key: cord-0846934-jxejwz7b
authors: McManus, Kathleen A; Srikanth, Karishma; Powers, Samuel D; Dillingham, Rebecca; Rogawski McQuade, Elizabeth T
title: Medicaid Expansion’s Impact on Human Immunodeficiency Virus Outcomes in a Nonurban Southeastern Ryan White HIV/AIDS Program Clinic
date: 2020-12-28
journal: Open Forum Infect Dis
DOI: 10.1093/ofid/ofaa595
sha: 05c77b3b18ba03472f77e75d54aedbafa734f46c
doc_id: 846934
cord_uid: jxejwz7b

BACKGROUND: Although the Ryan White HIV/AIDS Program supports high-quality human immunodeficiency virus (HIV) care, Medicaid enrollment provides access to non-HIV care. People with HIV (PWH) with Medicaid historically have low viral suppression (VS) rates. In a state with previously high Qualified Health Plan coverage of PWH, we examined HIV outcomes by insurance status during the first year of Medicaid expansion (ME). METHODS: Participants were PWH ages 18–63 who attended ≥1 HIV medical visit/year in 2018 and 2019. We estimated associations of sociodemographic characteristics with ME enrollment prevalence and associations between insurance status and engagement in care and VS. RESULTS: Among 577 patients, 151 (33%) were newly eligible for Medicaid, and 77 (51%) enrolled. Medicaid enrollment was higher for those with incomes <100% federal poverty level (adjusted prevalence ratio, 1.67; 95% confidence interval [CI], 1.00–1.86) compared with others. Controlling for age, income, and 2018 engagement, those with employment-based private insurance (adjusted risk difference [aRD], −8.5%; 95% CI, −16.9 to 0.1) and Medicare (aRD, −12.5%; 95% CI, −21.2 to −3.0) had lower 2019 engagement than others. For those with VS data (n = 548), after controlling for age and baseline VS, those with Medicaid (aRD, −4.0%; 95% CI, −10.3 to 0.3) and with Medicaid due to ME (aRD, −6.2%; 95% CI, −14.1 to −0.8) were less likely to achieve VS compared with others. CONCLUSIONS: Given that PWH who newly enrolled in Medicaid had high engagement in care, the finding of lower VS is notable. The discordance may be due to medication access gaps associated with changes in medication procurement logistics.

Before the Patient Protection and Affordable Care Act (ACA), it was estimated that approximately 20% of people with human immunodeficiency virus (PWH) in the United States were uninsured [1] . As of July 1, 2020, 37 states and the District of Columbia have expanded Medicaid through the ACA in an effort to increase access to healthcare for people with low incomes [2] . In 2018, approximately half of PWH in Medicaid expansion states were covered by Medicaid, and in Medicaid nonexpansion states, only one third of PWH had Medicaid [1] . Given the large numbers of PWH relying on Medicaid, it is important to understand and quantify the rates of viral suppression (VS) associated with Medicaid and Medicaid expansion, especially as the United States aims for the UNAIDS goal of 90-90-90 and the federal government's "Ending the HIV Epidemic" initiative [3, 4] . Viral suppression is an essential outcome given that the individual benefits from improved health and the public benefits through averted human immunodeficiency virus (HIV) transmissions [5] [6] [7] . When a PWH becomes virally suppressed, it averts 1.8 infections per 100 person-years [8] , and each averted HIV infection saves $402 000 [9] .

Before the ACA and Medicaid expansion, PWH with Ryan White HIV/AIDS Program (RWHAP) support and Medicaid historically had low VS rates (65%-68%) [10, 11] . For 2015, the first full year of Medicaid expansion, among 40 United States jurisdictions, the sustained VS rate in jurisdictions that expanded Medicaid was higher (51.4%) than the rate in jurisdictions that did not expand Medicaid (47.9%) [12] . From other recent aggregate data presented, the post-ACA sustained VS rates for PWH with Medicaid are reported to be 59%-60% [1] . To date, there has been little published about VS rates associated with Medicaid, specifically after Medicaid expansion, using patient-level data and adjusting for individuals' characteristics.

Beginning in 2014, all PWH with low-incomes who were Virginia AIDS Drug Assistance Program (ADAP) clients were offered individual private plans or Qualified Health Plans (QHPs) purchased on the ACA Marketplace and funded by the Virginia ADAP. This program resulted in a significant increase in VS [13] . During 2018, 70.6% of Virginia ADAP clients had ADAP-supported QHPs [14] . When Virginia expanded Medicaid in January 2019, many PWH transitioned from ADAP-purchased QHPs to Medicaid. More states are now expanding Medicaid and will shift PWH off of ADAP-purchased QHPs to Medicaid [2] . In addition, employment changes brought on by the coronavirus disease 2019 (COVID-19) pandemic [15] may result in more PWH transitioning from individual and/or employer-based private insurance to Medicaid. It will be critical to ensure that gains in VS made while on QHPs are preserved for PWH transitioned to Medicaid Expansion.

Changes in insurance or in methods of accessing care and medications can result in gaps and unintended outcomes. Although not HIV-specific, a study of patients with chronic conditions who had Medicaid demonstrated that changes in healthcare insurance plans created interruptions in coverage that resulted in worsened disease outcomes [16] . Given that PWH in our state were undergoing a healthcare delivery transition, we aimed to characterize those who transitioned to Medicaid among those newly eligible and compare their engagement in care and VS rates to those with other insurance coverage in a cohort from a nonurban Southeastern RWHAP clinic during the first year of their state's Medicaid expansion.

The studied RWHAP clinic is a nonurban academic-affiliated clinic at the University of Virginia. Clinic patients are eligible for RWHAP-supported services with an income below 500% of the federal poverty level (FPL). Approximately 97.7% of the clinic population received RWHAP-supported services during 2018-2019.

The study population includes all PWH over the age of 18 and under the age of 63 as of January 1, 2018, who had established care at the studied RWHAP clinic as of January 1, 2018, and had at least 1 medical visit at the studied RWHAP clinic in both 2018 and 2019. Data were collected from January 1, 2018 to January 15, 2020.

The design of the work was reviewed and approved by the University of Virginia Institutional Review Board for Health Sciences Research. Participant consent was not required because the University of Virginia Institutional Review Board for Health Sciences Research deemed that the project met the criteria of exempt research under 45CFR46.104(d) (4) iii.

Demographic characteristics included the following: age on January 1, 2018; sex; race/ethnicity; baseline income; baseline insurance status; rural residence; HIV risk factors; HIV/acquired immune deficiency syndrome (AIDS) diagnosis; baseline CD4 count; time since HIV diagnosis; time since first clinic visit; time since first antiretroviral therapy (ART) prescription date; baseline engagement in care; and baseline viral load status. All baseline characteristics were determined from 2018 data. Due to small numbers, transgender participants were excluded given privacy concerns.

Income was categorized based on the participant's FPL percentage according to their annual household income and household size. Residence was categorized as urban or rural using the Rural Urban Commuting Area (RUCA) approximations for each zip code [17] . For zip codes that were not categorized by RUCA, the National Center for Health Statistics' categorization was used [18] . For baseline insurance status, participants were categorized as having individual private insurance, employerbased private insurance, Medicare, Medicaid, or no insurance. The baseline viral load and CD4 count were defined by the first value measured between January 1, 2018 and December 31, 2018. For HIV risk factor, participants could report more than 1 risk factor, which included male-to-male sexual contact (man who has sex with men [MSM]), heterosexual contact, injection drug use (IDU), perinatal exposure, blood transfusion, and unknown. Using the RWHAP definition, baseline engagement in care was defined as at least 1 HIV medical visit in each 6-month period during 2018 with a minimum of 60 days between medical visits [19] .

Three primary outcomes were evaluated: (1) Medicaid enrollment in 2019 among those newly eligible for Medicaid due to Medicaid expansion and not on Medicare in 2018, (2) engagement in care for 2019, and (3) achieving and/or maintaining VS by the end of the study period. Participants who did not have Medicaid in 2018 and who had incomes ≤138% FPL in 2019 were considered to be newly eligible for Medicaid in 2019 due to Medicaid expansion. Engagement in care for 2019 was defined as described above. Viral suppression was categorized as undetectable if the last result between July 1, 2019 and December 31, 2019 was <200 HIV ribonucleic acid copies per milliliter [20] . If a participant had no viral loads collected in that window, a measurement within 15 days of the window (June 15-July 1, 2019 or January 1-15, 2020) was used with a preference for the latest viral load measurement. Our group chose to study outcomes timed close to the transition in healthcare, as we have learned from quantitative studies and from the perspective of PWH that changes in insurance status can result in quick impacts to care [21] [22] [23] .

Analyses were performed using R (R Foundation for Statistical Computing) and RStudio (RStudio, Inc.). For each of the outcome models, categories of covariates were collapsed as necessary to avoid sparse data bias, with the aim to have greater than 10 observations with the outcome for each covariate category, if possible.

For those newly eligible for Medicaid in 2019, we estimated the associations of select demographic characteristics (age, sex, race/ethnicity, baseline income, baseline insurance status, rural residence, specific HIV risk factors, AIDS diagnosis, and baseline engagement in care) with Medicaid enrollment prevalence. Because it was rare (2.2%) for a participant's primary insurance to switch from Medicare in 2018 to Medicaid in 2019, those who had Medicare in 2018 were excluded from enrollment analyses. We used g-computation with logistic regression to estimate risk ratios using the risk Communicator package [24] . Covariates that had prevalence ratios (PRs) that were statistically significant (P < .05) or of large magnitude (PR, <0.83 or >1.20) were included in the adjusted model.

The risk Communicator package was also used to estimate risk differences (RDs) in engagement in care by insurance status in 2019 (including Medicaid due to Medicaid Expansion). Select characteristics (age, sex, race/ethnicity, income in 2019, rural residence, specific HIV risk factors, AIDS diagnosis, and baseline engagement in care) were assessed to be included in an adjusted model. Covariates that had RDs that were statistically significant (P < .05) or of large magnitude (RD, <−5.0% or >5.0%) were included in the adjusted model. Finally, we estimated RDs in VS by insurance status in 2019 (including Medicaid due to Medicaid Expansion). Select characteristics (age, sex, race/ethnicity, income in 2019, rural residence, specific HIV risk factors, AIDS diagnosis, and baseline VS) were assessed to be included in an adjusted model for VS using the same criteria as described above. Income could not be included in the adjusted model because too few participants had a detectable viral load, such that the model did not converge.

The overall study cohort included 577 participants (Table 1) . Approximately half of the participants were in the 50-63 age Among those who did not start with Medicare (n = 455), 151 (33%) were newly eligible for Medicaid due to Medicaid Expansion, and of those eligible, 77 (51%) enrolled ( Figure 1 ). Baseline engagement in care for those newly eligible for Medicaid was 84.1% (89.6% for those who enrolled and 78.4% for those who did not enroll). Baseline VS for those newly eligible for Medicaid was 85.4% (84.4% for those who enrolled and 87.7% for those who did not enroll). Medicaid enrollment due to Medicaid expansion was more likely for those with a 2018 income less than 100% FPL ( Overall, 85.6% of participants were engaged in care in 2019 ( Figure 2 ). Those enrolled in Medicaid due to Medicaid Expansion had an engagement in care rate of 87.3% compared with 91.9% for individual private insurance, 89.6% for no insurance, 85.1% for employment-based private insurance, 84.1% for Medicaid, and 79.5% for Medicare. Engagement rates for those with Medicaid ( Figure 2 ) (adjusted RD [aRD], −3.8%; 95% CI, −11.2 to 3.7) and Medicaid Expansion (aRD, −1.9%; 95% CI, −10.6 to 5.6) were comparable to those with private insurance. Adjusting for age, income in 2019, and baseline engagement in care, being engaged in care in 2019 was an absolute 8.5% less likely (95% CI, −16.9 to 0.1) for those with employment-based private insurance and an absolute 12.5% less likely (95% CI, −21.2 to −3.0) for those with Medicare compared with those with individual private insurance.

For VS, 548 participants had available data, and 29 (5.0%) participants were excluded due to missing baseline viral load status (n = 1) and viral load outcome (n = 28). Overall, 94% of participants achieved VS (Figure 3 ). Those who newly enrolled in Medicaid due to Medicaid expansion had a VS rate of 85.2% Covariates that had PRs that were statistically significant (P < .05) or of large magnitude (<0.83 or >1.20) were included in the adjusted model.

Individual private 91 (91. 

Participants with Medicaid had the lowest VS rate of any insurance status with those with Medicaid due to Medicaid expansion faring the worst. Given that engagement in care was high for PWH who newly enrolled in Medicaid, the finding of lower VS is unexpected. Our hypothesis is that the discordance may be due to medication access gaps associated with changes in pharmacy logistics [23, 25] . Previous studies of PWH with Medicaid have found that once they discontinued ART, approximately half did not reinitiate an ART within 18 months, and those that did reinitiate had a median gap in ART of approximately 8 months [26] .

To put our observed VS rates associated with Medicaid (87.1%) and Medicaid expansion (85.2%) in context, our rates are higher than the rates for historical cohorts (65%-68%) [10, 11] . This may be because our cohort represents a population that had successfully linked to HIV care and had at least 1 HIV medical visit per year in 2 consecutive years. Recently, more studies have been using sustained VS over a year. A recent Kaiser Family Foundation analysis of the 2018 Centers for Diseases Control and Prevention's 2018 Medical Monitoring Project data demonstrated a sustained VS rate of 64% for those with Medicaid and RWHAP support and just 56% for those with Medicaid and no RWHAP support [1] . These VS rates are likely lower than our observed rates because the studied population included PWH who were not engaged in care, and they defined VS as all undetectable viral loads in a year. Given that our participants receive care in a RWHAP clinic, they would be most comparable to the Medicaid with RWHAP support group. The one other study that controlled for individuals' characteristics found a sustained VS rate of 67%. However, this study included PWH from 4 urban settings who had been engaged in care for 4 consecutive years and assessed VS more than 1 year after the insurance change [27] .

The observed VS rate of 94.0% for all participants in our cohort is quite high, even for PWH engaged in care. It is higher than the RWHAP 2018 VS rate of 87.1%, which is estimated to represent 69% of all RWHAP clients [28] . This reflects that study participants were receiving robust comprehensive HIV care through the studied RWHAP-funded clinic. However, it is concerning in a clinic that supports achievement of VS by more than 90% of PWH receiving care that there is a disparity in VS outcomes for those with Medicaid. Those with Medicaid are not achieving the UNAIDS goal of 90% VS [3] at the same rate as others, and, therefore, they are not experiencing the associated benefits of VS including longevity [5] and U = U, or the knowledge that they cannot transmit HIV to sexual partners [29] . In addition, this signals that there is a gap between the care supported by Medicaid and other insurance plans. Moreover, participants with no insurance, whom we presume relied solely on the RWHAP, achieved higher rates of VS, meaning that Medicaid performed worse for PWH relative to relying only on the RWHAP. This gap could be explained by paperwork burden, miscommunication of insurance information in the setting of lower health literacy, or the pharmacy logistics previously mentioned. In terms of possible interventions, a partnership between New York State and a managed Medicaid care plan offered intensified care coordination and peer support to PWH with detectable viral loads who were enrolled in their Medicaid plan; the partnership showed promising VS outcomes [30] .

However, in addition to clinic-level or state-level investigations and interventions, these disparities need to be investigated as a systemic issue with comparisons between states to identify best practices. Currently, a larger proportion of black PWH rely on Medicaid compared with white or Hispanic PWH [1] . Therefore, black PWH will be disproportionately affected by disparities in VS associated with Medicaid. There must be greater efforts to eliminate disparities, especially those that may stem from policies embedded within the structures that are meant to provide greater access to healthcare, such as Medicaid.

Our observed lower rate of VS among PWH with Medicaid compared with other insurance does not diminish the observed benefits of Medicaid expansion, from which PWH could benefit including, but not limited to, the following: reduction in mortality [31] and cardiovascular mortality [32] ; increased use of preventative care [33] ; decreased hospitalizations for ambulatory care-sensitive conditions [34] ; and associated decreases in inability to afford medical care, concern about paying medical bills, and taking less medication to save money [35] . Although the RWHAP supports high-quality HIV care, Medicaid enrollment improves access to non-HIV care and should continue to be supported by RWHAP.

Among those without Medicare, approximately half of those who were eligible enrolled in Medicaid. We hypothesize that the observed low uptake of Medicaid could have been due to (1) difficulty in reaching eligible people by mail so they may not have known that they were eligible or (2) reluctance to experience a change in insurance/medication logistics. This approximately 50% uptake is consistent with what we found in the first year of a previous health insurance transition for low-income PWH in Virginia [21] . Those with higher baseline 2018 incomes were less likely to enroll. This may be because our clinic's targeted outreach for Medicaid enrollment in 2019 focused on those who had incomes under 138% in 2018. Our clinic did have 24 participants who had incomes over 138% in 2018 whose incomes declined to under 138% in 2019, making them newly eligible for Medicaid. That type of income shift affected 16% of those who were Medicaid eligible in 2019. This highlights the importance of RWHAP clinics' knowledge about changes in a client's income and employment status, so that they can help to rapidly reassess eligibility for programs such as Medicaid or ADAP. Given that clinics likely do not have adequate staffing to proactively reach out to clients between visits, we also need to ensure that RWHAP clients know the benefits of contacting their clinic with any changes in income or employment. However, we must also recognize that sharing income information is often experienced as an unwanted intrusion. Although knowledge of income and income changes is essential to the current RWHAP income-based qualification requirement for access to healthcare, we should also continue to strive to make healthcare access more universally available.

We did not observe lower rates of engagement in care for PWH with Medicaid. Rather, those with employment-based private insurance and Medicare had lower rates of engagement in care. In general, employment is thought to be positively associated with progression along the continuum of HIV care [36] . However, there is not much known about how work environments and culture may impact a PWH's ability to engage in care. We hypothesize that those who are employed in positions that offer employment-based insurance may have little leeway to miss work for appointments. In addition, those with employment-based private insurance and Medicare may have to cover more of the financial burden of visits to the RWHAP clinic if they are not receiving RWHAP support either due to not qualifying or not filling out the paperwork. These costs may be a disincentive to engagement in care. Moreover, these 2 groups had to arrange their own transportation, and in a nonurban southern setting, transportation is a barrier to engagement in care [37] . Alternatively, it is also possible that those who were only recently provided access to Medicaid through expansion had multiple unmet medical needs resulting in more frequent visits while not necessarily achieving VS.

In terms of limitations, this study was performed at 1 RWHAP clinic and the results may not be generalizable. The VS outcome was limited to those who had a viral load result in 2018 and 2019, so there were some participants excluded from that analysis. However, it was less than 5%. Given the epidemiologic issues with the definition of sustained VS [38] , we opted to use an outcome of whether the last viral load was undetectable. In determining Medicaid eligibility, we did not account for citizenship or residency requirements. In terms of additional limitations, there could be unmeasured differences in terms of barriers faced between those who enrolled in Medicaid due to Medicaid expansion and those who were eligible but did not enroll. We could not adjust for differences in many social determinants of health, such as housing, alcohol use, education, internet access/literacy, mental health, transportation, stigma, language barriers. In the VS model, we could not control for income because it was collinear with insurance status. We were able to determine that having an HIV risk factor of IDU was not associated with differences in Medicaid enrollment, engagement in care, and VS, but we did not have current data about substance use.

With the economic issues related to COVID-19 affecting PWH [15] and the potential for additional states to expand Medicaid, we may see a substantial shift of PWH from private insurance to Medicaid. However, an ACA challenge has also been argued before the Supreme Court with a decision expected in June 2021. Understanding the impact of healthcare delivery changes on VS is crucial. We hope that our ongoing qualitative study will help to illuminate issues leading to the lower VS rate for PWH with Medicaid. Moreover, future studies with larger cohorts will need to examine how Medicaid expansion and state-level Medicaid policies (automatic renewals, work requirements, prior authorization requirements, step therapy, monthly fee-for-service prescription limits, etc) affect PWH's overall health and HIV outcomes and the goals of the "Ending the HIV Epidemic" initiative.

nsurance coverage and viral suppression among people with HIV

Status of state action on the Medicaid expansion decision

90-90-90: an ambitious treatment target to help end the AIDS epidemic

Ending the HIV epidemic: a plan for the United States

Comparison of overall and comorbidityfree life expectancy between insured adults with and without HIV infection

PARTNER Study Group. Sexual activity without condoms and risk of HIV transmission in serodifferent couples when the HIV-positive partner is using suppressive antiretroviral therapy

Risk of HIV transmission through condomless sex in serodifferent gay couples with the HIV-positive partner taking suppressive antiretroviral therapy (PARTNER): final results of a multicentre, prospective, observational study

Human immunodeficiency virus transmission at each step of the care continuum in the United States

Updates of lifetime costs of care and quality-of-life estimates for HIV-infected persons in the United States: late versus early diagnosis and entry into care

High rates of retention and viral suppression in United States HIV safety net system: HIV care continuum in the Ryan White HIV/AIDS Program

Ryan White HIV/AIDS Program assistance and HIV treatment outcomes

HIV care engagement and sustained viral suppression among persons with diagnosed HIV infection in Medicaid expansion and non-expansion states

Affordable Care Act qualified health plan enrollment for AIDS Drug Assistance Program clients: Virginia's experience and best practices

National RWHAP Part B and ADAP monitoring project annual report

Financial, food, and housing insecurity due to coronavirus disease 2019 among at-risk people with human immunodeficiency virus in a nonurban ryan white HIV/AIDS program clinic

Medicaid programme changes and the chronically ill: early results from a prospective cohort study of the Oregon Health Plan

Rural-Urban commuting area codes

NCHS urban-rural classification scheme for counties

Health Resources and Services Administration HIV/AIDS Ryan White & Global Program. HIV/AIDS Bureau performance core measures

Panel on Antiretroviral Guidelines for Adults and Adolescents. Guidelines for the use of antiretroviral agents in adults and adolescents with HIV. 2020. Available at

Affordable Care Act qualified health plan coverage: association with improved HIV viral suppression for AIDS drug assistance program clients in a Medicaid nonexpansion state

Evidence from a multistate cohort: enrollment in Affordable Care Act qualified health plans' association with viral suppression

Facilitators and barriers: clients' perspective on the Virginia AIDS drug assistance program's Affordable Care Act implementation

G-computation to estimate risk differences/ratios, rate differences/ratios, and other epidemiological effects

State Medicaid management of prescription drugs for HIV treatment and prevention

Factors associated with antiretroviral therapy reinitiation in Medicaid recipients with human immunodeficiency virus

Human immunodeficiency virus (HIV) viral suppression after transition from having no healthcare coverage and relying on Ryan White HIV/AIDS Program support to Medicaid or private health insurance

Ryan White HIV/AIDS Program annual client-level data report

HIV viral load and transmissibility of HIV infection: undetectable equals untransmittable

Enhanced care coordination improves HIV viral load suppression rates

Medicaid and mortality: new evidence from linked survey and administrative data

Association of Medicaid expansion with cardiovascular mortality

Third year of survey data shows continuing benefits of Medicaid expansions for low-income childless adults in the U.S

Medicaid expansion associated with reductions in preventable hospitalizations

Health and access to care during the first 2 years of the ACA Medicaid expansions

A scoping review of employment and HIV

National survey of US HIV clinicians: knowledge and attitudes about the Affordable Care Act and opinions of its impact on quality of care and barriers to care

What we talk about when we talk about durable viral suppression

We thank the University of Virginia Ryan White HIV/AIDS Program Clinic clients. We also thank the University of Virginia Ryan White HIV/ AIDS Program Clinic staff, especially the case managers, insurance navigators, and community health workers.Disclaimer. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.Financial support. This work was funded by the National Institute of Allergy and Infectious Diseases at the National Institutes of Health (Grant Number K08AI136644; to K. A. M.).Potential conflicts of interest. K. A. M. reports an investigator-initiated research grant from Gilead Sciences, Inc. and stock ownership in Gilead Sciences, Inc. R. D. reports an investigator-initiated research grant from Gilead Sciences, Inc. and consulting for Warm Health Technology, Inc. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.