The introduction of combination antiretroviral therapy (ART) produced a marked decline in AIDS-associated opportunistic illnesses, incidence of HIV-associated dementia, and mortality.1–3 However, lifelong ART will likely be needed. Although more potent agents, simpler regimens, and combination formulations have improved ART, adherence remains important.4,5 Suboptimal adherence is associated with incomplete HIV suppression, drug resistance, and treatment failure.6–8
Many factors contribute to ART nonadherence, including regimen complexity, pill burden, side effects, untreated depression, and substance abuse.4,9–12 Forgetfulness is the most frequently cited cause of suboptimal adherence, underscoring the potential impact of HIV-associated cognitive impairment on ART-taking behavior.4,13,14
Incidence of HIV-associated dementia has declined, but prevalence of less severe forms of HIV-associated neurocognitive disorder (HAND) is reported to be 21%–50%.15,16 HAND can affect multiple neuropsychological (NP) domains, including psychomotor skills, executive functioning, attention, information processing speed, and memory15,17,18 and may decrease adherence (reviewed in 19). Hinkin et al20 first described a direct association between global NP impairment and ART nonadherence. Subsequent studies have confirmed this observation, which seems to be driven by deficits in executive functions, attention/working memory, and prospective memory.20–23 Risk of nonadherence associated with NP impairment is independent of other risk factors, including demographics, psychiatric illness, and substance abuse.22 Importantly, NP impairment and nonadherence may be linked by a vicious cycle in which initial nonadherence results in ongoing HIV replication, which exacerbates NP impairment and results in worse adherence.24 Studies evaluating NP impairment and adherence have rarely used biologic end points like HIV RNA(18, reviewed in 19). Despite the importance of lifelong adherence, few studies have reported >1 month of follow-up. The importance of extended follow-up is highlighted by one study, where decline in global NP function over 6 months was associated with decreased ART adherence.25 We postulated that a longitudinal study that assessed performance on a comprehensive NP test battery, psychiatric illness, and substance abuse could better assess the impact of cognitive impairment on adherence and virologic suppression.
There is no universal standard for assessing adherence.26 Medication self-report and pill counts are convenient and low-cost but may overestimate adherence. NP impairment may affect accuracy of self-reported adherence. A recent review identified only 11 publications that used standardized NP tests and a direct measure of ART adherence.19 Medication Events Monitoring System (MEMS) caps provide detailed information about medication use but are costly and complicated. Using pharmacy refill records is relatively inexpensive and straightforward; their effectiveness is based on consistent use of the same pharmacy and the assumption that prescription filling and adherence are correlated.27,28 Pharmacy refill records have predicted plasma HIV RNA and immunologic responses in HIV-infected patients.28–30
We conducted a longitudinal analysis of research participants with well-characterized cognitive function at 2 academic HIV clinics, using pharmacy refill records as the primary measure of adherence. We hypothesized that NP deficits and mood and substance use would predict subsequent ART adherence and that adherence would be associated with virologic responses in plasma and cerebrospinal fluid (CSF).
CHARTER (CNS HIV Anti-Retroviral Therapy Effects Research) is a prospective observational study conducted at 6 US academic medical centers that enrolled 1597 participants from 2003 to 2007 and followed 600 participants longitudinally.15
Two CHARTER sites [Johns Hopkins University (JHU) and University of Washington (UW)] with access to pharmacy refill records conducted this substudy. Eligible participants had CHARTER baseline and 6 month visits that were no more than 9 months apart. Each site's Institutional Review Board approved this research; all participants gave written informed consent.
Neuromedical and Laboratory Assessments
Participants underwent comprehensive standardized evaluations of neuromedical, NP, and psychiatric status; substance use; and adherence at each visit.15 HIV infection was diagnosed by enzyme-linked immunosorbent assay with Western blot confirmation. Plasma and CSF HIV RNA were measured centrally by reverse transcriptase–polymerase chain reaction (Roche Amplicor, version 1.5, lower limit of quantification, 50 copies/mL, Roche Diagnostic Systems, Indianapolis, IN). Estimated duration of HIV infection and nadir CD4+ T-count were obtained by self-report.
Participants completed a comprehensive NP test battery assessing 7 cognitive domains: verbal fluency, executive functioning, speed of information processing, learning, memory, attention/working memory, and motor performance (see Appendix, Supplemental Digital Content, http://links.lww.com/QAI/A376).15 Raw scores were converted to demographically corrected standard scores (T-scores) using best available normative standards that adjust for age, education, gender, and ethnicity. We then applied an automated approach that yields a continuous estimate of functioning [Global Deficit Score (GDS)]. GDS emphasizes number and severity of deficits, giving less weight to average and above average performances.31 GDS of ≥0.5 classifies individuals as NP impaired. This approach was also used to generate deficit scores for each cognitive domain. An objective clinical rating algorithm was used to diagnose overall NP impairment (global impairment).32 Comorbidities that could contribute to NP impairment were classified as incidental (noncontributory), contributing, or confounding, as previously described.15,33 We used Hollingshead Two-Factor Index to assess social status.34 The reading subset score of the Wide Range Achievement Test, 3rd edition (WRAT), was used as an estimate of premorbid intellectual functioning.
Psychiatric and Substance Use Assessments
The Composite International Diagnostic Interview35 examined current (within the last 30 days) and lifetime (>30 days ago) DSM-IV diagnoses of major depression and substance abuse and dependence for alcohol and illicit substances. Depressive symptoms were collected using the Beck Depression Inventory-II.36 A semistructured interview examined quantity and frequency of alcohol and recreational drug use.
Investigators (A.S.A.A. and A.C.C.) reviewed ART regimens for irregularities in doses and regimens. Four participants who were prescribed ART for ≤30 days and one participant prescribed one agent not intended as monotherapy were excluded. For each regimen, a sentinel drug was identified according to an algorithm (Fig. 1). Sentinel drugs were selected based on their central nervous system (CNS) penetration–effectiveness (CPE) scores37 with the assumption that drugs with higher CPE would have greater suppressive impact on CSF (brain) viral levels and potentially improve cognitive status and adherence. If a regimen contained 2 drugs with identical CPE scores, the sentinel drug was selected by higher dosing frequency/pill burden and drug class. We assumed the sentinel drug was started the day after it was dispensed; if pills remained from a previous prescription, the new prescription was assumed to start after the current supply ran out. An injection or liquid dose was considered one “pill.” Doses less frequent than daily were calculated as fractions. Sentinel drugs prescribed as part of coformulated drugs were analyzed as individual agents. Pharmacy records were used to estimate ART adherence from 60 days before the baseline visit to the first follow-up visit (the 6-month visit). Adherence was computed as number of days for which a participant had sufficient supply to take prescribed doses, divided by total number of days for which the sentinel drug was prescribed. Nonsentinel drugs were not analyzed.
Self-reported adherence was determined by a 4-day report (proportion of prescribed doses taken) of the AIDS Clinical Trials Group Adherence Questionnaire.38
Transformations and nonparametric techniques were performed as needed. Parametric assumptions were verified. To stabilize variance, an arcsin transformation was applied to the pharmacy records-derived adherence measure, resulting in estimates weighted slightly more for longer exposure to ART (eg, 100% adherence over 8 months counted more than 100% adherence for 2 months).39 Two-sided testing, no multiple comparison adjustments, and a 5% significance level were used unless otherwise indicated. Site differences in baseline characteristics were tested using t-tests, Wilcoxon rank sum, χ2, or Fisher exact tests, depending on parametric test assumption validity. Baseline variables were screened individually for their association with adherence using linear regression, first unadjusted, then including terms for site and the variable's interaction with site. Terms with P < 0.10 were retained as candidates for multivariable modeling. Main effects were always included with interactions.
Multivariable linear regression models, all including a term for site, were constructed after omitting all but one of the mutually correlated candidate variables based on removal of the minimal number of variables while retaining the least mutually correlated variables according to P value (WRAT, age, and duration of antiretroviral drugs removed). Since identifying the relationship of NP deficits to ART adherence was a primary aim, GDS and its interaction with site, if significant, were included in all models. Subsequent models included all terms from the previous model except for the one with the largest P value ≥ 0.05; this process was repeated until only significant variables remained. After this model was produced, GDS was replaced with each NP domain deficit score separately and that domain was tested for its association with adherence; nonsignificant interactions were excluded from the final models. To assess relationships between adherence and changes in log-transformed viral load in subjects who had detectable baseline HIV RNA, linear regression was used, after examining scatter plots of baseline versus change for evidence of regression to the mean.40 If warranted, adjustment for baseline level and its interactions with adherence were included in the regression model. The associations between adherence and HIV RNA changes are described separately for subjects with detectable and undetectable baseline viral loads with median [interquartile range (IQR)]. For descriptive purposes, we dichotomized self-reported adherence at 95%, as has been done by others.6 For inferential analysis of self-reported adherence, we arcsine-transformed adherence and used the lowest self-report at either visit and analysed the data the same way as for pharmacy refill data on subjects with detectable baseline viral loads, except that adherence was dichotomized (100% vs <100%) and plasma HIV RNA change was compared between the 2 adherence levels using a Wilcoxon rank sum test.
Characteristics of the Study Population
Most participants were African American (56%) and male (79%) (Table 1). Median estimated minimum duration of HIV infection was 10.6 years. Median nadir and current CD4+ T-counts were 72 and 383 cells/mm3, respectively. At baseline, participants reported taking their current ART regimen for median of 8 months. Lifetime major depression (49%) and substance abuse/dependence (83%) were common. Half of participants were globally NP impaired. JHU participants were more likely to be African American, slightly older, and less likely to be male than UW subjects. The proportion of participants with baseline plasma HIV RNA <50 copies/mL was higher at JHU than at UW. The groups were similar in most other characteristics, but mean WRAT scores, used to estimate premorbid intellectual functioning, were higher at UW.
Antiretroviral Adherence by Pharmacy Refill Records
The mean (SD) adherence rate to the sentinel drug was 86.4% (18.5%); rates were similar at JHU [85.7% (16.9)] and UW [86.9% (19.6)].
Factors Associated With ART Adherence
Table 2 shows baseline characteristics (both unadjusted and controlling for site and interactions with site) and their association with adherence. Of note, current ART regimen duration and depression or substance use diagnoses were not associated with adherence. Table 3 shows the final multivariable model that indicates male sex, longer estimated duration of HIV, and better baseline global NP performance were significantly associated with better ART adherence. Relationships between GDS and adherence differed by site; at JHU but not UW, worse global neurocognitive performance (higher GDS) was associated with worse adherence.
Relationship Between Neurocognitive Domains and Adherence
After replacing GDS with each of the 7 NP domains, only one, working memory, showed a significant effect on adherence (P = 0.026, Table 3). Worse working memory was associated with poorer adherence. Duration of HIV infection was still associated with adherence in the model that included working memory, but sex was no longer a significant factor and was removed from the final model.
Relationship Between Adherence by Pharmacy Refill Data and Changes in Plasma and CSF HIV RNA
The relationships between adherence and changes in plasma and in CSF HIV RNA were nonlinear. An examination of scatter plots of baseline HIV RNA levels versus change in RNA showed evidence of regression to the mean for viral load in plasma but not in CSF.
Only 12 of 74 subjects with RNA measurements in CSF had detectable baseline levels. Among these participants, better adherence was associated with greater declines in CSF HIV RNA, especially in those with higher baseline viral load (adherence P = 0.058, RNA P = 0.013, interaction P = 0.08, Table 4). For example, mean [standard error (SE)] reduction in CSF HIV RNA (log10 copies/mL) was estimated to be −0.4 (0.7) for baseline RNA of 2 log10 and −1.1 (0.9) in subjects with baseline RNA of 4 log10 and approximately 75% adherence, whereas the mean (SE) changes estimated were −0.2 (0.8) and −2.4 (0.8) log10, respectively, with 100% adherence. Among the 58 subjects for which baseline CSF RNA results were available and undetectable, 52 (90%) had viral loads that remained undetectable at month 6 and had median (IQR) adherence of 96% (85%–100%); the remaining 6, with median (IQR) adherence of 93% (58%–100%), had increases of <1 log10 (N = 4), and between 1 and 2 log10 (N = 2) at month 6.
Evidence of regression to the mean precluded any adjustment for baseline plasma RNA levels.41 For the 46 subjects with detectable baseline plasma HIV RNA, adherence alone, estimated by pharmacy refill data was not associated with plasma viral load change (P = 0.719); 17 (37%) of these subjects had undetectable plasma viral load at the 6-month visit and median (IQR) adherence of 95% (67%–99%); the remaining 29 who had detectable plasma HIV RNA at 6 months had median (IQR) of 92% (78%–100%) adherence. Among the 34 subjects with undetectable plasma RNA at baseline, 29 (85%) had viral loads that remained undetectable at 6 months and had median (IQR) adherence of 95% (85%–99%); the remaining 5 had plasma RNA increases ranging from 0.05 and 3.2 log10 and median (IQR) adherence of 98% (52%–100%).
Antiretroviral Adherence by Self-Report
At baseline, the mean (SD) self-reported 4-day adherence for the 74 subjects for whom these data were available was 98.2% (5.7%), with no site differences. When adherence was defined by this metric as ≥95%, 65% (88%) and 69% (94.5%) of subjects reported being adherent at baseline and the 6-month visits, respectively, with no site differences.
Relationship Between Self-Reported Adherence and Changes in Plasma and CSF HIV RNA
With the lowest self-reported adherence at either the baseline or 6-month visit dichotomized as 100% (n = 39) vs <100% (N = 7) for the 46 subjects with detectable baseline plasma HIV RNA, there was no difference in plasma HIV RNA change between the 2 adherence groups (P = 0.748).
Among the 12 subjects with detectable CSF HIV at baseline, there was a significant effect of self-reported adherence on change in CSF HIV RNA (P = 0.0055) and significant effects of baseline CSF levels and the interaction of these 2 variables (P < 0.0001 and P = 0.0024, respectively). Those with higher baseline CSF HIV RNA and higher adherence had greater decreases estimated in their CSF HIV RNA over 6 months than those with lower baseline CSF viral load regardless of adherence.
The study used a novel metric to assess adherence and evaluated longitudinal plasma and CSF RNA changes. Data from these 2 US sites in the CHARTER study suggest that HIV-infected participants with greater overall NP deficits and specific deficits in working memory have lower ART adherence, as defined by our pharmacy refill record review. This adherence measure was focused on the drug with the maximal CNS ART penetration–effectiveness score.37 Notably, better adherence based on this metric was associated with decreasing levels of CSF HIV RNA over a 6-month time period.
Global NP impairment has been independently associated with ART nonadherence in studies incorporating multivariable analyses to account for other confounding factors. After controlling for age, current psychiatric disorders, neurologic conditions, and regimen complexity, global NP impairment correlated with ART nonadherence measured by MEMS caps over a 4-week period.22 In a different study, decreased NP function in adults older than 50 years was the sole predictor of ART nonadherence.21 Many studies relating cognition and adherence have been cross-sectional and unable to address causal associations. Does NP impairment result in suboptimal adherence, is poor adherence and consequent ongoing HIV replication causing NP impairment, or are both pathways contributing to a vicious cycle?24 Relatively few longitudinal studies have addressed this issue. Older age is often associated with better adherence. However, Hinkin et al41 reported that 83% of subjects older than 50 years with poor adherence were cognitively impaired. Over 4 weeks, nonadherence (by MEMS caps) was associated with poor baseline global NP functioning in a subset of 23 subjects with consistently poor adherence.42 Retrospective analyses of HIV-infected adults in Los Angeles showed that those with decline in global NP function over 6 months had a greater drop in ART adherence than those with stable function.25 Controlling for baseline factors, our findings suggest that global cognitive deficits were associated with poor ART adherence over the subsequent 6 months.
Some prospective studies have reported that specific cognitive domains are associated with ART adherence. Levine et al42 noted working memory and attention deficits in 27 subjects with poor weekend adherence in a 4-week study. A study of New York City men who have sex with men found worse performance on Trail Making Test Part B (which has a working memory component as well as flexibility of thinking and speed of information processing) was associated with worse adherence over 10 months.43 Prospective memory, the ability to “remember to remember” something in the future correlated with adherence in a clinical trial.44 Deficits in time-based prospective memory were independently predictive of poor ART adherence over 1 month.22 Thus, our finding that worse working memory, in combination with other factors, was associated with lower subsequent ART adherence is generally consistent with these reports and suggests that our results may be generalized to other US adults. Successful antiretroviral adherence is a complex task, requiring intention to take medication at specific time(s) per day, retaining this plan during daily activities, remembering dosing time and any associated instructions (eg, take with food or on an empty stomach), and executing the plan by taking the medication.22 Since others have described that working memory deficits can negatively affect higher level cognition, such as decision making45 and everyday functioning,46 it is not surprising that such deficits can negatively impact ART adherence.
In our study, better adherence was seen in men and those with longer estimated duration of HIV infection. Data on the relationship between sex and ART adherence have been inconsistent, although a recent review suggested that women have worse adherence than men, similar to our findings.47 In that report, depression and lack of support systems were associated with poorer adherence in women. Although some reports link depression to nonadherence, this relationship was not found in other studies or in ours.21,47 We did not collect information on support systems, although our subjects may have benefited from contact with CHARTER staff. In our multivariable model of specific NP domains, working memory moderated the significant association between sex and adherence (Table 3). Our observation that longer duration of estimated HIV infection was associated with better adherence may reflect survivorship bias.48 It is also possible that this association is confounded by age (ie, better adherence in older subjects, after correcting for NP function).
We found that worse NP functioning led to poorer adherence in JHU patients but had a negligible effect in UW participants. Although our multivariable analyses contained likely confounding factors, these analyses may not have adequately adjusted for all potential differences between sites. For example, UW subjects were more likely to be male, younger, and less likely to be African American. Although older patients usually have better adherence, age-related cognitive decline (ie, normal cognitive aging) might have exacerbated the impact of disease-related NP deficits on adherence at JHU.21 Higher level of premorbid intellectual function at UW, as indicated by higher WRAT scores, may also have moderated the impact of NP impairment. Finally, possible differences in adherence support services at the sites might also explain the greater impact of NP impairment on adherence at JHU.
We evaluated pharmacy refill records as an alternative monitoring tool because of concerns about the accuracy of self-reported adherence in NP-impaired patients. Like our pharmacy refill-based adherence data, self-reported adherence was significantly correlated with virologic suppression in CSF, even among the participants who had detectable CSF HIV RNA at baseline. Although pharmacy refill records have previously been correlated with plasma viral load suppression, we were not able to confirm such an association in our study. However, the 4-day self-report has recognized limitations, including overestimating adherence.5 We did a priori identify our pharmacy refill metric as the primary adherence metric for this study, and our refill records-based adherence metric was associated with reduction of CSF HIV RNA values in all subjects studied. Because CPE score was the primary criterion for identifying a sentinel drug in our study, the relationship between better adherence and reduction in CSF HIV suggests that the protection afforded by better CNS-penetrating drugs may have preserved cognition and consequently adherence. This possibility is supported by Cysique et al49 who reported that CPE score was an independent predictor of NP improvement in study participants with HAND who started ART.
Our results are limited by a relatively small sample size. In addition, these findings are most generalizable to patients obtaining medications at a single pharmacy. Filling prescriptions does not guarantee that medications are taken. Finally, factors such as stockpiling doses or obtaining medications from other pharmacies could have affected our refill metric. In spite of these limitations, we believe that this is the first prospective study with virologic outcomes using pharmacy refill records to quantify adherence to ART in HIV-infected patients who have undergone rigorous NP, psychiatric, and substance-use assessments.
In conclusion, poor adherence to ART in our study population was correlated with working memory and global NP deficits. Careful assessment of working memory and cognition before ART initiation and in patients with suboptimal virologic responses to ART may help identify persons in need of adherence support. Working memory training for stimulant addicts and strategies for rehabilitation of working memory deficits in persons with HIV infection have been described.50,51 Such strategies should be further investigated in persons with HAND to determine whether they can improve adherence and optimize benefits of ART.
The authors thank the study participants for their contribution. The authors would also like to thank Stephen Van Rompaey, PhD, Bradley Kosel, PharmD, and Mari Kitahata, MD, MPH for their assistance with accessing and cross-checking the UW pharmacy records; Deborah Lazzaretto and Chris Ake, formerly at the CHARTER Coordinating Center, who made contributions to the early stages of this project; and Steven P. Woods, PsyD, for helpful comments on the manuscript. The CNS HIV Anti-Retroviral Therapy Effects Research (CHARTER) group is affiliated with the Johns Hopkins University, Mount Sinai School of Medicine, University of California, San Diego, University of Texas, Galveston, University of Washington, Seattle, Washington University, St. Louis and is headquartered at the University of California, San Diego (UCSD) and includes Igor Grant, MD (UCSD, Director); Ronald J. Ellis, MD, PhD (UCSD, Codirector); Scott L. Letendre, MD (UCSD, Codirector); Ian Abramson, PhD (UCSD, Coinvestigator); Muhammad Al-Lozi, MD (Washington University, Coinvestigator); J. Hampton Atkinson, MD(UCSD, Coinvestigator); Edmund Capparelli, PharmD (UCSD, Coinvestigator); David Clifford, MD [Washington University, Site Principal Investigator (PI)], Ann C. Collier, MD (University of Washington, Site Co-PI), Christine Fennema-Notestine, PhD (UCSD, Core PI), Anthony C. Gamst, PhD (UCSD, Core PI), Benjamin Gelman, MD, PhD (University of Texas, Site PI), Robert K. Heaton, PhD (UCSD), Thomas D. Marcotte, PhD (UCSD, Core PI), Christina Marra, MD (University of Washington, Site Co-PI), J. Allen McCutchan, MD (UCSD, Site PI), Justin C. McArthur, MD (Johns Hopkins, Site PI), Susan Morgello, MD (Mount Sinai, Site Co-PI), David Simpson, MD (Mount Sinai, Site Co-PI), Davey M. Smith, MD (UCSD, Core PI), Michael J. Taylor, PhD (UCSD, Core Coinvestigator), Rebecca Theilmann, PhD (UCSD, Imaging Physicist), Florin Vaida, PhD (UCSD, Coinvestigator), Steven Paul Woods, PsyD (UCSD, Coinvestigator); Study Coordinators: Terry J. Alexander, RN (UCSD, Neuromedical Coordinator), Clint Cushman (UCSD, Data Manager), Matthew Dawson (UCSD, Neurobehavioral Coordinator), Donald Franklin, Jr. (UCSD, Center Manager), Eleanor Head, RN, BSN (University of Texas, Site Coordinator), Trudy Jones, MN, ARNP (University of Washington, Site Coordinator), Jennifer Marquie-Beck, MPH (UCSD, Recruitment Coordinator), Letty Mintz, NP (Mount Sinai, Site Coordinator), Vincent Rogalski, CCRP (Johns Hopkins, Site Coordinator), Mengesha Teshome, MD (Washington University, Site Coordinator), Will Toperoff, BS, ND (UCSD, Site Coordinator).
1. Hammer SM, Squires KE, Hughes MD, et al.. A controlled trial of low nucleoside analogues plus indinavir in persons with human immunodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less. AIDS Clinical Trials Group 320 Study Team. N Engl J Med. 1997;337:725–733.
2. Sacktor N, Lyles RH, Skolasky R, et al.. HIV-associated neurologic disease incidence changes: multicenter AIDS Cohort Study, 1990-1998. Neurology. 2001;56:257–260.
3. Sterne JA, May M, Costagliola D, et al.. Timing of initiation of antiretroviral therapy in AIDS-free HIV-1-infected patients: a collaborative analysis of 18 HIV cohort studies. Lancet. 2009;373:1352–1363.
4. Bartlett JA. Addressing the challenges of adherence. J Acquir Immune Defic Syndr. 2002;29(suppl 1):2–10.
5. Chesney MA. The elusive gold standard. Future perspectives for HIV adherence assessment and intervention. J. Acquir Immune Defic Syndr. 2006;43(suppl 1):149–155.
6. 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.
7. Raffa JD, Tossonian HK, Grebely J, et al.. Intermediate highly active antiretroviral therapy adherence thresholds and empirical models for the development of drug resistance mutations. J Acquir Immune Defic Syndr. 2008;47:397–399.
8. Bangsberg DR, Perry S, Charlebois ED, et al.. Non-adherence to highly active antiretroviral therapy predicts progression to AIDS. AIDS. 2001;15:1181–1183.
9. Bartlett JA, DeMasi R, Quinn J, et al.. Overview of the effectiveness of triple combination therapy in antiretroviral-naïve HIV-1 infected adults. AIDS. 2001;15:1369–1377.
10. Ammassari A, Murri R, Pezzotti P, et al.. Self-reported symptoms and medication side effects influence adherence to highly active antiretroviral therapy in persons with HIV infection. J Acquir Immune Defic Syndr. 2001;28:445–449.
11. Wagner GJ, Goggin K, Remien RH, et al.. A closer look at depression and its relationship to HIV antiretroviral adherence. Ann Behav Med. 2011;42:352–360.
12. Lucas GM. Substance abuse, adherence with antiretroviral therapy, and clinical outcomes among HIV-infected individuals. Life Sci. 2011;88:948–952.
13. Barfod TS, SØrensen HT, Nielsen H, et al.. “Simply forgot” is the most frequently stated reason for missed doses of HAART irrespective of degree of adherence. HIV Med. 2006;7:285–290.
14. Senkomago V, Guwatudde D, Breda M, et al.. Barriers to antiretroviral adherence in HIV-positive patients receiving free medication in Kayunga, Uganda. AIDS Care. 2011;23:1246–1253.
15. Heaton RK, Clifford DB, Franklin DR Jr, et al.. HIV-associated neurocognitive disorders persist in the era of potent antiretroviral therapy: CHARTER Study. Neurology. 2010;75:2087–2096.
16. Robertson DR, Smurzynski M, Parsons TD, et al.. The prevalence and incidence of neurocognitive impairment in the HAART era. AIDS. 2007;21:1915–1921.
17. Grant I, Atkinson JH, Hesselink JR, et al.. Evidence for early central nervous system involvement in the acquired immunodeficiency syndrome (AIDS) and other human immunodeficiency virus (HIV) infections. Studies with neuropsychologic testing and magnetic resonance imaging. Ann Intern Med. 1987;107:828–836.
18. Andrade AS, McGruder HF, Wu AW, et al.. A programmable prompting device improves adherence to highly active antiretroviral therapy in HIV-infected subjects with memory impairment. Clin Infect Dis. 2005;41:875–882.
19. Lovejoy TI, Suhr JA. The relationship between neuropsychological functioning and HAART adherence in HIV-positive adults: a systematic review. J Behav Med. 2009;32:389–405.
20. Hinkin CH, Castellon SA, Durvasula RS, et al.. Medication adherence among HIV+ adults. Neurology. 2002;59:1944–1950.
21. Barclay TR, Hinkin CH, Castellon SA, et al.. Age-associated predictors of medication adherence in HIV-positive adults: health beliefs, self-efficacy, and neurocognitive status. Health Psychol. 2007;26:40–49.
22. Woods SP, Dawson MS, Weber E, et al.. Timing is everything: antiretroviral nonadherence is associated with impairment in time-based prospective memory. J Int Neuropsychol Soc. 2009;15:42–52.
23. Zogg JB, Woods SP, Sauceda JA, et al.. The role of prospective memory in medication adherence: a review of an emerging literature. J Behav Med. 2012;35:47–62.
24. Ettenhofer ML, Foley J, Castellon SA, et al.. Reciprocal prediction of medication adherence and neurocognition in HIV/AIDS. Neurology. 2010;74:1217–1222.
25. Becker BW, Thames AD, Woo E, et al.. Longitudinal change in cognitive function and medication adherence in HIV-infected adults. AIDS Behav. 2011;15:1888–1894.
26. Simoni JM, Kurth AE, Pearson CR, et al.. Self-report measures of antiretroviral therapy adherence: a review with recommendations for HIV research and clinical management. AIDS Behav. 2006;10:227–245.
27. Kitahata MM, Reed SD, Dillingham PW, et al.. Pharmacy-based assessment of adherence to HAART predicts virologic and immunologic treatment response and clinical progression to AIDS and death. Int J STD AIDS. 2004;15:803–810.
28. Grossberg R, Gross R. Use of pharmacy refill data as a measure of antiretroviral adherence. Curr HIV/AIDS Rep. 2007;4:187–191.
29. Bisson GP, Gross R, Bellamy S, et al.. Pharmacy refill adherence compared with CD4 count changes for monitoring HIV-infected adults on antiretroviral therapy. PLoS Med. 2008;5:e109.
30. El-Khatib A, Katzenstein D, Marrone G, et al.. Adherence to drug-refill is a useful early warning indicator of virologic and immunologic failure among HIV patients on first-line ART in South Africa. PLoS One. 2011;6:e17518.
31. Carey CL, Woods SP, Gonzalez R, et al.. Predictive validity of global deficit scores in detecting neuropsychological impairment in HIV infection. J Clin Exp Neuropsychol. 2004;26:307–319.
32. Heaton RK, Grant I, Butters N, et al.. The HNRC 500-Neuropsychology of HIV infection at different stages. J Internat Neuropsychol Soc. 1995;1:231–251.
33. Antinori A, Arendt G, Becker JT, et al.. Updated research nosology for HIV-associated neurocognitive disorders. Neurology. 2007;69:1789–1799.
34. Hollingshead AB, Redlich FC. Social Class and Mental Illness: A Community Study. New York, NY: John Wiley; 1958.
35. World Health Organization. Composite International Diagnostic Interview, Version 2.1. Geneva, Switzerland: WHO; 1997.
36. Beck AT, Steer RA, Brown GK. Beck Depression Inventory–Second Edition Manual. San Antonio, TX: The Psychological Corporation; 1996.
37. Letendre S, Marquie-Beck J, Capparelli E, et al.. Validation of the CNS penetration-effectiveness rank for quantifying antiretroviral penetration into the central nervous system. Arch Neurol. 2008;65:65–70.
38. 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.
39. Zar JH. Biostatistical Analysis. 3rd ed. Upper Saddle River, NJ: Prentice Hall; 1996:282.
40. Barnett AG, van der Pols JC, Dobson AJ. Regression to the mean: what it is and how to deal with it. Int J Epidemiol. 2005;34:215–220.
41. Hinkin CH, Hardy DJ, Mason KI, et al.. Medication adherence in HIV-infected adults: effect of patient age, cognitive status, and substance abuse. AIDS. 2004;18(suppl 1):S19–S25.
42. Levine AJ, Hinkin HC, Castellon SA, et al.. Variations in patterns of highly active antiretroviral therapy (HAART) adherence. AIDS Behav. 2005;9:355–362.
43. Solomon TM, Halkitis PN. Cognitive executive functioning in relation to HIV medication adherence among gay, bisexual, and other men who have sex with men. AIDS Behav. 2008;12:68–77.
44. Contardo C, Black AC, Beauvais J, et al.. Relationship of prospective memory to neuropsychologic function and antiretroviral adherence. Arch Clin Neuropsychol. 2009;24:547–554.
45. Martin EM, Pitrak DL, Weddington W, et al.. Cognitive impulsivity and HIV serostatus in substance dependent males. J Int Neuropsychol Soc. 2004;10:931–938.
46. Heaton RK, Marcotte TD, Mindt MR, et al.. The impact of HIV-associated neuropsychological impairment on everyday functioning. J Int Neuropsychol Soc. 2004;10:317–331.
47. Puskas CM, Forrest JI, Parashar S, et al.. Women and vulnerability to HAART non- adherence: a literature review of treatment adherence by gender from 2000 to 2011. Curr HIV/AIDS Rep. 2011;8:277–287.
48. Glesby MJ, Hoover DR. Survivor treatment selection bias in observational studies: examples from the AIDS literature. Ann Intern Med. 1996;124:999–1005.
49. Cysique LA, Vaida F, Letendre S, et al.. Dynamics of cognitive change in impaired HIV-positive patients initiating antiretroviral therapy. Neurology. 2009;73:342–348.
50. Bickel WK, Yi R, Landes RD, et al.. Remember the future: working memory training decreases delay discounting among stimulant addicts. Biol Psychol. 2011;69:260–265.
51. Woods SP, Weber E, Cameron MV, et al.. Spontaneous strategy use protects against visual working memory deficits in older adults infected with HIV. Arch Clin Neuropsychol. 2010;25:724–733.