Introduction

Substance users remain a population of concern among persons living with HIV/AIDS (PLWHA). They suffer from adverse health outcomes [1], more rapid HIV disease progression [2], higher likelihood of disengaging from the HIV continuum of care [3], and a lower likelihood of receiving combination antiretroviral therapy (cART) than their non-using peers [4]. Active substance use has also been shown to decrease cART adherence among PLWHA, indirectly contributing to worse HIV outcomes [5–7]. In addition to the well-characterized prevalence of injection drug use (IDU) among PLWHA in various settings [8], non-injection drug use (NIDU) [9] and alcohol related disorders (ARD) may also be highly prevalent in this population [10–12], indicating a relatively large public health impact of these exposures.

More specifically, use of alcohol, heroin, and cocaine, and despite a lack of consensus in the literature, possibly methamphetamines and marijuana, has been shown to be associated with poor cART adherence [13]. However, a large meta-analysis estimated that 60 % of substance users had ‘optimal’ adherence to cART, suggesting no differences in adherence between substance users and non-users [14]. The authors noted that the analysis was limited by a high degree of heterogeneity between studies (including study time-frame, adherence measures, and routes/types of substance used). For example, compared to studies using narrower time-frames to measure adherence (e.g., 2 weeks), studies using a larger time frame had adherence estimates that were approximately 50 % lower. Differences in methods used to measure both adherence and substance use disorders present a continuous challenge in clinical practice, and in research that attempts to quantify the prevalence and impact of substance use.

Substance use impacts communities globally to varying degrees [15], yet most evidence relating to the use of psychoactive drugs and cART adherence comes from high-income countries. There is a paucity of information from Latin America, which is a barrier to the planning and provision of adequate treatment programs for co-morbid drug dependence in the region. In one of few studies from the region, ‘high-intensity use of alcohol’ alone, and the simultaneous use of alcohol and illicit drugs, were independently associated with an increased chance of cART non-adherence in a Brazilian cohort [16]. Similarly, having an alcohol use disorder was associated with non-adherence among HIV-infected Peruvian men who have sex with men [17].

Although substance use has been associated with decreased cART adherence in other populations, data more generalizable to persons from low- and middle-income countries (LMIC) are scarce. The aims of this study were (1) to describe the prevalence of recent alcohol use and NIDU among PLWHA receiving care at CCASAnet (Caribbean, Central and South America network for HIV epidemiology) sites; and (2) to analyze the association between active substance use and missing doses of cART.

Methods

Design

The CCASAnet cohort (www.ccasanet.org) is an ongoing observational research network established to collect clinical care data from HIV clinical and research sites in Central and South America and the Caribbean [18]. This cross-sectional study combined routine patient care data with data from a clinical screening for substance use that was implemented at six CCASAnet sites. The rapid screening tool (RST) is a short, self-reported questionnaire that evaluates the frequency of substance use and the number of missed cART doses in a 7-day recall period. It has been used by CCASAnet and Vanderbilt researchers and shown to correlate well with other measures [19] and was translated and back-translated by local research staff into Spanish or Portuguese. Clinic staff administered the questionnaire at patient visits. Basic demographic and clinical data were obtained from the CCASAnet data repositories.

Study Population

A convenience sample of PLWHA (≥18 years of age) who attended routine clinical visits at CCASAnet sites responded to the RST. Sites contributing data to this study included Fundacion Huésped, Buenos Aires, Argentina (FH-Argentina); Instituto Nacional de Infectologia Evandro Chagas, FIOCRUZ, Rio de Janeiro, Brazil (INI- Brazil); Fundación Arriarán, Santiago, Chile (FA-Chile); Instituto Hondureño de Seguridad Social and Hospital Escuela, Tegucigalpa, Honduras (IHSS/HE-Honduras); Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, Mexico City, Mexico (INCMNSZ-Mexico); and Instituto de Medicina Tropical Alexander von Humboldt, Lima, Perú (IMTAvH-Peru).

Between June 2012 and December 2013, 6476 questionnaires were completed by 3586 individuals. If a patient responded to more than one questionnaire, the most recent questionnaire was selected for analysis. cART data was extracted from the CCASAnet database and 219 individuals who were not on cART in the 7 days prior to completing the questionnaire were excluded, as well as 15 individuals for whom there was no information regarding missed cART doses, and 9 individuals for whom alcohol and drug use information (marijuana, cocaine, crack, heroin) was missing. The final sample comprised 3343 individuals.

Variables and Definitions

Demographic information included sex, age, and CCASAnet site. Clinical variables included transmission mode (i.e., the probable route of HIV infection) categorized as heterosexual, homosexual, IDU, or other/unknown (bisexual was categorized as homosexual for men, and heterosexual for women); the first available CD4 cell count since entry in the cohort; time on ART (in years); and undetectable viral load (defined as <50 copies/mL) measured within 10 days [19] of the RST questionnaire.

The responses to RST questions regarding the frequency of alcohol, marijuana, cocaine, crack and heroin use in the previous 7 days (i.e. In the last week (7 days), how many times did you use any of these substances? (a) marijuana, (b) cocaine, (c) crack, (d) amphetamines/methamphetamines, (e) heroin) were dichotomized as ‘yes’ (if the substance was used one or more times) and ‘no’ (if the substance use was not reported). Substance use was then classified into four mutually exclusive categories, as follows:

  • None no alcohol or illicit drug use was reported;

  • Alcohol use only one or more alcoholic drinks were reported, but no illicit drug use;

  • Illicit drug use only any reported use of marijuana, cocaine, crack, or heroin, but no alcohol use;

  • Alcohol and illicit drug use reported both alcohol and illicit drug use.

The main outcome measured was the report of any missed cART doses in the 7-day recall period.

Ethical Issues

The study was approved by institutional review boards at each participating site and at Vanderbilt University Medical Center. Written informed consent was required and obtained for participants at the IMTAvH site in Lima, Perú.

Statistical Analysis

The prevalence of alcohol use only, illicit drug use only, alcohol and drug use, as well as the prevalence of use of each illicit drug is presented overall and by site. Among individuals who reported any alcohol use, the median number of alcoholic drinks was calculated. Substance use, demographics, and clinical variables were compared between individuals reporting missed cART doses and those reporting no missed doses. Chi square tests were used for categorical variables and Kruskal–Wallis tests were used for continuous variables.

Logistic regression was performed to assess the association between substance use (considering the four groups: no substance use, alcohol use only, illicit drug use only, and alcohol + illicit drug use) and missed cART doses, controlling for demographics and clinical variables. Measures of association were adjusted odds ratios (AORs) and 95 % confidence intervals (95 % CI). Analyses were conducted in SAS 9.3 (SAS Institute Inc., Cary, NC) and scripts are posted at http://biostat.mc.vanderbilt.edu/ArchivedAnalyses.

Results

Among the 3343 patients included in the study, alcohol was the most frequently used substance reported at all sites. Overall, 29.3 % of the sample reported having at least one alcoholic drink in the 7-day recall period. Among alcohol users, the median (interquartile range [IQR]) number of drinks was three [2–7]. Five percent of the sample reported using an illicit drug in the past 7 days; marijuana was the most frequent (4.1 %), followed by cocaine (1.3 %) and crack (0.3 %). Two individuals reported heroin use. Methamphetamine use was not reported. The prevalence of substance use varied substantially across sites, as shown in Table 1. Only 9.7 % of participants at the site in Mexico had used alcohol in the past 7 days whereas 53.2 % of participants had used alcohol at the site in Chile. Illicit drug use was most common at the site in Argentina.

Table 1 Substance use in a 7-day recall period stratified by CCASAnet site, 2012–2013
Full size table

Most of the study population was male (76.2 %) and median age was 41 years (IQR = 34–49); 65 % were between 30 and 49 years at the time of the questionnaire. Approximately 17 % (n = 570) reported missing at least one cART dose in the 7-day recall period (Table 2). The proportions of men in each category of reported substance use were: 71.6 % among those with no substance use, 85.5 % among those with alcohol use alone, 90.6 % among those with NIDU alone, and 92.2 % among those with both alcohol use and NIDU (p < 0.001). The median age (IQR) was 42 years (35–50) for those who did not use substances, 41 years (34–49) for alcohol-only users, 38 years (31–46) for NIDU-only, and 36 years (32–45) for those who reported both alcohol use and NIDU (p < 0.001).

Table 2 Demographic and clinical characteristics stratified by missing cART doses in a 7-day recall period at CCASAnet sites, 2012–2013
Full size table

In univariate analyses, substance use, age, site, and transmission mode were associated with missed cART doses (Table 3). In adjusted analyses, substance use remained strongly associated with self-reported missed cART doses (p < 0.001). Compared to no substance use, patients reporting alcohol use (AOR = 2.5, 95 % CI: 2.0–3.1), any illicit drug use (AOR = 3.6, 95 % CI: 2.0–6.3), and both alcohol and drug use (AOR = 5.0, 95 % CI: 3.2–7.8) had higher odds of missing cART doses, as shown in Table 3. Study site (AOR = 1.87, 95 % CI = 1.17–3.01 for IHSS/HE-Honduras vs. FH-Argentina; AOR = 0.08, 95 % CI = 0.04–0.16 for INCMNSZ-Mexico vs. FH-Argentina), probable route of HIV transmission via IDU compared to heterosexual (AOR = 2.46, 95 % CI = 1.04–5.83), and younger age (each 10 year increase in age was associated with an AOR 0.88 in the chance of missing ART) were also associated with missed cART doses in the adjusted analysis. There was no evidence that the association between substance use and adherence differed by sex (interaction p value = 0.83).

Table 3 Factors associated with missing cART doses in 7-day recall period after logistic regression at CCASAnet sites, 2012–2013
Full size table

Viral load was not consistently measured at the time of the questionnaire because of clinic logistics. However, in a subset of patients who had a viral load measurement within 10 days of the study questionnaire (n = 480), self-reported missing of cART doses was associated with lower odds of an undetectable viral load (OR = 0.44, 95 % CI: 0.27–0.73, p = 0.001).

Discussion

The concurrent use of alcohol and other drugs has been associated with increased health and social problems among both PLWHA [2, 20] and people without HIV [21]. Despite these well known relationships, ours is the first study to evaluate alcohol use and NIDU in the context of cART adherence among PLWHA in care at diverse clinics in Latin America. The prevalence of self-reported substance use varied across sites, but at all sites the most frequently used substances were alcohol, marijuana, and cocaine. Self-reported use of alcohol or illicit drugs alone, and in combination, was associated with an increased odds of missing cART doses in our cohort, with persons reporting use of both alcohol and drugs having the highest likelihood of non-adherence.

Overall, the prevalence of marijuana, cocaine, and crack use was substantially lower than that reported in studies among PLWHA in other settings. These differences may be related to the timeframe utilized to measure substance use (e.g., use in the last 30 days [22] or the last 3 months [9]). Nonetheless, a study that used the same RST to evaluate substance use in a southeastern United States cohort found 12.1 and 3.3 % prevalence of marijuana and cocaine (or crack) use, respectively [19]. The lower prevalence observed in our cohort may reflect the variable distribution of substance use across geographical areas, but may also indicate that substance users within the studied regions are less likely to access health services. For example, in Brazil, there has been a decrease in injected cocaine use and an increase in crack-cocaine use over the past decade [23, 24] (with the prevalence of smoked cocaine use in 2012 estimated at 0.8 % in the general population [25]), and crack-cocaine users reported a low utilization of social and health services [26]. Crack use is also known to be associated with an increased risk of HIV infection among inner-city young adults in the United States [27], and nearly one-third of PLWHA who use crack have reported never accessing HIV care in other settings [28, 29]. Failure to fully engage substance users, including crack cocaine users, in the HIV continuum of care may jeopardize the individual health and population benefits of early cART initiation.

Regarding alcohol use, other studies have demonstrated that PLWHA with co-morbid ARD have a threefold increased risk of hospitalization compared with either HIV-negative or HIV-infected persons without ARD [30]. PLWHA may not perceive the medical risks associated with alcohol consumption [31] and ARDs often are underdiagnosed in health care settings [32]. As nearly 30 % of individuals included in our study reported having at least one alcoholic drink in the 7-day recall period, there is an important need for screening, diagnosing and, possibly, treating alcohol abuse and dependence in this population.

With respect to our primary study outcome, the lifesaving benefits of cART are strongly dependent on treatment adherence [33, 34]. A meta-analysis done in 2011 including studies from 20 countries estimated that only 62 % of people on cART reported adherence ≥90 % [35] and that the average proportion of adherence was higher in developing countries when compared to developed ones. In our study, self-reported missed cART doses in a 7-day recall period varied widely across CCASAnet sites, ranging from 1.6 % in Mexico to 33.0 % in Honduras. Although there are several methods for measuring adherence, and self-reported measures may overestimate the use of medication compared to microelectronic devices [7, 36], our study shows an association between self-reported missed cART doses and a lower probability of viral suppression. These findings further support the use of the RST as a reliable measure of adherence that may be incorporated into routine clinical care [19].

Though the diagnosis of substance use disorders may be challenging in busy clinical settings, the RST used in this study was a helpful screening tool for NIDU at CCASAnet sites. Beyond treatment that addresses substance use directly, recent work has evaluated interventions which may improve adherence among non-adherent substance users who have HIV [37]. Among these interventions, medication assisted therapy (mainly methadone substitution therapy for opioid dependence) and directly administered antiretroviral therapy were shown to improve short-term viral load and CD4 measures. More research is needed to evaluate long-term outcomes after these programs are implemented, including the delivery of behavioral interventions and adherence monitoring through the use of mobile technologies (such as cell phones and the internet) [37]. Considering the profile of substance use found in CCASAnet, alternative approaches that address NIDU and substances other than opioids are needed in Latin America.

Other factors associated with missing cART doses in a 7-day recall period in our study were age and transmission mode. Older individuals were less likely to report missing cART doses. This is in accordance with international studies that have shown worse adherence among young people [38–40], reinforcing the need to retain these individuals in care [41] and to design age-appropriate interventions to improve adherence. Although the number of individuals reporting IDU as possible route of HIV infection was low in our cohort (n = 30), they had an increased chance of missing cART doses compared to those reporting heterosexual HIV transmission, contrary to prior findings from other studies [14]. It is also important to note that CD4 at enrollment and time on cART were not associated with adherence in our study. The low median baseline CD4 among individuals in our study, whether missing cART doses or not (234 and 227 cells/mm3, respectively), may indicate late entry into care and reinforces the need for improved efforts towards early testing and linkage to care in the region.

Study strengths include the large number of patients recruited in low- and middle-income countries. To our knowledge, this is the first multisite study of substance use and adherence in Latin America. Patients were asked to recall their substance use and adherence over a 7-day period, but the data collection at each site was prospectively obtained among actively followed CCASAnet patients. The prospective nature of our data collection allowed us to capture active drug use.

This study had some limitations. First, the sample was non-probabilistic and results may therefore not be generalizable to the entire CCASAnet cohort, nor to the general population of PLWHA in Latin America. Second, given its cross-sectional nature, causality cannot be inferred within our study, and longitudinal studies evaluating how active substance use impacts patient adherence in LMIC are necessary. Third, other variables that are likely related to adherence, such as education and psychiatric comorbidities, were not measured. Fourth, although the prevalence of active use may be underestimated through use of the RST as a screening tool in this cross-sectional assessment, it is possible that more individuals who actively use drugs could be identified with systematic incorporation of the RST into routine follow-up visits. Further research is necessary to evaluate the predictive value of the RST to diagnose substance use disorders, but its implementation in routine care is an important step towards detecting the need of specific care delivery for substance users living with HIV. Further studies are also necessary to disentangle the complex scenario of substance use patterns (such as dependency, or binge drinking, for instance) and adherence among PLWHA in Latin America, as well as investigating other substances that may be used in some countries for cultural reasons.

In conclusion, implementation of the RST was feasible and useful for screening for substance use and missed cART doses in diverse Latin American settings. Although there were differences in the prevalence of substance use across sites, results from our analyses support the need for screening and interventions targeted toward alcohol and illicit drug use to improve cART adherence and, ultimately, HIV outcomes among PLWHA in Latin America.