MAJOR ARTICLE
HIV/AIDS
A Programmable Prompting Device Improves Adherence to Highly Active Antiretroviral Therapy in HIV-Infected Subjects with Memory Impairment Adriana S. A. Andrade,1 Henraya F. McGruder,3 Albert W. Wu,2,5 Shivaun A. Celano,2 Richard L. Skolasky Jr.,3,4 Ola A. Selnes,3,4 I-Chan Huang,5 and Justin C. McArthur3,4 Divisions of 1Infectious Diseases and 2General Internal Medicine, Department of Medicine, and Departments of 3Neurology, 4Epidemiology, and 5Health Policy and Management, The Johns Hopkins University, Baltimore, Maryland
Therapeutic failure occurs in approximately one-half of HAART recipients and is often associated with acquisition of resistance mutations by HIV because of incomplete adherence [1]. Patients frequently cite “forgetting” as a reason for nonadherence to HAART [2, 3]. HIV-infected patients with memory dysfunction are at increased risk of poor medication adherence, compared with their memory-intact counterparts [4–6]. Memory-assistive methods, such as use of prompting
devices, seem to be well suited to help individuals overcome forgetfulness [7]. To test our hypothesis that the use of a medicationreminder device by HIV-infected patients might enhance their adherence to HAART, we conducted a randomized, controlled trial of a programmable portable electronic device that gives verbal reminders to take medications and records dosing times for individualized HAART regimens. We examined whether use of this device increased adherence to HAART and improved virological and immunological outcomes.
Received 19 January 2005; accepted 10 May 2005; electronically published 5 August 2005. Reprints or correspondence: Dr. Adriana S. A. Andrade, Div. of Infectious Diseases, Dept. of Medicine, The Johns Hopkins School of Medicine, 1830 E. Monument St., Ste. 8074, Baltimore, MD 21205 (
[email protected]).
SUBJECTS AND METHODS
Clinical Infectious Diseases 2005; 41:875–82 2005 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2005/4106-0019$15.00
From 1999 through 2001, we enrolled 64 HIV-infected male and female subjects who were ⭓18 years of age,
Study Group
HIV/AIDS • CID 2005:41 (15 September) • 875
Downloaded from http://cid.oxfordjournals.org/ by guest on February 2, 2016
Background. Patients cite “forgetting” as a reason for nonadherence to highly active antiretroviral therapy (HAART). We measured the effect of a memory-prompting device on adherence to HAART in memory-intact and memory-impaired human immunodeficiency virus (HIV)–infected subjects. Methods. The study was a prospective, randomized, controlled trial involving 64 HIV-infected adults. The intervention was the Disease Management Assistance System (DMAS) device, combined with monthly adherence counseling. Control subjects received only adherence counseling. The DMAS was programmed with HAART regimen data to provide verbal reminders at dosing times. Adherence was measured for 24 weeks using electronic drug exposure monitor (eDEM) caps. Results. A total of 58 subjects completed the 24-week study period; 28 were HAART naive (12 DMAS users and 16 control subjects). Mean adherence scores did not differ significantly between DMAS users (80%) and control subjects (65%). Post hoc analysis of 31 memory-impaired subjects (14 DMAS users and 17 control subjects) revealed significantly higher adherence rates among DMAS users (77%), compared with control subjects (57%) (P p .001). However, analysis of memory-intact subjects showed that adherence was not significantly improved for DMAS users (83%), compared with control subjects (77%) (P p .25 ). At week twelve, 38% of the DMAS users and 14% of the control subjects had an undetectable plasma HIV RNA load (P p .014 ), and at week 24, the plasma HIV RNA load was undetectable for 34% of the DMAS users and 38% of the control subjects (P p .49). CD4+ cell counts did not differ between the study arms. Virological and immunological responses were not related to DMAS use in memory-impaired subjects. Conclusion. The DMAS prompting device improved adherence for memory-impaired subjects but not for memory-intact subjects.
Adherence Measurements and Intervention
able to self-medicate, and currently receiving care at The Johns Hopkins Moore (HIV) Clinic (Baltimore, MD). Subjects eligible for inclusion were either previously treatment naive and initiating HAART for the first time or antiretroviral experienced and switching HAART regimens. Among subjects in the latter group, we included only those who had received ⭐3 HAART regimens before study enrollment. Exclusion criteria were inability to self-medicate, presence of severe dementia, and institutionalization. There were no criteria for CD4+ cell count, plasma HIV RNA load, or HAART regimen. The Institutional Review Board at The Johns Hopkins University approved the study. Each participant gave written informed consent and received $25 per study visit. The Disease Management Assistance System (DMAS) Device
The DMAS (Adherence Technologies) is a portable (weight, 168 g; dimensions, 11.5 ⫻ 7.0 ⫻ 2.5 cm), battery-powered electronic device that uses a digital signal processor to produce a timed, programmed voice message that prompts subjects to take their antiretrovirals (figure 1). It records dosing times and dates when the subject pushes a response button, and the data can be uploaded and printed. The device can store up to 3 months of messages for up to 25 different medications. We have previously evaluated the feasibility of using the DMAS to enhance treatment adherence by HIV-infected subjects [8]. 876 • CID 2005:41 (15 September) • HIV/AIDS
Data Collection and Procedures
Baseline evaluation. Subjects were randomly assigned to intervention or control groups. All subjects participated in an individualized, 30-min adherence counseling session each month and received adherence feedback from a standardized transcript that provided general education about the barriers to adherence, the hazards of nonadherence, and their prescribed HAART regimen. Adherence counseling and feedback were provided by a clinical pharmacist with extensive experience in the field of HIV/AIDS. Patients in the intervention group were also given the DMAS device for 24 weeks. The DMAS device was programmed with reminder messages and dosing times for each medication in the HAART regimen. Devices were inspected monthly and reprogrammed when the
Downloaded from http://cid.oxfordjournals.org/ by guest on February 2, 2016
Figure 1. The Disease Management Assistance System device. When it is time for a remainder, the device beeps, and the light above the “play” button blinks. The patient then presses the “play” button to hear the voice message. The light above the “yes” button then blinks, and the patient presses this button, which prompts the device to record the time and day when the medication was taken.
Adherence to HAART was measured for 24 weeks using electronic drug-exposure monitoring (eDEM) caps (Aprex) [9]. We used the AIDS Clinical Trial Group (ACTG) Adherence Questionnaire as a secondary measurement of adherence [10]. Both tools were used to validate adherence rates derived from the response button of the DMAS device. Mean adherence scores obtained from eDEM caps were used to describe the impact of the DMAS on adherence behavior. At the initial study visit, subjects were trained in the use of the eDEM caps and were counseled not to remove extra antiretroviral doses for later use. Data on eDEM caps were retrieved monthly, and adherence results were reviewed with all participants. eDEM caps were preferentially placed on a medication bottle containing a protease inhibitor (PI) or a nonnucleoside reversetranscriptase inhibitor (NNRTI). For dual-PI regimens and ritonavir-boosted PI regimens, the cap was placed on the bottle containing the primary PI or on the bottle containing the PI associated with the largest number of pills/day. For nucleoside reverse-transcriptase inhibitor (NRTI)–only regimens, the cap was placed on the bottle containing the NRTIs that were administered twice daily. Adherence scores during the 4 days preceding the study visit were calculated as the number of recorded bottle openings divided by the number of doses prescribed per day, multiplied by the number of days monitored. Participants completed the ACTG Baseline Adherence Questionnaire during the initial study visit and the ACTG Followup Adherence Questionnaire during subsequent visits. Fourday average adherence was calculated as the number of prescribed doses minus the number of missed doses, divided by the number of prescribed doses. For the DMAS device, adherence was calculated as the number of times the response button was pressed divided by the number of medication prompts during the 4-day period preceding the study visit. Upon completion of the study, patients returned the eDEM caps and the DMAS devices.
Table 1.
Neuropsychological assessment battery.
Cognitive function tested
Neuropsychological test
Attention
California Computerized Assessment Package (Choice and Sequential)
Verbal memory Working memory Motor
Hopkins Verbal Learning Test—Revised Delayed Recognition Span Test Grooved Pegboard (Nondominant Hand); Trailmaking Test A Verbal Fluency; Trailmaking Test B
Executive NOTE.
Adapted from [11].
RESULTS Participant characteristics. Fifty-eight of the 64 enrolled subjects completed the 24-week study (figure 2). The distribution of HAART-naive subjects was not significantly different be-
Laboratory Procedures
CD4+ cell count and plasma HIV RNA load were assessed at baseline and at weeks 12 and 24 of follow-up. CD4+ cell counts were determined using flow cytometry assays, and plasma HIV RNA loads were measured with the Amplicor HIV-1 Monitor test (Roche Diagnostics) at baseline (lower limit of detection, !400 copies/mL) and with the UltraSensitive test (lower limit of detection, !50 copies/mL) thereafter. Statistical Analysis
Assuming an initial adherence rate of 65% in the control group, we estimated that 84 subjects would be required to detect a 25% difference in adherence rates between the control and
Figure 2. Flow of HIV-infected participants in a trial comparing the efficacy of the Disease Management Assistance System for improving adherence to HAART. HIV/AIDS • CID 2005:41 (15 September) • 877
Downloaded from http://cid.oxfordjournals.org/ by guest on February 2, 2016
HAART regimen was changed or replaced if they were lost or malfunctioning. Follow-up evaluations. The following 3 factors were assessed during the follow-up period: adherence, neuropsychological (NP) status, and mood disorder and substance use. To assess adherence, participants returned monthly during the study to have DMAS and eDEM caps adherence data uploaded and to complete the ACTG Follow-up Adherence Questionnaire. To determine the influence of NP status on adherence, we used validated NP tests to assess attention, memory, new learning, psychomotor speed, and executive functions (table 1) [11]. A trained tester administered the NP test battery twice during the study, at baseline and after 24 weeks of HAART. Appropriate normative data were used to define impairment for individual NP tests and summary measurements [12, 13]. NP impairment was defined as a score of 1.5 SDs below the mean, and memory impairment was defined as at least 1 Hopkins Verbal Learning Test–Revised subtest score of ⭓1.5 SDs below the mean. The principal outcome measure was a summary z score average of individual NP tests. Symptoms of depression were assessed using the Center for Epidemiologic Studies Depression (CES-D) scale. Subjects were categorized as having clinically significant symptoms of depression if their CES-D scores were ⭓16 [14]. Participants were also questioned about active illicit drug use and alcohol use during the past 4 days.
intervention groups, with a power of 80% and a 2-sided a of 0.05. Following a planned interim analysis, the assumptions were modified, and the necessary sample size was revised to 64 subjects (PASS statistical software; NCSS). All analyses were performed under an intention-to-treat paradigm by use of SAS software, release 8.02 (SAS Institute) [15]. Baseline differences between the 2 groups were tested using conventional parametric and nonparametric statistics. For continuous measures, distributional assumptions were checked [16]. In cases in which these assumptions held, parametric tests were used. Otherwise, nonparametric methods were used. Categorical and dichotomous measures were assessed using x2 analysis and Fisher’s exact test. Longitudinal differences between groups were tested using paired Student’s t tests to investigate differences at 2 time points or general estimating equations to examine trajectories over multiple time points [17]. A post hoc subgroup analysis was conducted using data from individuals with baseline memory impairment to determine whether the monitoring device had an impact in this population. A number of participants did not provide complete adherence information for all study periods, because of loss to followup, death, or mechanical malfunction of the DMAS and/or eDEM devices. The following a priori strategy was used: durations associated with scheduled treatment holidays or interruptions or with institutionalization and durations that were not monitored because of device malfunction or loss of electronic equipment were removed from the calculation of mean adherence scores. Participants who were lost to follow-up or died were considered to have provided data up to the point of last contact. Differences in the rates of loss to follow-up and death were calculated between the 2 study groups.
Table 2. Baseline demographic and clinical characteristics of participants in a study comparing the efficacy of the Disease Management Assistance System (DMAS) for improving adherence to HAART.
Characteristic
38 7 16 (55) 26 (90) 235 141 4.4 0.2
⫺2.17 ⫺1.59 ⫺3.01 ⫺0.48 ⫺2.05 ⫺0.45 ⫺2.21 ⫺2.14
1.80 1.56 4.18 1.13 2.08 0.97 1.91 2.26
⫺1.12 1.45 ⫺0.72 1.00 14 (48) 16 (55)
3 1 0 0 2
(10) (3) (0.0) (0.0) (7)
Control group (n p 29) 38 7 18 (62) 25 (86) 213 245 4.3 0.2
⫺1.95 ⫺2.21 ⫺1.65 0.10 ⫺1.71 ⫺0.47 ⫺1.87 ⫺1.89
1.66 2.02 4.44 1.30 4.18 1.86 2.45 2.34
⫺0.94 1.56 ⫺0.66 1.22 17 (59) 16 (55)
2 0 1 0 3
(7) (0.0) (3) (0.0) (10)
12 (41) 9 (31) 6 (21)
16 (55) 5 (17) 7 (24)
17.7 11.0 14 (48)
23.7 11.8 20 (69)
NOTE. Data are no. (%) of participants or mean value SD . No differences between the 2 groups were statistically significant, as shown by Student’s t testing. CALCAP RT, California Computerized Assessment Package; CDC, Centers for Disease Control and Prevention HIV disease; CES-D, Center for Epidemiologic Studies Depression; DRST, Delayed Recognition Span Test; GP, Grooved Pegboard (Nondominant Hand) Trailmaking Test A; HVLT, Hopkins Verbal Learning Test—Revised. a b c d
Data are for 28 subjects in the intervention group and 28 subjects control group. Data are for 17 subjects in the intervention group and 13 subjects control group. See Subjects and Methods for a description of the scoring system. Data are for 27 subjects in the intervention group and 28 subjects control group.
tween groups (P p .12), and antiretroviral drug classes were evenly distributed (data not shown). HAART regimens were changed for 3 patients in the intervention group and 4 patients in the control group. HAART was discontinued for 10 subjects in the intervention group and 8 subjects in the control group. Eight DMAS users and 5 control subjects did not reinitiate HAART by the end of week 24. Thus, 58 subjects—29 from each group—were included in the final data analysis. 878 • CID 2005:41 (15 September) • HIV/AIDS
Baseline demographic and clinical characteristics were similar for the 2 groups (table 2); 32 subjects (55%; 16 in each group) had AIDS. Twenty-eight subjects (48%; twelve DMAS users and 16 control subjects) were antiretroviral naive. Nine subjects (31%) in the intervention group and 5 (17%) in the control group were taking their first HAART regimen, and 6 (21%) and 7 (24%), respectively, were receiving their second HAART regimen. Mean plasma HIV RNA loads and CD4+ cell counts were comparable between the 2 study groups.
Downloaded from http://cid.oxfordjournals.org/ by guest on February 2, 2016
Age, years Male sex Black race CD4+ cell count, cells/mm3a a Plasma HIV RNA load, log10 copies/mm3 a,c Neuropsychological test, z score HVLT Trial 3 Delayed recall Recognition DRST GP (Nondominant Hand) Verbal Fluency Trailmaking Test A Trailmaking Test B CALCAP Choice Sequential Memory impaired AIDS, according to CDC classification Drug use during past 4 days before follow-up visit Marijuana Cocaine Heroin Amphetamine Methadone treatment HAART status at study entry Naive Receiving first HAART regimen Receiving second HAART regimen CES-D analysis Overall scored Score of ⭓16
Intervention group (n p 29)
Table 3. Results of an intent-to-treat analysis showing mean adherence scores during the 4 days preceding follow-up visits for 29 Disease Management Assistance System users and 29 control subjects Intervention group, mean percentage adherence SE
Control group, mean percentage adherence SE
4 8
76 5.7 75 6.3
69 5.2 68 6.1
12 16 20
87 4.9 83 5.9 80 5.8
64 6.3 62 7.4 61 7.3
24 Overall
80 6.3 80 2.3
64 8.2 65 2.7
Week after study entry
NOTE. By week 24, participants in the intervention group achieved a mean adherence score of 83%, compared with 77% for control subjects (P p .25). Analysis was conducted using general estimating equations, adjusting for plasma HIV RNA load, CD4+ cell count, depression, and drug abuse.
Figure 3. Results of an intent-to-treat analysis showing the overall mean log10 copies/mm3 change in plasma HIV RNA loads and absolute CD4+ cell counts from baseline to week 24 for patients who used the Disease Management Assistance System (DMAS) and for control subjects. At week 24, significantly more DMAS users than control subjects had a decrease in their plasma HIV RNA load of 11 log10 copies/mm3 (P ! .05). The analysis was conducted using Student’s t test. *P ! .05 relative to the control group. HIV/AIDS • CID 2005:41 (15 September) • 879
Downloaded from http://cid.oxfordjournals.org/ by guest on February 2, 2016
Adherence to HAART and analysis of biological markers. Partial eDEM cap adherence data were available for 11 subjects (6 DMAS users and 5 control subjects). Two DMAS devices were not returned; 7 had to be replaced because of electronic malfunction (i.e., they were not playing messages, were not beeping, or had nonfunctional batteries). Overall mean adherence scores measured by means of data from eDEM caps were not significantly different between DMAS users (80%) and control subjects (65%) (table 3). The effect size was 0.51 (95% CI for Cohen’s d, 0.26–0.76). At week 12, eleven DMAS users (38%) and 4 control subjects (14%) had undetectable plasma HIV RNA loads (defined as a plasma HIV RNA load of !50 copies/mL) (P p .014), and at week 24, plasma HIV RNA loads were undetectable for 10 (34%) and 11 (38%), respectively (P p .49 ). At week 12, sixteen subjects (55%) in the intervention group and 12 (41%) in the control group experienced a decrease in the plasma HIV RNA load of at least 1 log10 copies/mm3 (P p .098). At week 24, reductions in plasma HIV RNA loads were significantly greater in the intervention group, with 21 DMAS users (72%) and 12 control subjects (41%) experiencing a decrease of at least 1 log10 copies/mm3 (P p .02). Overall mean reductions in plasma HIV RNA loads from baseline to week 24 were ⫺2.10 and ⫺0.98 log10 copies/mm3 for DMAS users and control subjects, respectively (P p .02) (figure 3). The mean CD4+ cell counts (SD) in the intervention and control groups were 337 183 and 258 184 cells/mm3, respectively, at week 12 (P p .14), and were 301 172 and 250 172 cells/mm3, respectively, at week 24 (P p .28). The overall increase from baseline to week 24 was nonsignificantly greater in the intervention group, compared with the control group (68 vs. 45 cells/mm3; P p .62) (figure 3). The post hoc analysis of the impact of the DMAS on HAART
adherence was conducted in subgroups of memory-impaired and memory-intact patients. Baseline NP testing indicated that 31 participants (53%; fourteen DMAS users and 17 control subjects) had memory impairment. Among memory-impaired subjects, overall mean adherence was statistically significantly higher for the DMAS group (77%) than for the control group (57%) (P p .001) (table 4). This difference was still significant after adjusting for CD4+ cell count and plasma HIV RNA load. In contrast, among memory-intact subjects, adherence did not differ significantly between the DMAS group (83%) and the control group (77%; P p .25) (table 4). In the DMAS intervention group, there were no significant differences in the virological or immunological responses between memory-impaired and memory-intact subjects; mean decrements in plasma HIV RNA loads at week 24 were ⫺1.8 and ⫺1.5 log10 copies/mm3, respectively (P p .56 ). At week 24, mean CD4+ cell counts among memory-impaired and memoryintact subjects were 317 and 246 cells/mm3, respectively (P p .27). At week 20, a total of 88% of the patients in the DMAS group reported that they used the device some or all of the time to manage HAART use. Only 1 subject reported never having used the device. Skipping of medications was reported as being “much less than before” and the time at which HAART was taken was reported as being “much more consistent” by 56% of the subjects in the DMAS group.
Table 4. Results of an intent-to-treat analysis showing mean adherence scores during the 4 days preceding follow-up visits for 29 Disease Management Assistance System users and 29 control subjects, by memory status.
Week after study entry
Mean adherence percentage SE for subjects with impaired memory
Mean adherence percentage SE, for subjects with intact memory
Intervention group
Control group
Intervention group
Control group
4 8
70 8.8 75 8.8
63 6.8 62 8.4
82 7.0 70 9.4
77 7.7 76 8.5
12 16 20
81 7.3 72 9.3 73 7.5
56 8.4 55 9.8 49 8.7
88 7.7 91 6.2 86 8.9
77 8.1 76 9.6 80 9.3
24 Overall
79 10.8 77 3.4
56 10.6 57 3.5
82 8.0 83 3.3
76 12.5 77 3.5
We also investigated whether baseline NP performance predicted adherence to HAART in the intervention and control groups at week 24. At the 6-month evaluation, there was no correlation between adherence scores and NP test performance for the evaluations measuring attention, memory, new learning, psychomotor speed, and executive functions. Overall mean adherence scores measured by data from the eDEM caps and the DMAS devices for memory-impaired subjects were correlated (r p 0.758 ; P p .022). We found a low correlation between self-report and data from the DMAS device (r p 0.354; P p .002) and no significant correlation between self-report and data from the eDEM caps (r p 0.242; P p .624). DISCUSSION The results of this randomized, controlled trial suggest that an electronic verbal prompting device can improve adherence to HAART by HIV-infected subjects who have memory impairment. The effect of the DMAS device at 24 weeks was only evident in the memory-impaired group, resulting in a 20% increase in the adherence rate, compared with 6% for the memory-intact patients. Other studies have shown a deleterious effect of memory dysfunction on medication adherence in HIV-infected patients. In a cross-sectional study of HIV-infected subjects, Hinkin et al. [4] found that poorer cognitive function was associated with worse adherence. Albert et al. [5] found that deficits in memory and executive function compromised adherence to HAART. A number of electronic devices are being investigated as potential medication adherence–enhancing tools, but none have addressed the specific challenges faced by patients with HIV-associated memory impairment [18–25]. To our knowledge, this is the first prospective, randomized study to show 880 • CID 2005:41 (15 September) • HIV/AIDS
that a reminder device improves medication adherence in HIVinfected subjects with memory deficits. Because memory deficits may be an important obstacle to adherence in these patients [4–6], interventions targeting reinforcement may be particularly useful. Medication adherence has been shown to degrade over time, making it important to identify effective interventions to provide continued reinforcement beyond the initial period of treatment [26]. We tested a device that used reminders, because repetition has been shown to be important for successful adherence interventions and might be particularly beneficial to HIV-infected subjects with memory deficits [26]. In our study, the DMAS device helped memory-impaired patients compensate for their memory deficits by providing ongoing medication reminders between study visits. Reports from the DMAS group suggest that nearly all patients were willing to use the device and that they used it consistently. Although the DMAS resulted in improved adherence, the overall mean adherence score was only 77% for DMAS users with memory deficits, which is well below the score of 95% commonly cited as necessary for optimal viral suppression [27]. The modest enhancement in adherence to HAART could be explained by the high proportion of subjects in our study with memory deficits, compared with previous trials. Because we selected for self-medicating subjects with memory deficits, the 95% threshold for clinically significant improvement may have been unachievable. Furthermore, HIV-associated cognitive impairment may affect domains other than memory, including executive functioning and attention [6]. The problem of nonadherence in HIV treatment is complex, and a number of factors are known to influence adherence to HAART [2, 3, 28–34]. Symptoms of depression were prevalent in our study population, whereas drug use was less prevalent. However, because the mean CES-D scores and the frequency
Downloaded from http://cid.oxfordjournals.org/ by guest on February 2, 2016
NOTE. By week 24, adherence scores were higher in the intervention group, regardless of their memory function. However, among memory-impaired subjects, scores were statistically significantly greater in the intervention group (P p .001 ). Analysis was conducted using general estimating equations, adjusting for plasma HIV RNA load and CD4+ cell count.
disease caregivers might be unaware of the negative impact of subtle memory impairments on adherence to HAART [36]. This highlights the importance of adherence aids, such as the DMAS, that aim to bolster adherence. In conclusion, poor adherence to HAART may be explained in part by HIV-associated memory deficits. In this study, adherence was improved by a memory-prompting device. Patients with HIV-associated memory deficits may benefit from additional support to reach and sustain necessary levels of adherence to HAART. Practices that boost memory should be integrated with other adherence interventions in this patient population. Future trials are needed to confirm and explore the longerterm benefits of electronic-reminder devices to promote adherence, especially in persons with HIV-associated memory impairment. Acknowledgments We thank A. Letzt for providing the DMAS devices and technical support and Dr. D. McClellan for editorial assistance. Financial support. National Institutes of Health (grants NS44807 and NS049465 [to J.C.M.] and NS36519 [to Dr. Leon Epstein]) and the General Clinical Research Center of the Johns Hopkins Hospital (grant MO1RR00052). Merck Laboratories provided an unrestricted educational grant. Potential conflicts of interest. A.S.A.A. received grant and research support and honoraria from Pfizer/Agouron (unrelated study). O.A.S. is a consultant for Eisai Pharmaceuticals. S.A.C. is on the speakers’ bureau for GlaxoSmithKline and Roche Laboratories and consulted for Abbott Laboratories. A.W.W. received grant and research support from Boehringer Ingelheim and Pfizer and is a consultant for Bristol-Myers-Squib, AstraZeneca, Abbott, and Aventis. All other authors: no conflicts.
References 1. Deeks S, Beatty G, Cohen PT, et al. Viral load and CD4+ T cell changes in patients failing potent protease inhibitor therapy [abstract 419]. In: Program and abstracts of the 5th Conference on Retroviruses and Opportunistic Infections (Chicago). Alexandria, VA: Foundation for Retrovirology and Human Health, 1998. 2. Cheever LW, Wu AW. Medication adherence among HIV-infected patients: understanding the complex behavior of patients taking this complex therapy. Curr Infect Dis Rep 1999; 1:401–7. 3. Wu AW, Ammassari A, Antinori A. Adherence to antiretroviral therapy: where are we, and where do we go from here? J Acquir Immune Defic Syndr 2002; 31:S95–7. 4. Hinkin CH, Castellon SA, Durvasula RS, et al. Medication adherence among HIV positive adults: effects of cognitive dysfunction and regimen complexity. Neurology 2002; 59:1944–50. 5. Albert SM, Flatter SR, Clouse R, et al. Medication management skill in HIV: I. Evidence for adaptation of mediation management strategies in people with cognitive impairment. II. Evidence for pervasive lay model of medication efficacy. AIDS Behav 2003; 7:329–38. 6. Selnes OA. Neurocognitive aspects of medication adherence in HIV infection. J Acquir Immune Defic Syndr 2002; 31(Suppl 3):S132–5. 7. Van den Broek MD, Downes J, Johnson Z, Dayus B, Hilton N. Evaluation of an electronic memory aid in the neuropsychological rehabilitations of prospective memory deficits. Brain Inj 2000; 14:455–62. 8. Andrade A, Davis H, Wu AW, et al. Feasibility study of the Disease Management Assistance System: a potential adherence-enhancing device [abstract 298]. In: Program and abstracts of the 7th Conference
HIV/AIDS • CID 2005:41 (15 September) • 881
Downloaded from http://cid.oxfordjournals.org/ by guest on February 2, 2016
of active drug abuse did not differ significantly between groups, these findings do not explain the differences in adherence scores found in the DMAS users. An examination of the relationship between adherence and depressive symptoms failed to demonstrate a modifying influence of depressive symptoms in either treatment group. Regimen complexity and the prevalence of subjects with advanced versus nonadvanced HIV disease were also balanced between groups. We did not measure self-efficacy (i.e., one’s belief about their capability to adhere to HAART), motivation, drug-induced toxicity, and physician-patient relationship. Overall, patients in the intervention group had significantly better virological control at week 12, whereas changes in CD4+ cell count did not differ significantly between groups. Although the inclusion of both HAART-naive and HAART-experienced subjects may have increased the generalizability of these findings, existing mutations in HIV in HAART-experienced subjects could have attenuated the effects of the intervention and influenced the number of subjects achieving an undetectable plasma HIV RNA load and immunological recovery. Also, because patients were not randomized to receive specific antiretroviral regimens, the variety of HAART regimens with differing potency could also have affected the rates of viral suppression and immunological recovery. We did not find a significant difference in virological and immunological response between memory-intact and memoryimpaired participants. Also, we saw no direct correlation between baseline NP performance and adherences scores at 24 weeks. However, our study did not have sufficient power to quantify changes in biological markers stratified on the degree of memory deficit or to detect differences in NP performances. Our study had some limitations. First, the benefits of the memory-prompting device tested in this population were revealed in a post hoc analysis. Thus, it will be important to replicate these findings. Second, we did not have sufficient power to stratify subjects according to the degree of their memory deficit to further quantify the effects of the DMAS on adherence to HAART. Also, although the DMAS device was helpful for subjects with memory impairment, the difference in its usefulness between memory-impaired and memory-intact patients could not be fully tested. Third, substance abuse was relatively uncommon in our study, and therefore the results might not be generalizable to populations in which drug abuse is more prevalent. Finally, this study was not designed to examine the effects of non–memory-related cognitive functions on adherence. Thus, the positive impact of the DMAS device may be on other domains, in addition to memory. Future studies should consider these issues. As the longevity of people infected with HIV increases, the prevalence of milder forms of cognitive impairment may actually be increasing [35]. People in this group and their HIV
9.
10.
11. 12.
13.
14.
16. 17. 18.
19.
20.
21.
22.
882 • CID 2005:41 (15 September) • HIV/AIDS
23.
24. 25.
26. 27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
Opportunistic Infections (Seattle). Alexandria, VA: Foundation for Retrovirology and Human Health, 2002. Sanchez M, Sabundayo B, Langan S, et al. Predictive value of electronic monitoring of HAART adherence in patients with acute and early HIV infection [abstract 544-T]. In: Program and abstracts of the 9th Conference on Retroviruses and Opportunistic Infections (Seattle). Alexandria, VA: Foundation for Retrovirology and Human Health, 2002. Tuldra` A, Wu WA. Interventions to improve adherence to antiretroviral therapy. J Acquir Immune Defic Syndr 2002; 31(Suppl 3):S154–7. Bangsberg DR, Bronstone A, Chesney MA, Hecht FM. Computerassisted self-interviewing (CASI) to improve provider assessment of adherence in routine clinical practice. J Acquir Immune Defic Syndr 2002; 31(Suppl 3):S107–11. McDonald HP, Garg AX, Haynes RB. Interventions to enhance patient adherence to medication prescriptions. JAMA 2002; 288:2868–79. Paterson DL, Swindells S, Mohr J, et al. Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med 2000; 133:21–30. Aloisi MS, Arici C, Balzano R, et al. Behavioral correlates of adherence to antiretroviral therapy. Italian Cohort Naive Antiretrovirals Behavioral Epidemiology Study Group. J Acquir Immune Defic Syndr 2002; 31(Suppl 3):S145–8. Trotta MP, Ammassari A, Melzi S, et al. Treatment-related factors and highly antiretroviral therapy adherence. AdICoNA Study Group. J Acquir Immune Defic Syndr 2002; 31(Suppl 3):S128–31. Bouhnik AD, Chesney M, Carrieri P, et al. Nonadherence among HIVinfected injecting drug users: the impact of social instability. MANIF 2000 Study Group. J Acquir Immune Defic Syndr 2002; 31(Suppl 3): S149–53. Gifford AL, Groessl EJ. Chronic disease self-management and adherence to HIV medications. J Acquir Immune Defic Syndr 2002; 31(Suppl 3):S163–6. Guarinieri M. Highly active antiretroviral therapy adherence: the patient’s point of view. ICoNA Community Advisory Board. J Acquir Immune Defic Syndr 2002; 31(Suppl 3):S167–9. Murri R, Antinori A, Ammassari A, et al. Physician estimates of adherence and the patient-physician relationship as a setting to improve adherence to antiretroviral therapy. AdICoNA Study Group. J Acquir Immune Defic Syndr 2002; 31(Suppl3):S158–62. Starace F, Ammassari A, Trotta MP, et al. Depression is a risk factor for suboptimal adherence to highly active antiretroviral therapy. NeuroICoNA study Group. J Acquir Immune Defic Syndr 2002; 31: S136–9. Deutsch R, Ellis RJ, McCutchan JA, Marcotte TD, Letendre S, Grant I. AIDS-associated mild neurocognitive impairment is delayed in the era of highly active antiretroviral therapy. HNRC Group. AIDS 2001;15: 1898–9. Hinkin CH, Van Gorp WG, Satz P, et al. Actual versus self-reported cognitive dysfunction in HIV-1 infection: memory-metamemory dissociation. J Clin Exp Neuropsychol 1996; 18:431–43.
Downloaded from http://cid.oxfordjournals.org/ by guest on February 2, 2016
15.
on Retrovirus and Opportunistic Infections (San Francisco). Alexandria, VA: Foundation for Retrovirology and Human Health, 2000. Paterson DL, Potoski B, Capitano B. Measurement of adherence to antiretroviral medications. J Acquir Immune Defic Syndr 2002; 31(Suppl 3):S103–6. Chesney MA, Ickovicks JR, Chambers DB, et al. Self-reported adherence to antiretroviral medications among participants in HIV clinical trials: the AACTG adherence instruments. Patient Care Committee & Adherence Working Group of the Outcomes Committee of the Adult AIDS Clinical Trial Group (AACTG). AIDS Care 2000; 12:255–66. Lezak M. Neuropsychological assessment. 2nd ed. New York: Oxford University Press, 1983. Janssen RS, Cornblath DR, Epstein LG, McArthur J, Price RW. Human immunodeficiency virus (HIV) infection and the nervous system: from the American Academy of Neurology AIDS Task Force. Neurology 1989; 39:119–22. Brandt J, Benedict HRB. Hopkins Verbal Learning test—revised professional manual. Odessa, FL: Psychological Assessment Resources, 2001. Radloff LS, Locke BZ. The community mental health assessment survey and the CES-D scale. In: Weissman MM, Myers JK, Ross CG, eds. Community surveys of psychiatric disorders. New Brunswick, NJ: Rutgers University Press, 1985:177–89. Piantadosi S. Clinical trials: a methodological perspective. New York: John Wiley & Sons, 1997. Shapiro SS, Wilk MB. An analysis of variance test for normality (complete samples). Biometrika 1965; 52:591–611. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcome. Biometrics 1986; 42:121–30. Roter DL, Hall JA, Merisca R, Nordstrom B, Cretin D, Svarstad B. Effectiveness of interventions to improve patient compliance: a metaanalysis. Med Care 1998; 36:1138–61. Mannheimer S, Hirsch Y, el-Sadr W. The impact of the ALR alarm device on antiretroviral therapy adherence among HIV-infected outpatients in Harlem [abstract 32325]. In: Program and abstracts of the XIIth World AIDS Conference (Geneva, Switzerland). Stockholm: International AIDS Society, 1998. Safren SA, Hendriksen ES, Desousa, N, Boswell SL, Mayer KH. Use of an on-line pager system to increase adherence to antiretroviral medications. AIDS Care 2003; 15:787–3. Deschamps A, De Saar V, De Graeve V, et al. Diagnostic values of measurement methods of adherence to HAART in HIV-infected persons [abstract ThPeB2136]. In: Program and abstracts of the XIVth World AIDS Conference (Barcelona). Stockholm: International AIDS Society, 2002. Collier AC, Ribaudo H, Feinberg J, et al. Randomized study of telephone calls to improve adherence to antiretroviral therapy [abstract 540-T]. Adult AIDS Clinical Trials Group (ACTG) Protocol 746 Team. In: Program and abstracts of the 9th Conference on Retroviruses and