, we estimate that California’s budget cut resulted in a 2% annual increase in new cases in the 59 jurisdictions studied. Restoring state HIV prevention funding to these jurisdictions and allocating resources more strategically to the most cost-effective prevention programs and to populations at highest risk for transmission could have substantial public health benefits and achieve considerable progress toward meeting (but not fully achieving) national goals. Although reductions in funding for HIV prevention have the potential to increase HIV infections, analyses like those described in this paper and other resource allocation tools can help program planners maximize the number of HIV infections prevented given available funding. Important opportunities exist to prevent even more HIV infections and reduce HIV treatment costs by careful allocation of existing HIV prevention funds. Failure to respond to these opportunities in a strategic and timely manner will lead to new HIV infections that could have been prevented, increased health care costs, and an incalculable burden on the lives of the men and women who become infected with HIV, their families and communities.