During the last decade, advances in scientific understanding of HIV dynamics and pathogenesis, the development of widespread use of quantitative HIV-RNA tests, and the availability of powerful antiretroviral agents culminated in dramatic changes in HIV clinical care and improved clinical outcomes.1-3 However, a major obstacle to the long-term success of highly active antiretroviral therapy (HAART) regimens remains lack of consistent medication adherence.4-7 Maintaining maximum antiretroviral potency requires extremely high levels of adherence to HAART regimens. Levels of adherence less than 90% of prescribed doses are associated with dramatic decline in treatment response.8-10 Recent work suggests that although newer regimens that are more potent may permit lower levels of adherence, nonadherence remains strongly associated with mortality.11
Although there is intense interest in understanding the factors associated with medication adherence and in designing interventions to improve adherence, data from which to derive evidence-based practice recommendations are limited. Most published studies report descriptive or pilot projects and lack important components of methodological rigor such as adequate sample size, randomization to experimental or control condition, postintervention follow-up, intention-to-treat analyses, and biological outcome measures.12-14 In addition to these weaknesses in research design, the results of studies published to date are disappointing. Of 15 randomized controlled trials identified in a recent review,14 only 2 studies reported intervention effects for both behavioral and clinical outcomes at follow-up.15,16 More recently, Remien et al17 demonstrated the efficacy of a couple-based intervention to improve mediation adherence in a randomized controlled trial, although the intervention was not associated with changes in HIV viral loads or CD4 cell counts.
Strategies for improving patient adherence to antiretroviral regimens most frequently fall into 1 of 4 categories described by Simoni et al14: cognitive and behavioral, behavioral only, modified directly observed therapy, and affective. Most adherence interventions are based on well-established psychological theories of individual behavior such as the Health Belief Model,18,19 Social Learning and Self-efficacy Theory,20,21 and the client-centered counseling approach known as motivational interviewing.22-24 Although a few interventions seek to improve clinician awareness and sensitivity about the complex challenge of patient adherence,25 most aim to increase patient motivation, knowledge, and skills in medication taking.15,27-29
However, social context also exerts a profound influence on individual health and well-being. The vast majority of people requiring antiretroviral therapy live under conditions of great material privation where the healthcare infrastructure is fragmented or nonexistent. Even in the resource-rich parts of the world, those with HIV infection often exist at the margins of society, suffering the consequences of addictive disease, poverty, and institutionalized racism. To address the challenge of providing effective antiretroviral therapy in these conditions, programs to deliver adherence interventions in the community are needed.
Interventions aiming to improve adherence through community-based social support have proved feasible in Los Angeles,30 Haiti,31 Boston,31,32 and rural South Africa.33 However, the efficacy of these approaches has not been examined in a controlled trial. To address this issue, we designed and conducted the ATHENA study (Adherence Through Home Education and Nursing Assessment), a randomized clinical trial of a community-based intervention to improve medication adherence through home visits.
This 2-arm, randomized, controlled trial was conducted between September 1999 and January 2002. Potential subjects learned about the study from flyers distributed at HIV/AIDS outpatient clinics, community-based organizations, and patient support groups in Connecticut, primarily in New Haven and Hartford Counties, or they were referred by nursing or medical staff at ambulatory HIV clinics. Interested individuals called the study office to schedule an eligibility interview. Subjects were eligible if they were HIV-infected, had been prescribed combination therapy with at least 3 antiretroviral agents, and intended to take their medications. Individuals who had severe cognitive impairment, lacked a home in which to receive the intervention, were not personally responsible for medication self-administration, were enrolled in another study of medication adherence, or did not actually intend to take the medication that was prescribed were excluded. English and Spanish speakers were accepted.
Randomization and Study Protocol
After reviewing each eligible subject's antiretroviral regimen, one of the patient's medications, selected at random from all those that required twice a day administration, was placed into a vial that accepted a Medication Event Monitoring System cap (MEMS cap; AARDEX, Zurich, Switzerland). The MEMS cap is a medication bottle cap with an embedded microchip to record bottle openings. At the time of the study, once a day antiretroviral regimens were not available, and at least one medication requiring twice a day administration had been prescribed for all subjects.
Due to prescribing laws, transfer of medication and labeling was done by the subject and observed by the researchers. Child-safety caps were provided for households with children, but regular screw caps were available for those who were unable to manipulate the MEMS child-safety caps. Subjects were trained in the use of the MEMS cap and received instructions to transfer medication to the MEMS vial when the prescription was refilled. Study staff was available by telephone after the visit to answer any questions that arose. Subjects were contacted within 48 hours to ensure that they had no problems with the MEMS cap. Subjects who used pillboxes and were unwilling to place one of their medications into a MEMS cap and bottle, but wished to participate in the study, were directed to open the MEMS cap each time they removed a dose of the randomized medication from their pillbox.
After enrollment, all subjects used the MEMS cap for 4 weeks and then returned to the study office for baseline interviews, at which time data from the MEMS cap were uploaded to the study database, using the software affiliated with the MEMS system. During the 4 weeks between enrollment and the baseline interview, study staff requested viral load and CD4 lymphocyte test results from the subject's HIV clinical service provider.
At the conclusion of the baseline interview, subjects were randomized to either the intervention or the control condition. A stratified randomization procedure with block size of 10 was used. A table was constructed using SAS programming to distribute intervention and control assignments in random order and was stratified into 3 strata: less than 400 copies/mL, 400 to 10,000 copies/mL, and greater than 10,000 copies/mL. Subjects were randomized within 1 of these 3 strata according to the results of the most recent HIV RNA test within the previous 60 days. Randomization was conducted by the study's data manager. All other study personnel, except the home-visit intervention team, were blind to the subject's assignment throughout the course of the study.
Subjects assigned to the intervention arm received the intervention in their homes according to a schedule of declining frequency: weekly for 3 months, biweekly for 3 months, then monthly for 6 months. Subjects from both the intervention and control arm returned to the study office 3, 6, 9, 12, and 15 months after enrollment for data collection (Fig. 1). If subjects were not interviewed within 45 days of their scheduled study visit, it was considered a missing data point, and subjects were interviewed at the next data collection point. The study protocol was reviewed and approved by the Human Subjects Investigation Committee of Yale School of Nursing and at all recruitment sites. All subjects provided written informed consent and signed a release of information allowing ascertainment of the results of HIV RNA and CD4 cell tests from the subject's clinical care provider. A performance and safety monitoring board was not required for this study.
Data collection and management
Data were collected by trained interviewers who remained blind to the subject's group assignment. Structured, interviewer-administered questionnaires were used to assess demographic, clinical, and psychosocial variables. Each interview was reviewed for completeness and accuracy by a second member of the data collection team. Interview data were double entered into the electronic database.
Subjects brought their MEMS cap to each study visit. Data from the cap were uploaded to the database but were not reviewed with the subject. When necessary, due to loss or battery failure, a new MEMS cap was issued. The results of HIV RNA and CD4 lymphocyte counts were obtained by direct communication with each subject's source of clinical care.
Subjects received $25.00 for each study visit completed and an additional $10.00 if they remembered to bring the MEMS cap to the visit.
The measure of adherence was the data provided by the MEMS cap. MEMS data were uploaded into the MEMS software and imported into SAS. At each data collection point, summary scores were calculated for the percentage of prescribed doses taken since the previous interview date. For the baseline adherence measure, summary scores were calculated for the percentage of prescribed doses taken in the 14 days immediately before randomization.
If the original medication selected for MEMS cap monitoring was discontinued, an alternative was randomly selected among all of the subject's prescribed medications with twice a day dosing schedules, and the subject was instructed to move the newly chosen medication into the bottle with the MEMS cap.
The results of HIV RNA and CD4 lymphocyte tests were obtained for each subject directly from the subject's source of clinical care. Results were accepted for tests occurring within a 90-day window of the subject's interview and MEMS cap upload. When the study began, use of the Roche-Amplicor PCR technology, with a lower limit of 400 copies, was standard. During the course of the study, more sensitive techniques became available to maintain consistency; all results less than 400 copies were categorized as "undetectable."
Subjects who were randomized to the experimental condition received an adherence intervention with social and educational components, guided by the educational philosophy of Paolo Freire, which holds that true learning occurs through participation in dialogue between inquiring equals.34 The goal of the dialogue is to define problems and pose questions in such a way as to allow participants to reflect upon the root cause of the problem and to propose actions they can take to gain more control over the situation.
Freire's educational model comprises 3 stages. The process of moving from one stage to the next is dialectic; that is, problems and potential solutions emerge from the discussion at each stage and lead naturally to the next stage. For Freire, authentic learning requires not only dialogue and reflection but also action. Thus, this approach views the participants as self-conscious agents capable of transforming their world and making changes in their lives. The patient-learners define the content and outcome of their own learning; the nurse and peer educator serve as expert resources rather than as voices of authority. The result is learning situated within the specific local cultural context.
The intervention team consisted of a nurse and a community support worker. The team made a series of home visits (Fig. 1), during which they facilitated a dialogue encouraging subjects to identify individual and social factors that they perceived influenced their success with antiretroviral adherence. Although the content of the discussions in the home varied in response to the individual concerns of each subject, the format of the interactions was highly structured. The intervention teams met weekly to discuss the process of the intervention and to ensure fidelity to the theoretical framework and to the intervention protocols. The intervention program has been described in detail elsewhere.35
The home visits took place on a schedule of declining frequency over a period of 12 months. Visits were uniform in design and conducted according to a protocol that totaled 24 home visits. Because subjects were members of a socially vulnerable population with a high prevalence of marginal housing, intermittent incarceration, mental illness, and substance abuse, missed intervention visits were expected. If a subject was not available for a scheduled home visit, the visit was not rescheduled and the subject was seen at the next scheduled time.
Subjects were not reimbursed for participation in the home-visit intervention.
All subjects received care from HIV-dedicated clinical services which provided clinic-based adherence support as part of the usual care protocol. Although specific procedures varied between clinics, usual care included individual review of patient medications by the prescribing clinician and the clinic nurses, assistance with the development of individual medication schedules, identification of strategies to improve adherence, and individualized patient education regarding medication dose, side effects, and the need for adherence. A variety of patient education materials were available in all clinical sites.
A team of nurses and peer support workers who made home visits in pairs delivered the intervention. The personal background of the members of the intervention team reflected spectrum of characteristics related to race, ethnicity, gender, class, education, and sexual orientation of the population, and included HIV-positive individuals and individuals with expertise in substance abuse and mental health. Members of the intervention team participated in an initial 1-month training, which included both didactic information on HIV/AIDS, sexuality, substance abuse, and medication adherence as well as experiential role-plays in the use of Freirian educational techniques.
To ensure fidelity and consistency in the implementation of the intervention, team members maintained detailed narrative logs of the content and process of each home visit. The logs were reviewed immediately after the visit by fellow team members and weekly by an intervention supervisor who was not involved in direct delivery of the intervention. The interventionists met weekly with the intervention supervisor to discuss challenging cases and to review the principles and protocols of the intervention. Biweekly meetings with the principle investigator and semiannual consultation with a community-based psychologist and an adult educator also focused on the process of Freirian education.
Data Analysis and Power
The primary end point was improvement in adherence. Adherence was defined as the ration of the number of recorded MEMS cap openings to the number of openings to be expected if the medication were taken as prescribed. The sample size was established to allow detection of a difference of 20% in mean adherence between the intervention and control group. Sample size was calculated using adherence as a continuous variable.
We used intent-to-treat analysis for this randomized controlled trial. We analyzed the data according to randomization assignment.36 Subjects who were randomized to the intervention but did not receive the complete schedule of home visits were included in the intervention group at analysis. We also carried out an evaluable subset analysis using the number of home visits received as a covariate. Results of the subset analysis did not differ significantly from the intent-to-treat analysis.
Analyses were carried out in SAS (Cary, NC). General descriptive statistics were carried out on all variables of interest to describe the study population, to assess assumptions of normality, and to check the final data. Continuous data were categorized when appropriate to reflect clinically relevant strata. Specifically, HIV RNA was categorized as undetectable and detectable as well as <400, 400-10,000, and >10,000 copies/mL; CD4 lymphocyte counts were categorized as <200 and 200 or greater cells/mL; and adherence was categorized as <90% or 90% and greater.
Bivariate analyses were done using Student t tests and χ2 tests. Multivariate tests included extended Mantel-Haenzsel χ 2 tests for examining the relationship between 2 categorical variables controlling for other categorical covariates (namely, time) and mixed modeling to address continuous outcomes over time using different correlation structures and choosing the best model based on AIC scores.
Loss to follow-up during the latter portion of the trial resulted in missing data. We compared missing data between the control and intervention group at all time points after randomization, and there were no significant differences in the distribution of missing data between the 2 groups. Using mixed modeling, we were able to include all those randomized in the analysis, even if they had incomplete data at some point during the study. Conducting repeated measures analyses using a mixed model allows incorporation of subjects for whom there are incomplete data, including those lost to follow-up. Mixed modeling also allows flexibility in modeling the error. We used compound symmetry, autoregressive, and autoregressive moving average correlation matrices in our analysis. The autoregressive moving average structure proved the most appropriate for these data. We chose not to apply the last observation carried forward technique to missing data because this approach may bias the results if the missing data are treatment related.37
The study subjects included 171 HIV-infected adults; 48% (n = 83) were women, 57% (n = 99) were racial or ethnic minorities, 71% (n = 122) had a history of at least one mental illness, and 81% (n = 138) were either present or past illicit substance users. The sample characteristics by study group are presented in Table 1. The 2 groups were similar with respect to the majority of the study variables.
Subjects' antiretroviral regimens were typical of clinical practice in our region from 1999 through 2001. At baseline, 32% were on an nonnucleoside reverse transcriptase inhibitor (NNRTI)-based regimen and 43% on a protease inhibitor (PI)-based regimen. Boosted PI regimens were rare (6%) and consisted primarily of indinavir/ritonavir. At baseline, efavirenz accounted for 47% of the NNRTIs prescribed, whereas nelfinavir was used in 52% of the PI-based regimens.
Figure 2 shows the flow of study participants, including the number screened and participating at each data collection point. Overall retention at 12 months was 74%, 75% in the control arm, and 72% in the intervention arm. Because the study closed before all subjects had the opportunity to return for a 15-month visit, only 51% (48% in the control arm, 54% in the intervention arm) contributed data at 3 months after the intervention. There was no difference between the arms in regard to rate of retention in the study, nor was there a significant difference between the intervention and control arm in the number of missed interviews over the course of the study.
The median number of home visits was 19, with a range of 0 to 24, out of a possible 24 visits. Twenty-four percent (n = 21) of subjects in the intervention arm had 7 or fewer visits, and 12 subjects missed all 24 visits. Twenty-nine percent (n = 25) received at least 23 visits, with 18 subjects receiving the full 24-visit course of the intervention.
At baseline, mean adherence was 72% for subjects in the control arm and 69% for those in the intervention arm; this difference was not statistically significant. Median adherence for both groups was 81%. Forty-two percent of subjects in the control arm and 40% of those in the intervention arm demonstrated greater than 90% adherence at baseline. Subsequently, a larger proportion of subjects in the intervention group had adherence greater than 90% at each time point compared with those in the control group (Fig. 3). For example, at 12 months, 31% of the intervention group compared with 22% of the control group demonstrated MEMS adherence greater than 90%. At 15 months, 36% of the intervention group compared with 24% of the control group demonstrated adherence greater than 90%. The difference in proportions over time is statistically significant (Extended Mantel-Haenszel test: 5.80, P = 0.02).
When adherence was computed as a continuous variable, there was no difference between the 2 groups in change in adherence over time. Controlling for dose of the intervention (the number of intervention visits actually received) did not alter these results.
At baseline, the HIV RNA level was undetectable for 54% of controls and 52% of treatment subjects. The absolute CD4+ cell count was greater than 200 for 70% of controls and 76% of treatment subjects. At 12 and 15 months, there were no significant differences between the 2 arms in the proportion of subjects who had undetectable HIV RNA test results or CD4 cell counts greater than 200. This finding held true, even when the data were stratified by detectable and undetectable at baseline. However, regardless of treatment arm, there was an association between subjects with greater than 90% adherence and an undetectable viral load which was significant before randomization (P = 0.01), at 12 months (P = 0.003), and at 15 months (P = 0.02).
This study is one of the few rigorous, controlled trials of a community-based medication adherence intervention. It provides encouraging evidence to support the further development and evaluation of such programs. In this study, home visits by nurses and peer support workers were associated with medication adherence greater than 90% among a vulnerable population of people living with HIV/AIDS in northeastern United States. Most subjects in this study had a history of substance abuse, often with concomitant mental illness.
Five factors may have increased the difficulty of detecting a biological difference between the intervention and control group. Researchers should consider these factors in the design of adherence intervention studies in the future. First, this study accepted all comers who met the eligibility criteria, including those with both high and low initial adherence levels. Over half of subjects had an undetectable viral load at randomization, thus limiting the room for improvement among a large portion of subjects in both groups. Future trials of adherence interventions should consider limiting participation to individuals with adherence of less than 90% and detectable HIV viral loads.
Second, an increasing amount of adherence support was included in HIV clinical services over the course of the study, potentially attenuating the impact of the adherence intervention. Further, during this time, antiretroviral regimens became simpler and more powerful. Changing clinical practices are an inescapable element of clinical research, and study subjects should not be denied access to widely accepted improvements in care. Documenting the extent of such changes and their impact on the results of trials presents a major challenge to researchers.
Third, participation in an adherence research study may have improved adherence in the control group through a placebo-like effect. Subjects in the control arm reported that they looked forward to the regularly scheduled study interviews. Although the interviews were highly structured and did not include adherence counseling or therapy, the interviewers were professionals with warm personal styles and experience with substance abuse and mental health problems. The act of informing these professionals about their medication taking behavior may well have led to more self-awareness and improved adherence in the control group.
Fourth, although a consistently larger proportion of subjects in the intervention group demonstrated greater than 90% adherence, most subjects in both groups failed to achieve this important milestone. Therefore, the intervention, although effective, was not universally so. In future studies, it will be important to determine if there are specific individuals for whom this type of intervention is more or less effective and which specific elements of the intervention exert the most powerful effect on adherence behavior.
Fifth, although widely used in adherence research, MEMS caps are an imperfect measure of adherence behavior. In addition to numerous logistical challenges,38 it is possible that use of the MEMS caps themselves alters adherence behavior. The data regarding this possibility are conflicting. One study examining the effect of both monetary rewards and feedback based on MEMS caps found a short-term effect.26 In contrast, a study explicitly designed to evaluate the impact of electronic monitoring on adherence found no effect.39
For both groups in the ATHENA study, the instruction provided on the use of MEMS caps and the continuing use of the caps may have acted as an adherence intervention in its own right. Although the ATHENA research team did not provide feedback to subjects on their MEMS data, subjects were aware that the data would be uploaded at each data collection visit. The effect, if any, of the MEMS caps and data collection process would have been similar for both the intervention and the control groups. However, a positive effect of the measurement tool on adherence might have attenuated the differences between the 2 groups.
MEMS caps provide a wealth of detail regarding the individual pattern of medication bottle cap opening and thus, presumably, about individual medication-taking behaviors. However, most research reports, including this one, report these data aggregated over time and group. Because it is likely that, for most patients, adherence is not static but waxes and wanes over time, much rich individual detail is lost when MEMS data are aggregated. Preliminary, exploratory studies of data analysis strategies suggest that it is possible to characterize categories of adherence behavior patterns.40 If successful, these new approaches to analyzing electronic adherence data may improve our understanding of which categories of patients most benefit from specific interventions.
As pharmacologic agents have become more powerful, clinicians more sophisticated in their use and patients more exposed to their biological effects, the relationship between adherence and viral response has become complex. On the one hand, increasingly powerful regimens may be more forgiving of lapses in adherence; on the other hand, increasing numbers of drug-resistance mutations may be associated with higher levels of adherence.41 The failure of this study to demonstrate a robust intervention effect on biological outcomes is consistent with other studies,12-14,17 as is the consistency of the association between high adherence and undetectable HIV RNA.8-10 These findings confirm that adherence is not the sole determinant of biological response to antiretroviral therapy. Nevertheless, the association between adherence greater than 90% and nondetectable viral load suggests that interventions to improve adherence remain clinically relevant.
Adherence is likely to be a factor in the ultimate success of antiretroviral programs around the world. Globally, adherence rates are variable and often quite low.42 Given the large numbers of patients in established HIV epidemic areas and those affected by the newly emerging epidemics in Asia and Russia, effective adherence interventions are urgently needed. An approach that dichotomizes factors that influence adherence into "structural" versus "behavior" variables42 may be too simple because individual behavior occurs in a social context. A recent study of barriers to adherence in an Indian cohort identified not only the structural barrier of financial cost but complex and socially determined barriers such as stigma, fear of disclosure, and lack of family or community support.43 The Freirian approach to improving adherence evaluated in this study addresses individual behavior in the context of the community and emphasizes individual action to change structural, social, and individual barriers to health.34 Such an approach may be well suited to the heterogeneity of the individuals and communities who need this help, now and in the future.
1. Hogg RS, Heath KV, Yip B, et al. Improved survival among HIV
-infected individuals following initiation of antiretroviral
2. Mocroft A, Ledergerber B, Katlama C, et al. Decline in the AIDS and death rates in the EuroSIDA study: an observational study. Lancet
3. Palella FJ, Delaney KM, Moorman AC, et al. Declining morbidity and mortality among patients with advanced HIV
infection. N Engl J Med
4. Bartlett JA. Addressing the challenges of adherence
. J Acquir Immune Defic Syndr
5. Friedland GH, Williams A. Attaining higher goals in HIV
treatment: the central importance of adherence
6. Liu H, Golin CE, Miller LG, et al. A comparison study of multiple measures of adherence
protease inhibitors. Ann Intern Med
7. Wood E, Hogg RS, Yip B, et al. Effect of medication adherence
on survival of HIV
-infected adults who start highly active antiretroviral
therapy when the CD4+
cell count is 0.200 to 0.350 × 109
cells/L. Ann Intern Med
8. Gross R, Bilker WB, Friedman HM, Strom BL. Effect of adherence
to newly initiated antiretroviral
therapy on plasma viral load. AIDS
9. Paterson DL, Swindells S, Mohr J, et al. Adherence
to protease inhibitor therapy and outcomes in patients with HIV
infection. Ann Intern Med
10. Sethi AJ, Celentano DD, Gange SJ, et al. Association between adherence
therapy and human immunodeficiency virus drug resistance. Clin Infect Dis
11. Wood E, Hogg RS, Yip B, et al. Effect of medication adherence
on survival of HIV
-infected adults who start highly active antiretroviral
therapy with the CD4+ cell count is 0.200 to 0.350 × 109
cells/L. Ann Intern Med
12. Fogarty L, Roter D, Larson S, et al. Patient adherence
medication regimens: a review of published and abstract reports. Patient Educ Couns
13. Reynolds NR. Adherence
therapies: state of the science. Curr HIV Res
14. Simoni JM, Pantalone MS, Frick PA, Turner BJ. Enhancing antiretroviral adherence
: a review of published reports of RCTs and on-going NIH-funded research. In: Best Practices in the Behavioral Management of Chronic Disease, Vol. 2: Chapter 3
. Los Altos, CA: Institute for Disease Management; 2005.
15. Tuldra A, Fumaz CR, Ferrer MJ, et al. Prospective randomized two-arm controlled study to determine the efficacy of a specific intervention to improve long-germ adherence
to highly active antiretroviral
therapy. J Acquir Immune Defic Syndr
16. Pradier C, Carrieri P, Bentz L, et al. Impact of short-term adherence
on virological and immunological success of HAART: A case study among French HIV
-infected IDUs. Int J STD AIDS
17. Remien RH, Stirratt MJ, Dolezal C, et al. Couple-focused support to improve HIV
: a randomized controlled trial
18. Becker MH, Maiman LA. Sociobehavioral determinants of compliance with health and medical care recommendations. Med Care
19. Becker MH. Patient adherence
to prescribed therapies. Med Care
20. Bandura A. Social Learning Theory
. New York: General Learning Press; 1971.
21. Bandura A. Perceived self-efficacy in cognitive development and functioning. Educ Psychol
22. Miller W, Rollnick S. Motivational Interviewing: Preparing People to Change Addictive Behavior
. New York: The Guilford Press, 1991.
23. Rollnick S, Heather N, Bell A. Negotiating behavior change in medical settings: the development of brief motivational interviewing. J Ment Health
24. Emmons K, Rollnick S. Motivational interviewing in health care settings: opportunities and limitations. Am J Prev Med
25. Murri R, Antinori A, Ammassari A, et al. Physician estimates of adherence
and the patient-physician relationship as a setting to improve adherence
therapy. J Acquir Immune Defic Syndr
26. Rigsby MO, Rosen MI, Beauvais JF, et al. Cue-dose training with monetary reinforcement: pilot study of an antiretroviral adherence
intervention. J Gen Intern Med
27. DiIorio C, Resnicow K, McDonnell M, et al. Using motivational interviewing to promote adherence
medications: a pilot study. J Assoc Nurses AIDS Care
28. Farley J, Hines S, Musk A, et al. Assessment of adherence
to antiviral therapy in HIV
-infected children using the Medication Event Monitoring System, pharmacy refill, provider assessment, caregiver self-report, and appointment keeping. J Acquir Immune Defic Syndr
29. Murphy DA, Lu MC, Martin D, et al. Results of a pilot intervention trial to improve antiretroviral adherence
-positive patients. J Assoc Nurses AIDS Care
30. Wohl AR, Garland WH, Squires K, et al. The feasibility of a community-based directly administered antiretroviral
therapy program. Clin Infect Dis
31. Behforouz HL, Farmer PE, Mukherjee JS. From directly observed therapy to accompagnateurs:
enhancing AIDS treatment outcomes in Haiti and Boston. Clin Infect Dis
32. Behforouz HL, Kalmus A, Scherz CS, et al. Directly observed therapy for HIV antiretroviral
therapy in an urban US setting. J Acquir Immune Defic Syndr
33. Jack C, Lalloo U, Abdool Karim Q, et al. A pilot study of once daily antiretroviral
therapy integrated with tuberculosis directly observed therapy (TBDOT) in a resource limited setting. J Acquir Immune Defic Syndr
34. Freire P. Pedagogy of the Oppressed
. New York: Continuum, 1986.
35. Williams AB, Burgess JD, Danvers K, et al. Kitchen table wisdom: a Freirian approach to medication adherence
. J Assoc Nurses AIDS Care
36. Grady D, Cummings SR, Hulley SB. Designing an experiment: clinical trials II. In: Hulley SG, Cummings SR, Browner WS, Grady D, Hearst N, Newman TB eds. Designing Clinical Research
, 2nd ed. Philadelphia: Lippincott Williams & Wilkins, 2001:163-164.
37. Chow SC, Lliu JP. Design and Analysis of Clinical Trials: Concepts and Methodologies
, 2nd ed. New Jersey: Wiley, 2003:518-520.
38. Bova CA, Fennie KP, Knafl GJ, et al, Dieckhaus KD, Watrous E, Williams AB. Use of electronic monitoring devices to measure antiretroviral adherence
: practical considerations. AIDS Behav
39. Wagner GJ, Ghosh-Dastidar B. Electronic monitoring: adherence
assessment or intervention? HIV Clin Trials
40. Knafl GJ, Fennie KP, Bova C, et al. Electronic monitoring device event modelling on an individual-subject basis using adaptive Poisson regression. Stat Med
41. Bangsberg DR, Moss AR, Deeks SG. Paradoxes of adherence
and drug resistance to HIV antiretroviral
therapy. J Antimicrob Chemother
42. Gill CF, Hamer DH, Simon JL, et al. No room for complacency about adherence
therapy in sub-Saharan Africa. AIDS
43. Kumarasamy N, Safren SA, Raminani SR, et al. Barriers and facilitators to antiretroviral adherence
among patients with HIV
in Chenai, India: a qualitative study. AIDS Patient Care STDs