In this issue, Simoni and colleagues present the first published meta-analysis of randomized controlled trials (RCTs) of interventions for adherence to antiretroviral therapy for HIV. Their analyses summarize a decade of intervention research, coinciding with the era of highly active antiretroviral therapy (HAART). Across 19 RCTs with more than 1800 participants, those who received an adherence intervention were 1.5 times as likely to report 95% adherence and 1.25 times as likely to achieve an undetectable viral load as participants in comparison conditions. The magnitude of this aggregated intervention effect is encouraging, but the authors also appropriately indicate that further work is needed. The cumulative evidence has reached a point at which it is important to take stock of the progress, examine the strengths and limitations of the evidence, identify critical research directions, and consider implications for practice.
In the most recent updates of the Department of Health and Human Services guidelines for the use of antiretroviral agents in HIV-1-infected adults and adolescents, including an adherence supplement published in October 2004,1 recommendations for providers regarding adherence were constrained primarily to general strategies for monitoring and intervention. The panel cautioned that “interventions to improve adherence for HAART are insufficiently characterized and understood, and additional research regarding this topic is needed.” The response to this call for action has been robust; 75% of the RCTs in the meta-analysis were published since 2003, and providers now have more specific recommendations to guide adherence interventions in their practices. In this commentary, the key findings of the meta-analysis are discussed, heterogeneity in the individual studies is detailed, and the results are placed in the context of the domestic and international pandemic.
KEY FINDINGS IN THE META-ANALYSIS
The meta-analysis indicates a significant overall intervention effect, which is an important message for the field. As with any meta-analysis, however, the capacity to answer more complex questions about the state of the science is limited by the number of available studies and their individual features. There is no “gold standard” for how many individual studies are necessary to conduct an informative quantitative synthesis. For this meta-analysis, the number of studies, their collective methodologic quality, and the authors' choice of meta-analytic strategies all meet acceptable standards; thus, the overall findings can be interpreted with confidence. Sufficient detail is provided about the methodologic features of each RCT (eg, inclusion criteria, retention rates) to indicate adequate rigor. Results of sensitivity analyses to test whether any single study accounts for the overall effect and an analysis to test for publication bias also bolster confidence in the main findings. Moreover, the authors use outcome variables for the meta-analysis that are conservative and, whenever possible, rely on objective assessments. They also directly contacted investigators for data not available from publications. Adherence outcomes are analyzed with a dichotomized 95% cutoff, the extent of electronic monitoring of pill taking is reported, and intervention effects on viral load are examined. These selections match the current standards in the field against which to measure intervention success.
The decision to use 95% as the cutoff to define intervention efficacy warrants additional comment. As the authors assert, the choice of this cutoff is defensible because seminal studies with protease inhibitor therapy indicated that this adherence level was associated with the best virologic outcomes.2 There is increasing evidence that the relation among adherence, viral response, and resistance is more complex than first posited, however. More potent combination regimens that involve nonnucleoside reverse transcriptase inhibitors (NNRTIs) may be more forgiving of moderate (eg, 75%-95%) levels of adherence.3 Moreover, different therapies may confer varying risks for treatment resistance depending on the level and pattern of adherence behavior, and these associations may be nonlinear.4 For certain regimens, resistance may be more likely at the highest levels of adherence, and for other regimens, resistance may only occur at the poorest adherence levels. As clinical outcomes for newer regimens are reported, adherence intervention trials should have higher proportions of patients taking these regimens; thus, it is important to re-examine the evidence base frequently. Such accumulative data and analyses may help providers to make better informed treatment choices, in collaboration with their patients-choices that may currently be challenging because of paradoxes in the emerging literature. Investigators conducting adherence intervention trials should more systematically collect and report data on treatment regimens, and subsequent meta-analysts may consider a lower cutoff criterion for efficacy or analyze adherence data as a continuous variable.
The absolute proportion of participants in the RCTs who achieved optimal adherence is also noteworthy: 62% of intervention arm participants and 50% of control arm participants reported 95% adherence. Even though the meta-analysis is primarily concerned with relative efficacy between intervention arms, the overall efficacy is remarkable, given the literature for other health conditions and the life context for many persons living with HIV. Among patients in care for other chronic diseases, recent reviews report average adherence rates ranging from 43% to 78%.5 Further, although a detailed description of all participants was beyond the scope of the meta-analysis, examination of the individual studies reveals that samples included patients living with complex challenges. Patients with alcohol and drug use were targeted,6-8 memory impairment was sometimes evident,9 and samples often consisted largely of lower income ethnic minority patients at high-volume urban care clinics, where depression and other circumstances (eg, intermittent employment, housing issues) can interfere with treatment.10 In adherence intervention trials for other diseases, participants with substance use problems and/or mental illness are often excluded.
For the viral load outcome, the overall effect size was marginally significant, but, again, more than 60% of participants in intervention conditions achieved an undetectable viral load. The authors speculate that clinical or biologic factors such as preexisting resistance may explain why the effect on viral load was not stronger. These explanations are plausible, but it is also striking that relatively few of the studies restricted inclusion to patients with some risk marker for poor adherence (eg, 3 studies intervened only with patients with a detectable viral load). In most of the trials, patients were not specifically recruited because they had demonstrated adherence problems. Including patients whose adherence is already fairly good may compound a trial's inability to detect an intervention effect on viral load, because this outcome may already be influenced by other clinical and biologic factors. Although the analysis to test whether this inclusion factor moderated outcome was not significant, the results of all the stratification analyses should be cautiously appraised because of the relatively small number of studies in the subgroups. As studies are added, such analyses should have increased power to detect differences; thus, moderating factors that are potentially important may become more apparent. It may also be useful to have stricter criteria for studies to be included in subsequent meta-analyses (eg, minimum sample size).
Because the moderator analyses are not definitive as a result of the number of studies, the heterogeneity of the samples and intervention modalities can be viewed from several perspectives. The authors' positive interpretation is that a wide variety of interventions may be efficacious, noting a successful couples-focused intervention10 and other successful brief pharmacist-delivered didactic sessions for individuals.11,12 In terms of specific tools for providers, this positive perspective should be welcomed as long as the caveats about these studies are appreciated. For example, the follow-up assessment periods were relatively short and findings should not be generalized beyond the setting and samples of the individual studies. An alternative perspective is that the range of efficacious and nonefficacious interventions is so wide that it is not yet clear what mechanisms are responsible for the observed successes or failures. As is often the tension in clinical research involving urgent patient needs, it is extremely useful to have interventions that can be implemented today, but future public health gains depend on further research to determine why some interventions work and others do not.
RESEARCH DIRECTIONS AND IMPLICATIONS FOR PRACTICE
Another value of this meta-analysis is that significant research gaps are made apparent. Theoretic, methodologic, and other limitations in the literature are identified. Not all limitations are repeated here. As the authors point out, all studies evaluated interventions directed to patients, despite evidence that multiple factors can affect adherence, including provider-, system-, and contextual-level variables. Relatedly, with a few exceptions, the RCT investigators drew on variants of social-cognitive theory for the interventions' conceptual bases. Further theoretic development and testing should yield interventions designed to address other levels of influence that may improve the potency and sustainability of interventions. Alternatives could then be offered in settings in which time and resources may not allow the degree of contact dictated by individual-level protocols. It is encouraging that many other types of intervention modalities and levels are being tested in trials that are underway.13 For example, Milam et al14 tested the efficacy of a clinic-level intervention that combined changes in clinic structure to emphasize the importance of monitoring adherence, written materials, and brief messages delivered by providers at each visit. In clinics in which these protocols were implemented, patients who demonstrated high adherence rates at baseline (>95%) were more likely to maintain high adherence rates at follow-up compared with similar patients in comparison clinics.
An additional finding from the trial by Milam et al14-that the clinic-based intervention was not efficacious for patients whose initial adherence was <95%-points to another set of research opportunities related to individualizing care to match patient needs. In the study by Milam et al,14 it seems that the intervention helped excellent adherers to maintain high adherence rates but that patients with <95% adherence could have benefited from additional attention. In most settings in which treatment is delivered, efficacy and cost-effectiveness of adherence interventions would be optimal if tailored to individuals' varying needs. In a conventional RCT design, however, a single manualized intervention is usually developed and implemented over a specified duration and may end before a patient has reached the desired criterion (eg, the percent adherence goal). Typically, the intervention is also administered to available and consenting participants, such as all patients at a particular clinic. As noted earlier, many of the RCTs did not recruit participants on the basis of risk for nonadherence. Therefore, some patients were probably not suited for the intervention they received because the intensity or duration was excessive. Conversely, in RCTs with a relatively brief intervention, patients with greater needs may have been underserved.
A trial designed to test a tailored approach with a menu of interventions matched to patient needs (high-, moderate-, and low-intensity interventions), delivered on a continuous basis when indicated, has not yet been conducted. Directly observed therapy (DOT) could be considered as the highest intensity intervention but is perhaps indicated for relatively few patients. Although initial RCTs of DOT demonstrate efficacy15 and DOT can be cost-effective under some circumstances,16 observed dosing and intensive case management are unlikely to be needed or desired by all patients. In many cases, patients may only need a low-intensity intervention with brief motivational messages, information, or a reminder device to maintain high levels of adherence; thus, scarce clinic or community resources can be spared for those patients who need additional assistance. For those patients who need a moderate degree of assistance, a multiple-session intervention may be indicated that addresses their specific risk factors for nonadherence, such as mental health issues, a lack of social support, substance use, or competing life stressors.
Research on adherence could also continue to broaden beyond solely medication taking. This recommendation is not a criticism of the meta-analysis or the efforts in the RCT studies targeting medication adherence. Certainly, the best outcomes are observed when adherence levels are high, and studies designed to improve clinical markers through adherence interventions are still needed, especially with enhancements as previously noted. Nevertheless, many patients never initiate care after learning of their HIV serostatus,17 and many individuals who attempt to engage in HIV treatment do not attend enough appointments to benefit. In a recent RCT, a case management intervention to enhance linkage to HIV care among newly diagnosed persons with HIV was conducted in Atlanta, Baltimore, Los Angeles, and Miami, and results were significant.18 Compared with standard of care, a higher proportion of patients in the case management condition (60% vs. 78%) reported seeing an HIV provider at least once within 6 months. To complement the ongoing medication adherence studies, innovative interventions to increase consistent access and linkage to HIV care are needed. Especially if medication advances with once-daily dosing become increasingly available (reducing regimen complexity as an adherence barrier), retention in HIV treatment and important ancillary services should become even more critical. Macrolevel contextual factors are likely to be even more important as determinants for initiation and retention in care; thus, again, the recommendation for system- or community-level interventions is underscored. Future meta-analyses may be possible where such variables as “attended clinic appointment” or “completed referral to drug treatment” are additional outcomes of interest.
TRANSLATIONAL RESEARCH AND SYSTEMS TO GUIDE INTERVENTION IMPLEMENTATION ARE NEEDED
The importance of this meta-analysis should not be underestimated, given that providers in the United States and abroad are in immediate need of evidence-based interventions for their patients. Without additional attention to translational research, however, patients and treatment providers may not benefit from the findings. Whether it is called translational research, operations research, implementation research, technology transfer, systems science, dissemination research, or effectiveness studies, advances in these areas for HIV treatment adherence have not yet been fully realized. The science of technology transfer in this domain remains underdeveloped; thus, potential research questions are offered here for discussion. Because most adherence interventions are likely to be implemented within existing treatment settings, there are lessons to be learned from technology transfer research in other disease areas in which behavioral and social interventions are integrated into medical care (eg, behavioral management of diabetes, exercise and nutrition interventions in cardiac care settings). These studies should be mined for insights to assist rapid transfer of adherence interventions from research to practice, but this literature often suffers from some of the same limitations. For example, little is known about contextual issues that affect intervention adoption, adaptation, and effectiveness.
Organizational leadership, clinic care personnel turnover, and organizational size are factors that may influence the implementation of new interventions.19 Are there other organizational factors that may be related to adaptation of efficacious interventions and whether or not ongoing outcomes assessment is an institutional priority? Similarly, services-related research question s are understudied; for example, how financing of organizations and reimbursement affect adherence intervention implementation. Issues of staff training, organizational resources, and intervention cost are interrelated, and each individually and in combination may affect intervention success. How best to measure success also needs examination; success could be defined as whether or not an intervention is adopted and sustained for a specified time period, and/or success could be measured through direct assessment of clinic patient outcomes.
Effectiveness of HIV treatment adherence interventions implemented outside rigorous RCTs may rely on how well fidelity to the intervention protocols can be maintained in the face of competing clinic demands. Whether intentional or unintentional, interventions are going to be modified in real-world settings. Yet, it is an open question as to how much change an intervention can tolerate before it no longer has the essential elements of the originally tested version or how much an intervention should be tailored to suit the local population. For example, attempts to modify an adherence intervention intended primarily for pharmacists to deliver to English-speaking patients and altered for use by peers to deliver messages in a setting with primarily Latino women and their families should be evaluated. Public health may be best served if questions such as these are answered using a theoretic framework that advances general understanding of translational processes in addition to the specifics of adaptation and fidelity for HIV treatment adherence.
Finally, there currently is no formal process to guide the implementation of efficacious adherence interventions into community and clinical practice, such as the system that has been established for efficacious HIV prevention interventions.20 To meet the demand for technology transfer, relevant federal agencies should cooperatively dialogue with researchers, providers, and consumers to decide how best to develop and implement a pipeline for adherence interventions.
SPECIAL FINAL COMMENT: INTERNATIONAL NEEDS FOR ADHERENCE INTERVENTION RESEARCH
Much has been written about the importance of global HIV prevention21 and how HIV prevention should accompany HIV treatment rollout.22,23 Few would deny the importance of HIV prevention or treatment, and unanimity seems to be growing that it is essential to succeed in both domains and to integrate efforts whenever possible. It has not been as clear that expansion and integration of HIV treatment adherence interventions in these countries are viewed as essential. Perhaps attention to treatment adherence intervention has suffered because so much energy has been expended on scaling up treatment capacity or because early reports from some resource-constrained settings indicated good adherence rates.24,25 Regarding Africa, adherence issues have been at the center of some debate about treatment access and treatment resistance; perhaps this controversy has distracted from dialogue about which adherence interventions should be implemented when the need arises.
Regardless of how the debate about adherence and treatment access is settled in Africa, what is not debatable is that there were no published RCTs of adherence interventions in developing countries for Simoni et al to include in the meta-analysis. After 10 years of HAART, the only published reports of adherence RCTs have been conducted in the United States and Europe. Moreover, outside the United States and Europe in developed and developing countries in which HIV treatment is becoming more widely available (eg, Africa, Asia, Caribbean), descriptive data and studies of adherence predictors are also in their early stages. Without rapid attention to these issues, decisions about adherence interventions to accompany treatment are being made in the absence of RCT data. In the absence of empiric support, interventions are likely to be implemented that are created with well-intentioned clinical judgment or largely adapted from Western interventions that are not derived from theoretic models grounded in the country or region. These interventions may or may not be efficacious, but this is going to be difficult to assess. In many instances, settings may not have resources for evaluation or evaluations may use designs that do not meet the usual standards that determine guidelines for implementation.1,19
This is not to say that data derived from non-RCT designs cannot provide valuable sources of information to guide intervention implementation. There are influential models for HIV treatment and adherence that have been developed within resource-constrained settings, in collaboration with communities and evaluated using other designs and metrics. For example, in Haiti, accompagnateurs play an important role in patients' adherence to treatment, and before and after designs with outcomes of significant weight gain and the lack of morbidity and mortality are used to demonstrate effectiveness.26,27 Variations of this model using peers or community health workers are now being tested in other international28 and US settings.29 It is also important to note that there are often other compelling reasons, in addition to evidence from RCTs, that particular interventions are chosen and implemented, such as clinical duty, ethical obligation, equity and justice, or patient preference.30
In summary, because so little is known about adherence in many international settings, recommendations for research directions are not likely to be as specific as the suggestions cited earlier. Rather, it is sufficient to conclude that nearly all aspects would benefit from further research as soon as possible but that special attention may be warranted for some topics that are unique to settings in which treatment is now becoming available. Some of the first descriptions of adherence rates for children in resource-constrained settings suggest lower rates than has been observed for adults.31,32 Reasons for nonadherence attributable to the absence of a caregiver, the child forgetting, or drug stock exhaustion are early indications that interventions may need to have features to meet the special needs of children. Finally, antiretroviral therapy rollout is occurring rapidly in many countries. In Uganda, for example, the Joint Clinical Research Center is now treating more than 35,000 patients with antiretroviral therapy across the country, and this number should quickly increase as patients begin therapy earlier in the disease course.33 There is an urgent need to implement monitoring and adherence interventions that are affordable, brief, flexible, responsive to evolving service situations, and sustainable. A coordinated system for intervention and evaluation is being developed, but stigma, patient isolation, mobile populations, access, transportation, and other competing life needs present barriers on a scale not previously faced. These are the challenges to which the next generation of international adherence intervention research must respond, working in concert with providers, patients, government funders, donor agencies, and policy makers.
The author is privileged to work closely with many exceptional researchers and treatment providers; thus, this commentary benefits from their careful thinking and ideas. Discussions occurred at the Yale University meeting where this supplemental issue was put into motion and during the March 2006 National Institute of Mental Health/International Association of Physicians in AIDS Care International Conference on HIV Treatment Adherence.
1. Panel on Clinical Practices for the Treatment of HIV Infection. Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents. Available at: http://www.aidsinfo.nih.gov/
2. Paterson DL, Swindells S, Mohr J, et al. Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med
3. Maggiolo F, Ravasio L, Ripamonti D, et al. Similar adherence rates favor different virologic outcomes for patients treated with nonnucleoside analogues or protease inhibitors. Clin Infect Dis
4. Bangsberg DR, Moss AR, Deeks SG. Paradoxes of adherence and drug resistance to HIV antiretroviral therapy. J Antimicrob Chemother
5. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med
6. Margolin A, Avants SK, Warburton LA, et al. A randomized clinical trial of a manual-guided risk reduction intervention for HIV-positive injection drug users. Health Psychol
7. Rotheram-Borus MJ, Swendeman D, Comulada S, et al. Prevention for substance-using HIV-positive young people: telephone and in-person delivery. J Acquir Immune Defic Syndr
8. Samet JH, Horton NJ, Meli S, et al. A randomized controlled trial to enhance antiretroviral therapy adherence in patients with a history of alcohol problems. Antivir Ther
9. Andrade ASA, McGruder HF, Wu A, et al. A programmable prompting device improves adherence to highly active antiretroviral therapy in HIV-infected subjects with memory impairment. Clin Infect Dis
10. Remien RH, Stirratt MJ, Dolezal C, et al. Couple-focused support to improve HIV medication adherence: a randomized controlled trial. AIDS
11. Knobel H, Carmona A, Lopez JL, et al. Adherence to very active antiretroviral treatment: impact of individualized assessment. Enferm Infecc Microbiol Clin
12. Rathbun RC, Farmer KC, Stephens JR, et al. Impact of an adherence clinic on behavioral outcomes and virologic response in treatment of HIV infection: a prospective, randomized, controlled pilot study. Clin Ther
13. Simoni JM, Pantalone DW, Frick P, et al. Enhancing antiretroviral adherence: a review of published reports of randomized controlled trials and ongoing NIH-funded research. In: Trafton JA, Gordon W, eds. Best Practices in the Behavioral Management of Chronic Disease
. 2nd ed. Los Altos, CA: Institute for Disease Management; 2005:70-95.
14. Milam J, Richardson JL, McCutchan A, et al. Effect of a brief antiretroviral adherence intervention delivered by HIV care providers. J Acquir Immune Defic Syndr
15. Mitty JA, Mwamburi DM, Macalino GE, et al. Improved virologic outcomes and less HIV resistance for HAART-experienced substance users receiving modified directly observed therapy (MDOT): results from a randomized controlled trial. Presented at: 13th Conference on Retroviruses and Opportunistic Infections; 2006; Denver.
16. Goldie SJ, Paltiel AD, Weinstein MC, et al. Projecting the cost-effectiveness of adherence interventions in persons with human immunodeficiency virus infection. Am J Med
17. Giordano TP, Visnegarwala F, White AC Jr, et al. Patients referred to an urban HIV clinic frequently fail to establish care: factors predicting failure. AIDS Care
18. Gardner LI, Metsch LR, Anderson-Mahoney P, et al. Efficacy of a brief case management intervention to link recently diagnosed HIV-infected persons to care. AIDS
19. Gandelman A, Rietmeijer CA. Translation, adaptation, and synthesis of interventions for persons living with HIV: lessons from previous HIV prevention interventions. J Acquir Immune Defic Syndr
. 2004;37(Suppl): S126-S129.
20. Centers for Disease Control and Prevention. Diffusion of effective behavioral interventions for HIV prevention. Available at: http://www.effectiveinterventions.org
. Accessed May 12, 2006.
21. Stover J, Bertozzi S, Gutierrez JP, et al. The global impact of scaling up HIV/AIDS prevention programs in low- and middle-income countries. Science
22. Valdiserri RO. International scale-up for antiretroviral treatment: where does prevention fit? J Acquir Immune Defic Syndr
23. Maintaining anti-AIDS commitment post “3 by 5.” Lancet
24. Nachega JB, Stein DM, Lehman DA, et al. Adherence to antiretroviral therapy in HIV-infected adults in Soweto, South Africa. AIDS Res Hum Retroviruses
25. Byakika-Tusiime J, Oyugi JH, Tumwikirize WA, et al. Adherence to HIV antiretroviral therapy in HIV-positive Ugandan patients purchasing therapy. Int J STD AIDS
26. Farmer P, Leandre F, Mukherjee JS, et al. Community-based approaches to HIV treatment in resource-poor settings. Lancet
27. Koenig SP, Leandre F, Farmer PE. Scaling-up HIV treatment programmes in resource-limited settings: the rural Haiti experience. AIDS
. 2004;18 (Suppl):S21-S25.
28. Sarna A, Luchters S, Geibel S, et al. A randomized controlled trial evaluating a directly administered antiretroviral therapy (DAART) intervention to promote adherence to ART in Mombasa, Kenya. Presented at: NIMH/IAPAC International Conference on HIV Treatment Adherence; 2006; Newark.
29. Behforouz HL, Farmer PE, Mukherjee JS. From directly observed therapy to accompagnateurs: enhancing AIDS treatment outcomes in Haiti and in Boston. Clin Infect Dis
30. Paltiel AD. Cost-effectiveness of adherence interventions in HIV. Presented at: NIMH/IAPAC International Conference on HIV Treatment Adherence; 2006; Newark.
31. Elise A, France AM, Louise WM, et al. Assessment of adherence to highly active antiretroviral therapy in a cohort of African HIV-infected children in Abidjan, Cote d'Ivoire. J Acquir Immune Defic Syndr
32. Pierre RB, White YRG, Steel-Duncan J. Adherence to antiretroviral drug therapy in children with HIV/AIDS in Jamaica. Presented at: NIMH/IAPAC International Conference on HIV Treatment Adherence; 2006; Newark.
33. Mugyenyi P. Advancing HIV treatment and adherence in Africa. Presented at: NIMH/IAPAC International Conference on HIV Treatment Adherence; 2006; Newark.