330 likes | 451 Views
Adherence: The Achilles Heel of Clinical Trials. David Bangsberg, MD, MPH Director Massachusetts General Hospital Center for Global Health Professor Harvard Medical School Professor Harvard School of Public Health Visiting Professor Mbarara University of Science and Technology
E N D
Adherence: The Achilles Heel of Clinical Trials David Bangsberg, MD, MPH Director Massachusetts General Hospital Center for Global Health Professor Harvard Medical School Professor Harvard School of Public Health Visiting Professor Mbarara University of Science and Technology March, 2014
Acknowledgements • Jessica Haberer • Steve Safren • Alex Tsai • Christina Psaros • Ingrid Katz • Elizabeth Garrett-Mayer
PrEP: A Case Study • Adherence and heterogeneity • Determinants of PrEP Adherence • Measuring adherence to guide interpretation of efficacy vseffectiveness • Post hoc analysis to estimate efficacy conditioned on adherence • Adherence intervention to improve efficacy estimates
Sources of Heterogeneity • Drop-out • Cross-over • Adherence • Other health-related behaviors related to adherence
Adherence and Efficacy in PrEP Trials Donnell et al CROI 2012 Grant et al N Engl J Med 2010 Van Damme et al CROI 2012 Paxton et al FDA 2012
PrEP: a Case Study • Adherence and heterogeneity • Determinants of PrEP Adherence • Measuring adherence to guide interpretation of efficacy vseffectiveness • Post hoc analysis to estimate efficacy conditioned on adherence • Adherence intervention to improve efficacy estimates
PREP adherence is opportunity to mitigate tension and strengthen relationship “Discordance dilemma” Excellent PrEPAdherence Explained by Relationship Dynamics (Ware et al JAIDS 2012) PrEP Resolves Tension in a Committed HIV Discordant Sexual Relationship
PREP adherence is opportunity to mitigate tension and strengthen relationship “Discordance dilemma” Excellent PrEPAdherence Explained by Relationship Dynamics (Ware et al JAIDS 2012) PrEP Resolves Tension in a Committed HIV Discordant Sexual Relationship Love
PREP adherence is opportunity to mitigate tension and strengthen relationship “Discordance dilemma” Excellent PrEPAdherence Explained by Relationship Dynamics (Ware et al JAIDS 2012) PrEP Resolves Tension in a Committed HIV Discordant Sexual Relationship Discord and Distrust
PrEP Adherence During Relationship Discord and Distrust Courtesy of Fran Priddy IAVI
Women’s Experiences with Oral and Vaginal Pre-Exposure Prophylaxis: The VOICE-C Qualitative Study in Johannesburg, South Africa (Arianevan der Straten et al PLoS One 2014) • Qualitative interview of 102 VOICE participants. • ambivalence toward research • preserving a healthy status • managing social relationships
PrEP: A Case Study • Adherence and heterogeneity • Determinants of PrEP Adherence • Measuring adherence to guide interpretation of efficacy vseffectiveness • Post hoc analysis to estimate efficacy conditioned on adherence • Adherence intervention to improve efficacy estimates
Measuring Adherence • Patient report • Longer intervals (Lu AIDS and Beh 2008) • Rating scales (Lu AIDS and Beh 2008; Deschampes AIDS Pt Care STD 2008) • Pill counts • Clinic (Donnell AIDS and Beh 2013) • Unanounced at home (Bangsberg AIDS 2000) • Unannouned via telephone (Kalichman JGIM 2007) • Drug level • Blood (Liechty AIDS 2004) • Hair (Liu PLoS One2014) • Pharmacy refill (Grossberg J ClinEpi 2004) • Electronic • Flash memory/MEMS (Paterson Ann Int Med 2000) • Real-time wireless (Haberer AIDS and Beh 2011)
PrEP: A Case Study • Adherence and heterogeneity • Determinants of PrEP Adherence • Measuring adherence to guide interpretation of efficacy vseffectiveness • Post hoc analysis to estimate efficacy conditioned on adherence • Adherence intervention to improve efficacy estimates
High Level of Protection When Tenofoviris Detected in Blood Donnell et al CROI 2012 Abstract 30 Grant et al N Engl J Med 2010
Adherence to Antiretroviral Prophylaxis for HIVPrevention: A Substudy Cohort within a Clinical Trial ofSerodiscordant Couples in East Africa(Haberer et al PLoS Med 2013) • 3 Site substudy of Partners PrEPStudy: • RCT of TNF/FTC, TNF, placebo • 1,147 HIV serodiscordantcouples in Kenya and Uganda • Adherence measures • unannounced home-based pill counts • electronic pill bottle monitoring. • Intensified counseling for unannounced pill count adherence < 80%. • 53% male, median age: 34 yr, and median partnership duration: 8.5 years.
Adherence to Antiretroviral Prophylaxis for HIVPrevention: A Substudy Cohort within a Clinical Trial of SerodiscordantCouples in East Africa(Haberer et al PLoS Med 2013)
PrEP Efficacy • HIV-1 infections • 14 in 404 participants on placebo (333 person-years) • 0 infections in 750 participants on active drug (616 person-years) • PrEP efficacy within the sub-study was 100% (95% CI 87-100%)
Impact of partial adherence in randomized controlled trials (Weiss et al EmergThemes Epidemiol 2008)
Impact of partial adherence in randomized controlled trials (Weiss et al Emerg Themes Epidemiol 2008) Partners PrEP Substudy Adherence: 96% Efficacy: 100% TDF2 Adherence: 79% Efficacy: 62% IPrex Adherence: 51% Efficacy: 44% Voice Adherence: 25% Efficacy: 4% Fem Prep Adherence: 21% Efficacy: 6%
Impact of partial adherence in randomized controlled trials (Weiss et al Emerg Themes Epidemiol 2008) Partners PrEP Substudy Adherence: 96% Efficacy: 100% TDF2 Adherence: 79% Efficacy: 62% IPrex Adherence: 51% Efficacy: 44% Voice Adherence: 25% Efficacy: 4% Fem Prep Adherence: 21% Efficacy: 6%
PrEP: A Case Study • Adherence and heterogeneity • Determinants of PrEP Adherence • Measuring adherence to guide interpretation of efficacy vseffectiveness • Post hoc analysis to estimate efficacy conditioned on adherence • Adherence interventions to improve efficacy estimates
PreEP Adherence Intervention Design (Psaros and Safren IAPAC 2013) • Based on Lifesteps, an ART treatment adherence intervention (Safren et al., 1997; 2001; 2007) targeted to HIV-negative participants with low (<80%) unannounced pill count adherence • Counseling occurs in two phases: • (1) with individual on PrEP • (2) with their HIV infected partner (optional) • Number of sessions tailored to needs of participants (minimum of 1) • Median number of sessions completed = 10 (IQR 5, 16) • Median length of sessions: • Session one: 40 minutes (IQR 30,50) • Session length decreased to median of 20 minutes (IQR 15,30) by session four.
Intervention Results • 168 (14.6%) of participants triggered the intervention due to <80% unannounced pill count adherence Counseling occurs in two phases: • 9 were ineligible for follow-up analysis • Of remaining 159: • 154 (91.7%) received at least one intervention session • 46 (94.8%) had adequate MEMS data to examine adherence following the intervention.
Enrolled in ancillary adherence study prior to July 10, 2011 (N= 1,147) Did not trigger intervention (N= 979) Triggered but ineligible for follow-up analysis* (N=9) Triggered intervention (N= 168) Completed only one session (N= 2) Did not receive at least one intervention session (N= 5) Received at least one intervention session (N= 154) Completed 2-4 sessions (N= 25) Completed >4 sessions (N= 127) Adequate MEMS data for analysis** (N= 146) Supplemental content figure I. Participant flow • Taken off drug due to exit from study (N=1), seroconversion (N=1), or triggering immediately before unblinding of the placebo arm during the parent trial (N=7) • **Had MEMS data available for the trigger interval, the interval between the trigger and intervention (minimum of 3 days), and following the first intervention session.
Mean crude and predicted adherence by months since the first intervention session (Psaros and Safren IAPAC 2013)
Conclusions • Edward Koop • Drugs don’t work if people don’t take them
Conclusions • Edward Koop • Drugs don’t work if people don’t take them • Corollary • It’s as important to find out if people will take a drug as it is to find out if it works
Conclusions • Edward Koop • Drugs don’t work if people don’t take them • Corollary • It’s as important to find out if people will take a drug as it is to find out if it works • Phase II adherence behavior finding studies
Conclusions • Edward Koop • Drugs don’t work if people don’t take them • Corollary • It’s as important to find out if people will take a drug as it is to find out if it works • Phase II adherence behavior finding studies • More attention to evidence based approaches to monitoring and supporting adherence to improve efficacy estimates