Enrollment Length, Service Category, and HIV Health Outcomes Among Low-Income HIV-Positive Persons Newly Enrolled in a Housing Program, New York City, 2014–2017

Abstract

Objectives. To evaluate the impact of duration and service category on HIV health outcomes among low-income adults living with HIV and enrolled in a housing program in 2014 to 2017.

Methods. We estimated relative risk of engagement in care, viral suppression, and CD4 improvement for 561 consumers at first and second year after enrollment to matched controls through the New York City HIV surveillance registry, by enrollment length (enrolled for more than 1 year or not) and service category (housing placement assistance [HPA], supportive permanent housing [SPH], and rental assistance [REN]).

Results. The SPH and REN consumers were enrolled longer and received more services, compared with HPA consumers. Long-term SPH and REN consumers had better engagement in care, viral suppression, and CD4 count than controls at both first and second year after enrollment, but the effect did not grow bigger from year 1 to 2. HPA consumers did not have better outcomes than controls regardless of enrollment length.

Conclusions. Longer enrollment with timely housing placement and a higher number and more types of services are associated with better HIV health outcomes for low-income persons living with HIV with unmet housing needs.

Persons living with HIV (PLWH) who have addressed their need for food, clothing, and housing have better adherence to HIV treatment (called antiretroviral therapy, or ART), lower HIV viral loads (VLs), higher CD4 T-cell counts, better mental health status, and reduced HIV risk behaviors.^1–4 PLWH who maintain an undetectable VL for at least 6 months cannot transmit HIV sexually, which has been promoted in the international “Undetectable = Untransmittable” or “U = U” campaign.⁵ CD4 count indicates immune health and is influenced by HIV VL, ART adherence, and quality of life (e.g., income, stress, exercise, sleep, and substance use).^6–11 Meeting basic needs, including housing, is an important component of the National HIV/AIDS Strategy for the United States.¹²

In New York City (NYC), housing programs specifically for low-income PLWH include Housing Opportunities for Persons With AIDS (HOPWA), Ryan White Part A Housing Program, and NYC HIV/AIDS Services Administration. Together they provide emergency housing, supportive housing, housing subsidies, or housing placement assistance to more than 30 000 PLWH per year, some of whom enroll in multiple programs simultaneously. NYC HIV/AIDS Services Administration serves the most New Yorkers and is funded largely by the city and state, while HOPWA is the only US federal program dedicated to the housing needs of low-income PLWH.^13,14 Funded by the federal Department of Housing and Urban Development, NYC HOPWA received $44 million in 2017 to 2019 and serves more than 3000 households in NYC and the tristate area annually, a substantial portion of the national allocation of $300 million, which serves 45 000 households annually. Several studies have suggested that HOPWA services improve housing status and engagement in HIV care but not CD4 count nor VL.^15,16 A randomized controlled trial of an enhanced HOPWA housing placement assistance model with 1-year supportive case management showed that faster housing placement was associated with better viral suppression status when compared with usual services.¹⁷ However, the impact of enrollment length and type and amount of housing and ancillary services upon HIV outcomes remains unclear.

In terms of study design, most HIV housing-related studies are observational studies (e.g. cohort, case–control, and cross-sectional studies), but many of them lack a comparison group or are compared with a citywide or nationwide population.² Proportions of PLWH engaged in care and virally suppressed tend to increase in NYC¹⁸ and the United States over time; therefore, it is difficult to attribute improvements in health outcomes to housing interventions alone without a control group. Furthermore, PLWH with unmet housing needs are a vulnerable population and not comparable with the general HIV population.

We present a retrospective matched cohort study, aiming to compare HIV health outcomes among NYC HOPWA enrollees with a similar population who were not HOPWA enrollees, stratified by enrollment length and service category. We also present housing outcomes and type and amount of housing and ancillary services among HOPWA enrollees.

METHODS

The NYC Department of Health and Mental Hygiene (DOHMH) directly oversees HOPWA contracts in NYC, aiming to reduce homelessness, connect and retain consumers in medical care, improve adherence to ART, and reduce HIV transmission by stabilizing housing and addressing other basic needs. The actual services are provided by community-based organizations with expertise in providing housing and HIV-related supportive services. All 3 HOPWA service categories—housing placement assistance (HPA), supportive permanent housing (SPH), and rental assistance (REN)—serve low-income PLWH and are used across the country. PLWH may apply to any or none. Each category is offered through a different set of local organizations and provides different amounts and types of services, described in Table 1. DOHMH monitors the HOPWA contracts, including for habitability of placed consumers’ housing units.

TABLE 1—

Descriptions of the 3 Service Categories of the New York City Housing Opportunities for Persons With AIDS Program Since 2014

	Housing Placement Assistance	Supportive Permanent Housing	Rental Assistance
Target population	Low-income persons living with HIV who live in emergency housing or independently and need services to secure permanent housing	Low-income persons living with HIV who have a demonstrated need for supportive services	Low-income persons living with HIV who need long-term rental assistance and can live independently
Intensity and duration of engagement between provider and recipient of housing services	Preplacement: high; Postplacement: low for 12 mo, then services end	High, long-term	Low, long-term
Housing services provided	Locating and establishing appropriate permanent independent or supportive housing, and training about living independently	Providing permanent supportive scattered-site or supportive congregate housing, with lease in the name of the agency, or in a building owned by the housing services provider, and training about living independently	Providing rental subsidies to establish or maintain permanent independent housing, with lease in the name of the consumer
Services in addition to those related to housing	Case management, medical accompaniment, and advocacy	Case management, medical accompaniment, advocacy, and comprehensive support services, including counseling for substance use, mental health, and health promotion	Case management and advocacy

Data Sources

We retrieved enrollment and closure dates, program type, housing placement dates, receipt of rental assistance, and other service data from housing administrative databases. We obtained VL, CD4 count, and variables used for matching from the NYC HIV surveillance registry, which is a population-based registry of all diagnoses of AIDS and HIV infection in NYC reported to the DOHMH. The registry contains select demographic, HIV transmission risk, and clinical information on persons diagnosed with HIV receiving care in NYC, as well as all diagnostic tests, VL tests, CD4 counts, and HIV genotypes carried out in NYC.¹⁹ Records of HOPWA consumers are matched to the registry quarterly by a deterministic matching algorithm based on name, birth date, Social Security number, and limited human review to ensure matching quality.

Inclusion Criteria

The treatment group consisted of PLWH who were newly enrolled in any HOPWA service category any time between July 1, 2014, and December 31, 2015, and received at least 1 service during 2-year follow-up after enrollment. The control group consisted of PLWH who were not enrolled in HOPWA any time before December 31, 2017 (the end time point of study), and shared similar baseline covariates, selected from the registry by a matching technique.

Both treatment group and control group also met all the following criteria: (1) diagnosed as HIV-positive before 2013, to have preenrollment measurement; (2) no evidence of death by December 31, 2017; and (3) lived in the 5 boroughs of NYC during 2013 to 2017 (the registry only has laboratory tests conducted in NYC).

Coarsened Exact Matching

Matching promotes comparability between cases and controls and reduces variability and systematic differences attributable to background variables that are not of interest to the investigators.²⁰ We used the coarsened exact matching technique because it is the most natural for data with continuous, discrete, and mixed variables, and faster than other methods and thus better for large data sets.²¹ We “coarsened” each variable, or recoded by grouping substantively indistinguishable values (e.g., we coarsened age to 5-year groups), then those data underwent exact matching (e.g., persons were matched on age group) to determine matches and prune unmatched units.²²

We matched 2 controls per exposure unit in the study to improve statistical power and precision without introducing bias.^20,23–26

Matching variables were

1. baseline clinical variables: VL and CD4 count within 12 months before enrollment, as they are the most important covariates affecting postenrollment HIV outcomes;

2. demographics: race/ethnicity, age, gender, place of birth (United States vs elsewhere), and neighborhood poverty level of zip code of residence, which make NYC HOPWA consumers different from the general HIV population, and are also associated with engagement in care and viral suppression^16,27;

3. HIV transmission risk: people with substance use are less likely to be engaged in HIV care and virally suppressed²⁷;

4. receipt of services from another housing program: more than one third of all PLWH in NYC receive some form of housing service (primarily rental subsidies from NYC HIV/AIDS Services Administration), and approximately 70% of HOPWA consumers receive services from NYC HIV/AIDS Services Administration; and

5. HIV diagnosis year: recently diagnosed persons have more treatment options over the course of their infection relative to their years of infection²⁸ and thus may achieve better health outcomes.

We coarsened VL to 6 categories (missing, ≤ 200 copies/mL, 201–1499 copies/mL, 1500–9999 copies/mL, 10 000–99 999 copies/mL, ≥ 100 000 copies/mL), CD4 count to 4 categories (missing, < 200 cells/µL, 200–499 cells/µL, and ≥ 500 cells/µL), age to 12 categories, and HIV diagnosis year to 4 categories (before 1995, 1996–2000, 2001–2005, and 2006 and after). We then conducted exact matching between the treatment and the control group on as many matching variables as possible. We balanced all baseline variables between groups by χ² test after matching. Controls were assigned the same “enrollment” dates as their matched HOPWA consumers for any relevant analysis.

Outcomes

Postenrollment binary clinical outcomes were (1) engagement in care (EiC), (2) viral suppression (VS), and (3) improvement to or maintenance of optimal CD4 count. We defined EiC as having at least 1 CD4 count or VL during a given period. We defined VS as the last quantitative VL during a given period being less than or equal to 200 copies per milliliter. We categorized CD4 count as the aforementioned 4 levels²⁹; we defined a successful outcome as improvement from one CD4 count level to another from baseline to after enrollment (e.g., from missing to < 200 cells/µL or from 200–499 cells/µL to ≥ 500 cells/µL) or maintenance of an optimal CD4 count level (≥ 500 cells/µL). We measured EiC and VS status at year 1 (month 0–12 after individuals’ enrollment dates), and year 2 (month 13–24). We measured CD4 count indicator from baseline to 12 months and from baseline to 24 months, because CD4 count improves more slowly than the other 2 outcomes.

Housing outcomes were (1) whether consumers had major housing needs met within 2 years after enrollment, (2) time to meeting major housing needs, (3) the average number of services received after meeting housing needs, and (4) types of services received. We defined meeting major housing needs as placement in permanent housing for HPA, placement in supportive housing for SPH, and receiving rental assistance for REN. We only measured housing outcomes among the treatment group.

Statistical Analysis

We assessed differences in categorical variables by χ² test and difference in means by t test with equal or unequal variance based on data distribution. We conducted descriptive analyses to compare housing outcomes by enrollment length and service category among the treatment group.

We used conditional Poisson regression with robust errors variance to test if the treatment group was more likely than the control group to achieve better HIV outcomes at the end of year 1 and end of year 2 after enrollment for the matched pair study.³⁰ Time to EiC, ART adherence, VS, and improved CD4 count vary by individual. Most people who start ART achieve VS within 6 months, but others may face barriers to maintaining adherence and experience virologic failure.^31,32 We were interested in whether HOPWA services can help more people maintain or achieve optimal outcomes in year 2 than year 1. Therefore, if any outcome showed significant improvement at both first and second year, we further measured if the difference between groups in the outcome was significantly greater in year 2 than in year 1 by running the regression with a main effect of group, a main effect of time (year 1 and 2), and an interaction term of time and group.

Stratification

We stratified all analyses by HOPWA enrollment length and service category. We classified enrollment length as “long term” if 1 or more years and “short term” if less than 1 year (Figure A, available as a supplement to the online version of this article at http://www.ajph.org). For the small portion of consumers enrolled in more than 1 service category during the 2-year follow-up, enrollment length was total unique days across all their categories, and service category was the initial enrollment. A sensitivity analysis conducted for consumers enrolled in only 1 category showed results similar to the complete analysis, suggesting it was legitimate to assign everyone their initial service category.

We conducted all of the analyses in SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

There were 756 HOPWA consumers newly enrolled in HOPWA between July 1, 2014, and December 31, 2015, who received at least 1 service; 116 were diagnosed with HIV in 2013 or later and were excluded from the study, 25 consumers died, and 54 moved out of NYC before December 31, 2017. Therefore, we included 561 HOPWA consumers in the treatment group: 269 long-term consumers and 292 short-term consumers (Figure A).

The majority of the 561 HOPWA consumers were male, Black or Latino, men who have sex with men or persons with a history of injecting drugs, lived in a medium- to very-high-poverty area (≥ 10% below federal poverty level, according to US Census Bureau 2009–2013 and 2010–2014 American Community Survey), received a housing subsidy from another housing program, and were diagnosed with HIV after 2006. Only the proportion born outside the United States was imbalanced between long-term (44%) and short-term (24%) consumers. The 1122 PLWH in the control group had similar characteristics after matching (Table 2; see Table A, available as a supplement to the online version of this article at http://www.ajph.org, for full table).

TABLE 2—

Baseline Characteristics Between Housing Consumers Living With HIV and Control Group Persons Living With HIV by Enrollment Length: New York City, 2013–2017

	Long-Terma Consumers, No. (%)	Controls, No. (%)	P	Short-Terma Consumers, No. (%)	Controls, No. (%)	P
All	269 (100)	538 (100)	. . .	292 (100)	584 (100)	. . .
Demographic characteristics
Genderb			.82			.9
Cisgender man	179 (66.5)	362 (67.3)		188 (64.4)	383 (65.6)
Cisgender woman	85 (31.6)	169 (31.4)		98 (33.6)	191 (32.7)
Transgender woman	5 (1.9)	7 (1.3)		6 (2.1)	10 (1.7)
Race/ethnicity			> .99			> .99
Non-Hispanic Black	126 (46.8)	252 (46.8)		167 (57.2)	333 (57.0)
Hispanic	132 (49.1)	264 (49.1)		108 (37)	217 (37.2)
Non-Hispanic White	8 (3.0)	16 (3.0)		15 (5.1)	30 (5.1)
Asian/Pacific Islander	2 (0.7)	4 (0.7)		2 (0.7)	4 (0.7)
Multiracial	1 (0.4)	2 (0.4)		0 (0)	0 (0)
Age category, y			> .99			> .99
18–24	12 (4.5)	24 (4.5)		17 (5.8)	33 (5.7)
25–44	107 (39.8)	214 (39.8)		118 (40.4)	236 (40.4)
45–64	142 (52.8)	284 (52.8)		149 (51.0)	299 (51.2)
≥ 65	8 (3.0)	16 (3.0)		8 (2.7)	16 (2.7)
Birth country			> .99			> .99
United States	115 (42.8)	237 (44.1)		174 (59.6)	353 (60.4)
US dependency	19 (7.1)	37 (6.9)		25 (8.6)	45 (7.7)
Foreign country	118 (43.9)	230 (42.8)		71 (24.3)	142 (24.3)
Unknown	17 (6.3)	34 (6.3)		22 (7.5)	44 (7.5)
HIV-related medical characteristics
Baseline engagement in carec			.48			.06
No	4 (1.5)	12 (2.2)		3 (1.0)	18 (3.1)
Yes	265 (98.5)	526 (97.8)		289 (99.0)	566 (96.9)
Baseline viral suppressiond			.77			.45
No	59 (21.9)	123 (22.9)		74 (25.3)	162 (27.7)
Yes	210 (78.1)	415 (77.1)		218 (74.7)	422 (72.3)
Baseline CD4 count category (cells/mm3)e			.84			.69
Missing	6 (2.2)	18 (3.3)		7 (2.4)	23 (3.9)
< 200	28 (10.4)	53 (9.9)		41 (14.0)	82 (14.0)
200–499	92 (34.2)	180 (33.5)		110 (37.7)	211 (36.1)
≥ 500	143 (53.2)	287 (53.3)		134 (45.9)	268 (45.9)

Note. Controls were sampled 2:1 from the New York City HIV surveillance registry and matched to individual Housing Opportunities for Persons With AIDS consumers on clinical and demographic characteristics.

^aLong-term consumers were consumers who were enrolled for no less than 1 year, and short-term consumers were consumers who were enrolled for less than 1 year.

^bPersons not known to be transgender were classified as cisgender (i.e., not transgender).

^cBaseline engagement in care was defined as having at least 1 VL or CD4 count within 12 mo before the enrollment month.

^dBaseline viral suppression was defined as the last quantitative viral load within 12 mo before the enrollment month was ≤ 200 copies/mL.

^eBaseline CD4 count category was based on the last CD4 count within 12 mo before the enrollment month.

Housing Outcomes

Every consumer participates in an intake assessment at enrollment with a case manager in the agency or at home, and a reassessment, typically every 3 months after enrollment. Assessments (i.e., intake and reassessments) involve a structured conversation with the case manager to check eligibility and needs. Therefore, long-term consumers received more assessments than did short-term consumers during the study period and also received significantly more follow-up services after housing needs were met (63.1 vs 18.9; P < .001; t test with unequal variance).

The 3 service categories are designed to meet different housing needs, and enrollment lengths and doses and types of services differ across categories. During the study period, 49% of HPA recipients had their housing needs met in a median of 89 days (Table 3). SPH and REN were more likely to enroll consumers for longer and met most consumers’ housing needs within about a week. SPH provided the most follow-up services to consumers after their needs were met (mean per consumer = 118; 95% confidence interval [CI] = 103, 132), while HPA provided the least (mean = 19; 95% CI = 16, 23). Advocacy for housing, health care, and entitlements was the service type SPH and HPA consumers received most; the second highest service type for SPH consumers was life skills management, such as individual and group health promotion and education, and independent living skills; and the top service category for REN consumers was rental assistance eligibility verification.

TABLE 3—

Housing Needs and Services Among Low-Income Persons Living With HIV, Within 2 Years After Enrollment, by Housing Program Service Category: New York City, 2013–2017

	Housing Placement Assistance (n = 384)	Supportive Permanent Housing (n = 92)	Rental Assistance (n = 85)
Housing needs were met,a no. (%)	187 (49)	74 (80)	77 (91)
Days from enrollment to housing needs met, median	89	8	0
No. of servicesb per consumer before housing needs were met, mean (95% CI)	27 (25, 30)	15 (10, 20)	2 (1, 3)
No. of servicesb per consumer after housing needs were met, mean (95% CI)	19 (16, 23)	118 (103, 132)	46 (42, 50)
Top 3 types of servicesb after housing needs were met	Advocacyc	Advocacyc	Rental assistance eligibility verificationd
	Service plan development or updatee	Life skills managementf	Rental assistanceg
	Counselingh	Counselingh	Service plan development or updatee

Note. CI = confidence interval.

^aMajor housing needs were defined as placement in permanent housing (either supportive housing or independent living) for housing placement assistance, placement in supportive housing for supportive permanent housing, and receipt of rental assistance for rental assistance.

^bThe services did not include intake assessments or reassessments. When counting the number of services, the numbers reflected the number of encounters, rather than the number of unique types of services.

^cAdvocacy includes advocacy for housing, health care, entitlements, and other.

^dRental assistance eligibility verification includes the application review process to determine and verify client eligibility for Housing Opportunities for Persons With AIDS–funded rental assistance. Also included is the review process to verify continual eligibility for recipients of Tenant-Based Rental Assistance.

^eService plan development or update includes identification, development, and update of long-term and short-term goals and the action steps (including timeframes) necessary to achieve these goals.

^fLife skills management includes individual or group health promotion and education, housing readiness workshops, and independent living skills.

^gRental assistance includes long-term rental assistance, short-term rental assistance, and start-up rental assistance.

^hCounseling includes individual or group counseling for mental health, substance use, and other supportive counseling.

HIV Health Outcomes

Long-term HOPWA consumers achieved better outcomes in EiC, VS, and CD4 count than their controls at both first year and second year after enrollment. At second year, long-term consumers maintained greater than 99% EiC, while among their controls it decreased from 98% at baseline to 93% (relative risk [RR] = 1.07; 95% CI = 1.04, 1.10); they were 10% more likely to be virally suppressed (87% vs 79%; RR = 1.10; 95% CI = 1.04, 1.17), and 16% more likely to achieve a better or maintain an optimal (i.e., ≥ 500 copies/µL) CD4 count level (65% vs 56%; RR = 1.16; 95% CI = 1.05, 1.28) than their controls (Table 4; see Table B, available as a supplement to the online version of this article at http://www.ajph.org, for full table). Compared with controls, short-term consumers only had higher EiC in year 1 (98% vs 93%; RR = 1.06; 95% CI = 1.03, 1.09); in year 2, when short-term consumers’ enrollments were closed, there were not significantly better HIV outcomes.

TABLE 4—

Engagement in Care, Viral Suppression, and CD4 Count Category Improvement by Enrollment Length and Service Category at Baseline and at First and Second Year After Enrollment for Housing Consumers Living With HIV and Control Group Persons Living With HIV: New York City, 2013–2017

	Long-Term Consumers vs Controls, RR (95% CI)	Short-Term Consumers vs Controls, RR (95% CI)
Overall
Engagement in carea
First year	1.05 (1.02, 1.08)	1.06 (1.03, 1.09)
Second year	1.07 (1.04, 1.10)	1.02 (0.98, 1.06)
Viral suppressionb
First year	1.09 (1.03, 1.16)	1.04 (0.96, 1.12)
Second year	1.10 (1.04, 1.17)	0.98 (0.91, 1.07)
CD4 countc
First year	1.13 (1.04, 1.24)	1.11 (1.00, 1.23)
Second year	1.16 (1.05, 1.28)	1.06 (0.95, 1.18)
Housing placement assistance
Engagement in carea
First year	1.03 (0.99, 1.07)	1.06 (1.02, 1.09)
Second year	1.07 (1.02, 1.11)	1.03 (0.99, 1.07)
Viral suppressionb
First year	1.04 (0.95, 1.14)	1.05 (0.97, 1.15)
Second year	1.05 (0.96, 1.15)	0.99 (0.91, 1.09)
CD4 countc
First year	1.12 (0.99, 1.28)	1.13 (1.00, 1.27)
Second year	1.12 (0.97, 1.30)	1.09 (0.97, 1.23)
Supportive permanent housing
Engagement in carea
First year	1.06 (1.01, 1.11)	1.06 (0.94, 1.19)
Second year	1.09 (1.03, 1.15)	0.93 (0.82, 1.04)
Viral suppressionb
First year	1.18 (1.01, 1.38)	0.88 (0.66, 1.19)
Second year	1.26 (1.10, 1.46)	0.88 (0.64, 1.23)
CD4 countc
First year	1.11 (0.94, 1.31)	1.11 (0.74, 1.68)
Second year	1.27 (1.03, 1.57)	0.96 (0.58, 1.59)
Rental assistance
Engagement in carea
First year	1.08 (1.03, 1.14)	1.08 (0.99, 1.18)
Second year	1.05 (1.01, 1.09)	1.03 (0.98, 1.08)
Viral suppressionb
First year	1.12 (1.04, 1.21)	1.06 (0.89, 1.26)
Second year	1.08 (0.99, 1.17)	1.03 (0.84, 1.25)
CD4 countc
First year	1.18 (1.01, 1.39)	0.97 (0.78, 1.20)
Second year	1.15 (0.97, 1.36)	0.88 (0.64, 1.20)

Note. CI = confidence interval; RR = relative risk. First year = enrollment to 12 months. Second year = 13 months to 24 months. Controls were sampled 2:1 from the New York City HIV surveillance registry and matched to individual Housing Opportunities for Persons With AIDS consumers on clinical and demographic characteristics.

^aEngagement in care was defined as having at least 1 VL or CD4 count during the given period.

^bBaseline viral suppression was defined as the last quantitative viral load during the given period was ≤ 200 copies/mL.

^cCD4 count was categorized as 4 categories: missing, < 200 cells/µL, 200 cells/µL–499 cells/µL, and ≥ 500 cells/µL, and a successful outcome was defined as the last CD4 count category improved from lower level to higher level from baseline to first year or baseline to second year after enrollment, or maintenance of an optimal CD4 count level (≥ 500 cells/µL).

Among the 3 service categories, REN consumers performed best in both years. For example, at second year, 100% of long-term consumers were engaged in care, 97% were virally suppressed, and 72% had achieved better CD4 count levels or maintained the optimal level. SPH long-term consumers increased suppression rates from 70% at baseline to 83% at year 1 and 87% at year 2 (compared with 66%, 70%, and 69%, respectively, for controls), which were the greatest increases among the 3 service categories. However, HPA consumers only achieved better outcomes in EiC in the first year for short-term consumers and the second year for long-term consumers; neither VS nor CD4 count showed significant improvement, and this service category had the worst outcomes among all long-term consumers (Table 4; see Table B for full table).

Among outcomes with improvement at both first and second year, the difference between groups was never significantly bigger in the second year than the first (i.e., no interaction term of time and group achieved statistical significance [data not shown]).

DISCUSSION

We found strong evidence that low-income PLWH who received services through HOPWA SPH and REN for more than 1 year were more likely to be engaged in care and virally suppressed and to achieve better CD4 counts than similar populations who did not receive these services. This is consistent with previous studies showing that housing assistance increased access to and retention in medical care and treatment.¹⁶ The effect of housing on HIV outcomes for people who were enrolled for less than 1 year was minimal, suggesting that maintaining housing services for more than 1 year may be essential for low-income PLWH with unmet housing needs.

HPA consumers had suboptimal postenrollment HIV outcomes, possibly exacerbated by prolonged unmet housing needs and limited ancillary services.³³ In our study, only half of HPA consumers’ housing needs were met within 2 years, half waiting 3 months and 10% for longer than a year. In addition, each consumer received only 19 services on average after placement, compared with an average of 118 for SPH and 46 for REN. Placed HPA consumers did not achieve better health outcomes than their controls, either (e.g., 80% suppressed at year 1 for placed HPA long-term consumers and 76% for their controls; χ² = 0.6; P = .4). The results echo a previous randomized controlled trial, in which usual services in HPA did not optimize the effect on VL.¹⁷ In contrast, SPH and REN consumers whose needs were met more quickly and who received more services showed improved HIV outcomes. Locating and securing appropriate housing can be difficult for low-income PLWH because of individual larger socioeconomic challenges.²

HPA consumers, by definition, have unmet housing needs at enrollment, which may partially explain their worse housing outcomes compared with SPH and REN consumers. HPA’s focus is locating housing; that HPA consumers with those needs met had no better health outcomes than their controls may highlight the importance of follow-up and supportive services—components of SPH and REN. In this context, housing placement alone may not be enough to improve health outcomes.

Although significant results were shown at first and second year after enrollment for long-term consumers, the effect did not grow over time regardless of service category, suggesting that current HOPWA housing services may successfully provide a short-term effect on HIV outcomes but fail to work for people who did not achieve optimal health outcomes in the first year. There are many other individual and social factors that affect HIV outcomes, such as adherence to ART, substance use, mental health issues, housing conditions, and social environments after placement.^2,27 Such barriers can be difficult to overcome. Besides providing housing assistance, housing case managers need to be able to identify barriers for each person and provide differentiated services or ensure consumers successfully connect with appropriate services.

Limitations and Strengths

Our study had some limitations. First, the NYC registry only has laboratory tests from people who are engaged in care in NYC. Although we limited our analyses to people who presumably lived in NYC throughout the entire study period, some people may have moved or might be engaged in care outside of NYC. However, less than 3% of PLWH move out of NYC every year (Qiang Xia, NYC DOHMH, e-mail communication, April 6, 2020), and given the wide availability of HIV care in NYC, PLWH in NYC may be unlikely to receive care outside the city. Second, the results may not be fully generalizable to other cities in the United States that do not have the abundance of HIV housing and HIV care programs that NYC does. Third, although we matched on numerous characteristics including neighborhood poverty level and receipt of services from another housing program, we did not have information for potential controls about their individual income or general interest in housing services, so it is possible that they differed from HOPWA consumers in income or housing need. Fourth, movement across care, VL, and CD4 count categories does not reflect within-category changes, such as increased care frequency or CD4 count remaining less than 200; we did not assess within-category changes and do not know whether analyzing HIV outcomes as continuous would have affected conclusions.

Our study had numerous strengths. First, laboratory test data were from the registry rather than self-reported, ensuring complete and accurate measurement of HIV outcomes and allowing us to follow up all persons for 2 years. HIV-related laboratory tests are electronically reported and estimated to be greater than 97% complete, with minimal differences in completeness by subgroup (Qiang Xia, NYC DOHMH, e-mail communication, October 28, 2016). Second, we constructed a comparable control group so that we could attribute the better HIV outcomes to the HOPWA housing intervention. Third, conducting a randomized controlled trial of housing is costly. A well-designed matched cohort study can generate comparable results with randomized controlled trials,²⁰ but the literature of conducting such studies is sparse and typically brief³⁴; therefore, for researchers who would like to conduct HIV housing-related research but with limited resources to conduct a randomized controlled trial, our study presents an example of a matched cohort study by constructing a control group via registry data.

Public Health Implications

Longer enrollment, timely placement, and a higher number and more types of follow-up services are important factors associated with better HIV health outcomes for low-income PLWH with unmet housing needs. Decision-makers such as federal health and housing program administrators and providers who serve this population should consider enhancing housing models to encourage these factors to further improve health outcomes.