Simple assessments of adherence to antiretorviral therapy predict virologic failure in HIV+ patients in Lusaka, Zambia Ronald A. Cantrell, MPH University of Alabama at Birmingham Centre for Infectious Disease Research in Zambia I’d like to thank the organizers of the conference for the opportunity to present this research. Today I will be discussing how simple assessments of adherence to antiretroviral therapy predict virologic failure in HIV+ patients in Lusaka, Zambia.
Background High levels of adherence to antiretroviral therapy (ART) necessary for reliable viral suppression Routine viral load (VL) testing not available in many settings Limited options for second-line therapy Evaluation of adherence critical Preservation of first-line regimens is important to programmatic success As we all know, consistently high levels of adherence to antiretroviral therapy are necessary for reliable viral suppression, and consequent prevention of resistance, disease progression, and death. In resource-limited settings where routine viral load testing is not widely available and there are limited options for second-line therapy, the evaluation of adherence to ART is critical, particularly in the public health approach, where preservation of the first line ART regimen is paramount to programmatic success.
Reported Adherence Measures in Sub-Saharan Africa Are Varied Several studies have looked at adherence in Sub-Saharan Africa in the past 5 years. This Meta-analysis was published last year and identified 9 peer-reviewed articles from 7 African countries in which adherence was the primary or secondary outcome. As you can see, there is no standardized measure of adherence. 8 of the nine studies relied on patient self report, but with several different variations on what is reported, such as past 3 days, 7 days or even month, and the threshold for defining adherence ranges from 80% to 100%. Mills et al JAMA 2006
Different Measures of Adherence Correlate with Viral Suppression Pharmacy claim data have been used to evaluate adherence to NNRTI based therapy in a private sector, managed care setting in South Africa Nachega et al Ann Intern Med 2007 Does adherence correlate with good outcomes. Simple measure in private care and more complex measure in US both show correlation with virologic suppression. Some studies have compared adherence to virologic suppression. Nachega used pharmacy claims in a private sector, manage care program to define adherence (the number of months with pharmacy claim data divided by total number of months on therapy) and reported a dose response relationship. Paterson measured adherence using a Microelectronic Monitoring System among 99 patients in 2 US hospitals. While they did show a dose response relationship, I think the small scale of the study demonstrates the difficulty in resource limited settings. MEMS TrackCap system has been used to evaluate adherence to PI regimens in the US Paterson et al Ann Intern Med 2000
Hypothesis Patients with poor adherence, as estimated by simple, routinely collected pharmacy prescription refill data and by patient self-report, will have detectable plasma viremia.
CIDRZ/MOH ART Program Population: 11.5 million (1.6 in Lusaka) 59,768 Enrolled On ART Zambia’s population of 11.5 million persons is among the world's poorest and most severely affected by AIDS. In the capital city of Lusaka, 22% are estimated to be infected with HIV virus. ART services began in primary health care centers in the Lusaka District in May 2004, with funding from the President's Emergency Plan for AIDS Relief (PEPFAR), and has expanded rapidly over the past 3 years. Our organization has supported the Ministry of Health in enrolling more than 80,000 ART naïve patients into HIV care. Population: 11.5 million (1.6 in Lusaka) HIV Prevalence: 16% (22% in Lusaka) 21,677 Enrolled Not On ART Stringer et al JAMA 2006
Assessment of Patients With Clinical or Immunologic Treatment Failure Because care is primarily provided by non-physician clinicians, care in the Lusaka ART program uses an algorithmic approach. The efficacy of ART is monitored with clinical exam and CD4 count. While the capacity to measure viral load began in December 2006, viral Load testing is certainly not routinely performed. Virologic Failure is assumed when patient meets criteria for BOTH Clinical and Immunologic Failure, but when these results differ, viral load testing performed. This is our study population. While we have Pharmacy refill data on all patients enrolled in the program, viral load results are only available for this population (POINT TO THE BOTTOM RIGHT OF THE SLIDE). ------------------------------------------------ Care in the Lusaka ART program uses an algorithmic approach modified from the WHO Adult ART guidelines (2006). Clinical treatment failure is defined as: A. New or recurrent WHO Stage 3 or 4 event signifying HIV disease progression at least 6 months after starting ART. B. Must rule out IRIS. C. However, discretion is needed by the clinician as some stage 3 and 4 illnesses (i.e. Pulmonary TB or uncomplicated lymph node TB) may not be clinical treatment failure. Immunologic Treatment Failure is defined as: A. Failure of CD4 count to improve, specifically, CD4 increase of <50 cells/µL after 6 months of ART OR Persistent CD4 count <100 cells/µL after 12 months of ART. B. Fall of CD4 count after improvement, specifically, Return or fall to pre-ART baseline after 6 months OR CD4 drop by >30% from CD4 peak on ART. C. However, Concomitant infection must be ruled out. NOTE: CHANGES from the WHO 2006 Guidelines are A)the CD4 rise of <50 cells is not in the WHO guideline and B) the CD4 drop by 50% (vs. 30% we are using) is in the guideline – identify more patients.
Methods Inclusion Criteria Outcome (Detectable Viremia): Exposures On ART for at least 100 days Viral load according to the District treatment failure algorithm Outcome (Detectable Viremia): Defined as > 400 copies/mL Exposures Age, Body Mass Index, Disclosed HIV status to spouse/partner, Spouse/partner tested for HIV, Adherence support (“buddy”) CD4+ lymphocyte count, Hemoglobin WHO Stage
Methods Adherence Pharmacy Refill Data Self-Reported Data Average number of days late for pharmacy visits as a proportion of total time on ART Medication possession ratio (MPR) divided into commonly reported thresholds (<80%, 80-94%, 95-99%, 100%) Self-Reported Data Total number of doses missed in 3 days preceding pharmacy appointment Any vs. None We measured adherence 2 different ways, first with pharmacy refill data and then again with self report data. In the Lusaka ART program, all patients are tracked via an electronic medical record, or SMARTCARE system. When patients report to the pharmacy to receive their medication, the dispensation is recorded along with the date they are due to return to the pharmacy. If they do not show up on time, we know and can calculate the number of days late for pharmacy as a proportion of the total time on therapy. This serves as a proxy for possession of ART. We converted this proportion into a medication possession ratio and then divide it into commonly used strata: (<80%, 80-94%, 95-99%, 100%). Adherence counselors speak with the patient at every pharmacy visit and again in the home if they are more than 5 days late for a clinical appointment.
Cohort Characteristics N=753 When you look at these characteristics broken down by adherence category, none of them were significantly different with the exception of the presence of adherence support.
Viral Suppression by Medication Possession Ratio in Lusaka Shape of our curve in public-sector program in Zambia looks like Nachega’s in the managed care setting in SA. 245 of 753 patients tested (33%) had detectable VL. 11
Relative Risk of Detectable Viremia by Medication Possession Ratio 245 of 753 patients tested (33%) had detectable VL. *adjusted for baseline CD4+ lymphocyte count, adherence support, and age
Relative Risk of Detectable Viremia by Self-Reported Missed Doses *adjusted for baseline CD4+ lymphocyte count, adherence support, and age 245 of 752 patients tested (33%) had detectable VL.
Limitations Both adherence measures are surrogate markers for true adherence May not be generalizable: viral loads not routinely ordered No confirmatory viral load measurement Strengths: MPR may overestimate adherence, but not as grossly as other common (cheap) method used in developing world: i.e. self-report. Operational context of large public sector program. VL data analyzed reflect the rationing strategy utilized in clinical care in the resource-poor setting. 14
Conclusions Adherence based on pharmacy refill data predicts virologic failure Self-reported missed doses does not appear to predict virologic failure in a dose response relationship While few people reported missing ART doses, those with positive responses were more likely to have detectable viral load.
Acknowledgements Zambian Ministry of Health The Lusaka Urban District Health Management Team Co-authors Collaborators Patients