We used the PICOTS criteria (population, intervention, comparator, outcomes (results), temporality, setting of intervention) to select our articles. In relation to the PICO, our population was made up of patients living with HIV, the intervention was antiretroviral (ARV) treatment, the result was adherence to antiretroviral therapy and the intervention environment was West Africa. We have given priority to single-proportion studies, so no comparator has been defined. The studies were taken into account if they included the following information: articles published between 1 January 2016 and 31 December 2023 and whose study topics were selected from the implementation of the ‘test and treat’ strategy of the WHO after June 2016 under antiretroviral treatment, residing in West Africa and in whom adherence to treatment with ARV drugs was measured. We did not specify standardized measurement tools for measuring adherence. All measurement methods were taken into account. Articles published in 2016 and after, if the topics were selected before 2016, were excluded from the study. Qualitative studies, as well as studies that did not estimate the rate of adherence, were also excluded from the study. Literature reviews, systematic reviews, meta-analyses were also excluded from our study.

We have mainly searched the online bibliographic databases for article selection. The bibliographic databases used were PubMed, Scopus, Web of Science, Google Scholar. Last search: 15 January 2025. The Reference lists of included studies were screened. Grey literature was excluded.

The writing of this systematic review and meta-analysis followed the standards of the PRISMA Guide (Preferred reporting items for systematic reviews and meta-analyses). The search equation was formulated by combining MeSH keywords using the Boolean operators OR and AND. The following search query has been formulated:

(“HIV” OR “acquired immune deficiency syndrome virus” OR “acquired immunodeficiency syndrome virus” OR “aids virus”) AND ((“medication adherence” OR “Drug Adherence” OR “medication compliance” OR “drug compliance” OR “medication persistence”) OR (“patient compliance” OR “client compliance” OR “user compliance” OR “patient adherence” OR “client adherence” OR “patient cooperation” OR “therapeutic compliance” OR “treatment compliance”)) AND ((“Treatment Adherence and Compliance” OR “Therapeutic Adherence and Compliance” OR “treatment adherence”) AND (“highly active antiretroviral therapy” OR “HAART” OR “Combination Antiretroviral Therapies” OR “Anti-retroviral agents” OR “ARV” OR “antiretrovirals”) AND (“West Africa” OR “Western Africa” OR “Benin” OR “Burkina Faso” OR “Cabo Verde” OR “Cote d’Ivoire” OR “Ivory Coast” OR “Gambia” OR “Guinea-Bissau” OR “Liberia” OR “Mali” OR “Mauritania” OR “Niger” OR “Nigeria” OR “Senegal” OR “Sierra Leone” OR “Togo”).

Three reviewers independently screened titles, abstracts, and full texts using Abstrackr. Disagreements were resolved by consensus or by a fourth reviewer.

The first step consisted of selecting the articles by title, then by summary, then the articles were exported to Excel. Once on Excel we proceeded to remove duplicates from the titles. After removing the duplicates, we proceeded with another selection by reading the full texts. The selection of articles is presented in the form of a flow chart.

The data extraction was performed according to PICOTS. The study population concerned patients living with HIV, the main measure was the proportion of adherence between 2018 and 2023. The variables collected concerned the authors, the date of publication, the year, the type of study, the design, the population studied, the number of subjects in the study, the proportion of observant patients, and the country in which the study was conducted. The methods for calculating adherence were not taken into account; instead, we considered the estimated adherence proportions reported in the studies.

The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) assessment grid was used to evaluate checks related to subject selection, confusion, follow-up, and adherence measurement method.

Primary outcome: ART adherence rate with 95% confidence intervals.

We conducted a meta-analysis of proportions following the PRISMA 2020 recommendations. The proportions extracted from each included study were combined to estimate an overall proportion with its 95% confidence interval. We used the software [11]. Given the bounded nature of the proportions (between 0 and 1), a stabilized variance transformation (logit type or double Freeman-Tukey arcsine) was applied in order to reduce the heterogeneous variance between studies. The overall estimate was then reconverted to the original scale for interpretation. A random-effects model was used (DerSimonian and Laird model) to account for inter-study heterogeneity. This model assumes that the proportions vary not only due to sampling error, but also due to actual differences between studies. Heterogeneity was assessed using the I² index and the Cochran Q test. One I² > 50% was considered moderate heterogeneity to be elevated. An exploration of sources of heterogeneity was carried out by subgroup analyses. The results were synthesized using forest plots, indicating individual proportions, confidence intervals, and weight of each study in the analysis. Publication bias was assessed using funnel plots.

Funnel plots and Egger’s test used.

GRADE framework applied to overall and subgroup findings.

PROSPERO: CRD420251008258.

Seven studies were included in this review following the selection process. Figure 1 shows the flow diagram of study selection.

Figure 1. Flow diagram for the identification and selection of articles included in this review.

The adherence research included two studies involving children and five involving adults, Table 1 presents the characteristics of the different studies selected for analysis.

Table 1. Characteristics of articles included in the systematic review on adherence to antiretroviral treatment in West Africa.

The assessment of risk of bias showed an overall level of very low to moderate quality of evidence. Table 2 presents the detailed risk of bias assessment of the studies.

Table 2. Evaluation of the risk of bias in included studies according to the grading of recommendations assessment, development, and evaluation (GRADE).

Adherence ranged 5.9% - 89.5%.

The study involved 4436 patients, and the overall adherence rate was estimated at 63%, CI95% [0.45 - 0.79]. The subgroup analysis noted a proportion of 78%, IC95% [0.66 - 0.88] in adults and a proportion of 23% IC95% [0.00 - 0.71] in children. The analysis showed a strong heterogeneity of study outcomes (I² = 99.3%, p < 0.001). Figure 2 presents the overall and subgroup estimates of adherence and heterogeneity analysis.

Figure 2. Forest plot showing overall and subgroup estimates of adherence and analysis of heterogeneity of antiretroviral treatment among patients living with HIV in West Africa.

Funnel plot asymmetry; Egger’s test p < 0.05.

Significant heterogeneity was observed (p < 0.01), indicating variable effects in terms of extent and/or direction of the effects. Figure 3 illustrates the funnel diagram used to assess potential publication bias.

Figure 3. Funnel plot of study publication bias assessment.

Low certainty (GRADE) due to bias, heterogeneity, and imprecision.