Psychosocial factors in adherence to antiretroviral therapy among HIV-positive people who use drugs

Abstract

Objectives

Sub-optimal adherence to antiretroviral therapy (ART) among HIV-infected people who use illicit drugs (PWUD) remains a significant concern, and there is a lack of effective adherence interventions for this population. Therefore, we sought to identify psychosocial determinants of optimal adherence, including adherence self-efficacy and outcome expectancies, with the aim of informing interventions designed to improve adherence among PWUD.

Methods

From December 2005 to November 2013, data were derived from the AIDS Care Cohort to evaluate Exposure to Survival Services (ACCESS), a prospective cohort of PWUD in Vancouver, Canada. Multivariable generalized estimating equations (GEE) analysis was used to identify longitudinal factors independently associated with ≥95% adherence to ART.

Results

Among 667 participants, including 220 (33%) women, 391 (59%) had ≥95% ART adherence at baseline. In multivariable GEE analysis, adherence self-efficacy (Adjusted Odds Ratio [AOR] = 1.16, 95% Confidence Interval [CI]: 1.11 - 1.21 per 10-point increase) was independently and positively associated with adherence, while negative outcome expectancy (AOR = 0.95, 95% CI: 0.93 - 0.98) was negatively associated.

Conclusions

In light of the ongoing challenges associated with ART adherence among HIV-positive PWUD, and our findings of associations between adherence, self-efficacy and outcomes expectancies, tailored intervention strategies based on constructs of social learning theory should be implemented and evaluated in an effort to improve adherence among HIV-infected PWUD.

Introduction

The introduction of antiretroviral therapy (ART) has significantly impacted and altered the course of HIV disease (Hammer, 1996). Shown to reliably suppress levels of plasma HIV RNA, optimal treatment has led to large reductions in HIV-related morbidity and mortality (Hogg et al., 2001; Egger et al., 2002). In order to be fully effective and for long-lasting clinical success, an optimal level of adherence, commonly defined as participants who have received their ART ≥95% of the time, is required. It has been previously shown that this validated measure of adherence predicts virologic suppression and survival (Wood et al., 2003). On the other hand, non-adherence detrimentally affects virologic control, and subsequently, increases the risk of disease progression, as well as the transmission of HIV to others (Deeks, 2003; Montaner, 2011; Paterson et al., 2000). Further, sub-optimal levels of adherence increase the risk of viral mutations, which can lead to treatment failure and the transmission of drug-resistant virus to others (Tang, 2012).

Despite the benefits of ART on survival and quality of life, a large body of research has shown that people who use illicit drugs (PWUD) often exhibit lower levels of adherence in comparison to individuals from other groups of people living with HIV (Bruce, & Altice, 2007; Wood, Montaner, Tyndall, Schechter, O'Shaughnessy, & Hogg, 2003). Previous studies have identified numerous factors that increase the risk of non-adherence among PWUD, including high-intensity drug use (Palepu, Tyndall, Yip, O'Shaughnessy, Hogg, & Montaner, 2003), co-morbid psychiatric disorders (Arnsten et al., 2007), as well as social and structural factors (e.g., incarceration and homelessness) (Tapp et al., 2011).

One possible factor that may affect ART adherence is self-efficacy; that is, one's belief in their own ability to complete a certain task. This psychosocial concept, derived from Bandura's social learning theory (Bandura, 1977), has influenced the way clinicians and public health officials view and design health-related interventions (Bandura, 2001). Bandura posits that although heightened awareness and knowledge of health risks are important components for change, effective self-regulation and behavioral change also requires self-motivation and self-guidance. Further, a lack of a sense of self-efficacy can lead to discrepancies between knowledge and action-taking. As such, by enhancing individuals' perception of self-agency, one can exercise more control over their health, thus leading to better health outcomes. Results from past studies have supported the relationship between higher self-efficacy and higher adherence rates to health and medical regimens, including ART adherence (Bandura, 1990; Colbert, Sereika, & Erlen, 2013; Kerr, et al., 2004; Johnson et al., 2006; Trovato et al., 2012). However, these analyses have been limited to cross-sectional designs and the observed relationships have not been assessed longitudinally in community-derived samples of PWUD. Another psychosocial construct that has been examined in relation to ART adherence to treatment is outcome expectancies regarding ART (Reynolds et al., 2004), with past work suggesting a strong relationship between negative beliefs about ART and non-adherence (Kerr, et al., 2004).

Given preliminary evidence suggesting that psychological constructs derived from social learning theory have been shown to be associated with adherence to different medical regimens, including ART among PWUD (Tyer-Viola, Corless, Webel, Reid, Sullivan, & Nichols, 2014), this study sought to build on this work by longitudinally assessing the relationships between adherence self-efficacy, outcome expectations, and adherence to ART among a community-recruited cohort of HIV-positive PWUD in Vancouver, Canada; a setting where all HIV/AIDS treatment and care is delivered at no-cost through the province's universal healthcare system. It is hypothesized that there would be a positive and independent association between adherence self-efficacy, and adherence to ART, whereas negative outcome expectations would yield a negative and independent association between adherence self-efficacy, and adherence to ART. Based on the adherence literature, although lower levels of self-efficacy and adherence have been observed among PWUDs compared to non-drug using populations (Sharpe et al., 2004), this study hypothesizes that these psychological constructs will predict adherence among PWUD in the same manner as non-drug using populations. Further, this study considers the role of self-efficacy, as it has been demonstrated to be a key construct in the development of interventions specifically tailored to not only PWUD, but also people living with HIV (Bandura, 1990; Bandura, 1999). Lastly, although several socio-demographic and behavioral variables (i.e., ethnicity, gender and drug-use patterns) have been shown in the literature to be associated with ART adherence (Golin, et al., 2002; Simoni et al., 1999; Sharpe et al., 2004), and therefore have been included as variables that might confound the relationship between self-efficacy and adherence.

Methods

The AIDS Care Cohort to evaluate Exposure to Survival Services (ACCESS) is an observational prospective cohort of HIV-seropositive illicit drug users in Vancouver, Canada, and has been described in detail elsewhere (Strathdee, et al., 1998; Wood, et al., 2004). In brief, individuals were eligible for ACCESS if they were aged 18 years or older, HIV seropositive, used any illicit (injection or non-injection) drugs other than cannabis in the last month,and provided written informed consent. At baseline and semi-annually thereafter, participants answer a standardized interviewer administered questionnaire and provide blood samples for serologic analysis. To compensate participants for their time, participants received a stipend of $30 CDN at each study visit. This study has received approval from Providence Health Care/University of British Columbia's Research Ethics Board.

The current study included individuals who had received ≥ 1 days of highly-active antiretroviral therapy (HAART) at the time of the baseline interview. Baseline HAART-naïve individuals who initiated HAART during the study period were included from their next follow-up interview following initiation. The primary outcome of interest was ART adherence, defined as the number of days that ART was dispensed divided by the total number of days since a participant initiated HAART, capped at 180 days. This outcome was dichotomized as ≥95% vs. <95%. It has been previously shown that this validated measure of adherence predicts virologic suppression and survival (Wood et al., 2003). ART dispensation was observed using a confidential record linkage with the Drug Treatment Program (DTP), a department of the British Columbia Centre for Excellence in HIV/AIDS (BC-CfE) which provides medication for all people living with HIV/AIDS in British Columbia. Through the DTP linkage, full retrospective and prospective HIV clinical profile for each participant were obtained, including all CD4 cell and plasma HIV-1 RNA viral load measurements conducted through the study or regular clinical care.

The present study considered a range of socio-demographic and behavioural variables, including: year of interview; age; gender (male vs. female); Caucasian ancestry (yes vs. no); currently living in unstable housing (yes vs. no); involvement in sex work, defined as exchanging sex for money, shelter, drugs or other commodities (yes vs. no); incarceration, defined as being in detention, prison, or jail overnight or longer (yes vs. no); and current enrollment in methadone maintenance therapy (MMT) (yes vs. no). Other variables included: adherence self-efficacy; negative outcome expectancy (per 10-point increase); and CD4 cell count (measured in cells per/μL). Drug use was considered using a four-level variable where abstinence constituted the reference category and was compared to the influence of any exclusive injection drug use (e.g., heroin, cocaine, or crystal methamphetamine injecting), exclusive non-injection drug use (e.g., crack or crystal methamphetamine smoking) and both non-injection and injection drug use during the previous 180 days. The Adult AIDS Clinical Trial Group (AACTG) subscale that asks participants to indicate reasons for missing doses of ART was also included. This scale has been well described in other studies (Chesney, et al., 2000). CD4 cell count refers to the median of all measurements taken in the previous 180 days. All variables refer to activities or experiences in the 180 days preceding the study interview, unless otherwise stated.

The eight-item adherence self-efficacy measure (ASEM), which was used to assess self-efficacy, also included a measure of negative outcome expectation. As is common in the measurement of self-efficacy, items were tailored to reflect the specific behaviours and skills under study (Bandura, 1997), and included six items assessing adherence efficacy expectations and two items assessing self-regulatory efficacy. These constructs have previously been found to be conceptually and behaviourally distinct, and were therefore measured separately (Bandura, 2001). However, within the context of this study, both efficacy expectations and self-regulatory efficacy will be referred to as, and encompassed within, adherence self-efficacy. The self-efficacy measure asks individuals to rate their confidence in their ability to engage in a range of behaviors specific to adherence to ART (Kerr, et al., 2004). Negative outcome expectancy was based on a 100-point scale for the question: “Taking HIV medications will make me feel sicker.” The internal consistency reliability of the ASEM was assessed with Cronbach alpha using the study sample. The Cronbach alpha was 0.85 for the self-regulatory efficacy subscale and 0.83 for the self-efficacy subscale. All variables refer to activities or experiences in the 180 days preceding the study interview, unless otherwise stated.

As a first step, the baseline characteristics of the explanatory variables were examined, stratified by ART adherence. Specifically, the Mann-Whitney test for continuous variables and Pearson's Chi-squared test for categorical variables were used to assess the relationships among the characteristics and ART adherence. Then, for the bivariable and multivariable analyses, generalized estimating equations (GEE) with logit-link function and exchangeable correlation structure was used to longitudinally examine the associations between the explanatory variables and ART adherence during the study period. GEE models can account for the correlation within repeated measures for each subject and produce valid estimates of standard error (Liang, & Zeger, 1986).

To fit the multivariable model, all explanatory variables that were significant in the bivariable analyses were considered for inclusion in the full multivariable model. A backward model selection procedure was used to identify the model with the best overall fit as indicated by the lowest quasi-likelihood under the independence model criterion value (Hanley, Negassa, & Forrester, 2003). All tests were two-tailed and the statistical significance was measured at p < 0.05. All statistical analyses were performed using the SAS software version 9.3 (SAS, Cary, NC).

We have also conducted a subanalysis in which we used lagged explanatory variables. In other words, we assessed the relationship between the value of the explanatory variable measured in the previous observation period (i.e., between 180 and 360 days prior to the current study interview) and the dependent variable measured in the previous six months. The sub-analysis was restricted to study visits in which the previous observation period was not missed, for a total of 3460 observation periods from 595 participants.

Results

A total of 667 of HIV-positive individuals were enrolled in the study between December 2005 and November 2013. Among this sample, 220 (33.0%) were female, 369 (55.3%) self-reported Caucasian ancestry, and the median age at baseline was 43 years (interquartile range [IQR] = 37 – 48). In total, the analytic sample contributed a total of 4127 observations, or a median of 6 study visits (IQR = 3 – 9). Of the 667 participants in the study, 595 (89.2%) returned for at least one follow-up visit, with a median of 5 follow-up visits (IQR = 3 – 8). The number of participants who contributed to each number of study visits are as follows: 1 visit = 72 (10.8%); 2 visits = 64 (9.6%); 3 visits = 62 (9.3%); 4 visits = 57 (8.5%), 5 visits = 57 (8.5%); 6 visits = 59 (8.8%); 7 visits = 58 (8.7%); 8 visits = 62 (9.3%); 9 visits = 32 (4.8%); 10 visits = 38 (5.7%); 11 visits = 36 (5.4%); 12 visits = 28 (4.2%); 13 visits = 24 (3.6%); 14 visits = 11 (1.6%); 15 visits = 6 (0.9%); and 16 visits = 1 (0.1%). Of 4127, 2563 (62.1%) six-month interview periods were characterized by ≥ 95% ART adherence. At baseline, 391 (58.6%) individuals had achieved optimal ART adherence. After excluding baseline observations where HAART was initiated within 180 days of the study interview date, which excluded 17 participants who had no follow-up visit, 558 (85.9%) of 650 participants have achieved ≥95% ART adherence at some point during the study period.

The baseline characteristics of all participants stratified by ≥95% ART adherence in the previous six months are presented in Table 1. The results of the bivariable and multivariable GEE analyses are presented in Table 2. In the multivariable GEE analysis, variables that were positively associated with greater likelihood of optimal ART adherence included: adherence self-efficacy (Adjusted Odds Ratio [AOR] = 1.16, 95% Confidence Interval [CI]: 1.11 – 1.21), age (AOR = 1.03, 95% CI: 1.02 – 1.04), current enrollment in MMT (AOR = 1.40, 95% CI: 1.15 – 1.71) and CD4 cell count (AOR = 1.19, 95% CI: 1.13 – 1.25). Conversely, exclusive non-injection drug use vs. abstinence (AOR = 0.63, 95% CI: 0.47 – 0.85), concurrent non-injection and injection drug use vs. abstinence (AOR = 0.62, 95% CI: 0.47 – 0.82) and negative outcome expectancy (AOR = 0.95, 95% CI: 0.93 – 0.98) were negatively associated with achieving optimal ART adherence.

Table 1. Baseline Characteristics Of Hiv-Infected People Who Use Illicit Drugs In Vancouver, Canada, Stratified By ≥95% Art Adherence (N = 667).

	≥95%ART Adherence*
Characteristic	Yes (%) (n = 391)	No (%) (n = 276)	p - value
Year of interview date
Median (IQR)	2009 (2007 – 2010)	2008 (2006 – 2010)	0.168
Age (in years)
Median (IQR)	45 (38 – 50)	41.5 (36 – 46)	<0.001
Gender
Male	282 (72.1)	165 (59.8)	0.001
Female	109 (27.9)	111 (40.2)
Caucasian ancestry
Yes	238 (60.9)	131 (47.5)	0.001
No	153 (39.1)	145 (52.5)
Currently in unstable housing
Yes	251 (64.2)	188 (68.1)	0.293
No	140 (35.8)	88 (31.9)
Sex work*
Yes	41 (10.5)	40 (14.5)	0.119
No	350 (89.5)	236 (85.5)
Incarceration*
Yes	47 (12.0)	28 (10.1)	0.450
No	344 (88.0)	248 (89.9)
Currently on methadone treatment
Yes	194 (49.6)	101 (36.6)	0.001
No	197 (50.4)	175 (63.4)
Drug use patterns*
Abstinent (Reference)	14 (3.6)	8 (2.9)
Injection only	28 (7.2)	17 (6.2)	0.911
Non-injection only	89 (22.8)	63 (22.8)	0.651
Both non-injection and injection	260 (66.5)	188 (68.1)	0.604
Adherence self-efficacy*
Median (IQR)	93.8 (80 – 100)	81.9 (67.5 – 93.8)	<0.001
Negative outcome expectancy*
Median (IQR)	0 (0 – 30)	15 (0 – 60)	<0.001
CD4 cell count (in cells/μL)*
Median (IQR)	350 (215 – 490)	257.5 (140 – 382.5)	<0.001

^*Activities in the previous 6 months, ART: Antiretroviral Therapy, IQR: Interquartile Range

Table 2. Bivariable and multivariable GEE analyses of factors associated with ≥95% ART adherence^* among HIV-infected people who use illicit drugs in Vancouver, Canada (n = 667).

	Unadjusted		Adjusted
Characteristic	Odds Ratio (95% CI)	p - value	Odds Ratio (95% CI)	p - value
Year of interview date
Per year increase	1.09 (1.05 – 1.13)	<0.001
Age
Per year older	1.04 (1.03 – 1.06)	<0.001	1.03 (1.02 – 1.04)	<0.001
Gender
Male vs. Female	1.31 (1.05 – 1.63)	0.018	1.24 (0.99 – 1.55)	0.063
Caucasian ancestry
Yes vs. No	1.34 (1.09 – 1.65)	0.006
Currently in unstable housing
Yes vs. No	0.93 (0.79 – 1.08)	0.339
Sex work*
Yes vs. No	0.85 (0.66 – 1.09)	0.207
Incarceration*
Yes vs. No	0.79 (0.62 – 1.01)	0.059
Currently on methadone treatment
Yes vs. No	1.38 (1.14 – 1.68)	0.001	1.40 (1.15 – 1.71)	0.001
Drug use pattern*
Injection only vs. Abstinent	0.76 (0.53 – 1.07)	0.116	0.78 (0.54 – 1.14)	0.201
Non-injection only vs. Abstinent	0.68 (0.51 – 0.89)	0.006	0.63 (0.47 – 0.85)	0.002
Both non-injection and injection vs. Abstinent	0.58 (0.44 – 0.76)	<0.001	0.62 (0.47 – 0.82)	0.001
Adherence self-efficacy*
Per 10-point increase	1.21 (1.16 – 1.26)	<0.001	1.16 (1.11 – 1.21)	<0.001
Negative outcome expectancy*
Per 10-point increase	0.93 (0.91 – 0.95)	<0.001	0.95 (0.93 – 0.98)	<0.001
CD4 cell count*
Per 100-cell/μL increase	1.23 (1.16 – 1.29)	<0.001	1.19 (1.13 – 1.25)	<0.001

^*Activities in the previous 6 months

ART: Antiretroviral Therapy, CI: Confidence Interval

Self-reported reasons for missing doses of ART during the study are presented in Table 3. As indicated, the most commonly cited reason (341 individuals, 51%) for missing doses of ART was the side effects that came with the medication. Finding ART's daily regimen hard to adhere to was reported by 231 (34.6%) individuals. Missing doses of ART due to the high number of pills required for optimal adherence was cited by 99 (14.8%) individuals, and 95 (14.2%) individuals reported suboptimal adherence to ART due to their current drug/alcohol use.

Table 3. Descriptive statistics for reasons for finding ART treatment hard to take at some point in the study (N = 667).

Characteristic	n (%)
Too many pills
yes	99 (14.8)
no	568 (85.2)
Currently on methadone treatment
yes	31 (4.7)
no	636 (95.3)
Sick from side effects
Yes	341 (51.1)
Gastro-intestinal symptoms*	292 (85.6)
Peripheral neuropathy*	33 (9.7)
Skin rash*	30 (8.8)
Sleep-related problems*	80 (23.5)
Dizziness/headaches*	25 (7.3)
Fatigue*	24 (7.0)
Other effects*	68 (19.9)
No	326 (48.9)
Hard to adhere daily
yes	231 (34.6)
no	436 (65.4)
Unstable housing
yes	53 (8.0)
no	614 (92.0)
Not close to pick-up location
yes	18 (2.7)
no	649 (97.3)
Using drugs and alcohol
yes	95 (14.2)
no	572 (85.8)
Hard to swallow pills
Yes	46 (6.9)
No	621 (93.1)
Other reasons
Yes	30 (4.5)
No	637 (95.5)

^*Percentages calculated from 341 participants who reported side effects

The subanalysis, which used lagged explanatory variables, was restricted to study visits in which the previous observation period was not missed, for a total of 3460 observation periods from 595 participants. The results that were obtained from this analysis were virtually identical to those from our primary analysis (data not shown).

Discussion

The present study observed a substantial proportion of HIV-positive PWUD who did not attain optimal adherence to prescribed ART during the study. In multivariable analyses, factors positively associated with ≥95% ART adherence included adherence self-efficacy, age, current enrollment in MMT, and CD4 cell count, while drug use patterns and negative outcome expectancy were negatively associated with optimal adherence to ART. Consistent with previous cross-sectional findings, CD4 cell count and age were found to be significantly and positively associated with ART adherence (Hinkin et al., 2006; Wood et al., 2004). This may in part reflect the greater instability and higher risk profile observed previously among young drug users (Hadland et al., 2012). However, given the widely described association between ART adherence and age, and CD4 cell count in the literature, the remainder of the discussion will be reserved for the other aforementioned significant variables. The most frequently reported reason for missing doses of ART was finding the treatment hard to take because of its side effects.

The psychological variable of adherence self-efficacy was found to be independently and positively associated with achieving optimal adherence to ART. This finding adds support to the existing studies that have successfully employed self-efficacy based interventions in order to address various health-related problems in general (Bandura, 1990; Trovato et al., 2012). One possible interpretation of this association is that those who possess higher levels of self-efficacy tend to score higher in the domains of choice, effort, persistence and coping exerted in the face of challenges (Bandura, 2001), all of which are relevant to adherence to ART. For example, it may be that individuals with higher self-efficacy were more likely to engage in activities that would encourage adherence to ART, including actively seeking to refill their prescription, and to persevere despite the side effects associated with the medication. The present study expands on the current ART adherence literature by being the first, to our knowledge, to demonstrate a longitudinal relationship between adherence self-efficacy and adherence to ART. As such, early assessment involving efficacy expectations can help identify persons at greater risk for lower levels of adherence, and interventions specifically targeting self-efficacy should be implemented to improve the likelihood of ART success. In order to encourage adherence, and in line with social learning theory from which self-efficacy is based, these interventions should focus on the development of skills and self-observation of accomplishments, initially involving modelling, guided practice, and then independent practice (Bandura, 1977). Lastly, given the independent effect of self-efficacy on adherence, our study suggests that improving adherence requires both behavioural change and cognitive change (i.e., perceptions of self-efficacy). As such, clinicians should make an effort to incorporate measures of both adherence-enhancing skill acquisition, and self-efficacy (i.e., enhancing perceptions of personal agency), in their practice.

Conversely, negative outcome expectations had an inverse relationship with adherence to ART. This study's measure of negative outcome expectation pertained to the belief that taking ART will make one sicker. It can be inferred that some individuals might associate negative outcomes with taking ART, despite the often observed efficacy of ART and the known consequences of untreated HIV disease. For example, Biswas and colleagues found that not having been told about the importance of ART was positively associated with non-adherence among a cohort of HIV-positive patients (Biswas, Dewan, Kant, Lal, & Rai, 2010). However, as has been previously documented (Catz, Kelly, Bogart, Benotsch, & McAuliffe, 2000; Kerr, 2004; Duran et al., 2001), it is also possible that PWUD with negative outcome expectations are concerned with the side effects of ART, rather than the belief that ART will fail to treat their HIV. In some instances, the side effects can be quite severe (Heath et al., 2001; Lugassy, Farmer, & Nelson, 2010), and are usually experienced earlier in treatment, whereas the clinical benefits of ART are generally experienced in the longer term. Given these dynamics, individuals may be prompted to skip doses of ART to avoid these side effects. This interpretation is supported by our finding that the most common self-reported reason for missing doses of ART was that treatment was, for many PWUD, hard to take due to its side effects. Therefore, our study highlights the need for interventions that focus on reducing the impact of side effects, as well as interventions that educate PWUD about the short-term nature and management of side effects. Indeed a small body of evidence suggests that interventions that involve increasing coping skills related to HIV treatment side effects are effective in mitigating the impact of side effects on ART treatment adherence (Johnson, Dilworth, Taylor, & Neilands, 2011). Further, psycho-educational methods focused on providing accurate information regarding HIV disease progression and emphasizing the clinical effects of ART could be implemented to change existing beliefs about the conditional relationship between ART and health (Tun, Celentano, Vlahov, & Strathdee, 2003).

The moderate rates of adherence to ART at baseline found in the current study are consistent with previous research that found that a significant proportion of PWUD failed to achieve adequate levels of adherence needed for viral suppression (Wood et al., 2003). However, it was also found that over 80% of PWUD have, at one point during the study period, achieved adequate adherence to ART. This finding is in line with Mann and colleague's study (Mann, Milloy, Kerr, Zhang, Montaner, & Wood, 2012) that reported an improvement in the levels of adherence to ART since the advent of combined ART treatment; 20% in 1996 to 66% in 2009. Although this trend can be attributed in part to improved tolerability and simplified ART regimens, there remains a high proportion of PWUD in the current study who have suboptimal levels of adherence. Enhanced efforts to support individuals in reaching optimal adherence are required.

This study found that PWUD who engaged in active drug use were less likely to meet adequate levels of adherence to ART. This is consistent with previous studies showing a relationship between drug use and non-adherence. More specifically, stimulant use, including cocaine and crystal methamphetamine, has been widely documented to adversely impact medication adherence (Gonzalez, Mimiaga, Isreal, Bedoya, & Safren, 2013; Ingersoll, 2004; Halkitis, Kutnick, & Slater, 2005). It has been suggested that disruptions to sleep and eating patterns, along with other environmental instabilities associated with the effects of stimulant use contribute to these observed suboptimal adherence rates. Research also supports the notion that opiate use can reduce levels of adherence, as it may be possible that frequent use of opiates can impair cognitive functioning (e.g., decision making and executive functioning), thus making it difficult for these PWUD to maintain optimal levels of adherence (Zeng et al., 2013). However, research in this area has yielded mixed and inconclusive results (Guerra, Sole, Cami, Tobena, & 1987; National Institutes on Drug Abuse, 1999e).

The finding that indicates that engaging in methadone maintenance therapy was associated with optimal adherence to ART is widely supported by the current literature (Palepu, et al., 2006). It is possible that adequate levels of adherence can be maintained through participation in MMT due to being engaged with the health care system where close monitoring through follow-up visits as well additional care to address potential barriers to ART adherence (i.e., co-occurring mental illness) are provided (Spire, Lucas, & Carrieri, 2007). Further, addiction treatment including MMT can improve access to ART, where co-administration of ART with daily dispensed MMT is possible (Berg, Mouriz, Li, Goldberg, & Arnsten, 2009). There are numerous health and social benefits derived from addressing opioid addiction through MMT, including increased employment, physical and mental health, and social functioning (Corsi, Kwiatkowski, & Booth, 2002; Fiellin, O'Connor, Chawarski, Pakes, Pantalon, &Schottenfeld, 2001; Zweben, &Payte, 1990). Given that MMT is a possible means for increasing ART uptake and adherence, concurrent delivery of ART and MMT should be encouraged in treatment interventions to improve the overall health and social outcomes of PWUD who are eligible for treatment.

This study has several limitations. First, despite extensive recruitment efforts, ACCESS is not a random sample, and therefore may not be generalizable to other drug-using or HIV-positive populations. Second, some data were collected via self-report and is thus vulnerable to response biases, including recall bias and social desirability. For instance, given the sensitive nature of some interview questions, respondents may be inclined to underreport stigmatizing behaviors such as illicit drug use and other high-risk behaviors. As a result, some of the findings may be conservative estimates. However, our outcome of interest was ascertained from the records of a comprehensive source of ART dispensation. Our use of scales may also elicit extreme responding, particularly for our psychosocial measures, as it has been previously shown that participants are more likely to respond in an extreme manner when it pertains to their own beliefs and motivations (Furnham, 1986). Third, given the non-randomized nature of this study, the relationships studied may be influenced by unmeasured confounders.

In conclusion, this study found moderate rates of adherence to ART among HIV-infected PWUD. Despite an observed trend towards greater proportions of individuals attaining optimal adherence during follow-up, there remains a large segment of the HIV-infected PWUD population who do not meet adequate levels of adherence to ART. In a multivariable model, self-efficacy and current enrolment in MMT was positively and independently associated with ≥95% ART adherence, while negative outcome expectation and active drug use was negatively associated with ≥ 95% ART adherence. This study highlights the role that psychosocial variables play in achieving adequate levels of adherence to medication. Future research should consider CD4 cell count and viral load as outcome variables of interest, as this would provide strong support for conducting clinical trials that target self-efficacy and outcome expectancies, with the aim of improving outcomes from HIV treatment. Lastly, these findings emphasize the potential importance of employing psychosocial measures to screen for risk of non-adherence, as well as the implementation of adherence interventions based on social learning theory.