Reynolds, Nancy R PhD*; Testa, Marcia A PhD†; Su, Max ScD‡; Chesney, Margaret A PhD§; Neidig, Judith L PhD∥; Frank, Ian MD¶; Smith, Scott PhD#; Ickovics, Jeannette PhD**; Robbins, Gregory K MD††; for the AIDS Clinical Trials Group 731 and 384 Teams
HIV-related morbidity and mortality have been dramatically improved in populations treated with potent combination antiretroviral therapy.1 Although different factors may influence the success of combination antiretroviral treatment over time, medication adherence has emerged as the single most important component; lower levels of adherence have been repeatedly linked with a loss of viral suppression and immune function.1 In addition to an undesirable therapeutic outcome, poor adherence is problematic in the context of clinical trials testing antiretroviral regimens because trials are powered under the assumption that treatment and control groups will be compliant with study medications. Less than perfect medication adherence may introduce biases or result in underpowered studies.2-6
Poor adherence to medication is a major obstacle to successful treatment outcomes, yet estimates suggest that at least 10% of patients with HIV in the United States report missing 1 or more medication doses on any given day, and as many as one-third to one-half of patients with HIV report missed doses in the past 2 to 4 weeks.1 Imperfect adherence to prescribed medications is not unique to persons living with HIV; nonadherence to therapies for chronic illness is a long-standing problem that is recognized as one of the most perplexing problems in health care today.7,8 Unfortunately, unlike most medications, even brief periods of nonadherence to antiretroviral regimens can render them permanently ineffective, with reduced chances of success using other options.9-11
The issue of why individuals do not take their medicine as prescribed, even when it may prolong or extend their lives, belies simple explanations. No single factor adequately accounts for adherence behavior over time; adherence to prescribed antiretroviral medication regimens is increasingly understood as a complex problem that may be influenced by a number of interacting situational and personal characteristics. Although a large body of research has improved an understanding of factors related to adherence behavior, there is currently little evidence supporting specific approaches for improving adherence to antiretroviral regimens.1,8,12
Guided by the tenets of self-regulation theory,13-16 which provides a dynamic framework for understanding the complexities of behavior, ACTG 731 was designed to test whether proactive telephone calls would improve medication adherence outcomes over time. The calls were tailored to the individual and structured to address common barriers to adherence and promote effective self-care strategies.
Study Design and Population
A randomized controlled trial was conducted to examine how standard patient education in combination with structured proactive telephone support affects adherence to combination antiretroviral therapy, as compared with a control group using standard patient education alone.
The study, ACTG 731, was 1 of 2 adherence intervention substudies of an AIDS Clinical Trial Group (ACTG) protocol. The parent protocol, ACTG 384, was a large multicenter trial that evaluated different strategies of initiating antiretroviral treatment in HIV-1-infected individuals (HIV-1 RNA ≥ 500 copies/mL) with <7 days of previous antiretroviral therapy and no acute illnesses or opportunistic infections. ACTG 384 trial participants were randomized to 6 different treatment strategies. Initial study regimens were composed of a dual nucleoside combination (zidovudine/lamivudine or didanosine and stavudine) combined with efavirenz, nelfinavir, or both, and subjects who experienced a virologic or regimen failure were switched to another study regimen designed to avoid cross-resistance. The primary endpoint was time to failure of the second of 2 consecutive 3-drug regimens or failure of the first 4-drug regimen. The study design and primary findings have previously been reported.17,18
Enrollment and Monitoring
ACTG 731 was open to ACTG 384 participants at 5 AIDS Clinical Trial Unit sites in the United States: Ohio State University, University of North Carolina, University of Pennsylvania, Washington University, and University of Nebraska. Inclusion criteria for ACTG 731 included enrollment into ACTG 384 and ability to participate in telephone calls. Subjects were enrolled to ACTG 384 between October 1998 and November 1999 and were followed for a median of 2.3 years.17 All participants signed an approved informed consent for both ACTG 384 and this adherence substudy (ACTG 731).
Clinical assessments, standard laboratory safety monitoring (complete blood count, liver function test, electrolytes, blood urea nitrogen/creatinine ratio, lipase), and plasma HIV-1 RNA measurements were obtained at screening; at entry; at weeks 4, 8, 12, 16, 20, and 24; and every 8 weeks thereafter per the ACTG 384 protocol.17,18 HIV-1 RNA levels were measured with the Roche Ultrasensitive Amplicor assay version 1.0 (Roche Diagnostics, Indianapolis, IN) with a lower limit of detection of 50 copies/mL at a central laboratory.17
Self-reported adherence was assessed with the ACTG Adherence Questionnaire19 at weeks 4, 16, 32, 48, and 64. The questionnaire queries the patient on the number of missed doses of a medication during each of the 4 days before a clinic visit (eg, “How many doses of efavirenz did you miss yesterday, the day before yesterday, 3 days ago, and 4 days ago?”). In addition, adherence is assessed with several, mostly Likert-type, general questions regarding adherence with the daily schedule (“Most study medications need to be taken on a schedule…. How closely did you follow your specific schedule over the last four days?” “Do any of your medications have special instructions?… If so, how often did you follow those instructions over the last four days?” “Some people find that they forget to take their pills on the weekend days. Did you miss any of your medications last weekend - last Saturday or Sunday?” “When was the last time you missed any of your medications?”).
After consent and pretreatment assessments were completed, participants were randomized to: (1) telephone support and standard care patient education, or (2) standard care patient education alone (the control).
Standard Patient Education (Control Condition)
All participants received standard ACTG patient education by a registered nurse or pharmacist at his or her participating site using ACTG Patient Care Committee education materials. Baseline face-to-face medication education included: (1) the dosage and timing of medications (with/without meals, time of day), (2) the importance of taking medication exactly as prescribed, (3) an ACTG site contact for problems, (4) strategies for promoting successful adherence, (5) potential significant side effects, and (6) recommended actions for problems. In addition, participants had scheduled face-to-face visits with ACTG providers per the ACTG 384 protocol. The control group did not receive regular, structured telephone calls.
Telephone Support (Experimental Condition)
In addition to the standard ACTG patient education, the telephone group received telephone calls from trained registered nurses at each of weeks 1 to 12, 14, and 16. The calls were delivered from a single, centralized site to prevent carryover effect from the intervention to the control group. Two registered nurse specialists, supervised by the investigators, were trained and then made the scheduled telephone interventions, and provided 24-hour coverage (structured weekly calls and 24-hour support available via a toll-free number).
Patients randomized to the telephone support group were asked for a telephone number and a convenient time to be reached by the trained registered nurse specialist. Participants were provided with the opportunity to ask questions regarding their medicines when contacted by the trained registered nurse specialist, or they could contact the trained registered nurse specialist at any other time as necessary (during the 16 weeks of intervention delivery) via a toll-free number. These calls were independent from the ACTG or clinical care (providers were not given information about patient treatment assignment).
At each of the regularly scheduled calls (weeks 1 to 12, 14, and 16), the registered nurse specialist attempted to contact the participant by telephone. Up to 6 attempts were made per scheduled contact point. Calls were not scheduled on days of parent protocol clinic visits. During scheduled calls, the registered nurse specialist inquired whether the patient was having any difficulty taking medicines regularly and as prescribed. If the patient was not having difficulties, the registered nurse specialist commended his or her efforts, inquired if she or he had any questions, and concluded the call. If the patient was having difficulties, the registered nurse specialist inquired as to the nature of the difficulties she or he was experiencing, recorded responses on a standardized form, and made recommendations tailored to the patient's area of difficulty using a standardized script.
If the problem was related to difficulties learning the regimen, remembering the medications, or integrating the regimen into his or her life, the registered nurse specialist made appropriate suggestions as explicated in the Strategies for Adherence instruction script (eg, “Many people find it helpful to use a daily activity, one done everyday without fail, as a prompt to take medications.”)
If the problem was related to difficulties with stress or depression, the registered nurse specialist made appropriate suggestions as explicated in the Strategies for Recognizing and Handling Depression script (eg, “Persons living with HIV can experience many kinds of stress which can make them feel sad, pessimistic and sometimes overwhelmed. While these feelings are very normal reactions, if they are not recognized and managed, they can interfere with sleeping, eating, concentrating and properly taking your medications.… Persons living with HIV often do a number of different things to maintain their mental health such as talking about their feelings with other people who also have HIV, exercising regularly, getting individual counseling, etc.…Do you do any of these things?”)
If the problem was related to difficulty with symptoms, the registered nurse specialist assessed symptom severity (grade 1, 2, 3, or 4). If the symptoms were grade 3 or grade 4, the registered nurse specialist counseled subjects to contact their providers for recommendations regarding grade 3 and grade 4 side effects and notified the participant's ACTG site. If symptoms were grade 1 or grade 2, the registered nurse specialist reassured the subject about the likely time-limited nature of particular symptoms, encouraged him or her to continue taking medications as prescribed, and suggested appropriate self-management strategies as explicated in the Strategies for Symptom Management instruction script.
The scripts were developed using best available evidence and were reviewed and approved by ACTG providers before implementation. The registered nurse specialists were guided by an effort to (1) assess what knowledge was needed by the subject regarding his or her medications and symptoms and provide or reinforce relevant information; (2) suggest feasible behaviors that the subject could engage in to access support and enhance adherence; and (3) provide reassurance and coach the subject to understand and cope with distressing affective states.
To maintain the integrity of the primary ACTG 384 analyses, objective, longitudinal indicators of adherence and health outcomes (immunologic and virologic) were not fully available for ACTG 731 analyses until ACTG 384 analyses were complete.
Data obtained from the 4-day adherence recall (item 1 of the ACTG Adherence Questionnaire) were used to derive summary measures of adherence rate for each assessment period and mean rate of adherence across weeks 4 to 64. Self-reported adherence rate was expressed as the percent of the prescribed regimen taken [1 − (proportion pills missed) × 100], where percentages closer to 100% indicate better adherence.
Adherence rates across weeks 4 to 64 were analyzed with a linear model for repeated measures with mixed effects for treatment, time, and treatment-by-time interaction. The correlation structure of within-subject repeated measures, determined by comparing Akaike's information criterion (AIC) from competing covariance structures, was used to achieve valid inferences on the fixed effects. Of special note to this analysis, no baseline adherence measures were available because subjects were not taking antiretroviral medications before randomization. Therefore, a significant treatment-by-time interaction or a significant treatment effect would indicate that treatment affects adherence. Given an overall difference in adherence over time between treatments, between-group mean difference at each week was examined using univariate analysis of variance (ANOVA). The longitudinal model was fit using maximum likelihood estimation, which takes into account the correlation among the repeated measures. Subjects missing 1 or more adherence scores were included in the analysis with no imputation for missing scores. This estimation technique provides valid estimates of the model parameters even if the “missingness” is dependent on the observed data rather than completely random.
Because the adherence rates were high across treatment groups-the majority of subjects reported perfect adherence-a post hoc analysis was conducted to consider adherence characteristics, in addition to the adherence rates (based upon 4-day recall) already analyzed. In keeping with a recent analysis that demonstrated a superior measure of adherence when all of the ACTG adherence questionnaire items are taken into account,20 reliabilities, factor analysis of the items, and effects of treatment on composite adherence scores over time were examined. The composite adherence scores were taken as the first 2 factor scores from a principal components analysis of 7 adherence items and were analyzed using the same longitudinal model as used for adherence rates.
During the analysis, it was determined that 13 persons assigned to the standard group (n = 55) at one of the participating sites incorrectly received treatment. Their data was included in the intent-to-treat (ITT) analyses, the primary analysis. A secondary on-protocol analysis was also performed that censored these subjects from the analyses. Note that the appropriate codes and instructions for randomization were sent to the clinical sites, and the subjects at each site were randomized correctly. However, a clinical site incorrectly provided the centralized site with the contact information for the 13 subjects (and those randomized to the intervention group) indicating they should receive the experimental condition. Because this error was unintentional and without systematic reason, we believe that the resulting protocol violations are correctly handled by an ITT analysis. This does force the treatment effect toward the null hypothesis, and unfortunately means a loss of power in both the ITT and on-protocol analysis. However, when the intent-to-treat treatment effects and the on-protocol treatment effects fall in the same direction, this provides assurance that the protocol deviations are not responsible for our findings.
A total of 109 subjects enrolled in ACTG 731; 55 were randomized to standard care and 54 to the telephone arm (Fig. 1). They were 85% male, 51% white (non-Hispanic), 43% black (non-Hispanic), with median CD4 cells/mm3 of 234.5 and median HIV RNA of 442,406 copies/mL. There were no significant differences in demographic measures or baseline personal and situational characteristics between the 2 randomized groups (Table 1).
The number of subjects reporting adherence decreased over time: n = 109 at baseline; n = 75 at week 64. The decrease was largely due to participants leaving the main study (ACTG 384) or reaching ACTG 384 study endpoints (rates were similar in both studies). Attrition rates in the 2 groups were similar, which was confirmed with a Cox regression of the time to off-study comparing the controls to the intervention arm (hazard ratio = 1.13, P = 0.52).
The majority (70% overall) of the scheduled telephone calls were completed each week as planned. The calls lasted an average of 7.9 minutes ± 2.5 over the 16 weeks (Table 2).
Primary Analyses (Intent-To-Treat)
Mean rates of self-reported adherence were high in both the telephone and standard care group across weeks 4, 16, 32, 48, and 64. The mean adherence rate (across weeks 4 to 64) was 97.5% ± 4.8, with the telephone group having higher mean adherence (98.0% ± 4.6) than the standard care group (97.0% ± 5.0). The repeated measures analysis of weeks 4 through 64 adherence rate scores was fit using an unstructured correlation matrix for within-subject measures. The treatment-by-time interaction was not significant (P = 0.37) and was not included in the final model. A significant treatment effect was observed (P = 0.023), indicating that adherence was higher in the telephone groups (Fig. 2). Mean rates of adherence were consistently higher in the telephone group at each time point and were significantly higher at week 64 (99.7% vs. 97.3%, P = 0.032).
In a post hoc analysis, longitudinal models for the first and second principal component factor scores showed significant benefit for telephone intervention during weeks 4 through 64 (P = 0.023 and 0.019, respectively). Means values for both factor scores were consistently higher across time in the telephone group.
Time to Virologic Failure
Comparing time to primary 384 study endpoint, the Kaplan-Meier survival curve for the telephone group (ITT) remained above the standard care group across weeks 20 to 64, suggesting a lower risk for regimen failure (Fig. 3). The Cox proportional hazards model, controlling for baseline RNA stratification, CD4 count, gender, age, race/ethnicity, and randomized ART treatment arm, maintained the directionality of this effect; however, the telephone intervention was not a statistically significant predictor of failure (hazard ratio = 0.68; 95% confidence interval: 0.38 to 1.23, P = 0.21). Using the observed hazard ratio, an accrual period of 2 years, and follow-up of 160 weeks, the calculated power for this analysis was 0.31.
Secondary Analyses (On Protocol)
The on-protocol findings, censoring the 13 misallocated patients (41 standard care/54 telephone arm), were similar to the ITT analyses and also showed a significant treatment-by-time interaction. The mean adherence rate (across weeks 4 to 64) was 97.7% ± 4.5, with the telephone group having higher mean adherence (98.0% ± 4.6) than the standard care group (97.4% ± 4.5). The repeated measures analysis of weeks 4 through 64 adherence rate scores was fit using an unstructured correlation matrix for within-subject measures. A significant treatment-by-time interaction was found between weeks 4 and 64 (P = 0.039). Like the ITT analysis, mean adherence rates were also significantly different at week 64 and favored the telephone group (99.7% vs. 97.3%, P = 0.022).
This randomized controlled trial compared antiretroviral adherence and clinical outcomes among ACTG 731 participants randomized to receive either proactive telephone support in addition to standard patient education or standard patient education alone. The mean rates of antiretroviral adherence were high in both the standard patient education and the standard education plus telephone support groups. Nevertheless, the statistically significant improvement in adherence in the treatment group provides support for the efficacy of a central telephone support intervention. The positive effect of the telephone intervention is further supported by the fact that composite adherence scores, taken as the first 2 factor scores from a principal components analysis, found significant intervention benefit.
Although the high rates of adherence limited our ability to demonstrate the full potential of this intervention, the fact that we were able to show significant differences in this highly adherent population is promising. Despite considerable interest in adherence interventions, to date there have been few clinically practical advances in this area. The adherence rates among participants of ACTG trials were unknown at the planning and start-up phase of this trial. Rates of adherence have since been demonstrated to be high in ACTG trials in contrast to those observed in general HIV-infected clinic populations.1 Even with this ceiling effect, the higher rates of adherence demonstrated in the telephone group in contrast to the standard care group suggests that even greater benefits may be realized among general clinic populations, where self-reported adherence rates are substantially lower. Moreover, current evidence suggests that significant long-term survival benefit and cost effectiveness may be realized with even modest improvements in adherence to antiretroviral medications (eg, Freedberg et al21).
A self-regulation framework derived from prior conceptual and empiric work provided direction for structuring the proactive telephone calls.13 Self-regulation theory emphasizes the importance of not only providing pertinent information, but addressing necessary skills and affective processes that may influence a health behavior change. Precepts of self-regulation theory suggest that acquisition of this repertoire will likely require a process of conceptual change and reinforcement over time.13 Many attempts have been made to promote adherence by providing information.8 However, research conducted across different chronically ill population groups has shown that while knowledge of a regimen is a necessary component of adherence, educational strategies alone are not sufficient to promote sustained adherence behavior.8,22,23 This telephone intervention was designed not only to provide information to enhance subjects' comprehension of the antiretroviral regimen and realistically prepare them for the experience, but to recognize, self-manage, and effectively solve problems that commonly threaten adherence over time.
Findings from this study are noteworthy in that adherence did not drop off in the telephone group following completion of the intervention at week 16 (a phenomenon that has been observed in other adherence intervention studies1,24). In contrast to the standard care group, where adherence tended to vacillate and decline over time, high levels of adherence in the telephone intervention group were sustained at weeks 32, 48, and 64. This suggests the intervention may have been effective in teaching effective self-regulating, problem-solving strategies that were maintained over time by participants even after the supportive telephone calls had ended. This also suggests that some coping skills may have been taught. Tailored counseling regarding strategies to remember and integrate medication-taking into daily living, interpret and self-manage problematic symptoms, and understand and cope with distressing affective states was provided and reinforced over time.
The strength of the study findings may also be attributable to the centralized approach used for delivery of the telephone intervention. Other trials (eg, Collier et al25) have been designed so that clinicians within a site deliver both standard care to control participants and the telephone intervention to persons assigned to the treatment condition. The centralized approach used in this study controlled for within-site dispersion of the intervention, and may also have been effective in controlling for provider bias by limiting within-site provider knowledge of treatment assignment. The treatment group also may have profited by having 24-hour support available via a toll-free number in addition to the scheduled calls. Although few unscheduled calls were initiated by the treatment group participants, they may have benefited by knowing the nurse specialist support was readily available if needed.
Findings from this study, though promising, must be regarded with caution. A difference for the primary clinical endpoint favorable to the intervention group was observed, and a 95% confidence interval that included values that would likely be considered clinically meaningful supports the potential of the intervention; however, these findings were not statistically significant. The results from the Cox model show the confidence interval around the hazard ratio to be quite large (0.38 to 1.23), indicating that the sample size may have been underpowered to properly test this endpoint. There is a need to replicate the intervention with a larger sample that is more representative of average adherence behavior found outside of ACTG trials to fully demonstrate clinical benefits; with a less adherent population, there is greater opportunity for improvement. Further, there is a need to examine the mechanisms of change. An ongoing trial is testing the intervention approach in a diverse sample of rural and urban AIDS Drug Assistance Program (ADAP) participants (RO1 NR0510).
In summary, despite the need for interventions to improve adherence, data from which to derive evidence-based practice recommendations are limited. Findings from this study indicate that a tailored, proactive telephone intervention that can be centralized may be ideal for translational research. Additional research to further substantiate the effectiveness of the approach in large samples with diverse characteristics is warranted.
We wish to thank the ACTG 384 team AIDS Clinical Trials Unit personnel and study volunteers for their contributions to this project.
1. Reynolds NR. Adherence to antiretroviral therapies: state of the science. Curr HIV Res. 2004;2:207-214.
2. Goetghebeur E, Molenberghs G, Katz J. Estimating the causal effect of compliance on binary outcome in randomized controlled trials. Stat Med. 1998;17:341-355.
3. Imbens GW, Rubin DB. Bayesian inference for causal effects in randomized experiments with noncompliance. Ann Statist. 1997;25:305-327.
4. Little RJ, Rubin DB. Causal effects in clinical and epidemiological studies via potential outcomes: concepts and analytical approaches. Annu Rev Public Health. 2000;21:121-145.
5. Rubin DB. More powerful randomization-based p-values in double-blind trials with non-compliance. Stat Med. 1998;17:371-385; discussion 387-389.
6. Robins JM, Finkelstein DM. Correcting for noncompliance and dependent censoring in an AIDS Clinical Trial with inverse probability of censoring weighted (IPCW) log-rank tests. Biometrics. 2000;56:779-788.
7. Sabate E, ed. Adherence to Long-Term Therapies: Evidence for Action. Geneva, Switzerland: World Health Organization; 2003.
8. Haynes RB, McDonald H, Garg AX, et al. Interventions for helping patients to follow prescriptions for medications. Cochrane Database Syst Rev. 2002;2:CD000011.
9. Descamps D, Joly V, Flandre P, et al. Genotypic resistance analyses in nucleoside-pretreated patients failing an indinavir containing regimen: results from a randomized comparative trial: (Novavir ANRS 073). J Clin Virol. 2005;33:99-103.
10. King MS, Brun SC, Kempf DJ. Relationship between adherence and the development of resistance in antiretroviral-naive, HIV-1-infected patients receiving lopinavir/ritonavir or nelfinavir. J Infect Dis. 2005;191:2046-2052.
11. Lucas GM. Antiretroviral adherence, drug resistance, viral fitness and HIV disease progression: a tangled web is woven. J Antimicrob Chemother. 2005;55:413-416.
12. Simoni JM, Frick PA, Pantalone DW, et al. Antiretroviral adherence interventions: a review of current literature and ongoing studies. Top HIV Med. 2003;11:185-198.
13. Reynolds NR. The problem of antiretroviral adherence: a self-regulatory model for intervention. AIDS Care. 2003;15:117-124.
14. Alonzo AA, Reynolds NR. The structure of emotions during acute myocardial infarction: a model of coping. Soc Sci Med. 1998;46:1099-1110.
15. Leventhal H, Benyamini Y, Brownlee S, et al. Illness representations: theoretical foundations. In: Petrie KJ, Weinman JA, eds. Perceptions of Health and Illness: Current Research and Applications. Singapore: Harwood Academic Publishers; 1997:19-45.
16. Cameron LD, Leventhal H. The Self-Regulation of Health and Illness Behaviour. New York: Routledge; 2003.
17. Robbins GK, De Gruttola V, Shafer RW, et al. Comparison of sequential three-drug regimens as initial therapy for HIV-1 infection. N Engl J Med. 2003;349:2293-2303.
18. Smeaton LM, DeGruttola V, Robbins GK, et al. ACTG (AIDS Clinical Trials Group) 384: a strategy trial comparing consecutive treatments for HIV-1. Control Clin Trials. 2001;22:142-159.
19. Chesney MA, Ickovics JR, Chambers DB, et al. Self-reported adherence to antiretroviral medications among participants in HIV clinical trials: the AACTG Adherence Instruments. AIDS Care. 2000;12:255-266.
20. Reynolds NR, Sun J, Nagaraja HN, et al. Optimizing measurement of self-reported adherence AIDS Clinical Trials Group Adherence Questionnaire: a cross protocol analysis. J Acquir Immune Defic Syndr. (in press).
21. Freedberg KA, Hirschhorn LR, Schackman BR, et al. Cost-effectiveness of an intervention to improve adherence to antiretroviral therapy in HIV-infected patients. J Acquir Immune Defic Syndr. 2006;43(Suppl 1):S113-S118.
22. Roter DL, Hall JA, Merisca R, et al. Effectiveness of interventions to improve patient compliance: a meta-analysis. Med Care. 1998;36:1138-1161.
23. Wright MT. The old problem of adherence: research on treatment adherence and its relevance for HIV/AIDS. AIDS Care. 2000;12:703-710.
24. Rigsby MO, Rosen MI, Beauvais JE, et al. Cue-dose training with monetary reinforcement: pilot study of an antiretroviral adherence intervention. J Gen Intern Med. 2000;15:841-847.
25. Collier AC, Ribaudo H, Mukherjee AL, et al. A randomized study of serial telephone call support to increase adherence and thereby improve virologic outcome in persons initiating antiretroviral therapy. J Infect Dis. 2005;192:1398-1406.
© 2008 Lippincott Williams & Wilkins, Inc.