Improving health equity and ending the HIV epidemic in the United States: a distributional cost-effectiveness analysis in six cities

Abstract

Background

In the United States, Black and Hispanic/Latinx individuals continue to be disproportionately impacted by HIV. Applying a distributional cost-effectiveness framework, we estimated the cost-effectiveness and epidemiological impact of two combination implementation approaches to determine the approach that best meets objectives of improving population health and reducing racial/ethnic health disparities.

Methods

We adapted a dynamic, compartmental HIV transmission model to characterize HIV microepidemics in six US cities: Atlanta, Baltimore, Los Angeles, Miami, New York, and Seattle. We considered combinations of 16 evidence-based interventions to diagnose, treat and prevent HIV transmission according to previously-documented levels of scale up. We then solved for optimal combination strategies for each city, with the distribution of each intervention implemented according to 1) existing service levels (proportional services approach) and 2) the racial/ethnic distribution of new diagnoses (between Black, Hispanic/Latinx, and white/other individuals; equity approach). We estimated total costs, quality-adjusted life years (QALYs), and incremental cost-effectiveness ratios of strategies implemented from 2020–2030 (health-care perspective; 20-year time horizon; 3% annual discount rate). We estimated three measures of health inequality (Between-Group Variance, Index of Disparity, Theil Index), incidence rate ratios, and rate differences for the selected strategies under each approach.

Findings

In all cities, optimal combination strategies under the equity approach generated more QALYs than those with proportional services, ranging from a 3.1% increase (95%CrI:1.4%-5.3%) in New York to more than double (101.9% [75.4%-134.6%]) in Atlanta. Compared to proportional services, the equity approach delivered lower costs over 20 years in 5/6 cities; lower cost differences ranged from $22.9M (95%CrI: -$5.3M-$55.7M) in Seattle to $579.8M (95%CrI: $255.4M-940.5M) in Atlanta. The equity approach also reduced incidence disparities and health inequality measures in five of six cities.

Conclusion

Equity-focused HIV combination implementation strategies that reduce disparities for Black and Hispanic/Latinx individuals can significantly improve population health, reduce costs, and drive progress toward Ending the HIV Epidemic in America.

Introduction

Despite great progress towards HIV epidemic control in the United States, racial/ethnic minorities continue to be disproportionately impacted by HIV.¹ In 2018, Black/African American (Black) and Hispanic/Latinx individuals accounted for 64% of all people with HIV (PWH) yet only 31% of the total US population.² Racial/ethnic disparities in health and HIV are the contemporary ramifications of historic structural racism that have manifested in modern health and social policy.³ Racial/ethnic inequities are seen in housing, employment, poverty, and health insurance coverage.⁴ Combined with increased stigma and medical mistrust,⁵ these factors contribute to racial/ethnic disparities in engagement at every step of the HIV continuum of care,⁶ beginning with lower awareness and uptake of preventive interventions⁷ and persisting with lower engagement in antiretroviral therapy (ART)⁸ and viral suppression. These disparate levels of healthcare access perpetuate racial/ethnic HIV disparities, which necessitate health policy change to correct the course of the HIV epidemic in the US.

The nation’s Ending the HIV Epidemic (EHE) strategy has ambitious goals to reduce HIV incidence by 75% nationally by 2025, and 90% by 2030.⁹ The EHE aims to achieve these goals by targeting federal resources towards priority county- and state-level jurisdictions, though the initiative has no explicit targets or indicators for reducing racial/ethnic disparities. In a series of recent articles,^10–15 we conducted cost-effectiveness analyses of HIV combination implementation strategies in six cities accounting for 24% of all people living with HIV in the US, applying conventional health-maximizing principles. In these analyses, we scaled-up implementation of evidence-based interventions among racial/ethnic and HIV transmission risk groups proportional to their existing levels of access, representing current social and structural constraints on access to care, thus implying higher levels of scale-up for groups with greater baseline access. However, even with proportional scale-up to near-ideal levels (i.e., interventions reaching 90% of their target populations), we found that EHE goals will likely not be reached by 2030 without further reductions in new transmissions among Black and Hispanic populations.¹³ Others have argued that allowing proportional scale up of interventions in settings with substantial inequities in current access to HIV prevention and treatment between racial/ethnic groups¹⁶ may reinforce racial disparities in HIV incidence.¹⁷

Reducing HIV-related health disparities is one of four goals of the recently updated National HIV Strategic Plan,¹⁸ signalling a need for implementation strategies focused on improving equitable access to HIV prevention and care. Distributional cost-effectiveness analysis (DCEA) is a new methodological framework that combines the dual objectives of maximising health and reducing health inequities when evaluating population health strategies.¹⁹ DCEA has been used to compare alternative public health policies that seek to increase total population health and improve health equity between demographic subgroups (e.g., socioeconomic status, race/ethnicity) by quantifying the distribution of health outcomes across groups. To this end, we seek to build on previous economic analyses of HIV combination implementation strategies by comparing the impact of two policy approaches to reach the EHE initiative targets. We apply a DCEA framework to estimate the distribution of health gains and determine the approach that best addresses the dual objectives of maximizing population health and reducing racial/ethnic disparities in HIV in the US.

Methods

Model description

We used a dynamic, compartmental HIV transmission model calibrated to replicate the city-level HIV micro-epidemics in Atlanta, Baltimore, Los Angeles (LA), Miami (Dade County), New York City, and Seattle (King County).^{10, 11, 15} Focal cities were selected for their diverse demographics and ranging disparities in HIV diagnoses (Table 1).

Table 1.

Demographic and HIV-related characteristics in the six US focal cities.

	Atlanta (GA)	Baltimore (MD┼)	Los Angeles (CA┼)	Miami (FL)	New York City (NY┼)	Seattle (WA┼)
Total Adult Population in 2016	3,812,143	1,874,601	6,964,983	1,821,311	5,865,683	1,503,497
Adult population in 2016**
Black or AA	1,336,469 (35%)	553,665 (30%)	568,815 (8%)	296,354 (16%)	1,304,687 (22%)	95,550 (6%)
Hispanic or Latinx	391,265 (10%)	102,495 (5%)	3,385,948 (49%)	1,246,583 (68%)	1,703,286 (29%)	137,818 (9%)
Non-Hispanic white/other	2,084,409 (55%)	1,218,441 (65%)	3,010,220 (43%)	278,374 (15%)	2,857,710 (49%)	1,270,129 (84%)
Projected adult population by 2040
Black or AA	1,800,171 (34%)	650,255 (35%)	460,413 (7%)	295,396 (14%)	1,312,629 (22%)	131,365 (8%)
Hispanic or Latinx	1,037,139 (20%)	154,281 (8%)	3,609,531 (53%)	1,600,165 (76%)	1,995,693 (33%)	274,266 (16%)
Non-Hispanic white/other	2,386,944 (46%)	1,055,221 (57%)	2,783,824 (41%)	217,487 (10%)	2,738,171 (45%)	1,328,009 (77%)
Proportion of HIV diagnoses in 2018***
Black or AA	73.0%	75.6%	22.8%	28.8%	45.9%	22.5%
Hispanic or Latinx	7.6%	6.1%	49.2%	59.2%	36.4%	17.9%
Non-Hispanic white/other	19.5%	18.3%	28.0%	12.0%	17.8%	59.6%
Proportion of non-elderly without health insurance in 2019, state-level**
Black or AA	14.8%	6.2%	6.4%	17.3%	6.5%	9.0%
Hispanic or Latinx	34.2%	21.4%	13.7%	21.2%	11.3%	18.8%
Non-Hispanic whiteb	12.4%	3.8%	5.1%	13.3%	3.9%	5.5%
Poverty status in 2018, %**c
Black or AA	15.2%	18.0%	20.7%	24.4%	20.0%	23.9%
Hispanic or Latinx	17.7%	14.2%	16.8%	15.1%	23.7%	14.7%
Non-Hispanic whiteb	6.7%	5.7%	9.7%	10.7%	11.0%	6.4%

AA African American

^*Study boundaries may not reflect county or metropolitan statistical area boundaries, and are detailed in a previous publication [11].

^**Data sourced from the US Census Bureau’s 2018 American Community Survey which reflects on the past 12 months

^***Data sourced from local surveillance reports. Diagnoses for Atlanta were from 2017, the most recent reported year.

^bOther races not shown.

^cPoverty status is defined by the US Census Bureau as annual income compared to a threshold determined by family size and age of householder, updated to account for the cost of living.

^┼State has adopted the Affordable Care Act

The model tracked HIV-susceptible individuals through infection, diagnosis, treatment with ART and ART dropout. In each city, the adult population (aged 15–64) was stratified by biological sex (male, female), HIV risk group (men who have sex with men [MSM], people who inject drugs [PWID], MSM-PWID, and heterosexuals), race/ethnicity (Black, Hispanic/Latinx, and non-Hispanic white/other [white]) and sexual risk behavior level (high- vs. low-risk). The model accounted for heterogeneity in the risk of HIV transmission (via assortative and proportional partnership mixing),¹⁵ aging (via differential maturation and mortality rates), and observed racial/ethnic inequities in access to health and prevention services, including HIV testing, ART, syringe service programs (SSP), medication for opioid use disorder (MOUD), and targeted pre-exposure prophylaxis (PrEP) for high-risk MSM. The model parameters were informed by an extensive evidence synthesis, detailed elsewhere.¹⁰

We considered 16 evidence-based interventions with established effectiveness data and promising scalability according to three pillars of the EHE strategy to prevent (SSP, MOUD and targeted PrEP), diagnose (six distinct HIV testing interventions), and treat HIV (six distinct ART initiation, retention and re-initiation interventions).¹¹ We considered only levels of scale-up that had been documented in the public domain, thus constraining our focus to a single level of scale-up for each intervention. The costs attributable to each intervention included costs of implementation, delivery, and sustainment. Details on the methods used to estimate the effectiveness, scale of delivery, and costs of each individual intervention have been documented elsewhere.¹¹ To characterize the scenario of maintaining baseline service levels (hereon referred to as the ‘status quo’), we updated the service levels used in prior analyses using most recent reports (up to 2019) for PrEP, MOUD, SSP, HIV testing levels, and ART engagement (Supplemental Material and Supplement Table 1, pp 2–4).

Implementation policy approaches

To demonstrate the impact of an equity-focused implementation policy, we compared two scale-up approaches: proportional to baseline service levels (proportional services approach) and proportional to new HIV diagnoses (equity approach). The proportional services approach assumed that increases from existing service levels were implemented proportionally across race/ethnic groups, implying a higher scale of delivery for groups receiving greater service levels at baseline (Figure 1). This policy represents an estimate of the expected level of scale-up that can be achieved within current social and structural constraints on access to care, and has been detailed previously.¹⁴ The equity approach was grounded in the equity principle of proportionate universalism, which suggests all groups receive care, with the amount of additional care being proportional to need or disparity.²⁰ The approach involved increases from existing service levels that were proportionate to the distribution of new HIV diagnoses in 2018 (2017 for Atlanta), stratified by race/ethnicity and sex (Table 1).

Figure 1. Schematic representation of the proportional services and equity approaches to scaling-up interventions from current status quo levels, illustrated with targeted pre-exposure prophylaxis (PrEP) scale-up for high-risk men who have sex with men in Baltimore.

Both policy approaches increased the overall level of service delivery by an equal amount with the key difference being the distribution across racial/ethnic group. Neither approach decreased service levels or coverage for any racial/ethnic group compared to the status quo. We used a similar approach for each individual intervention in each city with increases in services modifying the corresponding model parameters for each racial/ethnic group (when applicable). Detailed information on coverage by racial/ethnic group for each intervention in each city under the status quo, proportional services approach and equity approach are available is Supplement Tables 1–3.

Both policy approaches increased the overall level of service delivery for each intervention by equal amounts (e.g., the additional number of people receiving an HIV test, initiating ART or PrEP); the key difference was the distribution across racial/ethnic groups for each sex. Neither approach decreased absolute service levels for any racial/ethnic group compared to the status quo. All service levels provided in Supplement Tables 1–3 (pp 4–6). Finally, in the absence of evidence on costs for implementing an equity-focused approach, we assumed no additional costs for such additional efforts.

Cost-effectiveness analysis

For each city and scale-up approach, we estimated the total costs and QALY gains for all combinations of the 16 interventions (excluding combinations that would not practically be implemented jointly, such as two HIV testing interventions delivered in primary care), for a total of 23,040 unique combinations. We constructed health production functions specific to each approach, representing the combination strategies that provide the greatest incremental health benefits for a range of investment levels. We then identified the combination strategy producing the greatest health benefits while remaining cost-effective (referred to as ‘optimal’ hereafter), by estimating incremental cost-effectiveness ratios (ICERs) for successive combination strategies along each health production function, compared with the next most costly strategy. Although no explicit threshold exists in the US, we defined cost-effective strategies as those with an ICER below $100,000/QALY, consistent with efforts in 2016 to approximate the threshold according to the opportunity costs of displacing existing services.²¹ All combination strategies were sustained for a period of 10 years (2020–2030) to match EHE targets.⁹ Model outcomes were estimated over a 20-year time horizon (2020–2040) to capture long-term individual health benefits and second-order transmission effects. We conducted the cost-effectiveness analysis from the healthcare perspective, adhering to best practice guidelines of the Second Panel on Cost-Effectiveness in Health and Medicine.²² Model outcomes included quality-adjusted life-years (QALYs) and total costs (2018$US), both reported using a 3% annual discount rate.²³ We performed probabilistic sensitivity analyses using the 2,000 best-fitting calibrated parameter sets for each city, where non-calibrated parameters were sampled from previously developed distributions for each parameter.¹⁵ We present resulting median estimates with 95% credible intervals.

Inequality analysis

We assessed the distribution of cumulative QALY gains (per 100,000 population at 2030) and new HIV infections across racial/ethnic groups for the optimal combination strategy under each of the policy approaches for each city.

In recognition of the implicit value judgments underlying different measures of inequality in health distributions,²⁴ we calculated three summary measures of health inequality suitable for nominal groups: Between-Group Variance, the Index of Disparity, and the Theil Index.²⁵ Between-Group Variance is an absolute, weighted measure, whereas the Theil Index is a relative, weighted measure of inequality. Both indices weight all individuals equally and therefore implicitly assume that inequalities are more significant when they affect a larger proportion of the total population.²⁴ In contrast, the Index of Disparity is a relative, unweighted measure of inequality. The units of analysis are the racial/ethnic groups; all groups are weighed equally regardless of relative population size. Therefore, this measure implicitly assumes that inequalities that disadvantage a smaller group are equally significant as those that disadvantage a larger group. Further details on calculations for each summary measure are provided in Supplement Table 4 (p 7). All measures produce positive values, in which zero indicates no inequality and higher values indicate higher inequality.

We generated health-disparity impact planes to illustrate potential trade-offs between maximizing total health benefits and reducing health inequality. For each policy option, we quantified total health in terms of QALYs incremental to the status quo. We then calculated change in inequality as the percentage change between each policy scenario compared to the status quo, for each summary inequality measure.

To assess disparities in new HIV infections between racial/ethnic groups, we calculated pairwise incidence rate ratios and rate differences between Black and Hispanic versus white populations for 2020 and 2030. The rate ratio, a relative measure, values strict egalitarianism, as it does not contextualize total incidence rates.²⁴ The rate difference is the absolute difference between incidence rates, capturing racial disparities and level of HIV incidence. Neither measure weights the disparity by population size. Finally, to compare the impact of policy options on progress towards EHE targets,⁹ we estimated new HIV infections for 2030 under the status quo and each scale-up approach.

Sensitivity Analysis

Given the likelihood of higher costs for engaging underserved and hardly reached populations under the equity approach, we conducted a probabilistic threshold sensitivity analysis on the costs attributable to interventions. For each of the city-level optimal strategies under the equity approach, we estimated the net health benefit compared to the proportional services approach, accounting for health opportunity costs and increasing intervention costs (implementation, delivery, sustainment) when services were delivered to Black or Hispanic/Latinx individuals. Net health benefits were defined as incremental health benefits (QALYs) minus health opportunity costs (incremental costs divided by the decision threshold ($100,000/QALY)). We considered intervention costs that increased by factors ranging from one to ten to determine the threshold at which the equity approach may no longer provide additional net health benefits compared to the proportional services approach.

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

RESULTS

Optimal strategies under each implementation policy approach

Under both equity and proportional services approaches, all six cities’ optimal strategies included expansion of MOUD (both buprenorphine and methadone), EMR testing reminders, nurse-initiated rapid testing and case management for ART initiation (Supplement Figure 1, p 15). Expansion of opt-out testing (ER and primary care) and care coordination was not included in any city or implementation policy approach.

Optimal combination strategies derived from the proportional services and equity approaches differed in all but Miami and NYC (Supplement Figure 1, p 15). For Atlanta, the equity approach added three treatment engagement interventions (Electronic Medical Records (EMR) ART engagement reminders, expanded enhanced personal contact and re-linkage programs) but excluded SSP expansion. Baltimore added Electronic Medical Records (EMR) ART engagement reminders and SSP expansion. PrEP expansion was added in Los Angeles’s optimal strategy under the equity approach, while RAPID ART initiation was no longer included in Seattle’s optimal strategy under the equity approach.

Health benefits and costs

The equity approach was estimated to generate more health benefits for all cities compared to the proportional services approach (Table 2). Total additional QALY gains ranged from 3.1% (95% Credible Interval (95%CrI): 1.4%-5.3%) in NYC to more than double (101.9%; 75.4%-134.6%) in Atlanta. These relative QALY gains were largest among Black individuals in Atlanta (from 5,296 to 12,620 QALYs, a 138.3% increase), Baltimore (2,705 to 4,578; 69.2%) and Seattle (371 to 539; 45.4%), whereas Hispanic/Latinx individuals had the largest relative increases in Baltimore (245 to 1,295; 428.1%), Seattle (310 to 615; 98.6%) and Los Angeles (11,167 to 13,319; 19.3%). Changes in incremental health benefits across racial/ethnic groups were largest in Atlanta, Baltimore, and Seattle. For instance, Black individuals in Atlanta now received 77.1% of all incremental health benefits compared to 65.3% previously). Conversely, the share of health gains among racial/ethnic groups varied no more than 3% between policies in Los Angeles, Miami, and NYC.

Table 2.

Incremental costs and quality-adjusted life year (QALY) gains of optimal combination implementation strategies under alternate scale-up approaches in six US cities (2020–2040)^*.

	Atlanta	Baltimore	Los Angeles	Miami	New York City	Seattle
Status Quo Cost (billions)	271.1	131.6	423.0	128.7	439.3	94.7
	(250.7, 292.6)	(114.6, 151.5)	(378, 472.7)	(119.4, 138.5)	(382.6, 500.6)	(83.7, 106.6)
Status Quo QALYs (millions)	65.8	27.6	101.4	28.9	88.1	23.0
	(65.7, 65.8)	(27.6, 27.7)	(101.3, 101.5)	(28.8, 28.9)	(88, 88.2)	(23, 23)
Incremental Cost compared to status quo (US$, millions)
Proportional Services Approach	−190.5	58.2	535.5	−879.7	1,359.7	123.9
	(−366.1, −28.9)	(−46.5, 172.8)	(105.5, 945.5)	(−1555.1, 160.3)	(847.8, 1894)	(71.9, 190.7)
Equity Approach	−781.4	−185.0	961.2	−968.1	1,217.0	99.6
	(−1207.1, −392.7)	(−332.7, −13.3)	(410.4, 1428.1)	(−1669.6, 131.2)	(689.9, 1756)	(43.6, 173.5)
Incremental QALYs compared to Status Quo
Proportional Services Approach	8,168.7	6,129	24,767	23,313	27,081	2,981
	(6449, 10279)	(5045, 7298)	(18573, 31706)	(15350, 30515)	(20297, 34268)	(2363, 3670)
Equity Approach	16,497	9,155	31,586	24,728	27,921	3,498
	(13033, 20377)	(7677, 10786)	(25228, 39399)	(16262, 32603)	(21003, 35381)	(2781, 4309)
Equity Approach compared to Proportional Services Approach
Incremental Cost (millions)	−579.8	−244.6	420.3	−77.6	−140.8	−22.9
	(−940.5, −255.4)	(−354.7, −123.8)	(191.9, 606.5)	(−169.2, −10.4)	(−191.9, −93.7)	(−55.7, 5.3)
Incremental QALYs	8,270	3,032	6,594	1,389	818	522
	(6,129, 10,956)	(2,243, 3,984)	(4,197, 11,453)	(686, 2,329)	(350, 1,496)	(368, 714)

QALYs quality adjusted life-years.

^*Results are the median (with 95% Credible Intervals) from 2,000 probabilistic sensitivity analysis runs using a 3% discount rate and presented in 2018 $US.

Compared to the status quo, optimal strategies under both scale-up approaches were cost-saving (i.e., implementing either approach for 10 years is estimated to generate more health benefits and cost less than maintaining baseline service levels) over 20-years (2020–2040) in Atlanta and Miami (Table 2). Under the equity approach, the optimal strategy for Baltimore was also cost-saving compared to status quo.

When comparing optimal strategies under each scale-up approach, the equity approach was estimated to generate more QALYs and lower costs than the proportional services approach in all cities but Los Angeles over the study period. The inclusion of PrEP expansion in the equity approach in Los Angeles increased both health benefits and costs compared to the optimal strategy under the proportional services approach. Otherwise, the equity approach was estimated to incur lower incremental costs across remaining cities over the study time horizon, ranging from $22.9M (95%CrI: -$5.3M-$55.7M) in Seattle (18.5% lower than the proportional services approach) to $579.8M (95%CrI: $255.4M-$940.5M) in Atlanta (four times lower than the proportional services approach).

Measures of inequality and incidence

The equity approach resulted in lower levels of health inequality across racial/ethnic groups for all three measures of inequality in all but Los Angeles (Supplement Table 5, p 8). As a result of the projected demographic changes by 2040 in LA (Table 1), health inequality increased from status quo under both policy options (Supplement Figure 2, p 16). This increase was driven by larger per capita QALY gains for Black individuals compared to others.

In all cities, Hispanic/Latinx-white HIV incidence rate ratios in 2030 were projected to be the lowest under the equity approach, nearly reaching parity in Baltimore, Los Angeles and Miami (Figure 2). Black-white incidence rate ratios in 2030 were also projected to be the lowest under this policy in all cities except for Los Angeles. The largest relative reduction for the Black-white incidence rate ratio was in Baltimore (from 5.87 in 2020 to 2.84 in 2030). Incidence rate differences were estimated to be lowest under the equity approach in all cities. The sole exception was Hispanic/Latinx-white differences in Miami (6.4 under equity compared to 6.1 under proportional services).

Figure 2. Estimated HIV incidence rate ratios with 95% credible intervals and incidence rate differences (IRD) at baseline and at 2030 under three policy approaches, in six US cities.

Projected HIV incidence in 2030 was lower in every city when services were scaled-up under the equity approach compared to the proportional services approach (Figure 3). Furthermore, the equity-focused approach achieved incidence reductions nearly meeting EHE targets in Baltimore.

Figure 3. Estimated number of new HIV infections in 2020 and in 2030 under the status quo and the proportional services and equity approaches, by race/ethnicity, with total percentage change*, in six US cities.

*The percentage change in new HIV infections is compared to 2020 estimates for each city.

Sensitivity Analysis

We found that the equity approach would deliver net health benefits in Atlanta, Baltimore and Seattle across the range of incremental intervention costs that we considered (Figure 4). Relaxing the assumption of equivalent costs, we found that delivering the selected interventions under the equity approach could cost from three times (Los Angeles) to five times more (Miami) and still produce positive net health benefits.

Figure 4. Estimated net health benefit* over 2020–2040 of the optimal strategies under the equity approach compared to the proportional services approach when interventions costs for the equity approach among Black and Hispanic/Latinx individuals are increased, in six US cities.

*Defined as incremental quality-adjusted life years minus health opportunity costs (incremental costs divided by the cost-effectiveness threshold of $100,000/QALY).

DISCUSSION

This modeling study suggests that an HIV combination implementation strategy designed to reduce racial/ethnic inequities in healthcare access could produce substantially greater health benefits at lower long-term costs in the US. Explicitly reducing racial/ethnic inequities in HIV service access may not only be an effective strategy to reach the EHE initiative’s targets and improve racial health equity, it may also be a more efficient one. As domestic funding for HIV has plateaued over the past decade,²⁶ additional upfront investment will be required to achieve EHE goals, with potential cost-savings occurring in the long run.¹⁴ Assuming no additional costs to implement an equity-focused approach, we estimated lower costs in five of six cities. On aggregate, the equity approach is estimated cost $1.066B less (in present value) than the proportional services approach over the 20-year study time horizon, thus freeing additional resources that could be used to reach and engage underserved populations. Prior applications of DCEA have weighed the trade-offs between maximizing population health and reducing inequity; however, in this infectious disease application the sexual network dynamics of HIV compound the underlying economic and epidemiological conditions to overcome this trade-off, making the equity approach also the more efficient approach so long as intervention costs did not increase more than threefold. Large inequities in healthcare access,^{6, 7} reinforced by racially segregated sexual mixing patterns,²⁷ have produced disparities in individual and population health outcomes between white, Black and Hispanic/Latinx Americans. This study supports calls to action to reduce racial and ethnic disparities in HIV-related health outcomes by establishing a cogent economic argument for targeting efforts to reduce these racial/ethnic health disparities.

Equity-focused strategies improving access to HIV prevention and care can not only have important impacts on reducing disparities in HIV incidence, but can also help reduce racial/ethnic health inequities overall. The equity approach increased population health and reduced overall racial/ethnic health inequality in four of six cities over the 20-year period. LA’s counterintuitive results were due to anticipated demographic shifts over the study period. We estimated that LA’s Black population would generate the highest per capita QALY gains regardless of the scenario. These benefits would be accrued earlier in the study period for this relatively small and declining Black population,²⁸ while gains in the growing Hispanic/Latinx population were accrued much later in the study period, and thus heavily discounted (following methodological conventions). These factors combined to increase the share of health benefits accruing to Black individuals compared to the other racial/ethnic groups in LA. Nonetheless, the equity-focused approach in LA resulted in the largest incremental health gains estimated for any city. With continuing racial/ethnic and socioeconomic health disparities across many disease areas, most recently in COVID-19 infection and mortality rates,²⁹ a coordinated, equity-focused public health implementation strategy could have a profound impact toward achieving population health equity.

Our findings demonstrate that localized HIV strategies should be guided by explicit racial/ethnic inequity reduction goals. Cities with the largest proportion of Black people (i.e., Baltimore and Atlanta) benefited most from an equity-focused approach in terms of increased health gains (49.5% increase in Baltimore, more than double in Atlanta) and HIV incidence relative to white populations. Furthermore, the optimal strategies under the equity approach in these cities notably featured more ART engagement interventions. Lower ART retention in Black compared to white populations³⁰ combined with racially segregated sexual mixing patterns,²⁷ both contribute to the exceptional value in improving ART retention and viral suppression among currently underserved Black and Hispanic/Latinx populations. In addition to individual health benefits, supressed HIV viremia can prevent long transmission chains, helping to curtail epidemics that might otherwise be sustained within marginalized populations.¹⁷

The recently updated HIV National Strategic Plan has overarching goals to reduce HIV-related disparities, with progress measured by viral suppression targets for Black and Hispanic/Latino MSM, Black women, and other priority populations.¹⁸ While achieving viral suppression targets would help reduce disparities in HIV incidence,¹⁹ adding indicators on new infections/diagnoses by race/ethnicity may be the additional incentive needed for public health departments and funders to pursue equity-driven implementation strategies addressing issues such as inadequate housing and poverty. To bolster efforts, greater intersectoral/interagency collaboration is needed to address the social determinants of health–another national strategic objective¹⁸–and address barriers to care that are sustained through structural racism. Models of care with nonmedical supplementary services funded through the Ryan White HIV/AIDS Program have achieved increased rates of viral suppression among participants who are largely low-income and marginalized racial/ethnic groups.³¹ Finally, it is imperative to safeguard and expand healthcare reforms such as the Affordable Care Act including Medicaid expansion which have helped to remove barriers to equitable HIV prevention and care for PWH through expanded health insurance coverage.³²

This study has limitations in the structure of our model and the underlying data on HIV epidemiology, service access and interventions considered.^{10, 11, 15} Specifically, the evidence base on HIV-related service access by race/ethnicity is incomplete; in jurisdictions where stratified data were not available we substituted the average value of individuals within each sex and risk group (Supplement Table 1, p 4). This likely underestimates the true extent of baseline racial/ethnic inequities in service access, thus further reinforcing our findings. We note that our baseline service levels were updated from our prior analyses, and this adjustment did not alter the composition of the optimal strategies identified for the proportional services approach.¹⁴ Second, in solving for the ‘optimal’ combination strategies for each city, we did not consider the full range of scale-up, nor have we considered the full range of the distribution of the increased scale of each intervention across racial/ethnic groups. We imposed practical constraints on the level of scale-up of each intervention (ie based on levels reported in the public domain) and set the distribution of these benefits across racial/ethnic groups at two points (in proportion with baseline service levels and in proportion with the number of new diagnoses) for illustrative purposes. Third, in the absence of evidence on costs of delivering increasingly high service levels to underserved populations, we assumed constant costs and constant effectiveness of the selected interventions. While scaling up interventions among marginalized and hardly reached communities is likely to result in higher marginal costs, we found that the value provided by the equity approach compared to the proportional services approach was robust to a wide range of cost increases for delivering these interventions. Finally, the interventions we considered should not be deemed exhaustive; implementation efforts should incorporate interventions that may better suit underserved populations as this evidence base develops.³³

This study compares the economic and public health impact of two combination implementation strategies with equal levels of scale-up distributed differentially across racial/ethnic groups. Our results demonstrate that approaches driven by equity principles have the potential to reduce racial/ethnic disparities both in HIV and overall health, and increase total population health at expected lower overall costs compared to a strategy that remains agnostic to inequities in access to health services.

Research in context.

Evidence before this study

We searched PubMed for articles from database inception up to January 20^th 2021, with the terms (“HIV”) AND (“Cost-effectiveness” OR “Cost effectiveness”) AND (“Health equity” OR “Health Status disparities*”) AND ((“combination”) OR (“local*” OR “focus*” OR “target”)) AND (“USA”). We found one study that discussed how the ethical foundations of cost-effectiveness analyses (maximizing population health) can conflict with other equity principles. Another study modelled the scale up of HIV treatment to meet WHO guidelines in South Africa, ranking combinations of interventions based on survival, cost-effectiveness and equity; the latter being a key consideration in the current study. However, no studies conducted distributional cost-effectiveness analyses (DCEA) of combination HIV prevention and care implementation strategies guided by explicit racial/ethnic disparity reduction principles.

Added value of this study

We build on previous economic analyses of HIV combination implementation strategies by using an emerging DCEA framework to compare the distributional impact of two implementation approaches for six US cities. We identified optimal combination strategies for a proportional services approach, which assumed that increases from existing service levels were implemented proportionally across race/ethnic groups, and an equity approach, which assumed that increases from existing service levels were proportionate to the distribution of new HIV diagnoses, stratified by race/ethnicity and sex. We found that the equity approach generated greater health gains than the proportional services approach in all cities, reduced disparities according to several measures, and had lower costs in five of six cities while remaining cost-effective in the remaining one. This study provides an economic argument supporting calls to action to improve racial/ethnic health equity.

Implications of all the available evidence

Our results indicate that HIV combination implementation strategies designed to reduce racial/ethnic inequities in HIV service access could produce substantially greater public health and economic value in the US. Equity-focused implementation strategies improving access to HIV prevention and care may not only have important impacts on reducing disparities in HIV incidence but can also help improve racial/ethnic health equity at the population level. To guide investment, DCEA methods should be applied in HIV and other contexts. To maximize impact toward ending HIV epidemics, decision-makers should deliver locally oriented, combination implementation strategies that focus on achieving racial/ethnic health equity.