HPTN 052 is a Phase III, randomized multicenter study in 1763 HIV-1 serodiscordant couples in 9 countries where the HIV-1–infected partner was randomized to early or delayed antiretroviral therapy (ART). The trial demonstrated that ART-prevented HIV-1 transmission.1 These results were consistent with observational trials.2–4
However, the ability of ART to suppress viral replication is entirely dependent on adherence to the medications. Low social support, depression, substance use, side effects, lack of counseling, and socioeconomic factors have been predictive of poor adherence to ART in diverse global settings.5–8 Virtually, all earlier studies have examined adherence to ART in individuals who were taking ART because it was indicated for more advanced HIV-1 disease, rather than to prevent transmission to a sexual partner. The present analysis from HPTN 052 among those who were randomized to early ART allowed us to evaluate predictors of adherence to medication in HIV-1–infected study subjects who were offered treatment directed toward HIV prevention.
Details of the HPTN 052 study design can be found both on clinicaltrials.gov NCT00074581 and in the primary outcome articles.1,9 Participants were enrolled from 13 sites in 9 countries (Botswana, Kenya, Malawi, South Africa, Zimbabwe, Brazil, India, Thailand, and United States). The study enrolled HIV-1 serodiscordant couples (1 partner is infected with HIV-1 and the other is not), where the HIV-1–infected partner had a CD4 cell count of 350–550 cells per cubic millimeter. Couples were supposed to have been in a stable relationship for at least 3 months, reported 3 or more episodes of vaginal or anal intercourse during the past 3 months, and willing to disclose their HIV-1 status to their partner. The study was unblinded. The partner infected with HIV-1 was randomized to early or delayed ART. Local institutional review boards or ethics committees approved the study at each site.
For this analysis, data were used only from participants randomized to the early treatment arm. Accordingly, all participants in the following analyses initiated ART on study enrollment, and all data were from visits after the enrollment visit (after ART treatment initiation) through the time when the results of the primary outcome were publically released. The median follow-up for this analysis was 2.1 years.
When the study began in 2005, visits were monthly, and beginning in mid-2008, they transitioned to quarterly. Adherence and psychosocial data were recorded at each visit. Participants were provided regular adherence counseling using an adapted version of the LifeSteps intervention10,11 as a base (see http://www.hptn.org/web%20documents/hptn052/hptn052adherencecounseling.pdf for the initial training material) and regular counseling on HIV risk reduction. The LifeSteps adherence counseling included a medical provider portion, which provided education about ART medications and adherence, and a counselor portion, which provided assistance to participants in devising a plan and a backup plan for potential impediments to adherence (eg, creation of a daily medication schedule, developing reminder strategies, handling slips, discussion of involving social supports), following a cognitive-behavioral/problem-solving approach. See Appendix (Supplemental Digital Content, http://links.lww.com/QAI/A654) for the checklists used by counselors and medical providers. Study drugs included a combination of lamivudine and zidovudine (Combivir), efavirenz, atazanavir, nevirapine, tenofovir, lamivudine, zidovudine, didanosine, stavudine, a combination of lopinavir and ritonavir (Kaletra and Aluvia), ritonavir, and a combination of emtricitabine and tenofovir (Truvada). A prespecified combination of these drugs was provided to participants at each visit. For participants with virologic failure, specified second-line treatment regimens were provided.
At every visit, participants completed an interviewer-administered adherence questionnaire in the local language (translated, translated back, discrepancies reviewed, and corrected), and pill counts were conducted, which yielded a nonadherence self-report score and a nonadherent pill count categorization, respectively. The additional psychosocial variables were collected at every quarterly visit. At all sites, self-report measures were translated and translated back to maximize accuracy. These measures are described below.
The adherence questionnaire began with a grid whereby each study drug was listed with the number of doses prescribed per day filled in by the study staff. Participants would answer the number of prescribed doses missed for “yesterday,” “2 days ago,” “the past 2 weeks,” and “the past 30 days.” We followed the methodology of Reynolds et al12 to calculate the adherence ratios for yesterday, 2 days ago, in the past 2 weeks, and in the past 30 days as 1 (number of doses missed for the period divided by the number of doses prescribed). They were then asked a series of questions, including when they last missed medications (within past week, 1–2 weeks ago, 2–4 weeks ago, or never, and skipped/not applicable), how many days they had missed taking all of their doses during the past 4 days (none, 1, 2, 3, and 4 days), and whether they missed any medications over the past weekend (yes/no for Saturday or Sunday). The 7 self-report questions (4 adherence ratios and the 3 questions above) were subjected to a principal component analysis (PCA) to construct a nonadherence factor approximating a continuous latent variable. This PCA was conducted separately for each study visit. Inspection of the scree plot13 and the result that only the first PC had an eigenvalue greater than 114 lead us to retain 1 principal component as the self-report PCA adherence score. This procedure of using the available self-report questions to create a self-report continuous PC and maximize variability has been successfully used in similar studies,12,15 and the resulting variable was relatively unimodally distributed.
Additionally, participants were asked about potential reasons for nonadherence. This involved a checklist for “never,” “rarely,” “sometimes,” and “often” and had 24 potential reasons for nonadherence, such as “forgot,” “side effects,” “transportation problems getting to the clinic,” and “lost pills,” which were generated from the study sites and using items from previous AIDS Clinical Trials Group (ACTG) trials.16 The most frequently reported reasons are described in the study results.
At each study visit, study nurses dispensed pills expected for the following quarter, and participants were instructed to bring any remaining pills to the clinic at the next visit for a pill count. The adherence percentage was calculated for each participant as dividing the total number of pills taken by the total number of pills that should have been taken since ART initiation. A binary adherence variable (<95% vs. 95%–100%) was then created.
The psychosocial interview included a modified version17 of the ACTG SF-2118 quality of life questionnaire. To simplify the analyses, and based on previous adherence research, only the general health perceptions and mental health subscales were included. We also included 1 question about general satisfaction with social support,16,17,19 which asked about overall satisfaction with social support from friends and families, ranging from 0 (very dissatisfied) to 3 (very satisfied). For substance use, there was a frequency question about binge drinking, asking how often participants drank 5 or more drinks of alcohol in the past month, ranging from never (0) to daily (6), a series of yes/no questions for various substances (eg, cocaine, heroin, marijuana), followed by the frequency question for the substance used most. At last, there were questions about sexual behaviors in the past week, which yielded a variable indicating whether participants reported any condomless sex acts in the past week.
HIV-1 plasma viral load was collected quarterly, and viral suppression was defined as HIV-1 plasma viral load <400 copies per milliliter.
Logistic (allowing for odds ratios) and linear regression (for continuous measures with an unstandardized beta as the parameter estimate) models using generalized estimating equation for repeated measurements were first fit to examine the associations between adherence measured by pill count and self-report with psychosocial and demographic predictors, respectively. In model building, all variables associated with the outcome with P < 0.1 from the univariate regression model were included in later multivariable regression models. Similarly, logistic regression models were fit to examine the association between the 2 indicators of adherence (pill count and self-report) on viral suppression status (suppressed vs. detectable) including the psychosocial and demographic predictors as covariates. All analyses were conducted using SAS version 9.2 (SAS Inc, Cary, NC).
Demographic variables for the sample are reported in the primary outcome article for HPTN 0521 and are included in the Supplemental Appendix (see Supplemental Digital Content, http://links.lww.com/QAI/A654). Table 1 presents descriptive baseline data on the psychosocial and demographic data for the present sample. General health perception and mental health scores were in expected ranges, and most participants reported that they were either somewhat or very satisfied with social support. A small minority reported substance use, although almost 20% reported binge drinking. Only 4.2% of the sample reported less than 100% condom use during sex.
Levels of Adherence and Reasons for Nonadherence
According to pill count, in the first month after ART initiation, 82.2% of participants were “adherent” (defined by 95% or greater levels) and 83.2% of participants were adherent 1 year after ART initiation. With respect to self-report items for how often they missed doses after ART initiation, 88.8% reported less than perfect adherence in the first month and 84.2% at 1 year. In the first month, the most frequent (more than 5%) reasons for nonadherence were forgot (40.4%), traveling away from home (19.3%), wanted to avoid side effects (17.0%), busy doing other things (9.4%), other illness or health problems got in the way (8.2%), and ran out of pills (6.4%). At 1 year, the most frequent reasons given for nonadherence were forgot (45.1%), busy doing other things (20.7), traveling away from home (22.6%), and ran out of pills (14%).
Longitudinal Models of Adherence
Pill Count Adherence
Tables 2 and 3 present univariate and multivariable logistic regression analyses of psychosocial and demographic predictors of pill count adherence, inclusive of corresponding odds ratios, confidence intervals, and significance levels. According to the estimates in both the univariate and multivariable analyses, having a higher mental health score was the only statistically significant psychosocial predictor associated with greater adherence as measured by pill count. Geographic region (specifically, Asia and Africa vs. America) was also associated with adherence, with both groups having higher pill count adherence than America. In the univariate analyses, in addition to higher mental health scores, higher general health perceptions, and lower levels of unprotected sex were associated with greater adherence by pill count.
Self-report (PCA) Adherence Score
Tables 2 and 3 also present univariate and multivariable linear regression analyses of psychosocial predictors of adherence measured by the self-report PCA scores, with corresponding unstandardized beta regression coefficient estimates, confidence intervals, and significance levels. Similar to the adherence by pill count, in both the univariate and the multivariable analyses, having a higher mental health score was statistically significantly associated with better adherence. However, in the analysis by geographic region, participants from Asia had lower self-reported adherence. Additionally, older age was associated with higher adherence in univariate analysis, but only marginally significant (P = 0.06) in multivariable analysis, Better social support (somewhat satisfied vs. very dissatisfied) was associated with better adherence in univariate analysis only.
Predictors of Viral Suppression
Both univariate and multivariable analyses showed consistent significant associations between the self-report PCA medication adherence score and viral suppression (Table 4). In the multivariable analysis of pill count adherence, those categorized as being 95%–100% adherent were 1.42 times more likely to be virally suppressed; analysis of the self-report PCA adherence score showed each unit increase in the PCA resulting in a 1.20-fold increased likelihood of being virally suppressed.
ART that suppresses HIV-1 replication reduces transmission of HIV-1 to a sexual partner; however, this benefit requires strict adherence to treatment. Treatment for prevention approaches generally requires initiating ART when patients are healthier and therefore at higher CD4 counts. One study20 compared 60 individuals in Uganda with CD4 <250 cells with those with ≥250 cells and found greater number of treatment interruptions and more uncontrolled virus in those in the higher baseline CD4 category. No studies have examined adherence to ART when the stated purpose of the ART was specifically to test whether it would reduce transmission of HIV-1 to a sexual partner.
In this analysis, we noted higher levels of adherence to ART than typically observed in the context of ART for treatment.21 In addition, such adherence in the infected person is greater than reported in an observational study of couples in Zambia.22 The high level of adherence in this report may have reflected the intense management in the conduct of the HPTN 052 study,1,9 the benefit of couples' counseling,23 feedback about viral load suppression, or altruism related to prevention of transmission of HIV (a potential benefit communicated during the informed consent process). Additionally, counselors used an evidence-based cognitive-behavioral counseling intervention (LifeSteps)10,24 as the basis of their adherence counseling training and used checklists and visit documentation to maximize the actual delivery of this counseling approach in these settings.
We found that the only psychosocial variable that predicted adherence in the multivariable models was the mental health score on the quality of life assessment. Although the association of mental health to adherence is consistent with meta-analytic work in individuals with HIV-1 showing an association of depression to nonadherence,25 the lack of association of variables such as substance use26–28 and social support29 differs from what has been reported in studies of adherence in patients prescribed ART for treatment.
Our HIV-1 serodiscordant couples were in relatively stable relationships, located in resource-limited countries where initiation of ART was recommended at lower CD4 cell counts than what was dictated by the study, and who were volunteering for an HIV prevention trial where provision of care might be greater than what would otherwise be received in the local setting. Accordingly, participants electing to take medications for prevention purposes, and/or for purposes of being in a prevention trial, may have higher motivation for health behaviors in general. Hence, as seen in this study, the strength of the association of some of the typical psychosocial variables to health behavior may be attenuated when the need for treatment is also not as strong.
Nevertheless, higher mental health scores remained independently associated with better adherence. A treatment study that complemented HPTN 052 and was conducted at the same sites, ACTG5175,30 found that illicit drug use and general health perceptions but not mental health scores were variables associated with adherence in longitudinal multivariable models.15 These results suggest that there may be different psychosocial variables predicting adherence when attempting to maximize ART for prevention rather than ART for treatment.
There are several limitations to this analysis. First, self-report and pill count adherence are not as objective as other indicators, such as medication event monitoring systems, other electronically monitored adherence devices, unannounced pill counts, or blood ART levels. However, the association of viral load to both pill count and self-report PCA adherence scores shows that these can be useful indicators of adherence.31 Second, this analysis mainly examined main effects and not interactions because of the wide number of potential variables possible for interactions, such as study site, HIV risk group, gender, and country. Third, although participants enrolled in the study as a prevention trial, it is possible that some participants' motivation to join was as a vehicle to gaining ART as the treatment for themselves. Hence, although the time the study began initiating ART was not considered necessarily beneficial to the patient, this could have affected the motivations to take ART and be adherent to ART.
The results of this analysis demonstrate a very high degree of adherence to ART, which correlates well with the durable suppression of viremia observed.1,9 It seems likely that high adherence can be expected in many groups of couples given the fact that ART dramatically suppressed HIV transmission in 10 of 12 observational studies.32 Adherence to treatment is likely optimized when evidence-based counseling is an ongoing part of provision of ART,33,34 when couples are counseled together, when the prevention benefit to the sexual partner is made clear, and when feedback about viral suppression is provided. These data suggest the potential utility of implementing these approaches in clinical settings.
1. Cohen MS, Chen YQ, McCauley M, et al.. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011;365:493–505.
2. Quinn TC, Wawer MJ, Sewankambo N, et al.. Viral load and heterosexual transmission of human immunodeficiency virus type 1. Rakai Project Study Group. N Engl J Med. 2000;342:921–929.
3. Donnell D, Baeten JM, Kiarie J, et al.. Heterosexual HIV-1 transmission after initiation of antiretroviral therapy: a prospective cohort analysis. Lancet. 2010;375:2092–2098.
4. Rodger A, Bruun T, Cambiano V, et al.. HIV transmission risk through condomless sex if HIV+ partner on suppressive ART: PARTNER study. Presentation at the Conference on Retroviruses and Opportunistic Infections; March 2014; Boston, MA.
5. Reda AA, Biadgilign S. Determinants of adherence
to antiretroviral therapy among HIV-infected patients in Africa. AIDS Res Treat. 2012;2012:574656.
6. Malow R, Dévieux JG, Stein JA, et al.. Depression, substance abuse and other contextual predictors of adherence
to antiretroviral therapy (ART) among Haitians. AIDS Behav. 2013;17:1221–1230.
7. Sabin LL, Desilva MB, Hamer DH, et al.. Barriers to adherence
to antiretroviral medications among patients living with HIV in southern China: a qualitative study. AIDS Care. 2008;20:1242–1250.
8. Wasti SP, Simkhada P, Randall J, et al.. Factors influencing adherence
to antiretroviral treatment in Nepal: a mixed-methods study. PLoS One. 2012;7:e35547.
9. Grinsztejn B, Hosseinipour MC, Ribaudo HJ, et al.. Effects of early vs. delayed initiation of antiretroviral treatment on clinical outcomes of HIV-1 infection: results from the phase 3 HPTN 052 randomized controlled trial. Lancet Infect Dis. 2014;14:281–290.
10. Safren SA, Otto MW, Worth JL, et al.. Two strategies to increase adherence
to HIV antiretroviral medication: life-steps and medication monitoring. Behav Res Ther. 2001;39:1151–1162.
11. Safren SA, Otto MW, Worth JL. Life-Steps: applying cognitive-behavioral therapy to patient adherence
in HIV medication treatment. Cogn Behav Pract. 1999;6:332–341.
12. Reynolds NR, Sun J, Nagaraja HN, et al.. Optimizing measurement of self-reported adherence
with the ACTG adherence
questionnaire: a cross-protocol analysis. J Acquir Immune Defic Syndr. 2007;46:402–409.
13. Cattell RB. The scree test for the number of factors. Multivariate Behav Res. 1966;1:245–276.
14. Kaiser HF. The application of electronic computers to factor analysis. Educ Psychol Meas. 1960;20:141–151.
15. Safren SA, Biello KB, Smeaton L, et al.. Psychosocial predictors of nonadherence and treatment failure in a large scale multi-national trial of antiretroviral therapy for HIV: data from the ACTG A5175/PEARLS trial. PLoS One. 2014;9:e104178.
16. Chesney MA, Ickovics 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 Trials Group (AACTG). AIDS Care. 2000;12:255–266.
17. Safren SA, Hendriksen ES, Smeaton L, et al.. Quality of life among individuals with HIV starting antiretroviral therapy in diverse resource-limited areas of the world. AIDS Behav. 2012;16:266–277.
18. Wu AW, Rubin HR, Mathews WC, et al.. A health status questionnaire using 30 items from the Medical Outcomes Study. Preliminary validation in persons with early HIV infection. Med Care. 1991;29:786–798.
20. Adakun SA, Siedner MJ, Muzoora C, et al.. Higher baseline CD4 cell count predicts treatment interruptions and persistent viremia in patients initiating ARVs in rural Uganda. J Acquir Immune Defic Syndr. 2013;62:317–321.
21. Mills EJ, Nachega JB, Buchan I, et al.. Adherence
to antiretroviral therapy in sub-Saharan Africa and North America: a meta-analysis. JAMA. 2006;296:679–690.
22. Wall K, Inambao M, Simpungwe K, et al.. From efficacy to effectiveness: ART uptake and HIV seroincidence by ART status among HIV discordant couples in Zambia. Presentation at the 20th International AIDS Conference; July 2014; Melbourne, AU.
23. Dunkle KL, Stephenson R, Karita E, et al.. New heterosexually transmitted HIV infections in married or cohabiting couples in urban Zambia and Rwanda: an analysis of survey and clinical data. Lancet. 2008;371:2183–2191.
24. Simoni JM, Chen WT, Huh D, et al.. A preliminary randomized controlled trial of a nurse-delivered medication adherence
intervention among HIV-positive outpatients initiating antiretroviral therapy in Beijing, China. AIDS Behav. 2011;15:919–929.
25. Gonzalez JS, Batchelder AW, Psaros C, et al.. Depression and HIV/AIDS treatment nonadherence: a review and meta-analysis. J Acquir Immune Defic Syndr. 2011;58:181–187.
26. Friedman MS, Marshal MP, Stall R, et al.. Associations between substance use, sexual risk taking and HIV treatment adherence
among homeless people living with HIV. AIDS Care. 2009;21:692–700.
27. Lehavot K, Huh D, Walters KL, et al.. Buffering effects of general and medication-specific social support on the association between substance use and HIV medication adherence
. AIDS Patient Care STDS. 2011;25:181–189.
28. King RM, Vidrine DJ, Danysh HE, et al.. Factors associated with nonadherence to antiretroviral therapy in HIV-positive smokers. AIDS Patient Care STDS. 2012;26:479–485.
29. Simoni JM, Pearson CR, Pantalone DW, et al.. Efficacy of interventions in improving highly active antiretroviral therapy adherence
and HIV-1 RNA viral load. A meta-analytic review of randomized controlled trials. J Acquir Immune Defic Syndr. 2006;43(suppl 1):S23–S35.
30. Campbell TB, Smeaton LM, Kumarasamy N, et al.. Efficacy and safety of three antiretroviral regimens for initial treatment of HIV-1: a randomized clinical trial in diverse multinational settings. PLoS Med. 2012;9:e1001290.
31. Kabore L, Muntner P, Chamot E, et al.. Self-report measures in the assessment of antiretroviral medication adherence
: comparison with medication possession ratio and HIV viral load. J Int Assoc Provid AIDS Care. 2014;14:156–162.
32. Muessig KE, Cohen MS. Advances in HIV prevention for serodiscordant couples. Curr HIV/AIDS Rep. 2014;11:434–446.
33. Center for Disease Control. Medication adherence
; effective interventions, HIV prevention that works. Available at: https://www.effectiveinterventions.org/en/HighImpactPrevention/BiomedicalInterventions/MedicationAdherence.aspx
. Accessed December 19, 2014.
34. Amico K; On behalf of the Guidelines Panel. Education and counseling to support adherence
with general population of PLWH guidelines and recommendations for implementation. 2012. Available at: http://www.iapac.org/ias/presentations/Amico_KR.pdf
. Accessed December 19, 2014.