The study is a randomized, non-blinded trial comparing DAART and SAT in HIV-infected participants who are receiving medication assisted treatment at 1 of 5 OTPs in Baltimore, Maryland. In the DAART intervention, study assistants observe morning doses of antiretroviral therapy (ART) on weekdays when participants attend the OTP; other doses are self-administered. Participants in the control arm self-administer all doses. The primary outcome is viral load suppression, measured at 3, 6, and 12 months. We provide DAART for up to 12 months, after which participants assigned to this arm convert to self-administration. We follow participants to 18 months to assess the potential persistence of any benefits realized during the intervention period.
We enrolled participants from 5 OTPs in Baltimore between May 2006 and May 2010. Initially, the study began with three sites (1, 2, and 3). However, we discontinued recruitment at site 3 due to slow enrollment and replaced it with sites 4 and 5 in August 2007 and August 2008, respectively. Three programs were hospital-affiliated OTPs (sites 1, 3, and 5), and two were independent programs (sites 2 and 4). The OTP censuses ranged from 153 to 1007. All sites provided methadone maintenance therapy, and site 1 also provided buprenorphine maintenance therapy. Sites 1 and 5 had on-site HIV care available throughout the study, sites 2 and 3 had on-site HIV care available for part of the study period, and site 4 did not have on-site HIV-care services.
Individuals were eligible for the study if they were 18 years of age or older, HIV seropositive, had received maintenance therapy with methadone or buprenorphine for > 3 weeks with no plans to discontinue, and had an identified HIV provider and active insurance coverage for ART. Moreover, to be included in the study, we required participants to be ART experienced or (if treatment naïve) eligible for ART according to March 23, 2004 Department of Health and Human Services guidelines for treatment [8
], which included history of opportunistic condition, HIV-related symptoms, CD4 count < 350 cells/mm3
, or HIV RNA > 55,000 copies/mL. ART regimens were prescribed by HIV providers prior to randomization and had to include ≥ 3 drugs, including a protease inhibitor, non-nucleoside reverse transcriptase inhibitor, abacavir, or integrase inhibitor. Additionally, we required verbal agreement to participate from participants' HIV medical providers. Exclusion criteria included stable ART use with HIV RNA < 500 copies/mL, ART dosed more frequently than twice daily, use of liquid antiretroviral preparations, participation in another HIV adherence program in which medication is directly administered, or fewer than 1.5 'active' drugs in the specified regimen - as predicted by prior genotypic resistance tests (if available) and mutation interpretation guidelines [9
Research assistants, employed full-time at the sites and familiar with the OTP clients, were responsible for recruitment and initial screening of participants. We posted study flyers at all sites and made regular presentations to counselors, intake specialists, and administrators to describe the purpose of the study and request referral of HIV-infected clients for screening. Additionally, we facilitated on-site HIV counseling and testing services at the sites to improve access to testing and to identify HIV-infected persons who were not aware of their status. The research assistants and study coordinator obtained and reviewed medical records and contacted HIV care providers to obtain their agreement to participate.
Human Subjects Protection
The study was approved by the Johns Hopkins Medicine Institutional Review Board, the University of Maryland Baltimore Institutional Review Board, and the Veterans Administration Research and Development Committee. Participants provided written informed consent, and were encouraged to take the consent document home to discuss with family or medical providers prior to joining the study. We also asked participants to sign releases for medical and substance abuse treatment records to permit confirmation of study eligibility and to record outcome data during the trial. To protect confidentiality, unique study identification numbers were used on all case report forms and clinical samples. We stored hard copies of study records in locked filing cabinets inside locked offices. We store electronic data on a password-protected laptop computer that was backed-up each night to a secure server. We obtained a Certificate of Confidentiality from the National Institute on Drug Abuse to further protect sensitive participant data.
To prevent knowledge of treatment assignment from influencing the selection of antiretroviral drugs, we required HIV providers to prescribe ART regimens prior to randomization. Prior to the study, one of the investigators (GML) generated random treatment assignments to DAART and SAT in a 1:1 ratio, stratified by study site and ART exposure at baseline (naïve or experienced), using a commercial statistical software package. Assignments were generated in blocks that randomly varied in size between 2 and 8. The treatment assignment list was incorporated into a Microsoft Access-based program that revealed individual assignments sequentially as new participants were enrolled. When enrollment criteria were fulfilled for a new participant, the study coordinator obtained the treatment assignment by activating the "randomize" function in the computer program. The assignment list was password protected and was not accessible to the research coordinator or research assistants, apart from revealing sequential treatment assignments at the times of randomization.
We conducted study visits at baseline, 3, 6, 12, and 18 months. Study visits were conducted at OTPs and included an update of contact information, a face-to-face interview, and collection of blood and urine specimens. We measured CD4 cell counts (flow cytometry) and HIV RNA levels (AMPLICOR HIV-1 Monitor Test, version 1.5, Roche Diagnostics, Basel, Switzerland) from blood. We stored plasma samples from baseline and follow-up visits so that acquisition of drug resistance during the study period can be assessed. Urine samples were screened for methadone, opiates, cocaine, oxycodone, and benzodiazepines by enzyme immunoassay. Results of urine drug screens were not released to anyone outside of the study without written consent from participants.
The interviews addressed demographic and socioeconomic information, perceived availability of social support, depressive symptoms (Center for Epidemiologic Studies Depression Scale [CESD] short form [10
]), anxiety symptoms (anxiety subscale of Brief Symptom Inventory [11
]), self-reported ART adherence (3-day and 2-week recall of missed doses [12
]), alcohol use (Alcohol Use Disorders Identification Test [13
]), drug use (Johns Hopkins HIV Clinical Cohort Instrument [12
]), health-related quality of life (visual analogue scale), and emergency department visits and hospitalizations in the prior 3 months.
The study coordinator conducted all study visits. This was done to both standardize data collection and to foster candid responses from participants about medication adherence, substance use, and other sensitive issues. The study coordinator had no role in participants' substance abuse treatment or in delivery of the DAART intervention, which was managed by research assistants at the sites. Additionally, study staff abstracted data from participants' OTP substance abuse treatment records and from HIV clinic records every 3 months, including changes in methadone or buprenorphine dose, urine drug test results, CD4 cell counts, HIV RNA levels, opportunistic conditions, and changes to ART. We made concerted efforts to maintain contact with participants (particularly if they left the OTP) and made arrangements for study visits to be completed in another venue if preferred. We updated telephone and mailing address contact information at each encounter and also collected contact information for relatives that would be likely to know participants' whereabouts. We reimbursed participants for completing study visits.
Two specialty pharmacies packaged medications for the DAART arm in single-dose clear plastic bags that were labeled with medication and dosing information as specified in Maryland State regulations. When participants attended the OTP for methadone or buprenorphine they went to a private office where a research assistant or a methadone nurse observed them take an ART dose. While other investigators have administered ART and methadone simultaneously from OTP dosing windows [6
], participants in our pilot project found that this approach was stressful and jeopardized their confidentiality [7
]. We provided participants with take-home ART doses (packaged identically to observed doses) for evenings (when required by a twice-daily dosing schedule), weekends, holidays, and weekdays when participants did not have to attend the OTP (i.e., methadone take-home days). We maintained close contact with the medical providers and asked them to notify us promptly of changes to the ART regimen and to send new prescriptions to us by facsimile so that medication could be prepared for observed dosing. DAART participants were also supplied with three days of "emergency doses" to use if they did not attend the OTP as scheduled. Emergency doses are replenished as needed.
SAT participants self-administered all ART doses. We did not provide adherence counseling, coaching, or feedback to SAT participants. SAT participants were free to engage in adherence programs offered by their HIV clinics.
We alerted patients and HIV medical providers about potential drug interactions between methadone and antiretroviral drugs that were prescribed. In particular, we assured that participants and providers were aware of the potential for non-nucleoside reverse transcriptase inhibitors to increase the metabolism of methadone, which can cause clinically significant opioid withdrawal symptoms [15
]. We also notified participants' counselors and OTP clinicians so that methadone doses could be adjusted in a timely fashion if needed.
Electronic Adherence Monitoring
We used MEMS VI Track Caps (Aardex Group, Zug, Switzerland) for electronic adherence monitoring (EAM) for the first 2 months of the study in both arms. Because these devices are not compatible with pill-boxes, the EAM substudy was not required if participants preferred to use pill boxes. We selected one medication from each participant's ART regimen for EAM according to the following selection criteria hierarchy: 1) medication dosed twice daily, 2) combination preparation, containing two or three antiretroviral drugs, 3) protease inhibitor (excluding ritonavir if only used for pharmacokinetic boosting), or 4) non-nucleoside reverse transcriptase inhibitor. Research assistants met with participants to initiate EAM when the new regimen was started. Research assistants instructed participants on the use of EAM devices (with verbal and written instructions), activated new EAM caps, completed quality control protocols, and placed caps on medication bottles. Participants returned for EAM visits at approximately 10 days, 4 weeks, and 8 weeks. At the each of these visits, research assistants downloaded data from the EAM devices, completed a debriefing form to identify non-adherence with use of the devices, reviewed instructions for proper use of EAM devices, performed quality control checks, counted leftover pills, and refilled the monitored medication with a new 30-day supply if needed (i.e., at 4-week visit). Research assistants did not discuss EAM data or provide feedback to participants. We reimbursed participants for attending EAM visits and returning EAM devices.
We modified the procedures described above for participants assigned to DAART. As with SAT participants, DAART participants took home an EAM device for their monitored medication and they were instructed to take medication from the EAM device whenever they took a "home" dose (e.g., evening or weekend doses). We used a second EAM device to monitor observed doses delivered at the OTP. For observed doses, the research assistant removed a dose from the EAM device and gave it to the participant to take (in addition to other medications used in the regimen). In subjects assigned to DAART, data from the "home" and "OTP" EAM devices will be combined for analyses.
Power and Planned Statistical Analyses
The primary study outcome is suppression of HIV RNA < 50 copies/mL. We hypothesize that DAART will increase rates of viral suppression compared to SAT. At study outset, we planned to enroll 200 participants to provide 80% power to detect a 20% difference in HIV RNA suppression between the study arms at 12 months. However, enrollment was slower than anticipated because limited numbers subjects met the eligibility criteria. When addition of new study sites failed to fully rectify slow recruitment, we revised our primary outcome to viral suppression at three time points during active intervention (3, 6, and 12 months), rather than at a single time point (12 months). Using a repeated measures analytic approach, we calculated that a sample size of 120, assuming 15% loss to follow-up (i.e. effective sample size of 50 in each arm) and an intra-subject correlation of 0.2, will provide greater than 84% power to detect a 20% average difference in viral load suppression between the arms, given SAT viral suppression rates between 25% and 40%.
Secondary outcomes include HIV RNA < 400 copies/mL, change in log10 HIV RNA, change in CD4 cell count, cumulative use of ART, retention to the OTP, and acquisition of new antiretroviral resistance mutations. We will also evaluate differences in viral suppression at 18 months (6 months following the conclusion of the intervention) to assess for persistence of interventional effects. Finally, we will compare EAM in the study arms. We will use chi-squared and Wilcoxon rank sum tests to compare categorical and continuous variables, respectively. We will use mixed effects logistic and linear models to assess longitudinal outcomes. The primary analysis of viral suppression will be intent-to-treat, with missing values excluded. In sensitivity analyses, we will consider missing values to be failures and construct bounds for the potential bias.
Participant Enrollment and Disposition
A total of 457 individuals were screened for the study (Figure .). Of these, 338 were ineligible after screening and record review and 12 declined to participate. The most common reason for ineligibility was current ART use with suppressed viral load (n = 237), followed by disengagement in HIV care, lack of medication insurance, or failure to obtain an ART prescription (n = 59). A total of 107 participants were enrolled, with 51 assigned to SAT and 56 assigned to DAART. Compared to participants enrolled to the trial, the 12 individuals who declined to participate were more likely to be from site 4 and less likely to be African American (Table). Joiners and non-joiners were similar with respect to sex and age.
Figure 1 Disposition of individuals screened for a trial comparing directly administered antiretroviral therapy to self-administered therapy in opioid treatment programs, Baltimore, Maryland. ART, antiretroviral therapy; DHHS, Department of Health and Human Services; (more ...)
Study sample characteristics
A median of 14 participants were enrolled at each of the five participating OTPs (range 8 to 39). Study participants were predominantly African American, approximately half were women, and the median age was 47 years (Table .). Fifty two percent of subjects had completed high school or a general equivalency diploma. Only 15% of participants were employed. Participants had received medication-assisted therapy at the OTP for a median of 11 months prior to enrollment, with a median methadone dose of 90 mg or a median buprenorphine dose of 19 mg. Forty percent and 20% were urine screen positive for cocaine and opiates, respectively, at study enrollment.
Baseline characteristics of HIV-infected participants enrolled in a randomized trial comparing directly administered antiretroviral therapy to self administered therapy in opioid treatment programs, Baltimore, Maryland.
In general, study participants had advanced HIV disease, 62% had a history of Centers for Disease Control and Prevention Category C opportunistic conditions [17
], the median nadir CD4 count was 102 cells/mm3
, the median current CD4 was 207 cells/mm3
, and the median HIV RNA levels was 4.7 log10
copies/mL. Twenty percent of subjects were ART naïve at enrollment, and the prescribed regimen was dosed once daily in 61% and twice daily in 39%. Nearly 80% of participants were treated with protease inhibitor-based regimens, the majority of which were ritonavir-boosted.