Optimized Multidisciplinary Care Teams Enhance Antiretroviral Therapy Adherence --What We Know Michael Horberg, MD MAS FACP Director, HIV/AIDS Kaiser Permanente.

Optimized Multidisciplinary Care Teams Enhance Antiretroviral Therapy

Adherence--What We Know

Michael Horberg, MD MAS FACPDirector, HIV/AIDS Kaiser PermanenteExecutive Director Research, Mid-Atlantic Permanente Medical GroupClinical Lead, HIV/AIDS, Care Management InstituteVice-Chair, HIV Medicine Association

HIVI HIV Initiative of Kaiser Permanente and Care Management Institute

Why The Care Team as Necessary?

• Doctors don’t always discuss adherence with the patient or don’t emphasize it enough Nachega, IAPAC/NIMH 2011; Golin, JGIM, 2004

• Patient treatment adherence is not static over time Leading to changes in viral control over time (Mugavero,

IAPAC/NIMH 2011)

• Care necessities evolve over time drug-drug interactions, co-morbidities do too These impact adherence also

• Structural issues can impact adherence Including transportation, ease of refills

Slide 2

HIV Demographics—for purposes of reference

United States VA

Kaiser Permanente (KP)

+ Group Health

Year 2006 2008 2010

Number HIV+ 1,100,000 (est.) 23,463 20,180

% Female 25% 3% *16%

% Black 50% 50% *18%

% Latino 20% 7% *15-25%

% >50 years of age 27% 64% 42%

Sources: CDC, KFF, VA, KP

*--Varies significantly by state

Slide 3

KP and GHC operate in 9 states plus DC.

KP HIV population rising annually; VA remains steady.

KP Non-NQF HIV Quality Measures

non-NQF Measures: All Sites Combined: 2007, 2008, 2009KPCO and KPHI data not available for 2007

55.4

88.6

27.1

61.8

93.8

59.3

88.8

25.7

60.5

94.0

61.7

87.5

25.6

62.4

94.3

0102030405060708090

100

HIV Testing Among STI Positives

(3 STI)

CD4 Measured in 90 days of Identified HIV+

CD4 < 200 Among Newly

Diagnosed HIV+

Adherenceto ART ≥90%

Median ARTAdherence

Perc

ent

2007 2008 2009

Our Non-NQF HIV Quality Measures

non-NQF Measures: All Sites Combined: 2007, 2008, 2009KPCO and KPHI data not available for 2007

55.4

88.6

27.1

61.8

93.8

59.3

88.8

25.7

60.5

94.0

61.7

87.5

25.6

62.4

94.3

0102030405060708090

100

HIV Testing Among STI Positives

(3 STI)

CD4 Measured in 90 days of Identified HIV+

CD4 < 200 Among Newly

Diagnosed HIV+

Adherenceto ART ≥90%

Median ARTAdherence

Perc

ent

2007 2008 2009

Care coordination key here.

Many person effort necessary—not justthe physician!

Our NQF/NCQA HIV Quality Performance

NQF Endorsed Measures--KP Performance

76.886.3

68.0

86.892.9

79.385.8

65.6

89.294.4

77.885.5

65.9

90.5 94.5

0.010.020.030.040.050.060.070.080.090.0

100.0

Retention In Care

CD4 Measured

PCP Prophylaxis

On HIV Treatment

HIV RNA <75/mL

Metric

Perc

ent S

ucce

ss

2007

2008

2009

Many team efforts here also.

Outcomes are a team effort.

(Re-)New Interest in “Medical Home”

• Emphasis on integrated, multi-disciplinary care (MDCT) HIV Specialist (ID or primary care) as “specialty leader” Case manager and care management Can be physically in one place or connected by technology

Linkage to inpatient and outpatient care, lab, pharmacy services, consults

• Has been an element in HIV care Essentially, how KP practices HIV medicine Ryan White C clinics, VA also

• Not much research Some research but pre-combination ART (Le, 1998, Sherer, 2002)

HIV specialist improved outcomes (Kitahata 2000, Delgado 2003)

HIV clinical pharmacist (Horberg 2007)

Slide 7

Multidisciplinary Care Team Components (1)

Potential Components: NOTE: Need for local considerations always

• HIV Specialist Can be Infectious Disease Specialist Or Primary Care with extended experience with HIV Care

• Care Coordination Often an RN, but not necessarily Consider PA, clinical pharmacist, other

• HIV Clinical Pharmacist• Nurse Case Manager

Slide 8

Multidisciplinary Care Team Components (2)

• Social Work Benefits Coordination Access to outside services

?Housing ?Legal

• Health Educator• Nutrition Service• Transportation Specialist• Identified Specialists in other disciplines

Oncology Gastroenterology

• Mental HealthSlide 9

KP: Provider Experience and Outcomes

•IF antiretroviral naïve: ↑ panel size has modest

effect on adherence and odds BLQ at 12 month

Years of provider experience or specialty no association

Significant association with ARV class (NNRTI) and year started greatest impact

•IF antiretroviral experienced:

↑ years provider experience associated with ↑ adherence and odds BLQ

No association with panel size or specialty

Significant association with older age, Caucasian, MSM, initiation after 2000Horberg, Hurley, Towner, Allerton, Tang, Catz, Silverberg, Quesenberry, IDSA Abstract 1131, 2010; IAS Abstract MOPE464,

2011

Observational cohort data

Slide 10

Clinical Pharmacists: Roles

• Can have many roles Adherence and adverse effect counseling Manage adverse effects and drug-drug interactions Ombudsman with dispense pharmacies Research staff Potentially case management

• Physicians average 13 minutes entire course of a patient’s care discussing adherence while pharmacists spend 0.5-1.5 hours per visit discussing adherence

Older data, need to update And this was adherence to antiretroviral medications

only

Golin, JGIM, 2004; Geletko, Am J Hlth Sys Pharm, 2002;Rathbun, Clin Ther, 2005

HIV Clinical Pharmacists Study

ARV Naïve ARV Experienced

12 Month

--Continuous Outcome +6.5% (.06) [NOTE: Far greater impact on poorer subpopulation]

+3.1% (.34)

-- OR ≥90% 1.23 (.47) 1.02 (.92)

24 Month

--Continuous Outcomes +8.7% (.01) +3.0% (.40)/ +10.6% (.001)*

--OR ≥90% 3.88 (<.0001) 1.39 (.29)/ 2.49 (.002)*

*--1st value is 0-50 patients; 2nd value is 51+ patients

• 3538 patients evaluated—1571 antiretroviral naïve and 1967 experienced patients

Adherence Results (multivariate analysis, p value):

Horberg, Hurley, Silverberg, Quesenberry, Kinsman, JAIDS, 2007; 44:531-539

HIV Clinical Pharmacists Study (2)

ARV Naïve ARV Experienced

Hospital Days ↑67% (<.0001) [NOTE: Far greater impact on poorer subpopulation] / ↑10% (.36)*

↓14% (.008)/↓38% (<.0001)*

ER Visits ↓2% (.89) ↓18% (.05)

Office Visits ↓12% (<.0001) [NOTE: Far greater impact on poorer subpopulation] / ↓1% (.71)*

↓22% (<.0001) / ↑3% (.16)*

*--1st value is 0-50 patients; 2nd value is 51+ patients

Utilization Results:

Slide 13

HIV Multidisciplinary Care Team Study (1)

• Research Question: What components of the HIV MDCT in combination are

associated with the greatest increases in adherence?

• Retrospective analysis of HIV+ patients in KP California (11,411) initiating a new ART regimen from 1996-2006. ARV Naïve: 7,597 patients ARV Experienced: 3,814 patients

• Measured 12 month adherence to ART regimen using pharmacy dispense/refill records

Horberg, Hurley, Towner, Allerton, Tang, Catz, Silverberg, Quesenberry, Treatment Adherence Conference, 2011

HIV Multidisciplinary Care Team Study (2)

Primary Predictor—Exposure to MDCT component by medical center (26):•HIV Specialist (y/n)•Nurse Case Manager•Non-Nurse Care Coordinator•Clinical Pharmacist•Social Work/Benefits Counselor•Dietician•Mental Health

Other Predictor Variables•Age•Gender•Race/Ethnicity (White, Black, Latino, Other)•HIV Risk (MSM, IDU, Heterosexual)•HCV+•ART Regimen Class•ARV Experienced•Year this ART regimen was initiated (temporal trend)•Medical Center (cluster variable)•Provider (cluster variable)

Slide 15

HIV Multidisciplinary Care Team Study (3)

• Classification and regression tree approach (recursive partitioning) to ascertain potential MDCT compositions associated with maximal mean ART adherence (CART Pro 6.0®, Salford Systems, San Diego, CA)

• From above, potential combinations tested in adjusted* mixed linear regression to determine which associated with maximal ART adherence

*--Clustering by medical center, provider, patient. Adjusted for ART experience, age, gender, race/ethnicity, HIV risk, HCV+, ART regimen class, temporal trend

HIV MDCT Study (4): Recursive Partitioning

* p < 0.05

Multiple team combinations possible with significant effect.

First branch is clinicalpharmacist.

HIV MDCT Study (5): Teams Determined from RPMDCT below not significantly different between them

Team Composition

Mean Adjusted Percent Increase

Adherence 95% Confidence Interval p Value

Specialist Only 74.4% 72.2-76.8%

Clinical Pharmacist Only +3.3% +0.1 to +5.8% 0.01

Pharmacist + non-RN care coordinator +8.1% +2.7 to +13.4% 0.003

Nurse + Social Work/Benefits +7.5% +5.4 to+9.7% <0.001

HIV Specialist + Mental Health +6.5% +2.6 to +10.4% 0.001

Pharmacist + Social Work/Benefits +5.7% +4.0 to +7.4% <0.001

Slide 18

CC = (0)

TerminalNode 1

Class = 0Class Cases %

0 1740 57.31 1295 42.7W = 3035.00

N = 3035

CC = (1)

TerminalNode 2

Class = 1Class Cases %

0 84 36.51 146 63.5

W = 230.00N = 230

RNCCCM = (0)

Node 3Class = 0

CC = (0)Class Cases %

0 1824 55.91 1441 44.1W = 3265.00

N = 3265

SWBC = (0)

TerminalNode 3

Class = 0Class Cases %

0 219 56.21 171 43.8

W = 390.00N = 390

SWBC = (1)

TerminalNode 4

Class = 1Class Cases %

0 648 46.31 752 53.7W = 1400.00

N = 1400

RNCCCM = (1)

Node 4Class = 1

SWBC = (0)Class Cases %

0 867 48.41 923 51.6W = 1790.00

N = 1790

RX = (0)

Node 2Class = 0

RNCCCM = (0)Class Cases %

0 2691 53.21 2364 46.8W = 5055.00

N = 5055

SWBC = (1)

TerminalNode 5

Class = 0Class Cases %

0 1148 51.41 1084 48.6W = 2232.00

N = 2232

SWBC = (0)

TerminalNode 6

Class = 1Class Cases %

0 299 41.61 420 58.4

W = 719.00N = 719

MH = (0)

Node 7Class = 0

SWBC = (1)Class Cases %

0 1447 49.01 1504 51.0W = 2951.00

N = 2951

MH = (1)

TerminalNode 7

Class = 1Class Cases %

0 426 40.61 624 59.4W = 1050.00

N = 1050

CC = (0)

Node 6Class = 1

MH = (0)Class Cases %

0 1873 46.81 2128 53.2W = 4001.00

N = 4001

CC = (1)

TerminalNode 8

Class = 1Class Cases %

0 972 41.31 1383 58.7W = 2355.00

N = 2355

RX = (1)

Node 5Class = 1

CC = (0)Class Cases %

0 2845 44.81 3511 55.2W = 6356.00

N = 6356

Node 1Class = 0

RX = (0)Class Cases %

0 5536 48.51 5875 51.5W = 11411.00

N = 11411

CC onlyp<0.05

RX=clinical pharmacist

RNCCCM=nurse case manager

CC=non-RN case coordinator

SWBC=social work/benefits counselor

MH=mental health worker

Reference group is HIV specialist only

Ref.RNCCM only RNCCM

+ SWBC

RX + SWBCRX onlyp<0.05

RX + MH

RX + CC

This Can Be Applied to Other Outcomes– Odds Maximal Viral Control

Horberg, Hurley, Towner, Allerton, Tang, Catz, Silverberg, Quesenberry, IAS Abstract MOPE422, 2011

CC = (0)

TerminalNode 1

Class = 0Class Cases %

0 1740 57.31 1295 42.7W = 3035.00

N = 3035

CC = (1)

TerminalNode 2

Class = 1Class Cases %

0 84 36.51 146 63.5

W = 230.00N = 230

RNCCCM = (0)

Node 3Class = 0

CC = (0)Class Cases %

0 1824 55.91 1441 44.1W = 3265.00

N = 3265

SWBC = (0)

TerminalNode 3

Class = 0Class Cases %

0 219 56.21 171 43.8

W = 390.00N = 390

SWBC = (1)

TerminalNode 4

Class = 1Class Cases %

0 648 46.31 752 53.7W = 1400.00

N = 1400

RNCCCM = (1)

Node 4Class = 1

SWBC = (0)Class Cases %

0 867 48.41 923 51.6W = 1790.00

N = 1790

RX = (0)

Node 2Class = 0

RNCCCM = (0)Class Cases %

0 2691 53.21 2364 46.8W = 5055.00

N = 5055

SWBC = (1)

TerminalNode 5

Class = 0Class Cases %

0 1148 51.41 1084 48.6W = 2232.00

N = 2232

SWBC = (0)

TerminalNode 6

Class = 1Class Cases %

0 299 41.61 420 58.4

W = 719.00N = 719

MH = (0)

Node 7Class = 0

SWBC = (1)Class Cases %

0 1447 49.01 1504 51.0W = 2951.00

N = 2951

MH = (1)

TerminalNode 7

Class = 1Class Cases %

0 426 40.61 624 59.4W = 1050.00

N = 1050

CC = (0)

Node 6Class = 1

MH = (0)Class Cases %

0 1873 46.81 2128 53.2W = 4001.00

N = 4001

CC = (1)

TerminalNode 8

Class = 1Class Cases %

0 972 41.31 1383 58.7W = 2355.00

N = 2355

RX = (1)

Node 5Class = 1

CC = (0)Class Cases %

0 2845 44.81 3511 55.2W = 6356.00

N = 6356

Node 1Class = 0

RX = (0)Class Cases %

0 5536 48.51 5875 51.5W = 11411.00

N = 11411

CC onlyp<0.05

RX=clinical pharmacist

RNCCCM=nurse case manager

CC=non-RN case coordinator

SWBC=social work/benefits counselor

MH=mental health worker

Reference group is HIV specialist only

Ref.RNCCM only RNCCM

+ SWBC

RX + SWBCRX onlyp<0.05

RX + MH

RX + CC

This Can Be Applied to Other Outcomes– Odds Maximal Viral Control

First branch is clinical pharmacist.

Different results when stratified by ARV status.

Discussion of MDCT Study Results

• First study to consider all elements of HIV MDCT interactively

• Clinical pharmacist is significantly associated with adherence improvement Confirms prior study

• Multiple combinations (5) are associated with significantly improved adherence Compared to HIV specialist only

• Likely some confounding by indication

Slide 21

Next Research Questions/Steps

• Test MDCT optimized teams in prospective study Also in different healthcare systems

• Do these teams also improve other outcomes? Accessing Care Retention in Care Mortality

• Who are the key personnel for linkage to care? Especially in non-integrated care systems

• What services are not addressed by these teams?

Slide 22

Thank you!

The great work continues.(Paraphrased from Angels in America)