Adherence with Antiretroviral Therapy in HIV-infected Drug Users - - PowerPoint PPT Presentation

Adherence with Antiretroviral Therapy in HIV-infected Drug Users

Julia H. Arnsten, M.D., M.P.H. Associate Professor of Medicine, Epidemiology, and Psychiatry & Behavioral Sciences Albert Einstein College of Medicine Montefiore Medical Center

Outline

• How much adherence is enough
• Measuring adherence
• Barriers to adherence
• Adherence over time
• Directly Observed Therapy

How much adherence is enough?

• To prevent viral replication
• To prevent disease progression
• To prevent the development of drug

resistance

Adherence and viral load

MEMS Adherence and VL

Adherence to HAART measured for 6 mos (%) N =91

79% 48% 32% 29% 18% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% >=95 90-94.9 80-89.9 70-79.9 <70

% undetectable viral load

Paterson DL et al. Ann Intern Med. 2000;133:21-30.

Electronic Monitoring (MEMs)

Pharmacy Adherence and VL

Adherence to HAART measured for 19 mos (%) N =886

64% 47% 24% 12% 84% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% >=95 90-94.9 80-89.9 70-79.9 <70

% undetectable viral load

Low-Beer et al, JAIDS. 2000;23:360

MEMS and Self-report and VL

80 61 33 10 62 33 13 10 20 30 40 50 60 70 80 90 >95 70-95 50-70 <50

Adherence to HAART measured for six months (%)

MEMS Self-report

N = 67 % undetectable viral load

Arnsten et al, CID. 2001;33:1417-1423

Self-report adherence and VL

47 41 17 7

10 20 30 40 50 60 70 80 90

% undetectable viral load

100 95-99 80-95 <80 Adherence to HAART over 6 months (%)

N = 173

Haubrich, et al, AIDS 1999

METHODS

• Design : Prospective 6 month study, 1998-2001
• Population: HIV+ current / former drug users recruited

from ongoing longitudinal cohort

• Monthly research visits

– Adherence measures: MEMS caps, Self-report – HIV viral load – Recent drug use – HIV specific attitudes and beliefs – Symptoms and side effects – Standard psychosocial measures

• Comparison of MEMS and self-report data for the day

and the week preceding each research visit

Calculation of Mean Adherence

• Adherence calculated as: No. of doses taken

x 100

• No. of doses prescribed
• Adherence was first calculated for each drug at each

monthly interview. Then, monthly mean adherence was calculated for all drugs taken. Finally, aggregate mean adherence was calculated for the entire study period.

• Aggregate mean adherence was calculated for each subject

for both MEMS and self-report, and compared between pre- determined sub-groups using t-tests and linear regression.

• HIV RNA (copies/ml) was measured at each interview, and

aggregate mean HIV RNA was calculated for the entire study period (excluding the baseline interview).

Adherence Indices

• One-day adherence

– SR and MEMS

• One-week adherence

– SR and MEMS

• Adherence during the entire study period

– MEMS

• Dose interval adherence

– % of days on which at least one dose was taken (MEMS) – % of days on which the correct number of doses was taken (MEMS) – % of days on which all doses were taken within 25% of correct dosing interval (MEMS) – % of doses taken “on schedule” (SR)

Baseline Characteristics (N = 115)

Sex: Male 61 Female 39 Race: Black 24 Hispanic 60 White 12 Mean age, y (range) 43 (23-61) Methadone maintenance 96 Marital status: Married 31 Separated/divorced/single 69 Heroin use 30 Cocaine use 33 Alcohol use 22 Receiving public assistance 99 Unemployed 87

CD4 Count and HIV RNA

Median length of HIV infection, y (range) 7.3 (0.4-13.8) Median CD4+ count, cells/mm3 (range) 324 (3-1512) CD4+ count, cells/mm3 <50 13 50-200 22 201-500 46 >500 19 Baseline HIV RNA, copies/ml <50 29 50-500 26 501-10,000 8 10,001-100,000 14 >100,000 9

Current and Prior ART Experience

No prior ART 15 1 - 3 prior ART medications 45 > 4 prior ART medications 42 Current Protease Inhibitor 79 TID Regimen 37 Mean number of different drugs in current regimen 3.0 (2-5) > 3 medications in current regimen 84

Self-Reported Adherence

• “I’d like you to focus just on yesterday. Did

you skip any of your pills yesterday?”

• “Still focusing on yesterday, can you tell me

exactly how many pills you took?”

• “Now, I’d like you to focus on the past week,

thinking back to to last ______. In the last week, not counting today, how many pills did you miss taking altogether?”

Mean Antiretroviral Adherence Rates by MEMS and Self-report: Past Week and Past Day

10 20 30 40 50 60 70 80 90 100

MEMS adherence, past week MEMS adherence, past day Self-reported adherence, past week Self-reported adherence, past day

% Adherence

53 + 34% 57 + 32% 78 + 23% 79 + 26%

% Adherence = doses taken/doses prescribed x 100 p<0.0001 for comparison of MEMS v. self-report

Arnsten et al, CID, 2001;33:1417-1423

Sensitivity and Specificity of Self-Reported Non-Adherence

• Sensitivity of Self-Report = 50%

Only 50% of subjects who are <80% adherent (by MEMS) report non-adherence

• Specificity of Self-Report = 95%

Only 5% of subjects who are >80% adherent (by MEMS) report non-adherence

• Self-report of non-adherence is insensitive but

highly specific, using MEMS as gold standard

Dose Interval Adherence

–% of days on which at least one dose was taken (MEMS) –% of days on which the correct number of doses was taken (MEMS) –% of days on which all doses were taken within 25% of correct dosing interval (MEMS) –% of doses taken “on schedule” (SR)

Dose Interval Adherence

Arnsten et al, CID, 2001;33:1417-1423 64 39 26 79

10 20 30 40 50 60 70 80 90 100

MEMS: % of days at least

• ne dose

taken MEMS: % of days correct #

• f doses taken

MEMS: % of days all doses taken within 25% of interval SR: % of doses taken on schedule

% Adherence

Objectively Measured Adherence Across Studies

Clinic pill count 80% MEMS 54% CID 2001 McNabb MEMs 45-64% AIDS 2002 Howard MEMS 53% CID 2001 Arnsten Clinic pill count 83% MEMS 63% Annals Int Med 2001 Liu MEMS 74% Annals Int Med 2000 Paterson Unannounced pill count 73% MEMS 67% AIDS 2000 Bangsberg

Comparing Methods Within Studies: Self-Report v. Pill Count

Pill Count 100 80 60 40 20 Patient Report 100 80 60 40 20

R sq = 0.72 Bangsberg et al JAIDS 2001:26:435

Comparing Correlations between Different Adherence Measures and VL

MEMS, Pill Count, and Self-Report Correlations with VL

Adjusted MEMS vs Log Viral Load Adjusted MEMS 100 80 60 40 20

Log Viral Load 6 5 4 3 2 1

Unannounced PC vs Log Viral Load Pill Count % Adherence 100 80 60 40 20

Log Viral Load 6 5 4 3 2 1

Patient Self Report vs Log Viral Load Self Report % Adherence 100 80 60 40 20

Log Viral Load 6 5 4 3 2 1

Adjusted MEMS R sq = 0.67 Pill Count R sq = 0.45 Self Report R sq = 0.36

Bangsberg et al AIDS 2000 14(4)357-66

Correlations between Adherence and VL

Arnsten et al, CID, 2001;33:1417-1423 Measure R P- value Self-Report: Past week

• 0.52

<0.001 Self-Report: Past Day

• 0.43

<0.001 MEMS: Past Week

• 0.55

<0.001 MEMS: Past Day

• 0.46

<0.001 MEMS: Entire Study Period (Median=165 days)

• 0.57

<0.001 MEMS: % of days with correct no. doses

• 0.60

<0.001 MEMS: % of days with all doses < 25% of correct interval

• 0.52

<0.001 MEMS: % of days with at least one dose

• 0.53

<0.001

Adherence Measures

Self Report

• Strengths

– Simple, practical – Cheap – Allows for collection of other important information

• Weaknesses

– Social desirability bias – Recall inaccurate over long time periods – No consensus on how to ask, how to score – Only certain time intervals can be assessed

Pill Counts

• Strengths

– Objective – Relatively cheap (but not that cheap!)

• Weaknesses

– Pill dumping – Can’t tell when the pills were taken – Difficult to be sure of the correct “start date” for the pill supply – Patients may use multiple pill containers

Electronic Monitoring (MEMS, eDEM)

• Strengths

– Objective – Sensitive for detecting non-adherence – Can evaluate adherence to dosing intervals – Historically, “gold standard” – Lots of data

• Weaknesses

– Expensive, bulky, cumbersome – “Pocket pills” – Biased selection of patients – Lots of data

Pharmacy-based Adherence

• Strengths

– Unobtrusive – Population-based

• Weaknesses

– Problems with claims data (e.g., erroneous, intermittent eligibility) – Difficult to distinguish discontinuing a drug under provider’s supervision from poor adherence – Misses drugs from clinical trials or samples from office

Adherence and HIV progression

Adherence and Survival

60 70 80 90 <90% 90% or more

% survival after 3 years

Adherence (self-report and pharmacy)

N = 1219

Garcia de Olalla et al, JAIDS. 2002;30:105-110

Adherence and Survival

Factors Associated with Risk of Death

Variable Adjusted RR P value Age (per year) 1.02 0.07 Adherence (per 10% increase) 0.83 <0.001 Lower median income 2.03 0.001 Baseline CD4 count (per 100 cell decrease) 1.53 <0.001

Wood et al, AIDS. 2002;16:2065-2072

Adherence and Drug Resistance

Viral Load and Adherence in Four Selected Patients at Six Monthly Timepoints

1 10 100 1000 10000 20 40 60 80 100 120 1 10 100 1000 10000 20 40 60 80 100 1 10 100 1000 10000 100000 20 40 60 80 100 1 10 100 1000 10000 20 40 60 80 100

Viral load (copies/ml) % A d h e r e n c e

Development of resistance -1 Poor adherence and high VL

• Baseline and follow-up viral load > 1000
• Mean adherence = 23% (range: 0 - 47%)
• 50% had WT virus at baseline and f/u
• 50% had high level resistance to one or

more currently prescribed drugs at baseline and developed no new mutations

• None developed new mutations during the

6 month protocol

Development of resistance - 2 Good adherence and high VL

• Baseline and follow-up VL > 1000
• Adherence > 80% (range: 83 - 99%)
• All had high level resistance to two of their

currently prescribed drugs at baseline and developed no new mutations during the 6 month protocol

Development of resistance - 3 Adherence and resistance after viral rebound

• VL < 400 with rebound to VL > 1000
• Mean adherence before rebound = 54%
• Mean adherence after rebound = 8%
• 50% developed resistance (M184V, K103N)

RESISTANCE AND ADHERENCE

• Among 30 subjects with persistent viral replication (despite HAART),
• nly 57% were resistant to >1 drug
• Among 17 subjects with resistance, those with poor adherence

(<50%) had fewer mutations than those with moderate adherence

1 2 3 4 5 6 Poor adherence Moderate adherence

Number of mutations among subjects with resistance

10 20 30 40 50 60 70

Resistant Not resistant

Per cent of subjects with persistent viral replication

Adherence & Genotypic Resistance

Adherence (%)

110 90 70 50 30 10

• 10

Number of Mutations

9 8 7 6 5 4 3 2 1

• 1

Correlation = 0.8 P < .0005 Howard et al, 40th IDSA, 2002

Adherence & Genotypic Resistance

Adherence (%)

100 80 60 40 20

• No. of mutations

9 7 5 3 1

• 1

Correlation = 0.59 p = 0.001

0 -

JC Walsh, K Hertogs, BG Gazzard ICAAC 2000 #699

New Drug Resistance Mutations by Adherence Level Over 6 Months

0% 20% 40% 60% 80% 100% 1 2 3 4 #New Drug Resistant Mutations (IAS-USA)

Pill Count Adherence > p=0.02

Bangsberg CROI 2002 546-T

Summary: Adherence and Resistance

• Higher levels of adherence are associated

with lower viral loads and better outcomes.

• Among individuals with persistent viremia,

higher adherence is also associated with greater resistance.

• The majority of subjects with persistent viral

replication have poor adherence. Resistance is less frequent in subjects with poor adherence, likely due to a lack of sufficient selective pressure.

Barriers to Adherence

MEMS Adherence Rates and Recent Drug or Alcohol Use

10 20 30 40 50 60 70 80 Active Cocaine No Active Cocaine Active Heroin No Active Heroin Active Etoh No Active Etoh % Adherence

Arnsten et al, JGIM, May 2002

Adherence to ART by Gender and by Measure

% >90% adherent

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 MEMS Self-report Pharmacy

women women women men men 6 months

1 Arnsten JH et al. CID, 2001.

• 2. Turner BJ et al. Medical Care, 2000.

Univariate Predictors of Adherence

Variable % Adherence P-value Gender: Male 73 0.04 Female 46 Drug Use: Former 72 0.02 Active 38 Social network: <5 72 0.04 >5 46 Housing: Stable 75 0.003 Unstable 42 Belief in efficacy: Yes 77 0.002 No 45 Hopeful attitude: Yes 75 0.002 No 44 Living with partner: Yes 73 0.04 No 46

Multivariate Predictors of Adherence

Variable % Adherence P-value Gender: Male 73 0.04 Female 46 Drug Use: Former 72 0.02 Active 38 Social network: <5 72 0.04 >5 46 Housing: Stable 75 0.003 Unstable 42 Belief in efficacy: Yes 77 0.002 No 45 Hopeful attitude: Yes 75 0.002 No 44 Living with partner: Yes 73 0.04 No 46

Factors Associated with Adherence Linear model, R2 = 0.25

0.04 12.3 Stable housing 0.002 21.2 HIV group Violent life events 0.03

• 13.3

Large social network Heavy alcohol use 0.002

• 19.5

Current drug use 0.01

• 14.6

Female NS 6.3 Depression

Factors Associated with Adherence Linear model, R2 = 0.25

0.04 12.3 Stable housing 0.002 21.2 HIV group Violent life events 0.03

• 13.3

Large social network Heavy alcohol use 0.002

• 19.5

Current drug use 0.01

• 14.6

Female NS 6.3 Depression

Factors Associated with Adherence – Men Linear model, R2 = 0.22

Stable housing 0.001 30.6 HIV group 0.03

• 29.2

Violent life events 0.02

• 18.9

Large social network Heavy alcohol use Current drug use Female NS 2.7 Depression

Factors Associated with Adherence – Women Linear model, R2 = 0.28

Stable housing HIV group 0.03

• 35.0

Violent life events Large social network 0.001

• 32.5

Heavy alcohol use Current drug use Female NS 11.8 Depression

Summary: Barriers to Adherence

• Active (but not former) drug/alcohol use
• Lack of belief in meds
• Poor self-efficacy or “fit” with routine
• Mental illness (esp. depression)
• Lack of HIV support
• Violent life events
• Unstable housing

Interactions between HAART and Methadone

• PIs

– RTV, IDV, and Kaletra decrease methadone levels

• NNRTIs

– both NVP and EFV decrease methadone levels

• NRTIs

– methadone decreases ddI AUC by 60% and d4T AUC by 18% – methadone increases AZT AUC by 40%

Methadone Serum Concentrations

100 200 300 400 500 600 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Time (hours) Methadone (ng/ml)

Methadone Methadone + HAART

McCance-Katz EF, Am Jour on Addictions 2002; 11(4):271-278

Opiate Withdrawal Severity Declines with Increased Methadone Dose During HAART

5 10 15 20 25 30 35 40 45 50 Baseline 1 2 3 4 5 6 7 8 9 10 11 12

Study Week

Methadone Dose Change From Baseline (mg)

1 2 3 4 5 6

Objective Opiate Withdrawal Scale Score Change from Baseline Dose Opiate Withdrawal Score Mean

McCance-Katz EF, Am Jour on Addictions 2002; 11(4):271-278

Adherence Over Time

METHODS

• Design: Prospective 6 month study
• Population: HIV+ women on combination ART

recruited from HER Study cohort

• Semiannual visits

– Standardized interview – CD4+ count and HIV viral load

• Monthly research visits

– Adherence to all ART with MEMS caps

Baseline Participant Characteristics (N=161)

Mean age (std. deviation) 41.3 (±6.8) Race: Black 111 (69) White 26 (16) Hispanic 22 (14) Employed 31 (19) Ever injected drugs 90 (56) Recent IDU 14 (9) Recent crack cocaine 21 (13) Recent alcohol use 28 (17) ART use: < 2 years 46 (29) >2-4 years 87 (54) >4 years 28 (17) Protease inhibitor 81 (50)

Antiretroviral Adherence Over Time

10 20 30 40 50 60 70 80

1 2 3 4 5 6

Months of Observation

Mean Adherence (%)

* * *

Howard et al. AIDS 2002;16:2175-82.

Distribution of Changes in Adherence between Consecutive Monthly Measurements

117 146 151 159 161 N =

Consecutive Monthly Comparisons

5 4 3 2 1 Change in Monthly Adherence (%)

30 20 10

• 10
• 20
• 30
• 40

Howard et al., AIDS 2002;16:2175-82.

Consecutive Monthly Adherence Measurements in 4 Subjects with a Mean Overall Adherence of 61%

20 40 60 80 100 20 40 60 80 100 20 40 60 80 100 20 40 60 80 100

Adherence (%)

Factors Associated with Worsening Adherence

• Hospitalization
• Change in regimen
• Duration of ART > 2 years in those with

a detectable viral load

Howard et al. AIDS 2002; 16:2175-82

Changes in Use of and Adherence to ART Stratified by Changes in Drug and Heavy Alcohol Use

Lucas et al. AIDS 2002;16:767-74

Antiretroviral Adherence Over Time: Impact of Active Drug Use

20 25 30 35 40 45 50 55 60 65 70 1 2 3 4 5 6

Months of observation Per cent adherence

No drug use Heroin and cocaine

Adherence Changes Over Time: Implications for the Clinician

• Assessment of adherence should be an
• ngoing process
• Adherence interventions needed throughout

the course of therapy

• Factors associated with worsening

adherence:

– Relapse in substance abuse – Depression – Change in ART regimen – Intercurrent illness

Adherence Changes Over Time: Implications for the Investigator

• Studies of adherence-enhancing

interventions must include longitudinal strategies

– intervention must promote consistent adherence over time (e.g. “boosters) – measurement of intervention effect must be long-term

Adherence-enhancing interventions

• Adherence counseling using cognitive-

behavioral and motivational interviewing techniques (even one session)

• Prompt, frequent, and intensive follow-up
• Multi-disciplinary adherence team
• Beepers, alarms, and watches
• Directly observed therapy

Why DOT for HIV is different than DOT for TB

• Frequency of medication-taking
• Duration of treatment
• TB is curable
• Differences in transmission routes have

different public health implications

Summary of DOT studies

Author N Setting Length % UD Conway, 2002 52 Methadone 18 mos 65 Mitty, 2002 25 Community 6 mos 63 Fischl, 2001 50 Prison 12 mos 100 Lucas, 2002 31 Methadone 8 mos 75 Babudieri, 2000 37 Prison 9 mos 62 Clarke, 2002 39 Methadone 12 mos 58

Summary: Does DOT Make Sense

• To date, HAART DOT programs have

targeted select populations

• But non-adherence extends beyond

these groups

• DOT should be based on individual

adherence, not membership in a particular group

Summary

• Objective measurement of adherence is

more highly correlated with viral load than self-report

• Modifiable barriers to adherence have

been identified

• Adherence changes over time
• Interventions that improve adherence

are emerging