A mega-aggregation framework synthesis of the barriers and facilitators to linkage, adherence to ART and retention in care among people living with HIV

Although access to HIV care has improved significantly over the past few years, people living with HIV (PLHIV) still face numerous challenges when it comes to initiating care and staying on treatment. Human immunodeficiency virus (HIV) represents one of the greatest global public health challenge in history, and since the beginning of the epidemic, approximately 78 million people worldwide have been infected with HIV and 35 million people have died [12]. The Joint United Nations Programme on HIV/acquired immunodeficiency syndrome (AIDS) (UNAIDS) set the global 90-90-90-target to combat HIV infection by 2020 [13]. The goal aims for 90% of all people to know their HIV status, of those who test positive, 90% should be linked to care, and of those being adherent to care, 90% will have achieved viral suppression. Nearly 37 million people were estimated to be living with HIV worldwide in 2017; however, only 60% were aware of their HIV status and only 49% of those who knew their status were accessing treatment [14]. The HIV burden varies considerably between countries, with regions in Africa having the highest HIV prevalence with HIV being the leading cause of death in South Africa [12]. To date there is no known cure for AIDS. However, being linked to care and adhering to antiretroviral therapy (ART) has been shown to improve quality of life for PLHIV, and in most cases provided similar life expectancy periods for those without HIV [15, 16].

PLHIV continue to be challenged by the complexities related to being HIV positive and integrating ART treatment and care into their daily lives. Unsuccessful interventions and the target driven 90-90-90 goals have increased researchers’ commitment to understanding the human experience of living with HIV and engaging in the HIV treatment cascade. Some reviews have focused only on prevention, adherence, linkage to care or retention in care. This can lead to an abundance of research in one area on the cascade and neglect of others. With the growing body of existing systematic reviews [17, 18], there is no coherent sense of what is already known across populations and settings, and there is uncertainty about the quality of the existing evidence. This can make it hard for policy makers and practitioners to make evidence-informed decisions. However, the increase in QES makes research innovation in the synthesis of qualitative review-level evidence possible. The overarching aim of this study is to debut the first application of mega-aggregation framework synthesis to qualitative systematic review-level evidence. This method of summarising QES in an overview will be applied to review-level evidence of barriers and facilitators of linkage to care, adherence to ART and retention in care for PLHIV in low-, middle- and high-income countries.

Overviews of reviews aim to provide a single synthesis or summary from multiple systematic reviews [19]. QES often have a theoretical underpinning to understand findings and interpret meaning. Qualitative research is usually positioned in the interpretive or critical-realist paradigm. Another approach to QES is meta-aggregation, which is based on the philosophy of pragmatism [20, 21] and users of this method aim for immediate usability of the review findings. This study proposes the application of mega-aggregation, which, unlike mega-ethnography, does not focus on the generation of new theory nor aims to provide deepened conceptual interpretations of findings [8], but rather aims to provide an overview of the existing evidence, identify evidence gaps and make recommendations for future research or immediate action [21, 22].

In the context of the pragmatic stance and the anticipated large number of existing systematic reviews, a predetermined theoretical framework [23] with broad categories was selected to guide the aggregation and synthesis within this overview, which built on the steps in methods development for conducting overviews [24], QES [25, 26], systematic review synthesis [19, 27, 28], meta-aggregation [9, 20, 21] and framework synthesis [29, 30]. The novel approach of mega-aggregation framework synthesis was developed and utilised to identify evidence gaps and to inform future research from the evidence collated within included systematic reviews. The mega-aggregative framework synthesis approach consists of eight distinct steps (Fig. 1). The steps are as follows: (1) identify a clearly defined review question and objectives, (2) identify a theoretical framework or model, (3) decide on criteria for considering reviews for inclusion, (4) conduct searching and screening, (5) conduct quality appraisal of the included studies (although some may prefer not too), (6) data extraction and categorisation, (7) present and synthesise the findings, and (8) transparent reporting.

The database search resulted in 2762 article citations and an additional seven reviews were identified through other sources (two within the reference lists of included reviews and five through other readings). After the removal of duplicates, 1921 citations were imported into Covidence and the title and abstracts were screened, resulting in 78 retrieved for full text review. Thirty-nine reviews were excluded (Additional file 7), most reviews did not fit the criteria of a systematic review (n = 10), did not contain qualitative primary studies or data (n = 4) or did not include the target population group (n = 1). We were unable to obtain the full texts for two reviews and are waiting on information from one author and have classified these three reviews as ‘awaiting assessment’. Three ongoing reviews or protocols were found in our search (Additional file 7). We included 33 [41‐73] systematic reviews in this overview. Figure 2 describes the flow of reviews through the different stages of this overview using the PRISMA flow diagram [74].

Included systematic reviews were published between 2006 to June 2018, peaking at 6 publications in 2018 (Fig. 3).

The included reviews (N = 33) synthesised primary studies that were conducted in both high-income countries and low- and middle-income countries with a large concentration of included primary studies being conducted in sub-Saharan Africa (Fig. 4). No reviews written in languages other than English were found.

Self-reported barriers and facilitators of 1,156,540 PLHIV (children and youth, and adults) are included in this overview. Some reviews included high-risk populations, such as pregnant and postpartum women, children and adolescents, commercial sex workers, men who have sex with men, transgender persons, prisoners, intravenous drug users and foreign nationals. Two reviews on children and adolescents included data from caregivers. Table 2 summarises the characteristics of included studies.

We applied the conceptual definitions of the outcomes as per the overview protocol and we found that 13 reviews addressed the outcome of linkage to ART, 29 addressed the outcome of adherence to ART and 11 addressed the outcome of retention in care (see Fig. 2). The method of synthesis of the reviews varied and included thematic analysis, thematic content analysis, content analysis, narrative synthesis, meta-synthesis and meta-aggregation. Details of data are extracted, and critical appraisal of each included review is available in Additional file 8.

Due to the different ways of reporting results in the included reviews, we discerned between two population groups in terms of age, children and youth, and adults.

We found overlap in the qualitative, quantitative and mixed methods primary studies included within the 33 systematic reviews (Additional file 9). Review authors used varying definitions of linkage to ART, adherence to ART and retention in care when considering studies for inclusion. One author may have used a primary study to synthesise evidence on linkage to ART and another author may have used the same study to synthesis evidence on adherence to ART. The primary studies within the systematic reviews were published between and 1995 and 2017. Of the 1153 primary studies in the systematic reviews, 826 were unique studies, of which 616 were included in only one review. We found that 139 of the studies were included in two reviews, forty-seven in three reviews, fourteen in four reviews, three in five, four in six, and one study was included in seven reviews and another across eight reviews.

We explored whether there was overlap in the search dates between the included reviews and found that most reviews searched between 2000 and 2013 with an average search period covering 13 years (Fig. 5). Eight studies [44, 45, 56, 58, 60, 65, 67, 71] conducted comprehensive searches up to a year before publication. One review [41], conducted as part of an online postgraduate degree programme, had very short search period of 6 months. We found considerable overlap in the search dates of included systematic reviews.

The methodological quality of the included systematic reviews varied across reviews. Details of the justifications for quality judgements are reported in Additional file 8. All but five reviews had clear research questions. Key methodological aspects that were appraised as good quality were the relevance of recommendations for policy, having a clear research question and relevant directives for future research. The key aspects that were assessed as poor for the included reviews were the sources used to search and the inclusion criteria of the reviews. Some reviews did not clearly report items, and as we were thus unable to make a judgement, we assessed them as ‘unclear’. Items concerning if the ‘process of data extraction was appropriate’ and if the ‘critical appraisal was conducted by two or more authors’ were mostly identified as ‘unclear’ across included reviews. We identified 110 (30.3%) items out a possible 363 as ‘unclear’, 70 (19.3%) items as ‘no’ and 183 (50.4%) of items as ‘yes’. Two reviews [62, 65] were rated as high quality, six reviews [41, 43, 46, 55, 61, 67] were rated as medium quality and 25 reviews [42, 44, 45, 47‐54, 56‐60, 63, 66, 68‐73, 75] were rated as low quality (Table 3).

We found 544 unique third-order concepts from the included systematic reviews and to retain the essence of the review authors’ interpretations, we extracted concepts verbatim. We then categorised and aggregated the evidence into the predetermined framework, namely, Kaufman’s Behaviour Change Model of HIV [23].

Several of the individual, interpersonal, health system and structural-level factors identified in this overview are well known and have been identified in previous literature [35, 75]. We identified key areas and population groups which were under-represented in the review literature and where syntheses of primary research may be needed to understand these factors better.

First, individual-level factors such as peoples’ fears, psychological distress, and beliefs about ART, and conflicting beliefs between cultural practices and medicine, were frequently reported; there was however little to no emphasis in qualitative reviews on the facilitators to mitigate these barriers in children or adult populations.

Second, interpersonal factors such as the dynamics of relationships within the household, and the fears, anxieties, experiences of disclosure and support from peers and support groups were identified both as facilitators and barriers to care; however, details about the complex relationship of these factors needs further exploration.

Third, models of care and service delivery practices adopted by health care centres were prominent themes: with long waiting times, the risk of unintentional disclosure and the negative treatment by health care workers reported as barriers. Skilled health care workers, targeted services for men and youth, and integration of care as facilitating care were identified. Further exploration of not only the healthcare workers, but also the clinic facility, both as a physical space and beliefs about the clinic, is needed.

Fourth, structural-level themes included transportation to the clinic, poverty and low-income levels linked to the lack of food. Emerging issues related to the sensorial experience of adherence, the physical environment and community structure has yet to be explored in qualitative systematic reviews related to HIV.

Fifth, although there is progress to achieving the 90-90-90 targets, this study identified many challenges experienced by PLHIV and the themes of self-reported barriers provide a rich overview of the interwoven complexities of the lives of PLHIV. Recent literature linking barriers and facilitators of linkage, adherence and retention in care to priority solutions propose differentiated service delivery, client-centred care, safe spaces for HIV treatment and care, adolescent friendly services, family centred services and advocacy for human rights, among others, to best serve the needs of PLHIV and reduce the burden on the health care systems. Investigation and scale up of interventions for PLHIV to promote their adherence must be intensified and barriers and facilitators even beyond the realm of interventions proposed including reminder devices, pill box organisers, reduced toxicity, mental health screening and fixed dosed combinations must be explored.

Sixth, this overview identifies the need for qualitative exploration of the agency of the material environment encountered by PLHIV, not included in existing reviews or in intervention research and will make valuable contributions to the body of knowledge. The exploration of agency goes beyond the identification and description of themes to exploring the understanding of the meaning ascribed to materials such as agency of the human body of PLHIV, the agency of matter such as money, transport vehicle, clinic environments, food, community structural factors and living spaces, context and agency of medications.

Lastly, this study identified the need for more high-quality review-level evidence for key populations, specifically, children and youth with perinatal HIV infections, as they were underrepresented in the data. Most included reviews were assessed as having low quality. Future qualitative systematic reviews in the field of HIV linkage, adherence and retention in care should follow a rigorous approach to minimise the risk of bias.

Through our worked application, we identified benefits of the mega-aggregative framework approach. First, the pragmatic philosophical underpinning promoted the aggregation of findings and a clear identification of where research is lacking, the quality of the research and the location of research across socio-economic statuses of countries. Some overviews only provide a synthesis without clear directives for policy and practice, and authors may see this as optional—leaving the reader to deduce or draw their own conclusions on the way forward. Using the mega-aggregative framework approach, the lines of action provide researchers and funders with clear guidance on where to direct funding and resources or to develop new research questions. It also provides practitioners with a comprehensive selection of the available evidence and the appraised quality of the evidence. A second benefit is that a mega-aggregation framework synthesis can be conducted in a relatively short amount of time. Other methods of synthesis can be iterative as review authors make sense of their own subjective interpretations. Using a framework with broad categories allows for linear working and promotes the comparing of findings between review authors. Using this method, the number of findings is counted and reported, with research gaps and recommendations for policy and practice being apparent immediately. Concrete time is then spent of the lines of action. This approach is feasible and can be especially attractive to novice researchers. A third benefit is the attention to the methodological quality of included systematic reviews; however, the benefits of the method proposed are not without its challenges or limitations. While there is ongoing debate [64, 65] about when, within which paradigmatic stance, and whether to conduct critical appraisal of included studies within a qualitative evidence synthesis, we included methodological appraisal as a step in the mega-aggregative approach. We found very few high-quality reviews examining self-reported barriers and facilitators to linkage, adherence to ART and retention in care for children and adults. In the results, we presented an aggregation of findings and the provided a separate paragraph detailing the findings of the high-quality included reviews. This does not imply that the primary studies included in the reviews are of low quality, as we only assessed the methodological quality of the systematic reviews. We used guidance from recent methodological papers [25, 26, 76‐78] in the field of critical appraisal to revise the JBI-SR Checklist [78] for our overview and to determine our decision rules for overall quality of the reviews. The appraisal of the reviews challenged us as authors to think reflexively about the domains that influenced the quality of the reviews and the items on the appraisal tool that explored these domains. As with overviews of qualitative systematic reviews, the area for method development in critical appraisals of QES is understudied and tools for critical appraisal of QES are still under development. Using the mega-aggregative framework approach can limit the author team in their creativity or conceptual modelling of new theories. The mega-aggregative framework approach uses an existing framework and pragmatic approach that can make the analysis process mechanical and sometimes force the data into predefined categories. When we refined the steps of the method, we included selection of the framework as a critical step as much of the synthesis is defined by it. Using a framework with broad categories is preferable. Additionally, we experienced challenges related to the definitions of the outcomes of interest, namely, linkage, adherence and retention in care, in the included qualitative reviews, even though these were pre-specified in the protocol. In quantitative research, measurements are used to determine adherence, whereas qualitative data may be broader. We discussed these in the author team and decided to use overarching definitions and broad categorisation. It is recommended that future qualitative research interrogate the conceptual definitions of linkage, adherence and retention in care for PLHIV. We were challenged by our own assumptions that the pragmatic aggregative nature of the proposed synthesis would be easily applied to sub-group analysis. We intended to explore the data by subgroups for children 0–13 years, young people and adolescents, perinatally infected adolescents and adult subgroup populations, but due to the different reporting methods and study designs, we were unable to provide a detailed synthesis of themes for these subgroups. We were also unable to group specific themes to specific population groups or contexts. However, we grouped children and youth as up to 24 years and adults as our second population group. Furthermore, we found a high proportion of reviews to be conducted in low- and middle-income countries. We conducted the aggregation for all studies and included evidence annexes indicating the socio-economic country grouping as low, middle, or high for each reference. Lastly, the synthesis of the findings was limited to the included review authors’ interpretation of the data and we included the primary study data as per the review authors’ description. It does become challenging for research synthesis as the production of QES is increasing and the risk of sampling and re-interpretation of the same primary studies for different outcomes and objectives through different paradigmatic lenses. We extracted themes from the reviews as per the authors’ interpretation. However, using guidance from meta-aggregation [9, 22], we only included themes that were considered unequivocal, meaning the themes were supported with evidence in the included review. We provided transparent decision-making notes and evidence annexes to this manuscript to contribute to the trustworthiness of the findings of this overview.