St Marys Hospital Ingrid V. Bassett, MD, MPH Massachusetts General Hospital Harvard Medical School May 25, 2010 Who Starts ART in Durban, South Africa?

St Mary’s Hospital

Ingrid V. Bassett, MD, MPHMassachusetts General Hospital

Harvard Medical SchoolMay 25, 2010

Who Starts ART in Durban, South Africa?…Not Everyone Who

Should

Conflict of Interest Disclosure Ingrid V. Bassett, MD, MPH

Has no real or apparent

conflicts of interest to report

Overview 2.9 million on ART in sub-Saharan Africa, but

6.7 million need it How much of this gap is due to failure to link to

care after a new HIV diagnosis? What can be done to improve linkage?

WHO, 2009

Road Map Retention versus Linkage Linkage in resource-limited settings Durban, South Africa study Strategies to improve Linkage

Poor retention well-documented

Systematic review sub-Saharan Africa of adult patient retention in ART programs

Non-research ART programs, 2000-2007 33 patient cohorts (>74,000 patients) Very high rates of loss ~40% not in care at 2 years:

Loss to follow-up 56% Death 40%

Rosen, PLos Medicine, 2007

Background: Poor retention substantial during scale-up Updated systematic review 2007-2009 >226,000 patients, 39 cohorts, 80% in Afr ~25% lost at 2 years, ~35% lost at 3 years Slightly better than before, but still

substantial losses after ART initiation Routine M+E and PEPFAR reporting focus

on ART patients

Fox, Trop Med & Intl Health, 2010

Retention

On ART

Linkage

What is known about rates of linkage to HIV care after a new HIV diagnosis?

What are risk factors for failing to enter care? Do these differ by setting?

HIV Test CD4 count Training 1,2,3 On ARTPsychosocialAssessment

Determinants of mortality and non-death losses: Cape Town

Community-based ART program, 2002-2005

1235 enrolled in ART program 121 died (46% pre-tx; 40% early tx) Risk: symptomatic disease and CD4 <100 After first year of ART, low mortality rate <1%/year

High risk of pre-ART and early ART deaths

Lawn, AIDS, 2006

What is rate of mortality and retention pre-ART in Uganda? TASO ART Clinic, Jinja, rural Uganda Focus on 4-8 week pre-ART screening period 26% of ART-eligibles did not finish screening

Risk: lower median CD4, male gender Increased over time with clinic expansion

Home visits to ascertain status 30% on ART with different provider 25% alive and not on ART (44% due to transport) 28% died 17% LTFU

High rate of pre-ART mortality

Amuron, BMC Public Health, 2009

Why failing to link in Malawi? Cross-sectional study Rural Malawi (MSF), 2004-2007 Defaulters missed appointment by >1 mo 874 adults pre-ART traced, 71% found:

51% dead, most within 3 months of last visit Reasons for defaulting: stigma, dissatisfaction with

care/staff, perceived improved health, transport costs

McGuire, Trop Med & Intl Health, 2010

From HIV test through ART start in Mozambique: a retrospective study Routine care data from 2 HIV care networks HIV tested 2004-2005, first year after free ART in

public sector 7005 with HIV

Only 56% enrolled ART clinic within 30d 1506 ART-eligible

Only 31% start ART within 90d of CD4

Micek, JAIDS, 2009

Failure to Link: Open questions What proportion of newly identified HIV-

infected don’t start ART? What are the risk factors for failing to link to care?

Few prospective data about losses and mortality before HIV clinic entry

Valuable to design interventions to improve linkage to HIV care

Who Starts ART in Durban, South Africa?

…Not Everyone Who Should Ingrid V. Bassett, MD, MPH

Susan Regan, PhDSenica Chetty, MSc

Janet Giddy, MBChB, MFamMedLauren M. Uhler, BA

Helga Holst, MD, MBADouglas Ross, MBChB, MBA

Rochelle P. Walensky, MD, MPHKenneth A. Freedberg, MD, MSc

Elena Losina, PhD St Mary’s Hospital

Background: HIV in South Africa > 5 million people HIV-infected Largest ART program in the world Only ~40% of HIV-infected who need ART

are receiving it Few data why HIV-infected fail to link to care

WHO 2009; PEPFAR 2008; Lawn, AIDS, 2008

Objectives To evaluate rates of ART initiation within 12

months of a positive HIV test in Durban, South Africa

To identify baseline factors that predict failure to be on ART at 1 year

Methods: Two Study Sites

Prospective, observational cohort Sites: Outpatient departments in Durban

McCord (urban) St. Mary’s Mariannhill (semi-rural)

Partially government subsidized Patients pay a fee for care PEPFAR-funded HIV clinics

Methods: Study population Adults (≥18y) English or Zulu speaking Enrolled prior to rapid HIV test Enrolled November 2006-October 2008 Follow-up through June 2009

Methods: Data collection Baseline enrollment interview 6, 12 month questionnaire Domains: demographic, geographic, clinical Electronic medical record review at

enrollment site: CD4, ART start

Methods: Two Outcomes1) Obtaining CD4 count within 90 days

2) ART initiation within 12 months for eligible patients CD4 ≤ 200/µl within 90 days of HIV test ART initiation at study site documented in medical record

Methods: Data analysis Predictors of failing to initiate ART evaluated

with multivariate logistic regression Kaplan-Meier curve of time to ART initiation Mortality pre- and post-ART initiation

Enrolled2,775

HIV-negative1,308

HIV-infected1,467

HIV Test

No test/result: 71Indeterminate: 6

54% HIV prevalence

Screened3,401

Results: Cohort enrollment

Bassett, AIDS, 2010

HIV-infected cohort characteristics

Female 54% Median age 34 yrs (IQR 28-

41) Median follow-up time 12 mos (IQR 8-

14) Follow up available 70%

Bassett, AIDS, 2010

HIV-infected1,467

CD4 countwithin 90 days

Yes607

No862

CD4<200/μl368

CD4≥200/μl237

59% no CD4within 90 days

Results: CD4 count within 90 days

61% CD4<200/µlART eligible at baseline

Unknown2

Bassett, AIDS, 2010

HIV Tested*

HIV+

CD4/results

Eligible for ART

Start ART

*Screened 11/06-10/08, enrolled in study and have known HIV status

368

How many start ART?

Failure to obtain CD4Failure to

start ARTwhen eligible

Bassett, AIDS, 2010

154

605

2,775

1,467

Results: Long delay from HIV diagnosis to ART start

P<0.001

0.00

0.25

0.50

0.75

1.00

Pro

port

ion

Sta

rtin

g A

RT

0 90 180 270 360Days Since HIV Testing

Males: 40% started ARTby 6 months

Females:55% started ARTby 6 months

Bassett, AIDS, 2010 days

P<0.001

Results: Predictors of failure to start ART within 12 months

Male gender RR 1.5 (1.1-2.1) No HIV+ family/friend RR 5.1 (1.8-14.9)

Adjusting for: age, CD4 count, prior HIV test, work outside the home

Bassett, AIDS, 2010

Results: High rate of mortality 15% of HIV-infected cohort (216 deaths/1,467) 21% of ART eligible cohort (76 deaths/368)

051015202530

<50 50-99 100-149 150-199 >200

% deadHIV+cohortwith CD4

P<0.001CD4 (/µl) strataBassett, AIDS, 2010

High rate of mortality pre-ART Most patients died pre-ART or with unknown

ART status

051015202530

<50 50-99 100-149 150-199 >200P<0.001

CD4 (/µl) strata

Overall

Pre-ART

% deadHIV+cohortwith CD4

Bassett, AIDS, 2010

Limitations Sites may not be representative of public

sector hospitals in South Africa 30% of pre-ART patients were unreachable Likely underestimates mortality and ART

initiation that occurred at non-study sites

Bassett, AIDS, 2010

Study conclusions Substantial pre-ART loss along care path Men less likely to initiate ART Severe immune suppression at diagnosis Long delays to ART initiation High rates of pre-ART mortality

Bassett, AIDS, 2010

Implications Promote early HIV diagnosis and care Monitor mortality and losses pre-ART Improve access for men Interventions needed to improve linkage to

care/minimize delays Following new HIV diagnosis After ART eligibility determined

Learning from on-ART strategies: patient tracking

2 ART facilities Lilongwe, Malawi, 2006-09 Patients who missed clinic appt >3 weeks 2,653 patients identified, 85% traced by phone

and home visit 30% died

1,158 found alive, not transferred 74% returned to clinic (women, age >39 at ART start)

Tweya, Trop Med & Int Hlth, 2010

Learning from on-ART strategies

“Patient tracers” phone calls or home visits to ascertain vital status, help subjects return to care

Proactive adherence support, including home visits, community-based collaborations

Transportation vouchers Eliminate co-pays Reliable (electronic) monitoring

Geng, JAMA, 2008; Ochieng, IAS, 2007; Tweya, Trop Med & Intl Health, 2010; Rosen, Trop Med & Intl Health, 2010; Etienne, Trop Med & Intl Health, 2010; Forster, Bull WHO,2008

Learning from on-ART strategies: efficiency

Johannesburg 4-month pilot study of telephone tracing by social worker

Average $432/patient returned to care

Lessons learned from this and others: Maintain updated contact information Capacity to know about clinic transfers Capacity to access national death registry

Rosen, Trop Med & Intl Health, 2010 Bassett, JAIDS, 2009; Mwanaga, CROI, 2008;Tweya, Trop Med & Intl Health, 2010;

Learning from US linkage to care trial HIV-infected, recently diagnosed, multi-site US RCT case management vs standard of care Primary outcome in-care at 12 months A higher proportion in intervention arm visited

HIV clinician at least once within 6 months (78% versus 60%, p < 0.01) and at least twice within 12 months (64% versus 49%, p < 0.01)

No similar RCT has yet been performed in resource-limited settings

Gardner, AIDS, 2005

Upcoming NIMH-funded linkage to care trial Multi-site RCT in Durban starting in 2010 Assess clinical impact and cost-

effectiveness of a health system navigator assigned in the outpatient setting

Navigator in-person, SMS, phone contacts Evaluate linkage to HIV care and TB

treatment completion

AcknowledgementsDurban Team Janet Giddy Senica Chetty Douglas Ross Lindeni Sangweni Aletta Maphasa Success Mncwabe Yolisa Mgobhozi Bongiwe Mdadane Matilda Mazibuko Helga Holst

Study participants at McCord and St. Mary’s Hospitals, Durban

US Team and Funders Rochelle Walensky Ken Freedberg Elena Losina Susan Regan Sarah Bancroft Harv Univ Program on AIDS Harvard CFAR AI60354 NIAID K23 AI068458 Harvard Catalyst Grant Upcoming trial: NIMH R01

MH090326

St Mary’s Hospital