Clients’ satisfaction with HIV differentiated service delivery models and associated factors in sub-Saharan Africa: a protocol for a systematic review and meta-analysis

Abstract

Background

Differentiated Service Delivery (DSD) models for HIV are antiretroviral therapy (ART) mechanisms for stable people living with HIV (PLHIV) and are meant to be patient-centered to achieve treatment goals. One key element to achieve treatment goals is clients’ satisfaction with DSD models. Despite extensive implementation and scale-up of DSD models in sub-Saharan Africa, satisfaction levels remain unclear. Thus far, no comprehensive systematic review and meta-analysis regarding clients’ satisfaction level with DSD models and associated factors has been published. Therefore, we aim to conduct a systematic review and meta-analysis (1) to establish the level of clients’ satisfaction with DSD models, (2) to determine which specific DSD models clients are most likely to be satisfied with, and (3) to identify factors associated with clients’ satisfaction with DSD models.

Methods

All eligible studies reporting empirical evidence will be identified using a predetermined search strategy in several electronic databases such as EMBASE, PubMed, Scopus, Web of Science, CINAHL, and ProQuest. Observational studies that have been conducted in sub-Saharan African countries regardless of the year of publication will be included in the systematic review and meta-analysis. We will conduct this review and meta-analysis according to the guidelines by Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA). A random effects model will be used to conduct the meta-analysis using Stata 19.5 software. Heterogeneity will be assessed by using Cochran’s Q test and the I² statistic. Bias will be evaluated using Egger’s and Begg’s tests as well as visually by funnel plot.

Discussion

To the best of our knowledge, this review and meta-analysis will be the first to be conducted in the sub-Saharan Africa region to synthesize empirical evidence regarding clients’ satisfaction with DSD models and associated factors, as well as identify common DSD models most clients are likely to be satisfied with. Collection of existing evidence will improve and guide implementation, scale-up, and adoption of different DSD models. Such knowledge can inform program managers and implementers which DSD models are likely to result in desired HIV treatment outcomes.

Systematic review registration

PROSPERO CRD420251085833

Supplementary Information

The online version contains supplementary material available at 10.1186/s13643-025-03062-0.

Introduction

Despite decades of effort in the prevention and control of Human Immunodeficiency Virus (HIV) infection, it still remains a worldwide public health problem with cumulative deaths of about 40.8 million as of the year 2024 [1]. The Sub-Saharan African region is disproportionately affected, with 25.6 million infections representing about 70% of the global burden of HIV infection despite the fact that it contains only 11% of the global population. With no cure, treatment remains the key intervention available to those infected, and since the introduction of free antiretroviral therapy (ART) in Sub-Saharan Africa, there is an estimated 30.7 million people receiving ART in the region [2]. Although many countries in this region have made tremendous progress towards HIV epidemic control, which requires meeting the Joint United Nations Program on HIV/AIDS (UNAIDS) commitments and the 95–95–95 targets—that 95% of people living with HIV are aware of their HIV status, 95% of those aware of their HIV-positive status are initiated on ART, and 95% of those on treatment achieve and maintain viral suppression by the year 2030—much still remains to be done in resource-limited settings [3, 4]. With the World Health Organization (WHO) increasing its response to HIV treatment and delivery of ART [5], it has enabled access to treatment to be scaled up, with about 77% of people living with HIV receiving ART by the year 2023, up from 67% in 2021 [1]. To reinforce the delivery of ART, a public health approach has been adopted, which simplifies and supports decentralization of care, community delivery, task sharing, and efficient procurement and supply management [6]. The reinforcement was much needed in Sub-Saharan Africa, where despite the progress made in increasing treatment, challenges such as a third of people disengaging from treatment within the first five years of treatment initiation remain [7]. Additionally, other challenges include increased distance to treatment facilities and certain age groups being disproportionately affected by non-uptake of treatment, late treatment initiation, increased risk of HIV transmission to other individuals, and quick progression to AIDS [8, 9]. To address these challenges and improve patient outcomes, WHO introduced a public health approach, which is a shift away from the traditional “one-size-fits-all” facility-based model, mainly undifferentiated for individual needs [10] to alternative approaches such as HIV differentiated service delivery (DSD) models that are more client-centered and efficient while not compromising on patient care [11]. DSD models are meant to simplify HIV treatment services across the HIV cascade to better reflect the preferences, expectations, and needs of people living with HIV. In addition, DSD models aim to reduce unnecessary burdens and costs not only on the client but on the health system as well [12–14]. Additionally, there is a positive impact on retention, increased peer support, higher viral load suppression, reduced waiting time and clinic visits, and extended time for ART refills [14–17].

One key factor that may contribute to improved desired treatment outcomes is clients’ satisfaction [18]. A higher level of clients’ satisfaction is associated with adherence to treatment, increased viral suppression, improved retention into care, increased peer support, and reduced morbidity and mortality. Studies are emerging which have reported high clients’ satisfaction with DSD models such as 91.2% in Cameroon [19], 64.2% in Uganda [20] and 95% in South Africa [21]. The recent unmitigated funding freeze by the United States government could have a seismic impact such as significant disruption in HIV treatment, reversal in the progress in HIV response, smaller amounts in the ART given, community health systems, and task-shifting DSD models may crumble, and these could disproportionately affect the sub-Saharan African region [22]. This happened at the time when almost every country in the region was scaling up DSD models, and various funders, policymakers, and governments are questioning whether clients on treatment are satisfied with the models. There is a need to assess clients’ satisfaction with DSD models in the context of sub-Saharan Africa and whether certain models result in higher satisfaction than others [23]. To the best of our knowledge, no systematic review and meta-analysis has been conducted to summarize the levels of satisfaction with DSD models in sub-Saharan Africa. Therefore, the main aim of this systematic review and meta-analysis is to identify and summarize the available empirical evidence to determine clients’ satisfaction with DSD models among people living with HIV as well as identify associated factors [24].

Methods

Development of review protocol and registration

The development of this systematic review and meta-analysis methodology followed the guidelines as stipulated by the preferred reporting items for systematic review and meta-analysis protocol (PRISMA-P) statement [25] and the MOOSE Guidelines for Meta-Analyses and Systematic Reviews of Observational Studies [26, 27]. Items in the checklist as required by the PRISMA-P were completed. We registered this protocol in the international prospective register of systematic review and meta-analysis (PROSPERO), and the registration number is CRD420251085833.

Planned data sources and search strategies

Search for data sources will be conducted by PK, AS, and TM in different databases. We plan to develop a specific search strategy and adjust it according to specific databases (i.e., use of operators and symbols), while the authors will conduct the main search, PK will oversee the overall search. We will search for peer-reviewed published studies written in English regardless of the year of publication. Additionally, a manual search of the reference lists of the studies that will be identified will be conducted. The databases such as EMBASE, PubMed, Scopus, Web of Science, CINAHL, and ProQuest will be searched. Each database will be searched systematically for relevant articles using a low-specificity and high-sensitivity approach by segregating the type of DSD models as well as whether the studies were conducted in urban or rural areas. A list of appropriate medical subject headings (MeSH), alternative spellings, abbreviations used, synonymous words that authors might have used will be accumulated. All possible MeSH words will be combined using ‘OR’, and ‘AND’ will be used to capture a large range of articles and screened to identify relevant ones. The search will utilize the following keywords (satisfaction, HIV with differentiated antiretroviral therapy delivery models, sub-Saharan Africa, or south of Africa). We conducted the initial literature search from 1 st April 2025 and we plan to update the search between November and December 2025.

Eligibility criteria

We will include observational studies such as cross-sectional, case–control, and cohort studies. We will consider studies that have defined satisfaction if clients were satisfied with all of the following criteria: (1) health worker confidentiality; (2) psychosocial support received; (3) have time for other priorities such as work, school, income-generating activities, or important daily routines; (4) health cost associated with the DSD model; (5) time spent traveling and waiting to receive ART services [20]. Additionally, we will include studies that have reported prevalence or satisfaction levels but not associated factors and exclude those that have reported satisfaction with a qualitative approach. Additionally, data from conference abstracts, grey literature, dissertations, theses, and unpublished works will not be included. For mixed methods studies that have reported both quantitative and qualitative findings, we will consider the quantitative findings only. We will assess the quality of articles using a checklist (Supplementary Table S1).

Selection of studies

All studies that will be retrieved will be exported in Endnote citation management software [28]. Three review authors (IF, TM, and AH) will screen for duplicated studies and transfer them to a new folder in Endnote citation manager. Additionally, the aforementioned authors will check the titles and abstracts of each study according to the inclusion criteria.

Measurement

Clients’ satisfaction with the DSD model will be measured as a composite variable using several indicators (health care worker confidentiality, time spent travelling and receiving ART services, psychosocial support provided, time for other priorities, and health cost). When a client is satisfied on all the indicators, it will be considered as satisfied; otherwise, not satisfied [13].

Data extraction

Data extraction will be conducted by three review authors (PK, AS, and TM) from independent databases using a predetermined data extraction form. The data extraction form will be pretested on relevant articles before being used on the included studies. We will ensure high interrater reliability by conducting training on the method of coding all variables, and this is likely to improve precision and reviewers’ agreement. The coding form will register data regarding patients’/clients’ satisfaction with DSD models, baseline information, factors associated with satisfaction, and types of DSD models. We will collect information about author name and publication year, country in which the study was conducted, study design, sample size, gender of the participants, age of participants (as continuous or categorical), response rate, data collection period, viral suppression status, how satisfaction was measured, presence of comorbid conditions, duration in HIV care (years), duration on DSD models (years), distance (km) to health facility, available friend/relative at ART delivery point, employment status, marital status, education level, and transport cost to access ART. Additionally, information regarding effect sizes such as odds ratio, correlations, hedges g, coefficients will be collected. In case of reviewer authors’ information differing, discussion will be held until an agreement is reached.

Meta-analysis

Relevant data from text, figure, and tables will be extracted, and where possible, the authors will be contacted for any missing data. We will extract study design, country of study, inclusion criteria, study period, method of definition of satisfaction, satisfaction level and 95% CIs, type of DSD models, age group of participants, and when DSD models were implemented.

Assessment of risk of bias will be conducted by reviewer author (PK) using a validated 11-item checklist (Supplementary Table S2) in keeping with the recommendation for prevalence studies [29]. The reviewer will use a guide to assess articles on a scale from 0 (lowest quality) to 14 (highest quality). After scoring, total risk of bias scores will be coded as follows: low risk (11 to 14), moderate risk (6 to 10), and high risk (0 to 5). Cohen’s kappa will be calculated to assess and quantify the interrater agreement [30]. The reviewer author will assess the risk of bias using a funnel plot. A symmetrical funnel plot with points representing individual studies evenly distributed around the central pooled effect line will suggest no significant bias. However, if an asymmetrical shape is obtained, it will suggest potential bias, but sometimes could be due to true heterogeneity or other biases. Additionally, since the funnel plot is not definitive, Egger’s regression test will be used alongside visual inspection for a more robust conclusion regarding publication bias [29]. However, Egger’s test cannot offer the reason for the asymmetrical funnel plot and there is potential for variation in its power due to the number of studies included, effect sizes, and variances [31]. We will interpret heterogeneity using the I² statistic, which is considered the proportion of total variation among studies due to differences in the estimates and not due to sampling errors [32]. Articles will be weighted using the inverse of the variance, which is basically relative to the sample size [33]. Since we hypothesize that the true effect size of differences among studies varies due to real differences (heterogeneity) in populations where studies were conducted or due to methods, not just by chance, we therefore propose to use the random effects model as opposed to the fixed effect model. Heterogeneity will be interpreted as follows: 25% will indicate low heterogeneity, 50% indicating medium heterogeneity, and 75% indicating high heterogeneity, respectively [34]. The difference between sub-groups such as DSD models and age groups will be conducted using a meta-regression model with Wald test for sub-group effect sizes as well as acting as moderators [35]. We will explore the association between satisfaction levels and other variables.

Discussion

The objective of this systematic review and meta-analysis is to synthesize empirical evidence regarding stable clients’ satisfaction with DSD models of HIV treatment in sub-Saharan African countries. Evidence suggests that DSD models for HIV are being implemented and scaling up in different sub-Saharan African countries and thus far, there is no systematic and meta-analysis evidence to show the overall clients’ satisfaction with DSD models and associated factors. With the gap in the clients’ satisfaction in the region, it can negatively affect the scaling up and HIV treatment outcomes. Establishing empirical evidence regarding clients’ satisfaction with DSD models maybe challenging due to different scales and cutoffs of satisfaction levels used by different studies. This challenge may present wider confidence intervals of the level of satisfaction [19, 36]. The clients’ satisfaction levels can be presented in different effect sizes such as incidence rate, incidence ratio, prevalence, and absolute numbers, but in this systematic review and meta-analysis, absolute numbers are not going to be used if the denominators for the population are not given since the magnitude cannot be calculated. We will use rates and ratios where possible to compare inter and intra-DSD models.

DSD models are meant to be client-centered and are expected to result in better HIV treatment outcomes. However, with routine implementation and scale-up of DSD models, in most instances this is not matched with monitoring and evaluation strategies needed to assess the impact of DSD models on key HIV treatment outcomes. The pressure to roll out and adopt DSD models poses valid questions about whether desired treatment outcomes are being achieved [37]. In places where satisfaction is high, DSD models are more likely to result in better HIV treatment outcomes partly due to reduced clinic visits, reduced illness episodes, enhanced psychosocial support, and reduced transport costs. Additionally, since DSD models are for stable clients and are in HIV care longer, they are more likely to be satisfied with the models [21].

DSD models focus on differentiating service provision to clients already on ART, and most stable clients in sub-Saharan Africa are at least on some form of a model. Clients’ satisfaction may affect desired treatment outcomes and the set HIV targets [38]. Therefore, clients’ satisfaction may affect the adoption of the models and other factors such as a higher level of stigma, lack of political commitment, different context-specific challenges, more resources maybe needed for some models than others.

Subsequently, assessment of the clients’ satisfaction with DSD models and the associated factors in SSA is important to inform HIV programmers and policy implementers on the overall satisfaction and which models have higher satisfaction than others as this can have implications on the desired HIV treatment outcomes as well as the 95–95–95 HIV target of the year 2030. Additionally, the findings can inform stakeholders, HIV program funders, and governments regarding key factors that are associated with clients’ satisfaction as well as different types of DSD models. Moreover, the information to be generated can be used by implementation scientists to understand and maximize context-specific adoption and adaptation of the different DSD models in different settings.

Abbreviations

DSD

Differentiated service delivery

HIV

Human immunodeficiency virus

ART

Antiretroviral therapy

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-analyses

MeSH

Medical Subject Heading

CI

Confidence intervals

WHO

World Health Organisation

Authors’ contributions

PK conceived and designed the systematic review and meta-analysis. PK, AS, IF, TM, and AH drafted the protocol manuscript, and PK is Kaonga et al. PK, AS, and TM developed the search strings. PK, AS, IF, TM, and AH extensively reviewed and incorporated intellectual inputs in the protocol manuscript development. All authors read and approved the final version of the protocol manuscript.

Funding

No funding sources.

Data availability

Not applicable.

Declarations

Ethics approval and consent to participate

The systematic review and meta-analysis do not require ethical approval as the data are already published and available. The results of the study will be reported according to the PRISMA guidelines [25].

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.