1. Housing is HIV Prevention and Care Angela Aidala, PhD Department of Sociomedical Sciences Center for Homelessness Prevention Studies Mailman School of Public Health, Columbia University HIV CENTER FOR CLINICAL AND BEHAVIORAL STUDIES – APRIL 10, 2008

2. INTRODUCTION The goals of this presentation are to: Examine the role of housing – or lack of housing - for the continuing HIV epidemic and associated health disparities Review findings from NYC and national studies for evidence to support or challenge a number of causal models which have been proposed to explain observed relationships between housing and HIV/AIDS Discuss implications for housing as a structural intervention to reduce the spread of HIV as well as to increase the health and longevity of HIV-infected persons

3. Housing as Structural Factor  There is increasing awareness of the association between housing and HIV infection –usually discussed in terms of ‘thehomeless’ as a ‘special population’ Different focus: homelessness/ housing instability is causally implicated in increased risk for of HIV disease and for the infected, causally implicated in their medical care and treatment outcomes Housing is a structural factor - an environmental or contextual influence that affects an individual’s ability to avoid exposure to HIV, or for HIV positive individuals to avoid exposing others to infection

4. NYC C.H.A.I.N. STUDY Community Health Advisory & Information Network (CHAIN) Project Multi-stage probability sampling -- Stage 1: HIV health and social service agencies -- Stage 2: Random selection from lists or sequential enrollment Initial recruitment 1994-95, n=700 Refresher cohort 1998, n=268 Unconnected to care, n=74 New cohort 2002, n=684 + 23 unconnected In-person comprehensive (2-3hr) interview every 12 mos High retention rate: 81% - 95% of eligible respondents at each wave Compares to surveillance data

5. HRSA SPNS/ HUD HOPWA Multiple Diagnoses Initiative  Interviews conducted with clients of demonstration projects providing health and social services to low income persons infected with HIV in 1996-2000  Programs targeted ‘hard to reach’ marginalized populations not often included in conventional clinical studies  Sequential enrollment procedure  Baseline information from 3191 clients from 24 projects and follow-up data from 891 clients from 16 projects  Compares to clients in publicly funded services – Ryan White, HOPWA, Medicaid NATIONAL EVALUATION STUDY

6. HUD/HOPWA HRSA/SPNS Demonstration Projects

7. MEASURING HOUSING STATUS  HOMELESS -- homeless -- sleeping in the street, park, abandoned building -- in a public place (e.g. subway) not intended for sleeping -- in a shelter for homeless persons -- in a limited stay SRO or welfare hotel -- in jail with no other address  UNSTABLY HOUSED -- in transitional housing, resident treatment, halfway house -- doubled up with other people  STABLY HOUSED --own, secure housing in regular apartment or house

8. BACKGROUND: HOUSING & HIV EPI  Co-occurrence of homelessness and HIV infection increasingly recognized -- The prevalence of HIV/ AIDS is 3 to 16 times higher among persons who are homeless or unstably housed than among persons with stable, adequate housing depending upon population and geographic area studied -- Behaviors that put persons at risk for HIV are more prevalent among the homeless; however among persons at highest risk due to IDU or high risk sex, those without a stable home are significantly more likely than others to become infected -- HIV infection risk factor for housing difficulties - 17% - 60% of all PLWHA report post diagnosis experience of homelessness/ unstable housing

9. HOUSING & HIV EPI  Housing need among NYC and national samples: -- Approx 50% each NYC cohort were homeless or unstably housed during the year they were diagnosed with HIV -- Over 60% experienced unstable housing or homelessness at least once over the course of their illness -- In the national study over 40% of clients at general medical or social service agencies were homeless or unstably housed at program enrollment -- In NYC, at any point in time 25%-35% of all PLWH are homeless or unstably housed –even more report housing problems -- From a system perspective NYC rates of housing need remain fairly constant over time: as some PLWH get housing needs met, others develop housing problems

10. Aggregate Rates of Housing Need Remain High Rate of Housing Service Need by Date of Interview – 1994 thru 1996 Rate of Housing Service Need by Date of Interview – 2001 thru 2003

11. STUDY QUESTIONS: HOUSING & HIV RISK STUDY QUESTIONS:  What is the association between homelessness/ unstable housing and HIV drug and sex risk behaviors among HIV positive people  Does the effect of housing on HIV risk behaviors remain when controlling for the concurrent receipt of medical care and other treatments and services?

12. ANALYSIS  Logistical regression used to compare the odds of risk behavior associated with different housing situations  Adjusted odds ratios show odds of risk behavior controlling for socio- demographics (age, race/ethnicity, gender) risk exposure group (sexual orientation, history of substance use) economic resources (education, income, primary language, incarceration experience), health status (CD4 count, mental illness) and receipt of health and supportive services (regular source of medical care, case management)  National study presents cross-sectional relationships at program entry  CHAIN study examines 1000- 4000 observation points -logistic regression equations used GEE procedures to adjust for dependency among multiple observations contributed by the same individual

13. FINDINGS: HOUSING AND RISK BEHAVIORS

14. HOUSING & RISK BEHAVIORS In both the NYC CHAIN sample and the national sample:  Significant differences in drug and sex risk behaviors are associated with current housing status  The association of risk behaviors with housing status remains controlling for a wide range of client demographic, health, and service use variables  There is an apparent “dose-relationship” with the homeless at greater risk than the unstably housed, and both of these groups at greater risk than the stably housed

15. ODDS OF RECENT HARD DRUG USE CHAIN SAMPLENAT’L SAMPLE Rate Adjusted Odds Ratio 1 Rate Adjusted Odds Ratio 1 STABLE HOUSING21%16% UNSTABLE HOUSING37%1.6035%2.05 HOMELESS53%3.4564%5.54 1 Odds of drug use past 6 mos by current housing status controlling for demographics economic factors, health status, mental health, and receipt of health and supportive services Note: All relationships statistically significant p<.01

16. ODDS OF RECENT NEEDLE USE CHAIN SAMPLENAT’L SAMPLE Rate Adjusted Odds Ratio 1 Rate Adjusted Odds Ratio 1 STABLE HOUSING4% UNSTABLE HOUSING12%2.8713%2.51 HOMELESS17%4.7427%4.65 1 Odds of needle use past 6 mos by current housing status controlling for demographics economic factors, health status, mental health, and receipt of health and supportive services Note: All relationships statistically significant p<.01

17. Unadjusted Odds Ratio Adjusted Odds Ratio 1 Rate STABLE HOUSING2% UNSTABLE HOUSING5%3.182.91 HOMELESS12%7.875.62 1 Odds of needle sharing past 6 mos by baseline housing status controlling for demographics economic factors, health status, mental health, receipt of health and supportive services All relationships statistically significant p<.01 ODDS OF RECENT NEEDLE SHARING NATIONAL SAMPLE

18. ODDS OF UNPROTECTED SEX PAST 6-12 MOS NATIONAL SAMPLE Unadjusted Odds Ratio Adjusted Odds Ratio 1 Rate STABLE HOUSING40% UNSTABLE HOUSING43%(1.11)(1.04) HOMELESS62%2.372.67 1 Odds of unprotected sex past 12 mos by baseline housing status controlling for demographics economic factors, health status, mental health, receipt of health and supportive services Note: All relationships statistically significant p<.05 except ( ) =ns

19. ODDS OF UNPROTECTED SEX PAST 6 MOS CHAIN SAMPLE Adjusted Odds Ratio 1 Adjusted Odds Ratio 1 Rate STABLE HOUSING13% UNSTABLE HOUSING15%(1.11)21%1.61 HOMELESS16%1.6929%2.30 1 Odds of unprotected sex past 6 mos by baseline housing status controlling for demographics economic factors, health status, mental health, receipt of health and supportive services Note: All relationships statistically significant p<.05 except ( ) =ns MenWomen

20. ODDS OF RECENT SEX EXCHANGE CHAIN SAMPLENAT’L SAMPLE Rate Adjusted Odds Ratio 1 Rate Adjusted Odds Ratio 1 STABLE HOUSING7%5% UNSTABLE HOUSING12%1.64 # 17%2.72 HOMELESS16%2.0421%3.84 1 Odds of needle use past 6 mos by current housing status controlling for demographics economic factors, health status, mental health, and receipt of health and supportive services Note: All relationships statistically significant p<.01 except # p<.05

21. HOUSING & MEDICAL CARE

22. HOUSING & HIV MEDICAL CARE STUDY QUESTIONS:  What is the relationship between unstable housing and access and engagement with medical care and treatments?  Does housing need predict receipt of medical care that meets good clinical practice standards?

23. HOUSING & MEDICAL CARE In both the CHAIN and the national samples  Unstable housing leads to delayed entry into care and to discontinuous care - recent breaks in care, dropping in and out of care and/or changing providers often  Homeless or unstably housed individuals are less likely than other PLWHS to be receiving medical care that meets minimum clinical practice guidelines  Homelessness /unstable housing is one of the most important barrierslimiting the use of antiretroviral combination therapy  High viral load, recent opportunistic infection, and hospitalization for HIV related disease are associated with homelessness/ unstable housing

24. EXPLANATION OF FINDINGS?

25. EXPLANATION OF FINDINGS  Accumulating evidence documents the association between housing and risk behaviors and medical care outcomes – mechanisms less often investigated  Need to understand the causal direction and the mechanisms linking housing and behaviors that put people at risk for HIV infection and/or poor medical care outcomes  Question: Does housing status influence individual risk behaviors and medical care outcomes, or are findings evidence of self-selection of “risky persons” into conditions of homelessness

26. RISKY PERSON Model RISKY DISPOSITIONS/ PERSONALITY. SOCIAL Exclusion RISKY BEHAVIORS: Drug use Risky sex Illegal activities UNSTABLE HOUSING ECONOMIC Marginalization HIV INFECTION

27. OPPOSING MODEL: STRUCTURAL CONTEXTS OF RISK Housing seen as ‘vector’ -- an intermediary by which the pathogenic inequality that inheres in broader economic and political structures is carried to a susceptible host Broader processes of inequality and exclusion lead to the deterioration of housing situations and neighborhood environments for members of excluded groups Lack of housing creates or maintains pervasive context of risk makes it “hard” to avoid risky situations or to use risk-reducing tools and institutions

28. Direct and Indirect Effects of Housing  Effects of Housing include: -- Neighborhood effects -- Stress producing or protecting environments and experiences -- “Social capital” resources or deprivations -- Identity and meaning -- Press of daily needs barrier to service use when available -- Structuring the private sphere – lack of housing is barrier to forming stable partner relations

29. RISKY CONTEXTS Model Economic Marginalization UNSTABLE HOUSING Pervasive Risk Competing Needs Few Personal Resources Few Community Resources Risky Behaviors Drug use High risk sex Demoralization Depression Anxiety Barriers to service use Social Exclusion HIV infection

30. Evidence? Reduction in Risk Behavior “Risky person” model assumes behavior follows person - the formerly homeless who receive housing will continue to engage in risky behavior, continue to remain marginal to systems of care Test: Examine change in risk behaviors associated with change in housing status For over time analysis, examined the odds of risk behavior associated with change in housing status compared to no change Longitudinal analysis included additional controls for baseline housing status, baseline drug or sex risk behavior, and receipt of mental health and/or alcohol or drug treatment in the interim between baseline and follow-up

31. PREDICTING HARD DRUG USE: NATIONAL SAMPLE Started Drug use Stopped Drug use Adjusted Odds Ratio T2 Drug Use 1 NO CHANGE7%6% IMPROVED HOUSING2%12%0.47 WORSE HOUSING9%5%1.38 1 Odds of Time 2 drug use by change in housing status controlling for Time 1 drug use, Time 1 housing status, demographics, economic factors, health, mental health, and receipt of health and supportive services Note: All relationships statistically significant p<.01

32. PREDICTING UNPROTECTED SEX LAST INTERCOURSE Started Unprotected Sex Stopped Unprotected Sex Adjusted Odds Ratio T2 Unprotected Sex 1 NO CHANGE25%7% IMPROVED HOUSING19%15%0.37 WORSE HOUSING25%11%(1.02) 1 Odds of Time 2 sex exchange by change in housing status controlling for Time 1 sex exchange, Time 1 housing status, demographics, economic factors, health, mental health, and receipt of health and supportive services Note: All relationships statistically significant p<.01 except ( ) =ns

33. PREDICTING SEX EXCHANGE: NATIONAL SAMPLE Started Sex Exchg Stopped Sex Exchg Adjusted Odds Ratio T2 SexExchg 1 NO CHANGE11%13% IMPROVED HOUSING5%34%0.46 # WORSE HOUSING54%1%3.99 1 Odds of Time 2 sex exchange by change in housing status controlling for Time 1 sex exchange, Time 1 housing status, demographics, economic factors, health, mental health, and receipt of health and supportive services Note: All relationships statistically significant p<.01 except # p<.10

34. Evidence? Improved medical care outcomes Test: Examine change over time in engagement with medical care associated with change in housing status Findings: Longitudinal analysis shows that unstable housing/ housing problems risk for dropping out of medical care or remaining out of care National study – examine change in housing status associated with change in medical care indicators CHAIN study – can examine effects of receiving housing assistance Longitudinal analyses control for socio-demographics (age, race/ethnicity gender) risk exposure group (sexual orientation, history of drug use) economic resources (education, income, primary language, incarceration experience), health status (CD4 count, mental illness) and receipt of supportive services (case management )

35. PREDICTING T2 MEDICAL CARE National Sample Unadjusted Odds Ratio T2 Outpatient Visits Adjusted Odds Ratio T2 Outpatient Visits 1 NO CHANGE IMPROVED HOUSING(1.16) 4.74 WORSE HOUSING(1.43)(0.30) 1 Odds of Time 2 outpatient visit past 6 months by change in housing status controlling for Time 1 outpatient use, Time 1 housing status, demographics, economic factors, drug use, health status, mental health, and receipt of case management services N= 399. Relationships statistically significant p<.05 except ( ) =ns

36. PREDICTING T2 MEDICATION USE National Sample Unadjusted Odds Ratio T2 ARV Adjusted Odds Ratio T2 ARV 1 NO CHANGE IMPROVED HOUSING3.21 6.22 WORSE HOUSING(0.63)(1.01) 1 Odds of Time 2 antiretroviral medication use by change in housing status controlling for Time 1 ARV use, Time 1 housing status, demographics, economic factors, drug use, health status, mental health, and receipt of case management services N= 192. Relationships statistically significant p<.05 except ( ) =ns

37. Access to Medical Care: CHAIN NYC Has Any Medical Care Appropriate Clinical Care HOUSING NEED 0.70 ** 0.71 *** HOUSING ASSISTANCE 2.42 *** 1.53 *** Low mental health functioning (0.85)0.80 ** Current problem drug use 0.74 * 0.73 *** Mental health services2.08 ***1.43 *** Substance abuse treatment (0.97) 1.28 * Medical case management (1.38) (1.09) Social services case management2.43 ***1.70 *** N=1651 individuals, 5865 observations, 1994 - 2007

38. Continuity of Medical Care: CHAIN NYC Continuity Of Any Medical Care Continuity Appropriate Clinical Care HOUSING NEED 0.83 * 0.71 *** HOUSING ASSISTANCE 1.22 * 1.53 *** Low mental health functioning (0.85)0.80 ** Current problem drug use (0.97) 0.73 *** Mental health services (1.13)1.43 *** Substance abuse treatment (0.98) 1.28 * Medical case management (0.89) (1.09) Social services case management (1.18)1.70 *** N=1295 individuals interviewed 2+ times, 53759 observations, 1994 - 2007

39. Entry into Medical Care: CHAIN NYC Entry into Any Medical Care Entry into Appropriate Clinical Care HOUSING NEED 0.44 ** (0.73) HOUSING ASSISTANCE 2.15 ** 1.88 *** Low mental health functioning (0.76) (0.74) Current problem drug use 0.36 *** 0.69 * Mental health services 2.79 ** (1.29) Substance abuse treatment (1.64) (1.46) Medical case management (1.35) (0.78) Social services case management 2.27 *1.84 ** N=557 individuals who were not in care at one or more interviews, 720 observations, 1994 - 2007

40. Summary HIV positive persons with housing problems are more likely to engage in sex and drug risk behaviors, are less likely to be engaged in appropriate medical care Overtime analyses show improvement in housing situation is associated with reduction in risk behaviors and positive change in medical care outcomes Data show strong and consistent relationship between housing and risk and medical care outcomes, regardless of other client characteristics, health status, or service use variables Findings suggest that the condition of homelessness, and not simply traits of homeless individuals, influences risk behaviors and service utilization

41. Limitations Self-reported measures Limited information about specific sexual behaviors and relationships Timing of events not exact National study not probability sample, substantial loss to follow-up Findings consistent with the argument that housing status influences outcomes observed but cannot establish causality

42. Provision of housing is a promising structural intervention to reduce the spread of HIV as well as improve the lives of infected persons More directly malleable ‘state’ of housing situation holds more promise for intervention than mechanisms far antecedent in psychological development or closer to biological bases of disease Housing is a strategic target for intervention by addressing more proximal consequences of broader economic, social, political or policy barriers that affect HIV prevention and HIV care Expensive but offset by social and economic costs of ongoing HIV transmission and HIV treatment failure among significant proportion of HIV infected population Policy Implications

43. HOUSING IS PREVENTION AND CARE

44. ACKNOWLEDGEMENTS o The CHAIN research was made possible by a series of grants from the US Health Resources and Service Administration (HRSA) under Title I of the Ryan White Comprehensive AIDS Resource Emergency (CARE) Act and contracts with the New York City HIV Health and Human Services Planning Council through the New York City Department of Health and Medical and Health Research Association of New York City. o The national, multi-site research project is an inter-agency collaboration between the U.S. Health Resources and Services Administration (HRSA), Special Projects of National Significance (SPNS) Program, and the U.S. Department of Housing and Urban Development (HUD), Housing Opportunities for Persons with AIDS (HOPWA) Program of the Division of HIV/AIDS Housing. o Additional funding for risk behavior analysis was provided by the Behavioral Intervention Research Branch, Division of HIV/AIDS Prevention, National Center for HIV, STD, and TB Prevention; U.S. Centers for Disease and Prevention (CDC) o The contents contents are solely the responsibility of the Researchers and do not necessarily represent the official views of the U.S. Health Resources and Services Administration, HUD, CDC, the City of New York, or the Medical and Health Research Association.