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 % undetectable viral load N =91 90% 80% 70% 60% 50% 40% 79% 30% 48% 20% 32% 29% 10% 18% 0% >=95 90-94.9 80-89.9 70-79.9 <70 Adherence to HAART measured for 6 mos (%) Paterson DL et al. Ann Intern Med. 2000;133:21-30.

Electronic Monitoring (MEMs)

Pharmacy Adherence and VL % undetectable viral load N =886 90% 80% 70% 60% 50% 84% 40% 64% 30% 47% 20% 24% 10% 12% 0% >=95 90-94.9 80-89.9 70-79.9 <70 Adherence to HAART measured for 19 mos (%) Low-Beer et al, JAIDS. 2000;23:360

MEMS and Self-report and VL N = 67 % undetectable viral load 90 80 80 MEMS 70 Self-report 60 61 50 40 62 30 33 20 33 10 13 10 0 >95 70-95 50-70 <50 Adherence to HAART measured for six months (%) Arnsten et al, CID. 2001;33:1417-1423

Self-report adherence and VL % undetectable viral load N = 173 90 80 70 60 50 40 30 47 41 20 10 17 7 0 100 95-99 80-95 <80 Adherence to HAART over 6 months (%) 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 3.0 (2-5) regimen > 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 100 90 79 + 26% 78 + 23% 80 70 % Adherence 57 + 32% 53 + 34% 60 50 40 30 20 10 0 MEMS adherence, MEMS adherence, Self-reported Self-reported past week past day adherence, past adherence, past week day % 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 100 90 79 80 64 70 % Adherence 60 50 39 40 26 30 20 10 0 MEMS: % of MEMS: % of MEMS: % of SR: % of doses days at least days correct # days all doses taken on one dose of doses taken taken within schedule taken 25% of interval

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

Comparing Methods Within Studies: Self-Report v. Pill Count 100 80 60 40 Patient Report 20 0 0 20 40 60 80 100 Pill Count 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 Unannounced PC vs Log Viral Load Patient Self Report vs Log Viral Load 6 6 6 5 5 5 4 4 4 3 3 3 Log Viral Load Log Viral Load Log Viral Load 2 2 2 1 1 1 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 Adjusted MEMS Pill Count % Adherence Self Report % Adherence Adjusted MEMS Pill Count Self Report R sq = 0.67 R sq = 0.45 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 -0.52 <0.001 interval 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 N = 1219 % survival after 3 years 90 80 70 60 <90% 90% or more Adherence (self-report and pharmacy) 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 1.53 <0.001 cell decrease) Wood et al, AIDS. 2002;16:2065-2072

Adherence and Drug Resistance