A Case-Control Study of Virologic Failure Among People Living with HIV Enrolled in a NYC-Based Medicaid Special Needs Plan

Abstract

Engagement in HIV care is essential for viral suppression, and care coordination programs have shown improved retention and adherence. In New York Medicaid HIV Special Needs Plans (SNP), health-homes provide enhanced care coordination for people with HIV (PWH) at risk of non-adherence; however, evidence of their impact on virologic outcomes is limited. Using a matched case-control design and 2016–2018 data from a New York City Medicaid HIV SNP, this study examined associations between health-homes and virologic failure. Cases had virologic failure, defined as two consecutive viral loads > 200 copies/mL. Controls matched 1:1 on age (18+), race, and second viral load date, had no virologic failure. Conditional logistic regression examined virologic failure and health-home use duration, Social Deprivation Index (SDI), gender identity, HIV-infection stage, SNP enrollment, comorbidities, and polypharmacy. There were 2,566 PWH (62% male, mean age 42.9 years, mean SDI 92.3). Virologic failure was associated with HIV Stage III (OR = 2.01, 95% CI: 1.61–2.49), substance use disorder (OR = 2.09, 95% CI: 1.68–2.58), and < 6 months of health-home use (OR = 2.08, 95% CI: 1.27–3.42). The association attenuated with 6 + months of health-home use (reference=none; OR = 0.98, 95% CI: 0.77–1.25). Polypharmacy (OR = 0.72, 95% CI: 0.59–0.89) and mental health diagnoses (OR = 0.74, 95% CI: 0.59–0.94) had lower odds of virologic failure. Although further research is needed to clarify the roles of polypharmacy and mental health care, sustained health-home use may reduce disparities in virologic failure among PWH at high risk of falling out of care.

Introduction

Human Immunodeficiency Virus (HIV) continues to pose significant public health challenges, with approximately 1.2 million people in the US currently living with HIV, 13% of whom do not know their HIV status, and 50% of whom are over the age of 50 [1, 2]. Adherence to antiretroviral therapy (ART) is critical for achieving undetectable viral load and preventing HIV transmission. Research has shown that approximately 4–8% of people with HIV experience virologic failure [3]. However much of the research on virologic failure was done prior to advancements in potency and effectiveness of ART. Historically, 90% adherence was necessary for achieving viral suppression, but recent evidence has demonstrated that 80% adherence may be sufficient with modern ART regimens [4]. Moreover, modern ART regimens are less burdensome, produce fewer side effects, and are more often prescribed as a single pill regimen, rather than the 3 or more pills in the past [5]. Despite these advancements, virologic failure remains a concern and a barrier to the success of the “Undetectable=Untransmissible” campaign for treatment as prevention of HIV in the US and the Joint United Nations Program on HIV/AIDS (UNAIDS) effort to end HIV as a public health threat by 2030 [6]. UNAIDS aims to have 95% of people know their HIV status, 95% of those with HIV engaged in care, and 95% of those engaged in care virally suppressed [7]. However, these goals are challenging to meet given the complex challenge of aging with HIV.

People with HIV (PWH) experience age-related comorbidities, syndromes, and often receive subsequent treatment at a much younger age than individuals without HIV [8]. This creates disproportionate prevalence of polypharmacy among PWH [9]. Contrary to research in a non-HIV population, polypharmacy among PWH has been associated with better health outcomes, including adherence to ART, and likely represents appropriate engagement in healthcare [10]. This does not however preclude PWH from having drug-drug interactions, food-drug interactions, or acquiring HIV drug resistance which are risk factors for virologic failure [5, 11]. Moreover, social determinants of health, such as lower average income, and higher likelihood of mental health conditions and substance use further pose challenges to maintaining viral suppression because they are associated with lower ART adherence [12–14].

Enhanced care coordination has been proposed as a potential intervention for improving ART adherence in individuals with HIV, because it can address person specific factors related to aging, social determinants, and polypharmacy. Medicaid, the dual state and federal health insurance for low-income adults, is the largest source of insurance coverage for PWH in the US, insuring up to 40% [15]. Medicaid plays a crucial role in providing HIV treatment services, particularly among adults not yet eligible for Medicare, which mostly insures those aged 65+. Because Medicaid is jointly funded by state and federal governments, its structure and implementation varies by state [16]. States like New York (NY) have incentivized managed care organizations to implement chronic disease Special Needs Plans (SNP), similar to those offered through Medicare, which provide care optimized for the treatment of a specific chronic disease, in this case HIV [17]. Medicaid redesign efforts in NY in 2011 resulted in the transition from targeted case management programs to health homes [17]. The health homes model incorporates care coordination at two levels. All clients enrolled in the SNP receive basic care coordination, while those at risk of falling out of care or poor adherence to ART are eligible for enhanced care coordination through the health home. This includes phone calls from a pharmacist if HIV treatment prescriptions are not filled, “robo” calls, and texts as reminders about appointments, and outreach from clinical staff on a regular basis to promote care engagement. As was demonstrated in a 2019 study of standard Medicaid managed care plan enrollees, targeted communication and peer support programs for PWH can mitigate virologic failure and improve engagement in care [18]. However, health home programs are opt-in, making the duration of participation a potentially critical factor for achieving and maintaining viral suppression among the most vulnerable enrollees. Currently there is little known about the association of health homes with virologic failure.

In this study, we examined the association between enhanced care coordination delivered through health homes and virologic failure among PWH enrolled in a New York City Medicaid HIV-SNP [19, 20]. Although virologic failure is influenced by multiple factors, including polypharmacy, social determinants of health, and comorbidities, enhanced care coordination through health homes is a unique intervention that can be leveraged to optimize care adherence among vulnerable PWH. Prior research has evaluated other forms of care coordination; however limited evidence exists on the role of care coordination through health homes on HIV outcomes, particularly the importance of sustained adherence to ART over time. The objective of this study was to establish if at a baseline, a health homes program conferred a benefit on participants at risk of falling out of care. We therefore hypothesized that health home enrollment would be more common among individuals with virologic failure, reflecting targeted enrollment of those most at risk.

Methods

Data Source and Study Design

We conducted a matched case-control study using 2016–2018 deidentified Medicaid claims data from Amida Care, a private, non-profit managed care organization in New York City for people with or at risk for HIV. Amida Care administers an HIV-focused Medicaid SNP that provides care coordination for all clients [21]. The data included Amida Care enrollment and termination dates, viral load test date and results, prescription and medical claims, enrollment and termination dates for health homes, and HIV status (a binary variable for diagnosis of HIV stage I/II or III; all had HIV). HIV stage I refers to acute infection, stage II to chronic infection, and stage III to acquired immunodeficiency syndrome (AIDS) [22]. Demographic information included age at most recent enrollment, race (White, Black, Asian, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, Unknown, Other), ethnicity (Hispanic, non-Hispanic), gender (man, woman, transgender/gender diverse), NYC borough (Manhattan, Queens, Brooklyn, Staten Island, Bronx) and zip code of residence.

Inclusion and Exclusion Criteria

Study participants were between the ages of 18 and 64 because after age 64 participants would transition to Medicare; n = 4 were excluded due to age below 18 years. Anyone without viral load tests or who had tests with missing data was excluded (n = 1,402). Cases were individuals with virologic failure, defined as two consecutive viral load tests of > 200 copies/mL within a four-month span at any point in our data (n = 1283) [23]. If there was more than one virologic failure event for an individual, the most recent occurrence of virologic failure was chosen for analysis. The date of the second failed test in the four-month span defined the index date. We required one year of lookback time from the second viral load test to assess exposures of interest (n = 985 had < 1 year lookback).

Measures

Social Deprivation Index (SDI) scores, enrollment length in Amida Care, polypharmacy, mental health and substance use disorder diagnoses, and health home participation were assessed as potential independent variables associated with virologic failure. The SDI score is a publicly available pre-calculated measure of area-level deprivation based on seven demographic characteristics: poverty, low educational attainment (less than 12 years), single-parent households, rental housing, overcrowded housing, households without a car, and nonemployment among adults under the age of 65 [24]. The SDI measure was linked to zip code of residence and ranges from 0 to 100 with higher scores indicating greater deprivation. Amida Care enrollment duration was calculated as the number of days from the most recent enrollment date to the index date and was centered for analyses. Polypharmacy was defined as the use of five or more different Anatomic Therapeutic classes (ATC-3), excluding antiretroviral medications used for HIV treatment, for at least 75% (68 days) of the 90 days prior to the index date [25]. Mental health (anxiety, depression) and substance use disorders (including alcohol, opioid, cannabis, cocaine, hallucinogen, nicotine, inhalant, sedative, hypnotic or anxiolytic stimulant, other psychoactive, and non-psychoactive substance use related disorders) were identified based on the presence of two ICD-10 codes at least seven days apart during the 90-days prior to the index date and selected based on prior literature [26]. Health home enrollment, representing the enhanced level of care coordination provided by Amida Care, was categorized as 0 to 6 months of enrollment, more than 6 months of enrollment or no enrollment which served as the reference. Sensitivity analyses were conducted with alternative categorizations of health home enrollment including more granular time-based groupings such as 0–3 months and 6–12 months, as well as quartile-based groupings. Having 3 categories was the only option with sufficient sample size for analysis, compared to a greater number of categories.

Matching Cases to Controls

Cases and controls were matched one-to-one using nearest neighbor matching based on age at enrollment in Amida Care, race, and index date. Controls were individuals who did not experience virologic failure during the study period, defined as having at least two consecutive suppressed viral load tests in a four-month span, < 200 copies/mL (n = 2994). The second suppressed test date was used as the index date for controls. The index dates for cases and controls were matched within ± 31 days. Balance diagnostics were conducted to determine the optimal matches between cases and controls.

Statistical Analysis

Univariable conditional logistic regression analyses were used to calculate odds ratios (ORs) with 95% confidence intervals for the association of each independent variable of interest with the outcome of virologic failure. Independent variables that were significant at the p < 0.05 level in the univariable analyses were included in multivariable conditional logistic regression models. Model fit was evaluated using the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC), with lower values indicating better model fit. All analyses were performed using R version 4.4.0. This study was reviewed and approved by the Institutional Review Board at Northeastern University (IRB#: 24-03-16).

Post hoc Analysis

To assess whether the association between health home enrollment and virologic failure varied by duration of home health enrollment, a post hoc contrast analysis was conducted comparing individuals enrolled for 6 + months to those enrolled for 0–6 months. The difference in log-odds between the two groups was estimated using model coefficients and variance-covariance matrix from the multivariable regression model. This approach allowed for a direct comparison between the two enrollment durations.

Results

There were 1,283 cases of virologic failure from 2016 to 2018 among PWH enrolled in Amida Care. These cases were matched with 1,283 unique controls drawn from 2,994 eligible PWH, resulting in a total study sample of 2,566 PWH. After matching, the standardized mean difference between cases and controls was 0.0037, indicating excellent covariate balance between groups. Table 1 presents the demographic and clinical characteristics of the cases and controls. Of the total study sample, 1,740 (68%) were Black, 1,632 (64%) were non-Hispanic, 1,600 (62%) were male, and 208 (8%) identified as gender diverse. The mean age was 43 (sd = 10) and social deprivation was high with a mean SDI score of 92.26 (sd = 13.91; range 1-100).

Table 1.

Demographic characteristics of cases and controls

	Casesa n (%)	Controlsa n (%)
n	1283	1283
Age (mean ± sd)	43 ± 10	43 ± 10
Days enrolled in Amida Care (mean ± sd)	1597.16 ± 881.20	1859.36 ± 924.19
SDI (mean ± sd)	93.34 ± 12.67	91.19 ± 14.98
Race
White	343 (27)	347 (27)
Black	872 (68)	868 (68)
American Indian/Alaska Native	24 (2)	24 (2)
Native Hawaiian/Pacific Islander	15 (1)	12 (1)
Asian	4 (0)	3 (0)
Unknown	21 (2)	27 (2)
Other	4 (0)	2 (0)
Ethnicity
Hispanic	482 (38)	431 (34)
Not Hispanic	793 (62)	839 (65)
Other	0 (0)	1 (0)
Unknown	8 (1)	12 (1)
Gender
Man	782 (61)	818 (64)
Woman	400 (31)	358 (28)
Gender Diverse	101 (8)	107 (8)
Borough
Manhattan	198 (15)	233 (18)
Bronx	666 (52)	574 (45)
Brooklyn	245 (19)	280 (22)
Queens	37 (3)	72 (6)
Staten Island	26 (2)	32 (2)
HIV Status
Stage 1 & 2	918 (72)	1059 (83)
Stage 3	365 (28)	224 (17)
Health Home Participation
6 + months	230 (18)	254 (20)
0–6 months	70 (5)	38 (3)
none	983 (77)	991 (77)
Polypharmacy
Yes	364 (28)	451 (35)
No	919 (72)	832 (65)
Mental Health Disorder b
Yes	234 (18)	268 (21)
No	1049 (82)	1015 (79)
Substance Use Disorder c
Yes	472 (37)	280 (22)
No	811 (63)	1003 (78)

^aAge and Race were used with the index date to match cases to controls 1:1

^bMental Health Disorder refers to diagnosis with an anxiety disorder or depressive disorder

^cSubstance Use Disorder refers to diagnosis with a drug use disorder or alcohol use disorder

In the observed data, 918 (71.55%) cases had HIV stages I/II compared to 1,059 (82.54%) controls. A greater proportion of cases lived in the Bronx (n = 666, 51.91%) than controls (n = 574, 44.74%). On average, cases were enrolled in Amida Care for fewer days (1,597.16, sd = 881.20) than controls (1,859.36, sd = 924.19). The overall number of individuals enrolled in health homes was similar; however, a higher proportion of cases were enrolled for less than six months (n = 70, 5.46%) compared to controls (n = 38, 2.96%).

Table 2 displays the results of the univariable and multivariable conditional logistic regression analyses. Gender was excluded from multivariable models as no significant association was found in the univariable regression. Diagnosis with a mental health disorder was also not significantly associated with virologic failure (OR 0.83, 95% CI: 0.69–1.02) but was retained in multivariable models due to its borderline significance and clinical relevance. In the multivariable regression, HIV stage III (previously termed AIDS) was positively associated with virologic failure compared to stages I/II (aOR 2.01, 95% CI: 1.61–2.49). Polypharmacy was associated with lower odds of virologic failure (aOR 0.72, 95% CI: 0.59–0.89). Higher SDI scores remained significantly associated with virologic failure, though the effect size was minimal (aOR 1.01, 95% CI: 1.00-1.02). Days enrolled in Amida Care had no effect (aOR 1.00, 95% CI: 1.00–1.00). Excluding enrollment length was considered; however, retaining the variable resulted in improved model fit, as indicated by lower AIC and BIC values. Mental health disorders were associated with reduced odds of virologic failure (aOR 0.74, 95% CI: 0.59–0.94), while substance use disorders were associated with increased odds of virologic failure (aOR 2.09, 95% CI: 1.68–2.58). When looking at enhanced care coordination through health homes, individuals enrolled for over six months were no more or less likely to experience virologic failure compared to those not enrolled (aOR 0.98, 95% CI: 0.77–1.25). In contrast, enrollment for zero to six months was associated with higher odds of virologic failure (aOR 2.08, 95% CI: 1.27–3.42) compared to those not enrolled. Post hoc analysis comparing the health home enrollment groups showed that individuals enrolled for more than six months had significantly lower odds of virologic failure compared to those enrolled for zero to six months (aOR 0.47, 95% CI: 0.28–0.80).

Table 2.

Conditional Logistic Regression Results

	Univariable Regression		Multivariable Regression
Exposure	OR (95% CI)	p value	aOR (95% CI)	p value
SDI Score	1.01 (1.00–1.02)	< 0.001	1.01 (1.00–1.02)	< 0.01
Gender
Gender Diverse	0.98 (0.72–1.32)	0.87	-	-
Woman	1.17 (0.99–1.40)	0.07	-	-
Man	REF	REF
Health Home Length
6 + months	0.91 (0.74–1.12)	0.38	0.98 (0.77–1.25)	0.9
0–6 months	1.88 (1.24–2.83)	< 0.01	2.08 (1.27–3.42)	< 0.01
None	REF	REF	REF	REF
HIV Stage
Stage 3	1.83 (1.52–2.21)	< 0.001	2.01 (1.61–2.49)	< 0.001
Stage 1 & 2	REF	REF	REF	REF
Time in Amida Care	1.00 (1.00–1.00)	< 0.001	1.00 (1.00–1.00)	< 0.001
Polypharmacy (68 + days)
Yes	0.77 (0.59–1.00)	0.05	0.72 (0.59–0.89)	< 0.01
No	REF	REF	REF	REF
Mental Health Disorder
Yes	0.83 (0.69–1.02)	0.08	0.74 (0.59–0.94)	< 0.05
No	REF	REF	REF	REF
Substance Use Disorder
Yes	2.18 (1.81–2.62)	< 0.001	2.09 (1.68–2.58)	< 0.001
No	REF	REF	REF	REF

Discussion

This study examined the role of enhanced care coordination through Medicaid health homes and other factors associated with virologic failure among PWH in NYC enrolled in a Medicaid HIV-SNP between 2016 and 2018. Consistent with our hypothesis, individuals enrolled in health homes for fewer than six months had higher odds of virologic failure, likely reflecting targeted enrollment of individuals at elevated risk of disengaging from care. However, important differences in late-stage HIV infection, polypharmacy, and comorbidities between individuals with and without virologic failure persisted and may have limited the full benefits of health home participation.

This study provides promising evidence for the role of enhanced care coordination through health homes in HIV care and viral suppression. In this study, individuals with virologic failure were more likely to have been enrolled in health homes for less than six months, compared to those enrolled for longer durations or not enrolled at all, likely reflecting targeted enrollment of higher-risk individuals. Moreover, the lack of statistically significant difference in virologic failure between those not using health homes and those with prolonged health home enrollment suggests that health homes and the case management they provide had likely achieved their goal of engaging vulnerable populations in healthcare mitigating poor outcomes, such as virologic failure. These findings highlight the importance of continuity and sustained duration of care coordination to stabilize treatment and support viral suppression among high-need populations. In practice, this suggests that health homes may require sufficient time to stabilize patient engagement, address potential barriers, and support ART adherence.

The potential benefits of sustained care coordination observed in this study are consistent with prior evidence from other care coordination models, including targeted case management (TCM). Care coordination through TCM has shown improved care engagement and treatment adherence, with TCM clients achieving CD4 count levels similar to non-TCM clients over a one-year period [27]. However, because health homes are designed to support individuals requiring care coordination and thus tend to enroll individuals already at higher risk of virologic failure [28], it may be important for future research to examine the population eligible for home health enrollment comparing those that do and do not participate. Developing a deeper understanding of who chooses to enroll, and the engagement of participants could support evidence-based refinements to eligibility criteria and retention strategies, which would ultimately strengthen the effectiveness of the health homes model.

Despite the potential benefits of the health homes program, HIV stage appeared to influence the likelihood of virologic failure in this study. Stage III HIV, formerly classified as AIDS, reflects advanced disease progression, which on average has slowed with the improvements in ART potency and availability [29]. Although older adults tend to be diagnosed at later stages of HIV infection and therefore progress to Stage III infection sooner [30], our sample had a mean age around 43 suggesting this is an unlikely explanation for late-stage disease we had in our sample. Instead, additional risk factors such as substance use and mental health conditions may play a larger role. This is supported by prior research across 14 states’ Medicaid HIV populations, which found these factors to be significantly associated with progression to later stage infection [31]. Future research should examine specific differences between those with late and earlier stage HIV to understand if HIV stage impacts benefits received from health home utilization.

The negative association of virologic failure with polypharmacy and mental health diagnoses in claims observed in this study demonstrated the potential benefits of care engagement for an older HIV population. These findings align with previous research on geriatric conditions and ART adherence, which demonstrated that polypharmacy was associated with better outcomes such as adherence to ART [10]. This contrasts with patterns seen in the general population where polypharmacy is typically associated with worse health outcomes [32]. This suggests that polypharmacy may reflect higher levels of healthcare utilization among PWH. The same might be said for mental health conditions, where seeing the claims for mental health conditions likely indicates care engagement and could explain the lower likelihood of virologic failure. This is supported by research showing that utilization of mental health services is associated with increased odds of retention in HIV care [33]. In this context, polypharmacy and claims for mental health conditions may serve as a proxy for consistent engagement in care and ongoing monitoring by a care provider, all of which may improve adherence to ART and in turn reduce the likelihood of virologic failure.

We strove to maintain strong internal validity, however, there are certain limitations in this study due to the nature of the data and study design. First, the Medicaid claims data used in this study was not collected for the original purpose of research and therefore there may be uncontrolled confounding. Though it can be assumed that generally the population is of lower socioeconomic status because it is a Medicaid plan, the unique SNP structure used by New York may limit generalizability to other states or Medicaid populations [34]. Second, as the cases and controls were matched on age and race, we were unable to assess potential associations between virologic failure and these demographic variables. Nevertheless, matching on these factors allowed us to focus on the effects of other variables where intervention is possible, such as health home involvement and polypharmacy.

Conclusion

Short term health home enrollment was associated with virologic failure while enrollment of 6 + months reduced disparities in virologic failure among Amida Care members at elevated risk of falling out of care. These findings suggest that sustained participation in enhanced care coordination through health homes may be necessary to realize its potential benefits for virologic outcomes, although further evaluation is warranted. Additionally, factors such as HIV stage, polypharmacy, mental health conditions, and substance use may inform targeted strategies to reduce virologic failure within managed care settings. Together, these results underscore the potential role of sustained health home care coordination in optimizing HIV care delivery and reducing disparities among high-risk populations. This study serves as evidence for pursuing future research examining the features of health homes most important for promoting ART adherence.

Author Contributions

Emma Kohrt & Brianne Olivieri-Mui: Conceptualization, Methodology, Formal analysis, Writing—Original draft preparation; Brandon Dionne: Validation, Writing—Reviewing and Editing. Mark Brennan-Ing: Writing—Reviewing and Editing. Robert Cavanaugh: Analytic Support, Validation.

Funding

Open access funding provided by Northeastern University Library. This research was funded by a Northeastern University PEAK grant 2024–2025, awarded to Emma Kohrt. Dr. Olivieri-Mui is funded by the National Institute on Aging K01AG077972-01A1.

Data Availability

These data are proprietary and not available outside the Amida Care Brookdale Research Consortium.

Declarations

Conflict of interest

Authors have no conflicts to report.

Ethical Approval

This study was reviewed and deemed exempt by the Northeastern University Institutional Review Board.