Improving mental health among people living with HIV: a review of intervention trials in low- and middle-income countries

People living with HIV (PLWH) experience greater psychological distress than the general population. Evidence from high-incomes countries suggests that psychological interventions for PLWH can improve mental health symptoms, quality of life, and HIV care engagement. However, little is known about the effectiveness of mental health interventions for PLWH in low- and middle-income countries (LMICs), where the large majority of PLWH reside. This systematized review aims to synthesize findings from mental health intervention trials with PLWH in LMICs to inform the delivery of mental health services in these settings. A systematic search strategy was undertaken to identify peer-reviewed published papers of intervention trials addressing negative psychological states or disorders (e.g. depression, anxiety) among PLWH in LMIC settings. Search results were assessed against pre-established inclusion and exclusion criteria. Data from papers meeting criteria were extracted for synthesis. Twenty-six papers, published between 2000 and 2014, describing 22 unique interventions were identified. Trials were implemented in sub-Saharan Africa (n = 13), Asia (n = 7), and the Middle East (n = 2), and addressed mental health using a variety of approaches, including cognitive-behavioral (n = 18), family-level (n = 2), and pharmacological (n = 2) treatments. Four randomized controlled trials reported significant intervention effects in mental health outcomes, and 11 preliminary studies demonstrated promising findings. Among the limited mental health intervention trials with PLWH in LMICs, few demonstrated efficacy. Mental health interventions for PLWH in LMICs must be further developed and adapted for resource-limited settings to improve effectiveness.


Introduction
Low-and middle-income countries (LMICs) bear a disproportionate burden of the world's HIV infections, with over 85% of the world's 35 million HIV cases located in LMICs (UN Joint Programme on HIV/ AIDS, 2014). Studies across multiple settings have consistently observed that people living with HIV (PLWH) experience greater psychological distress, such as depression and anxiety, as compared with the general population (Bing et al. 2001;Ciesla & Roberts, 2001). Evidence from LMICs, though limited, has confirmed high rates of mental disorders and psychological distress among PLWH (Breuer et al. 2011;Chibanda et al. 2014). Mental disorders in LMICs must be addressed due to their impact on the lives of PLWH, as well as HIV-related outcomes at an individual and population level (Hartzell et al. 2008).
Mental disorders and psychological distress not only compromise overall well-being and quality of life among PLWH (Bing et al. 2000), but also impact individuals' ability to engage effectively with HIV care, including their adherence to antiretroviral therapy (ART) (Collins et al. 2006;Mayston et al. 2012;Uthman et al. 2014). The ability to halt disease progression and achieve full viral suppression, which requires proper care engagement, is a primary predictor of HIV infectiousness (Cohen et al. 2011). Poor HIV care engagement attributable to unaddressed mental health needs among PLWH may thus hamper proposed efforts to use 'treatment as prevention' to curb the spread of the HIV epidemic Gupta et al. 2014). Additionally, mental disorders and psychological distress may be associated with HIV risk behaviors such as substance abuse, multiple sexual partners and unprotected sexual intercourse (Crepaz & Marks, 2001), further contributing to the forward transmission of HIV (Senn et al. 2010). Taken together, addressing mental health among PLWH appears to be a critical component of HIV treatment and prevention, and should be considered as part of population-level approaches to prevent HIV transmission, particularly in LMICs where the burden of HIV and its associated morbidities is high.
An emerging body of evidence, mostly from highincome settings, suggests that psychological interventions, primarily focused on treating depression and anxiety, can improve the mental health of PLWH (Crepaz et al. 2008;Brown & Vanable, 2011;Clucas et al. 2011;Harding et al. 2011;Sherr et al. 2011;Seedat, 2012;Spies et al. 2013;Wu & Li, 2013). Among the various treatment modalities, cognitive-behavioral interventions (CBIs) have received the most attention, with skills training and stress management CBIs that include 10 or more sessions demonstrating the greatest improvement in mental health symptoms. These interventions have been found to be equally if not more effective than pharmacological treatments Spies et al. 2013), though it has been suggested that pharmacological management can be beneficial as an adjunct treatment or combined with psychological approaches ). In addition, CBIs have been found not only to improve mental health symptoms in PLWH, but also to impact HIV-related clinical outcomes, including CD4 counts (Crepaz et al. 2008). Furthermore, a limited number of intervention studies have begun to examine the impact of mental health treatment on care engagement and risk behaviors in PLWH. For example, depression treatment has been found to enhance ART adherence (Sin & DiMatteo, 2014), and coping interventions to reduce traumatic stress have been found to decrease substance use ) and sexual risk behaviors (Sikkema et al. 2008). There is further evidence that psychological interventions can be delivered in community settings, which has the potential for broader reach (Wu & Li, 2013). Intervention studies published subsequent to existing reviews suggest an emerging focus on aging populations (Heckman et al. 2013), increased emphasis on treating traumatic stress and post-traumatic stress disorder (PTSD) (Pacella et al. 2012;Sikkema et al. 2013), and the use of telephone and web-based modalities (Hersch et al. 2013;Himelhoch et al. 2013;Drozd et al. 2014), as well as alternative therapeutic approaches that draw on mindfulness (Gayner et al. 2012;Gonzalez-Garcia et al. 2014) and expressive writing paradigms (Ironson et al. 2013;Carrico et al. 2015).
Despite promising intervention efforts and related evidence, most of what has been appraised to date has emerged from high-income countries (HICs). Although previous reviews have included a limited number of studies from LMICs (Crepaz et al. 2008;Clucas et al. 2011;Harding et al. 2011;Sherr et al. 2011;Seedat, 2012;Spies et al. 2013), thus far, no reviews have focused on interventions that broadly address mental health for PLWH in these unique settings. Despite the dual burdens of HIV and mental disorders in LMICs, and evidence of their synergistic negative effects, strategies to address the mental health needs of PLWH in LMICs have received only limited attention (Mayston et al. 2012;Chibanda et al. 2014). Mental disorders in these settings often are untreated due to a lack of behavioral and pharmacological treatment opportunities (Kakuma et al. 2011;WHO, 2001). This treatment gap has implications for both the well-being of PLWH as well as the effectiveness of national HIV treatment programs (Mayston et al. 2012).
There is a need to take stock of existing efforts to respond to mental health distress challenges among PLWH in LMICs, so as to inform the development and scale up of appropriate mental health services in these countries. The goal of this review paper is to synthesize findings from mental health intervention trials for PLWH in LMICs. This information will extend our knowledge of mental health interventions for PLWH beyond those that have been developed and tested in HICs, and will illuminate avenues for future intervention development, testing, and delivery in LMIC settings where there is greatest need.

Inclusion criteria
Studies were included in this review if they met the following criteria: (1) Described a trial evaluating a mental health intervention (e.g. CBT, coping, cognitive functioning, and pharmacological) among PLWH, (2) Intervention was implemented in either a LMIC, as defined by the World Bank (The World Bank Group, 2015), or one of the emerging-economy Brazil, Russia, India, China, and South Africa (BRICS) countries (), (3) Trial assessed at least one negative psychological state or disorder (e.g. depression and anxiety), and (4) Outcomes were assessed pre-and post-intervention.

Exclusion criteria
Studies with a heterogeneous (HIV-positive and HIV-negative) sample were excluded if mental health outcomes were not presented separately for the HIV-positive population. Trials with only substance use outcomes were not included in this review. Systematic and non-systematic review articles, as well as studies unavailable in English, qualitative exploratory studies, and studies not published in a peer-reviewed journal were also excluded from this review.

Search strategy
PubMed, EMBASE, and PsychInfo were searched between 9 and 11 November 2014. Limits to time period were not applied to the search. Standardized search terms and key words related to the constructs of (a) HIV or AIDS, (b) mental health, (c) intervention, and (d) LMIC/BRICS were used in all databases. For example, within PubMed, terms used to capture the construct of mental health included the following: mental health, mental*, depress*, anxi*, trauma*, PTSD, psycholog*, coping, stress, and psychiat*. Search terms for LMIC/BRICS were derived from the World Bank's classification of low-income, lowermiddle income, and upper-middle income economies (The World Bank Group, 2015). Conducting the search was a multi-step process. First, separate searches were conducted using HIV/AIDS, mental health, intervention, and LMIC/BRICs search terms, respectively. In the final step, results from these four separate searches were combined using 'AND' terms to capture manuscripts that possibly met study inclusion criteria. Where possible, studies with 'orphan' in the title were eliminated from the search using the term 'HIV or AIDS not Orphan' so as to more efficiently exclude studies focusing on HIV-impacted orphans who were not necessarily HIV-infected. Filters for the above constructs were applied within each database to restrict the search in the following ways: (1) HIV or AIDS terms were restricted to title only, (2) Mental health terms were restricted to title or abstract, (3) Intervention terms were restricted to title only, and (4) LMIC/BRICS terms were restricted to title, abstract or topic.

Study selection and data abstraction
A list of all titles and abstracts were considered independently by two researchers, eliminating those studies that did not appear to meet study inclusion criteria; the full study team then reached consensus on articles to be assessed for eligibility. The full text of all relevant articles were then independently reviewed by two researchers to determine inclusion, with >85% agreement between researchers. Discrepancies about the remaining studies were reconciled through discussion with the full study team. Data display matrices were used to extract data from included studies. Data extracted from studies were as follows: author, year, title, date of study, city/country, sample characteristics (age, sex, HIV-status, sample size of intervention and control conditions, mental health inclusion criteria), intervention characteristics (name, level, components, duration, deliverer), study design, evaluation design (control or comparison condition, follow-up, retention), outcome measures, and relevant findings.

Search results
Initial database searches yielded 454 records, and 369 unique records remained after elimination of duplicates across databases. Screening of titles and abstracts resulted in 44 articles. The full text of these 44 articles was reviewed, which yielded 25 articles (describing 21 unique studies) that met the inclusion criteria for this review (see Fig. 1). One additional article was identified through examination of the reference lists of relevant systematic and non-systematic reviews (Field & Kruger, 2008), for a total of 22 unique intervention studies included in this review. All included studies were published in peer-reviewed journals between 2000 and 2014. The summaries of the 10 randomized control trials (Table 1) and the 12 pilot or feasibility trials (Table 2) are presented separately.

Participants
A total of 2893 participants were reported across all trials, of which 1664 individuals participated in the experimental intervention condition. A median sample size of 233.5 at baseline (range 67-1200) was reported across all trials. All studies included HIV-positive participants, yet study samples were diverse. Four RCTs were designed for implementation within families Boivin et al. 2013;Eloff et al. 2014); no studies included only child or adolescent participants. Of the six remaining RCTs, four were designed for implementation with HIV-positive men and women (Olley, 2006;Peltzer et al. 2012;SeyedAlinaghi et al. 2012;Eller et al. 2013), and two were for women only Richter et al. 2014;Rotheram-Borus et al. 2014). Only one study had a specified mental health inclusion criterion, depressive symptoms (Eller et al. 2013). None of the RCTs utilized a mental disorder as inclusion criteria.

Study design
One RCT  randomized by cluster; all other RCTs randomized individual participants to the intervention or control condition. In all studies, outcomes were assessed post-intervention (median retention rate reported of 87.5%, range 57-98%), with most trials also reporting longer-term follow-up assessments between 2 and 18 months (median retention rate reported final follow-up of 82%, range 24-97%). (Note: outcomes in Tables 1 and 2 correspond to postintervention measurements, unless otherwise noted).

Intervention content and delivery method
The interventions reported in these studies had diverse characteristics. Most interventions were designed to be implemented with groups SeyedAlinaghi et al. 2012;Kaaya et al. 2013;  (2) adhering to PMTCT tasks; (3) establishing healthy daily routines; (4) infant feeding methods (5) economic support (6) social support; and (7)

Outcome measures
Mental health outcomes were reported in all studies (Table 1); all primary mental health outcomes were measured using standardized psychological symptom scales. None of the studies reported including a diagnosis of mental disorder. Outcomes related to depression were reported in all but two studies , SeyedAlinaghi et al. 2012). In the latter, composite mental health outcome measures were utilized. See Table 1 for mental health and related psychological assessments utilized across studies. Six studies (Olley, 2006;Li et al. 2010Li et al. , 2012SeyedAlinaghi et al. 2012;Boivin et al. 2013;Kaaya et al. 2013;Eloff et al. 2014) noted attention to cultural adaptation of measures or validation of measures in the country in which the study took place.

Intervention effects
Four of the 10 studies reported a significant betweencondition intervention effect for PLWH (Olley, 2006;Li et al. 2010Li et al. , 2011Li et al. , 2012Li et al. , 2014Richter et al. 2014;Rotheram-Borus et al. 2014). Further, Boivin et al. (2013) and Eloff et al. (2014), both family studies with HIV-infected caregivers and HIV-uninfected children, found significant effects in the child sample but nonsignificant effects in the HIV-infected adult caregivers; notably, these studies were designed to improve neurocognitive ) and resiliency ) outcomes in children. Four studies SeyedAlinaghi et al. 2012;Eller et al. 2013;Kaaya et al. 2013) resulted in positive but non-significant intervention effects on mental health outcomes.

Pilot, feasibility, and quasi-experimental studies
In addition to the 10 RCT studies described in detail above, 12 additional preliminary studies were identified that provide future directions for mental health intervention with PLWH in LMIC settings. These studies, details shown in Table 2, were either described by the authors as pilot trials (many with RCT methods) or were not conducted using an RCT design.

Study design
Five of the pilot studies randomized individual participants (Molassiotis et al. 2002;Chan et al. 2005;Bhana et al. 2014;Petersen et al. 2014;Pence et al. 2014) and one study randomized clusters (Nyamathi et al. 2012) to the intervention or control condition. Three studies were quasi-experimental, and did not randomize participants to condition (Jirapaet, 2000;Field & Kruger, 2008;Mundell et al. 2011). Three studies were preexperimental with a one-group, pre-posttest design (Adams et al. 2012a, Pence et al. 2014Yu et al. 2014).
Most of the studies (8 of 12) measured effects at posttest only and utilized sample sizes appropriate for pilot studies.

Control/comparison condition
Most studies (eight) used a comparison or control condition, with six of the 12 studies randomizing participants to condition (Molassiotis et al. 2002;Chan et al. 2005;Nyamathi et al. 2012Nyamathi et al. , 2013Ravaei et al. 2013;Bhana et al. 2014;Petersen et al. 2014). As shown in Table 2, these included, for example, standard of care, wait list control, and counseling as needed.

Outcome measures
All studies assessed mental health outcomes using various standardized psychological scales (Table 2). Depressive symptomatology was the most commonly assessed mental health outcome; one study (Pence et al. 2014) used major depressive disorder based on a symptom scale and physician assessment.

Preliminary intervention effects
Eleven of the 12 pilot, feasibility, or quasi-experimental studies demonstrated promising effects related to improvements in mental health (Jirapaet, 2000;Molassiotis et al. 2002;Chan et al. 2005;Field & Kruger, 2008;Adams et al. 2012a;Nyamathi et al. 2012Nyamathi et al. , 2013Ravaei et al. 2013;Pence et al. 2014;Petersen et al. 2014;Yu et al. 2014), or behavioral (Adams et al. 2012a;Bhana et al. 2014) outcomes. One study (Mundell et al. 2011) demonstrated mixed effects, with improvements in active coping and self-esteem in the intervention group, but not in depression or social support.

Discussion
There is an urgent need to address mental health in the context of HIV/AIDS in LMICs, which bear the brunt of global HIV infections. Although previous reviews (Crepaz et al. 2008;Brown & Vanable, 2011;Clucas et al. 2011;Harding et al. 2011;Sherr et al. 2011;Seedat, 2012;Spies et al. 2013;Wu & Li, 2013) have identified a large number of RCTs to improve mental health in high-income settings, they have only included a small number of intervention trials conducted in LMICs. The purpose of this review was to systematically identify intervention trials that have addressed mental health among PLWH in LMIC settings, and to synthesize the lessons learned from those studies. Our review identified 22 unique intervention studies, 10 of which were evaluated using rigorous RCT methodology. Although this systematic search included a range of terms for mental health, only intervention trials with depression, anxiety or overall psychiatric distress outcomes were identified. Despite the fact that only a small number of full scale trials provided evidence for improvements in mental health, the findings of the review point to opportunities for further research on interventions to address the mental health needs of PLWH in LMIC settings. The four RCT intervention trials in our review that demonstrated an impact on mental health primarily utilized a multi-component approach. Interestingly, the most robust outcomes, including over longer term follow up assessments, were found in community based trials Richter et al. 2014;Rotheram-Borus et al. 2014) that contextualized HIV/AIDS and mental health within family interactions or through peer support that addressed issues related to pregnancy and child outcomes. The inclusion of pilot studies and quasi-experimental trials in our review provided a framework for mental health interventions that are in the development phase and potentially proceeding to full scale trials. Although the study methods were acknowledged as less rigorous, many used RCT methodology albeit among small sample sizes, and all reported encouraging results using several innovative intervention approaches.
A number of studies, both RCTs and preliminary studies, were group-based interventions, utilizing a cognitive behavioral approach (including stress management and coping interventions), and often delivered in a task-shifting or task-sharing model with lay counselors or community health workers. Future directions explored in pilot trials included a measurement based care (MBC) stepped approach to antidepressant medication management (Trivedi et al. 2007;Adams et al. 2012b) that employed non-specialists to screen and monitor depressive symptoms, thereby supporting intervention at the clinic system level. These trials, as well as two others that evaluated psychotherapeutic approaches, were the only studies that used an inclusion criterion for mental disorder or depressive symptoms. Several of these pilot trials included an intervention focus or secondary outcome measure of adherence to ART. These study methods and intervention approaches point the way forward for mental health intervention trials with PLWH in LMICs, including integration with HIV care and treatment.
The studies we identified had methodological limitations similar to those documented in prior reviews of trials conducted primarily in HICs (Crepaz et al. 2008;Clucas et al. 2011;Sherr et al. 2011;Seedat, 2012;Spies et al. 2013). Of key importance only a few studies used mental disorder or above threshold symptom levels as an inclusion criterion for trial selection. This suggests that either the interventions may not have targeted PLWH experiencing significant psychological distress or that symptom levels were subthreshold, making it difficult to demonstrate an intervention effect. In addition, a limited number of studies provided information on the cultural appropriateness of the intervention or the adaptation and validation of the mental health measures utilized (Bass et al. 2007). Several other secondary limitations of the methodologies used in our included studies should be noted. Even among the RCTs, the majority of the intervention trials assessed either immediate post-or shortterm intervention effects, with longer term follow up assessments needed to determine sustainability of intervention effects. Most studies provided a general description of the intervention approach, but a detailed description of intervention components and fidelity to them in delivery would enhance the understanding of intervention outcome findings. Finally, intervention trials using non-specialists to deliver mental health interventions (i.e. a task-sharing model) did not provide adequate information on the training and supervision of the providers (Patel et al. 2007).
Despite the methodological limitations of the studies included in this review, it is encouraging to see an increasing number of mental health intervention trials for PLWH conducted in LMICs, which span a wide range of populations, countries, and intervention approaches. Additional lessons can be drawn from efficacious interventions for depression treatment in LMICs, (Bolton et al. 2003;Patel et al. 2007Patel et al. , 2009Rahman et al. 2008), even if not specific to PLWH. Such interventions have been integrated into routine health care, adapted to local cultural context, and implemented by non-specialists. The intervention approaches for treating depression varied, but reflect approaches identified in this review for PLWH, including a collaborative stepped care approach (Patel et al. 2007(Patel et al. , 2009, home-based individual CBT (Rahman et al. 2008), and interpersonal group therapy (Bolton et al. 2003). Although mental disorders in HIV care settings in LMICs often go undiagnosed due to lack of screening protocols (Breuer et al. 2014;Tsai, 2014), these studies suggest that integration of mental health screening and intervention into health care settings would be an effective approach for improving mental health among PLWH. Two of the preliminary studies we identified support the feasibility of a MBC approach for antidepressant medication in the HIV care setting (Adams et al. 2012b;Pence et al. 2014), similar to the stepped approach in the MANashanti Sudhar Shodh (MANAS) trial (Patel et al. 2007(Patel et al. , 2009), yet more medication-based or combined medication and therapy based trials are needed.
CBIs, including problem solving, skills training, and stress management, commonly used in both HIC and LMIC, support the use of cognitive-behavioral treatments as a key mental health intervention approach. However, there is a need for cultural adaptation and tailoring mental health idioms to the local context, which may preclude the ready transplant of existing CBIs for PLWH. Research trials are still needed to test factors such as required intervention length for feasibility, effectiveness and maintenance of effect, and delivery by non-specialists. There is a need for task-sharing and interventions that can be delivered by non-specialists that are brief and scalable, while providing supervision and fidelity monitoring.
While mental health interventions for PLWH should draw upon the broader evidence of efficacious mental health interventions, they must also remain attuned to issues that are unique for a population living with HIV. These include potential barriers related to HIV-related stigma (Skinner & Mfecane, 2004), substance abuse co-morbidities (Gonzalez et al. 2011;Kader et al. 2012), and other disorders prevalent among PLWH such as PTSD (Machtinger et al. 2012). Multilevel system-strengthening approaches that integrate mental health care into HIV care and prevention within health care and community based organizations has been recommended (Joska & Sorsdahl, 2012;Lund et al. 2014). Areas for future research in LMICs include the integration of mental health treatment with adherence, HIV care engagement, and HIV prevention Chibanda et al. 2014). Thus, one priority for future research is to conduct RCTs of mental health interventions that improve mental health and enhance HIV treatment and prevention.
The findings from this review of interventions to improve mental health among PLWH in LMICs also provide lessons learned and potential future directions to improve related efforts in HICs. Effective interventions in LMICs utilized family or multilevel interventions and were integrated within community based health careapproaches that contextualize mental health and provide an opportunity to address comorbidities. These approaches, in combination with addressing structural barriers to care such as poverty, health care access, and mental health care policies, are also relevant in HICs, especially in settings where health disparities clearly exist. Given the limited mental health resources in LMICs, including the absence of specialists trained in mental health treatment, the LMIC interventions incorporated approaches delivered by non-specialists. This task-sharing approach is also relevant and should be evaluated in HIC settings, particularly in communities and settings where access to mental health care is limited. Research methodologies related to monitoring intervention fidelity and supervision of non-specialists could enhance our understanding of key elements of effective interventions in both settings. Lastly, this review of interventions in LMIC settings offers lessons for the adaptation of available evidence-based interventions with attention to language, culture, and literacy, as well as feasibility of intervention length. These factors may also impact the effectiveness of mental health intervention for PLWH in HICs, and if better addressed, could improve the outcomes in higher income settings where disparities and cultural differences exist.
Our search strategy, while systematized, cannot guarantee the identification of all interventions to improve mental health among PLWH in LMICs, and omission of related intervention research is possible and may have influenced our conclusions. Unlike prior reviews Seedat, 2012;Spies et al. 2013), we intentionally cast a wide net that resulted in the inclusion of studies ranging from multilevel family and community based approaches to a group adherence intervention that also assessed the impact on depression. In addition, we included trials that reported mental health as secondary outcomes (4 of 22 studies). Although this approach broadly defined mental health and supports the importance of addressing mental health in context, a limited number of RCTs were identified, and only a portion of these trials provided evidence for the efficacy of the mental health intervention, even when looking for post-only mental health outcomes. Despite these potential shortcomings, this review provides an overview of the body of evidence available on mental health treatment of PLWH on LMICs, and offers suggestions for the path forward for understanding and addressing these needs.

Conclusion
There is a paucity of empirical data investigating the effectiveness of interventions for mental disorders and psychological distress among PLWH in LMICs. The available data are restricted to several RCTs with widely varied approaches and methodology; and to several smaller pilot, and innovative studies. It is not possible therefore to describe from such a review, the nature, content, and delivery of an ideal intervention. Key issues such as adapting the intervention to suit local culture, language, and resource-limitations are typically not addressed. We are some way from being able to define an ideal intervention, and so a framework for building interventions of this kind is a desirable next step.