| |
Optimal Allocation of Societal HIV Prevention Resources to Reduce HIV Incidence in the United States
|
| |
| |
Download the PDF here
Download the PDF here
"The model reflects the clinical reality that when everyone infected is virally suppressed, transmission comes to a halt....The largest reduction in HIV incidence over 10 years was observed in the ideal, unlimited-reach scenario that required nearly all persons infected with HIV to be diagnosed promptly and effectively treated to achieve and maintain viral suppression......The optimal allocation in the unlimited-reach scenario was associated with a decrease in 10-year cumulative HIV incidence of 94%, from 331 051 cases to 20 417 cases (or 2000 cases per year on average) compared with the current scenario. At the end of the 10 years, nearly all infections (> 99.7%) were diagnosed; however, among low-risk heterosexuals, only 97% of infections were diagnosed. Nearly all persons with HIV (> 99.7%) were linked to care, prescribed ART, and virally suppressed."
----------------------------------------
Commentary in AJPH
Toward a New Framework for Equity in Epidemic Allocations: Implications of HIV-Prevention–Allocation Misalignment
M. Reuel Friedman
The important and timely article "Optimal Allocation of Societal HIV-Prevention Resources to Reduce HIV Incidence in the United States" by Sansom et al. (p. 150) models different federal and private HIV-prevention resource allocation strategies to prioritize HIV funding through 2027. Modeling exercises are useful starting points for decision making yet may not fully incorporate real-world complexities because of model assumptions and limited quantifiable inputs. Models provided exclude multilevel interventions, policy and structural-level initiatives, and within-group cost differentiation, all key considerations for affecting communities at highest risk for HIV infection.
Relying solely on cost-effectiveness metrics in allocation modeling leads to overreliance on interventions that are the most efficiently deployed, thereby ignoring underserved populations who may require greater cost-per-person investments; in such cases, researchers have argued for a balance between efficiency and equity.6 Although Sansom et al. are unable to differentiate within-group cost-per-person metrics, it is likely that effectively reaching racial/ethnic minorities requires higher upfront costs. We communicate four suggestions for inclusion into the optimal allocations models promoted: (1) nesting analyses so that race/ethnicity, age, gender, and region are used to make allocation decisions; (2) analyzing risk group intersections (e.g., bisexually behaving men, MSM who inject drugs); (3) accounting for the effects of injectable preexposure prophylaxis on HIV-prevention success; and (4) design, refinement, and adoption of an EqEA framework.
The field of HIV prevention and care has never been more advanced or poised for success, yet we cannot succeed if we are myopic to viable, multilevel solutions. Resource allocation models must account for the historic, intersectional mechanisms that maintain HIV inequities among racial/ethnic and sexual and gender minorities. The proposed EqEA framework may help achieve Ending the HIV Epidemic endpoints and offers insights for other infectious diseases, such as directing COVID-19 prevention resources to minority communities wherein SARS-CoV-2 is exacting a disproportionately lethal toll and federal aid formulas for hospitals have large-scale racial biases.7 Adopting equitable allocation strategies will ensure that resources do not remain woefully misaligned and our systems do not exacerbate the well-defined shortcomings of decades of efforts.
---------------------
"The model reflects the clinical reality that when everyone infected is virally suppressed, transmission comes to a halt....The largest reduction in HIV incidence over 10 years was observed in the ideal, unlimited-reach scenario that required nearly all persons infected with HIV to be diagnosed promptly and effectively treated to achieve and maintain viral suppression......The optimal allocation in the unlimited-reach scenario was associated with a decrease in 10-year cumulative HIV incidence of 94%, from 331 051 cases to 20 417 cases (or 2000 cases per year on average) compared with the current scenario. At the end of the 10 years, nearly all infections (> 99.7%) were diagnosed; however, among low-risk heterosexuals, only 97% of infections were diagnosed. Nearly all persons with HIV (> 99.7%) were linked to care, prescribed ART, and virally suppressed."
Optimal Allocation of Societal HIV Prevention Resources to Reduce HIV Incidence in the United States
Stephanie L. Sansom, PhD, MPP, MPH, Katherine A. Hicks, MS, Justin Carrico, BS, Evin U. Jacobson, PhD, Ram K. Shrestha, PhD,
Timothy A. Green, PhD, and David W. Purcell, JD, PhD
Stephanie L. Sansom, Evin U. Jacobson, Ram K. Shrestha, Timothy A. Green, and David W. Purcell are with the Division of HIV/AIDS Prevention, Cen- ters for Disease Control and Prevention (CDC), Atlanta, GA. Katherine A. Hicks and Justin Carrico are with RTI Health Solutions, Raleigh, NC.
Published Online: December 16, 2020
Abstract
Objectives. To optimize combined public and private spending on HIV prevention to achieve maximum reductions in incidence.
Methods. We used a national HIV model to estimate new infections from 2018 to 2027 in the United States. We estimated current spending on HIV screening, interventions that move persons with diagnosed HIV along the HIV care continuum, pre-exposure prophylaxis, and syringe services programs. We compared the current funding allocation with 2 optimal scenarios: (1) a limited-reach scenario with expanded efforts to serve eligible persons and (2) an ideal, unlimited-reach scenario in which all eligible persons could be served.
Results. A continuation of the current allocation projects 331 000 new HIV cases over the next 10 years. The limited-reach scenario reduces that number by 69%, and the unlimited reach scenario by 94%. The most efficient funding allocations resulted in prompt diagnosis and sustained viral suppression through improved screening of high-risk persons and treatment adherence support for those infected.
Conclusions. Optimal allocations of public and private funds for HIV prevention can achieve substantial reductions in new infections. Achieving reductions of more than 90% under current funding will require that virtually all infected receive sustained treatment.
As HIV heads into its fifth decade in the United States, treatment has improved remarkably, so that even those diagnosed in their 20s can achieve nearly normal life expectancy, though at a lifetime cost approaching $500 000.1,2 The annual number of new infections has dropped precipitously from an estimated 130 000 in 1985, but has stalled at about 39 000 a year since 2013.3,4 An estimated 1.1 million persons are living with HIV, but only 86% are aware of their infection, and only 53% are receiving sustained treatment sufficient for transmission-eliminating, life-prolonging viral suppression.5
In 2019, the US Department of Health and Human Services (HHS) proposed the "Ending the HIV Epidemic: A Plan for America" initiative. This federal effort aims to reduce the annual number of new infections to fewer than 3000 or less than 1 per 100 000 population, which, per the World Health Organization, defines epidemic control. HHS plans to achieve this aim by coordinating the programs, resources, and infrastructure of its many agencies and offices.6 In addition to federal agencies, state and local governments and the private sector also provide significant support for HIV prevention and treatment.
Optimal resource allocation methods can help determine the most efficient use of HIV prevention funds to reduce new infections. Previous HIV resource allocation models have examined the most efficient use of funds from 1 or 2 federal agencies.7,8 However, given the ambitiousness of the current initiative to end the HIV epidemic, an evaluation of combined societal funding—public and private—may shed more light on whether and how elimination might be achieved. In this article, we estimate societal funding for HIV prevention and its optimal allocation to curtail HIV incidence in the United States.
DISCUSSION
Models, no matter the complexity or degree of validation, cannot fully represent the dynamics of HIV infection or capture the uncertainties inherent in HIV prevention program implementation. However, when tested carefully, modeling can provide insights into strategies more likely than others to achieve large reductions in HIV incidence. Our results suggest that the current estimated allocation of HIV prevention funds, if maintained over the next 10 years, is likely to be associated with stable incidence rates of approximately 33 100 cases a year. The current allocation spends a large proportion of prevention funding on testing low-risk heterosexuals and on PrEP for high-risk MSM. Although PrEP has been clinically proven to be highly efficacious in preventing acquisition of HIV among those susceptible,14–16 models comparing interventions show that it is less effective in reducing new HIV cases nationally than ensuring that those already infected cannot transmit to others by achieving and maintaining viral suppression with effective ART.9 Our analysis focused on the most efficient use of constant annual prevention funding to prevent new cases of HIV. Only after the most efficient interventions are funded are the remaining dollars shifted to less efficient interventions.
When compared with the current allocation, optimal allocations increased funding for screening populations at high risk of acquiring HIV and for interventions that move people along the HIV care continuum, especially those that support adherence to achieve and maintain viral suppression. The result was a surge in the percentage of persons with HIV whose infection was diagnosed and who were virally suppressed, and a sharp reduction in incidence over 10 years. The model reflects the clinical reality that when everyone infected is virally suppressed, transmission comes to a halt.25
We evaluated 2 optimization scenarios: a limited-reach scenario in which estimates of the maximum percentage of eligible persons who could be reached by each intervention reflected expanded efforts to serve such persons, and an idealistic, unlimited-reach scenario in which all eligible persons could be reached by each intervention. The largest reduction in HIV incidence over 10 years was observed in the ideal, unlimited-reach scenario that required nearly all persons infected with HIV to be diagnosed promptly and effectively treated to achieve and maintain viral suppression.
To more closely mimic how prevention programs are funded, we structured the model to allow for 2 consecutive 5-year allocations rather than a single 10-year allocation. In the limited-reach scenario, the optimal allocation for the first 5 years invested every dollar possible into screening all risk groups except low-risk heterosexuals, linking diagnosed persons to care and treatment, and supporting efforts to achieve and maintain viral suppression. Funding HIV screening and the HIV care–continuum interventions according to the optimal allocation, however, required only 51.6% of available prevention funds because of constraints on the number who could be reached. Sufficient funds thus were left over to allocate enough to support all persons eligible for SSPs. Even then, nearly half of all funds were unallocated, and most (46.0%) went to PrEP for high-risk MSM. In the unlimited-reach scenario, in which all eligible persons could be reached, the model increased allocations to HIV screening and to interventions that moved people along the care continuum, and these interventions absorbed all prevention funds, so that none were available for SSPs and PrEP.
In both the limited- and unlimited-reach scenarios, allocations in the second 5-year period served to shore up gains made in infections prevented during the first 5 years and to shift funds no longer required for screening, linkage, ART prescription, and achieving viral suppression into less cost-effective interventions. For instance, in the limited-reach scenario, even more funding was allocated to PrEP for high-risk MSM. In the unlimited-reach scenario, nearly half of all funds were allocated to screening low-risk heterosexuals, many of whom were unreachable in the limited-reach scenario.
Reductions in incidence over time resulted in reductions in annual HIV care and treatment costs, estimated at $35.2 billion per year on average from 2018 to 2027 in the current allocation scenario. In both the limited- and unlimited-reach scenarios, those costs rose above the average during the first 5 years to pay for the increased number of persons diagnosed and on ART and dropped below it during the final 5 years as the number of new HIV cases decreased. In the limited-reach scenario, they dropped 1.0% ($354 million/year) compared with current costs, and in the unlimited-reach scenario, they dropped 4.3% ($1.5 billion/year), indicating the large potential health care savings when HIV incidence drops.
Limitations
Our analysis has a number of potential limitations. We assumed that moving each PWH along each step of the care continuum required an average expenditure based on published cost data. However, for some people the move may have been costless, whereas for others it may have been more costly than we assumed. Because of lack of data, we did not increase intervention costs in either optimal allocation scenarios as higher percentages of eligible persons were reached or for subgroups that historically have been hard to reach. Better assessments of how intervention costs change for the hardest to reach will be important for understanding the full costs of HIV elimination in the United States.
We did not explicitly account for costs incurred as funds are transferred downstream from agencies to program providers, although these costs can be substantial. However, the Kaiser Family Foundation reported that the federal fiscal year 2018 request for domestic HIV prevention funds was $0.9 billion.26 Considering that our estimated $2.6 billion prevention cost included $1.4 billion in funds for PrEP, typically incurred by the private sector, our public sector funding was approximately $1.2 billion. This amount is reasonably consistent with the Kaiser estimate, although we used very different methods to derive it. We were not able to include some interventions that have been implemented in local communities; we call for additional scientific research to demonstrate the efficacy of these interventions in preventing HIV.
Estimation of Current Funds and Optimal Allocation
We derived total funding for each intervention by multiplying the cost per person served by the annual number served. For diagnosis in particular, we estimated the average cost per diagnosis for each risk group by dividing the costs of screening and diagnosis by the total number of diagnoses. For interventions that move PWH along the HIV care continuum, we assessed the average number of persons reaching each step of the continuum (e.g., linking to care, being prescribed ART, achieving and maintaining viral suppression) annually from 2018 to 2027. We determined the average annual number of persons reaching each step of the continuum by model calibration, so that the modeled number matched published HIV surveillance data on the care continuum in 2010, and either 2015 or 2016 (the most recent data for each step). We projected rates of change between the 2 time periods forward through 2027 in the current allocation. Per-person costs (Appendix, Table A) were based on published studies of interventions.
We based the per-person PrEP cost on the annual 2018 drug cost of $12 59920 plus an annual monitoring cost of $1431.21 The estimated number receiving PrEP in 2018 was 100 292.22 Thus, the estimated total cost of PrEP delivery in the United States was $1.4 billion. We estimated the per-person cost in 2018 for syringe services programs, $234, by using data on the median annual number of syringes used by PWID23 and the cost of injection equipment.19,24 The cost of needle-using equipment per injection itself was derived from the estimate of the total costs of SSPs nationally ($24.5 million) and the number of syringes distributed under those programs (45.9 million).19 All costs in the model were expressed in 2018 US dollars. We assumed that the current allocation of total HIV prevention funding remained fixed from 2018 through 2027 under the current allocation scenario.
Although not included in the optimization, we estimated care and treatment costs by disease stage and progress along the HIV care continuum. We assumed that everyone linked to care received care, and that those prescribed ART received ART, unless they dropped out of care. The per-person annual ART cost used in the model was $25 059.20 We included health-care utilization costs for HIV-related illness.
Using the estimated total prevention funds and the current allocation of those funds, we explored optimal allocations of the funding across interventions and populations in 2 scenarios: a limited-reach scenario in which estimates of the maximum percentage of eligible persons who could be reached by each intervention reflected expanded efforts to serve such persons, and an idealistic, unlimited-reach scenario in which all eligible persons could be reached by each intervention, given sufficient funding. Changes in allocations to interventions under the 2 scenarios slowed the annual rate of movement to related care continuum steps when funding decreased, and accelerated it when funding increased.
Assumptions about the expected proportion of eligible persons who could be reached under the limited-reach scenario fall between the proportion currently reached and 100% in the unlimited reach scenario (Appendix, Table A). To model the effect of prevention funding, which is typically provided in 5-year increments, we estimated the optimal allocation of these funds for the 5-year time periods 2018 to 2022 and 2023 to 2027. We reported the results of the 2 consecutive 5-year allocations that, when combined, produced the greatest reduction in new HIV infections over 10 years.
For our current allocation scenario, we estimated the number of HIV infections that would occur from 2018 through 2027 if the current allocation of total HIV prevention funding remained fixed throughout that period. Then, assuming the same amount of funding, we used optimization techniques (from MATLAB's Optimization Toolbox) and the HOPE model to estimate the 2018–2022 and 2023–2027 allocations that would prevent the most HIV infections from 2018 through 2027. We outline the full optimization formulation in Sections 2 and 3 of the Appendix.
Key outcomes included the optimal allocations to HIV screening, the HIV care–continuum interventions, PrEP, SSPs, and the resulting number of new HIV infections from 2018 to 2027. We projected changes in the proportions of PWH who had achieved each step along the HIV care continuum by 2027, and we noted changes in average annual treatment costs.
Uncertainty and Sensitivity Analyses
We conducted sensitivity and uncertainty analyses, and we present the methods and results of those analyses in Section 4 of the Appendix.
RESULTS
We estimated total 2018 national HIV prevention funding of $2.6 billion (Table 1). Among prevention interventions, we estimated 30.0% currently was allocated to HIV screening, including 25.3% for low-risk heterosexuals and 1.2% for high-risk MSM; 16.7% to interventions that move people along the HIV care continuum, including 5.7% and 9.7% to interventions that support adherence to care and treatment to achieve and to maintain viral suppression, respectively; 52.5% to PrEP, including 6.0% to high-risk heterosexuals and 46.4% to high-risk MSM; and 0.9% to SSPs. When we continued the estimated current allocation through 2027, the model projected a total HIV incidence over that period of 331 051 cases, or 33 100 a year on average (Figure 2; Table 2).
Limited-Reach Scenario
The optimal allocation for the limited-reach scenario was largely influenced by the percentage of eligible persons who we specified as reachable. For both 5-year time periods, the model allocated the maximum amount possible to 10 of the 14 interventions given the limit on the percentage of eligible persons who could be reached, indicating that even more would be spent on those interventions (and less on others) in the absence of those limits. The interventions funded to the maximum level included the screening of all risk groups except low-risk heterosexuals, all HIV care–continuum interventions, and SSPs.
For the first 5 years of the limited-reach scenario, the optimal allocation included 14.9% for screening, 36.7% for the HIV care–continuum interventions, 46.0% for PrEP, and 2.4% for SSPs. Major increases (defined as 5 or more rounded percentage points) in the proportion of prevention funding allocated to a particular intervention between the current and optimal scenarios during the first 5-year time period occurred in screening high-risk heterosexuals (2.4% to 9.3%), interventions that support adherence to care and treatment to achieve viral suppression (5.7% to 13.7%), and interventions that support adherence to care and treatment to sustain viral suppression (9.7% to 19.1%; Table 1). Major decreases occurred in screening of low-risk heterosexuals (25.3% to 0.4%) and PrEP for high-risk heterosexuals (6.0% to 0.0%). In the second time period, a major increase in funding, compared with the first 5-year period in the limited-reach scenario, occurred in PrEP for high-risk MSM (from 46.0% to 59.5%). A major decrease occurred in interventions that support adherence to care and treatment to achieve viral suppression (13.7% to 5.1%).
These consecutive 5-year optimal allocations were associated with a decrease in 10-year cumulative HIV incidence of 69% compared with the current allocation, from 331 051 cases to 103 359 cases (or 10 400 cases per year on average; Table 2). At the end of the 10 years, among all risk groups with the exception of low-risk heterosexuals, 99% of persons with HIV were diagnosed (for low-risk heterosexuals, 85% of those infected were diagnosed), 98% were linked to care, 98% had been prescribed ART, and 86% had achieved viral suppression.
Unlimited-Reach Scenario
In the optimal allocation for the unlimited-reach scenario, of the 14 interventions, 6 were funded for everyone eligible during the first 5 years and 7 during the second 5 years. During the first 5 years, fully funded interventions included screening of high-risk MSM and interventions that increase linkage to care at and after diagnosis, increase ART prescription, and support adherence to care and treatment to achieve and maintain viral suppression. During the second 5 years, interventions that were fully funded were the same as during the first 5 years but also included SSPs.
For the first 5 years of the unlimited-reach scenario, the optimal allocation included 35.4% for HIV screening, 64.6% for interventions that move people along the HIV care continuum, 0.0% for PrEP, and 0.0% for SSPs. Major increases in the proportion of prevention funding allocated to a particular intervention in the first 5 years of the unlimited-reach scenario compared with the first 5 years of the limited-reach scenario included screening high-risk heterosexuals (from 9.3% to 14.1%), screening high-risk MSM (from 2.7% to 16.2%), interventions that increase linkage to care after diagnosis (from 1.3% to 7.2%), interventions that support adherence to care and treatment to become virally suppressed (from 13.7% to 30.8%), and interventions that support adherence to care and treatment to remain virally suppressed (from 19.1% to 24.8%). A major decrease occurred in PrEP for high-risk MSM (from 46.0% to 0.0%).
In the unlimited-reach scenario, compared with the first 5-year time period, major increases in the allocation of prevention funding during the second 5-year time period included screening low-risk heterosexuals (from 0.0% to 48.8%) and SSPs (from 0.0% to 4.5%). Major decreases occurred in screening high-risk MSM (from 16.2% to 7.6%), interventions that increase linkage to care after diagnosis (from 7.2% to 0.1%), and interventions that support adherence to care and treatment to become virally suppressed (from 30.8% to 0.2%).
The optimal allocation in the unlimited-reach scenario was associated with a decrease in 10-year cumulative HIV incidence of 94%, from 331 051 cases to 20 417 cases (or 2000 cases per year on average) compared with the current scenario. At the end of the 10 years, nearly all infections (> 99.7%) were diagnosed; however, among low-risk heterosexuals, only 97% of infections were diagnosed. Nearly all persons with HIV (> 99.7%) were linked to care, prescribed ART, and virally suppressed.
|
|
| |
| |
|
|
|
