Table 4. Strength of Evidence and Generalizability/Transferability of Key Review Findings to High Prevalence Contexts.

Level of Influence	Key Review Finding	Strength of Evidence Summary	Generalizability/Transferability Summary
Individual	1a) Socio-demographic factors (i.e., age, educational level, residency) can influence ART initiation, adherence, and retention.	High Eight papers reported a range of socio-demographic factors. Age and educational level were most widely reported as specific findings in papers with strong quantitative designs, specifically that older women, or those achieving a higher level of education, were more likely to adhere. Residency was reported in two papers – one quantitative (women in central hospitals were less likely to disengage than women in smaller and more remote hospitals), and one qualitative/descriptive, focusing on challenges in rural communities around maternal roles.	Moderate This finding was reported across a range of contexts, and is generalizable, although issues around maternal role may be context-based.
	1b) Level of knowledge about health services, ART, and/or PMTCT can affect ART initiation, adherence, and retention.	Moderate Six papers reported on the association between knowledge (of health services, or ART) and adherence. Mixed method and quantitative designs reported higher levels of knowledge correlating with adherence. One qualitative study based on focus group discussions and one paper derived from a conference abstract confirmed the other studies’ findings. Knowledge of the referral process was only cited as an enabler in one mixed method paper.	High This finding was reported across a range of contexts, with no anomalies, and is likely to apply broadly, although evidence for knowledge of the referral process requires further exploration.
	1c) Women’s fears and perceptions of treatment, and the desire to maintain their roles and status within families, can affect ART initiation, adherence, and retention.	Moderate Eleven papers reported on these issues. Qualitative and mixed method studies with strong designs reported that a woman’s role in the family may conflict with her needs as a patient. A desire to protect children from HIV was reported in qualitative studies and one quantitative study, although the latter was only a second order interpretation.	Moderate Evidence relating to these factors was internally consistent within the review, although it did include second order interpretations. The particular forms these factors take are likely to be context-based.
	1d) Factors in a woman’s daily life can affect ART initiation, adherence, and retention.	High Challenges managing the practical demands of ART were reported in 12 quantitative or mixed method papers, and included day-to-day demands or (in one mixed method paper) lack of access to water and/or food. Scheduling problems, or being away from home, were frequently reported barriers to adherence. All findings were of first order interpretation.	Moderate Evidence relating to these factors was internally consistent within the review, although the particular forms these factors take are likely to be context-based.
	1e) Beliefs (e.g., religious beliefs, feeling healthy, and having a positive outlook) can affect ART initiation, adherence, and retention.	Moderate Seven papers reported that individual beliefs affected initiation, adherence, and retention. The finding about feeling too well to attend HIV services was only a second order interpretation in one quantitative paper and a report by one respondent in a qualitative study. Feeling ‘happy’ was significant in one quantitative study. The positive role of religion was a strong finding in two papers, though one (quantitative) was a second order interpretation. Advice to use traditional medicines instead of ART was only cited as a barrier in only one mixed method study, but it was a core finding.	Moderate Evidence was reported across a range of contexts, confirming the relevance of the broad finding, but specific examples are likely to be context-dependent.
	1f) Behavioral factors can be key barriers to ART initiation, adherence, and retention.	High Behavioral factors influencing ART initiation, adherence, and retention were reported in eight papers. Three quantitative papers with strong designs reported use of alcohol or illegal drugs as barriers to adherence. Four quantitative studies, and one mixed method study, reported evidence that forgetting, or misplacing, medication was a barrier to adherence.	High This finding was reported across a wide range of contexts; the two essential components can be generalized.
Interpersonal	2a) Relationships with partners can have a substantial influence on ART initiation, adherence, and retention.	High This finding was reported across 14 studies, with quantitative and qualitative designs. Five of the papers also reported on issues around gender dynamics and six of the papers reported on the benefits of disclosure to a partner to adherence. One quantitative paper reported a significant and contrary finding that not disclosing HIV status to a partner enabled adherence, but this was not reported elsewhere.	High This broad finding was reported across a wide range of contexts, arguing for a high validity. The one finding that non-disclosure to a partner enabled adherence highlights how the nature of interpersonal influence may vary by context.
	2b) Relationships within the family affect ART initiation, adherence, and retention.	Moderate This finding was reported in three qualitative papers and a mixed method study. Designs were robust and the evidence was descriptive in all four studies.	Low This finding was reported in a small number of largely descriptive papers.
Community	3) Stigma within a community can be a significant barrier to ART initiation, adherence, and retention.	High This finding was reported in 15 papers, almost all of which were based on qualitative studies. Data on stigma as a barrier typically came in the form of participant self-report. One quantitative study described self-disclosure as an enabler of access to HIV care and adherence to ART.	High This finding was reported across a wide range of contexts, especially studies in which women identified barriers to initiating, adhering to, or remaining on treatment.
Structural	4a) Higher participation in recommended health services leads to increased likelihood of ART initiation, adherence, and retention.	Moderate This was a core finding in six quantitative papers. Delivery in a health center was found to be an enabler in two studies, although one was derived from a conference abstract with limited detail. Two papers reported an association between ART adherence and receiving treatment (e.g., tuberculosis treatment or multivitamins) for other conditions.	Moderate This finding was reported across a range of contexts.
	4b) Logistical problems around access to services can be barriers to ART initiation, adherence, and retention.	Moderate Seven papers with a range of designs (i.e., four qualitative, two mixed method, and one quantitative) reported this finding. Transportation problems were described descriptively in four well-designed studies, and cost was reported as a limited finding in one quantitative study. Long queues at health facilities were report as barriers to initiation, adherence, and retention in four papers of differing methods.	Moderate This finding was reported across a range of contexts and designs. The finding is generalizable, though specific factors may be context-dependent.
	4c) Interactions with health workers are valued, and affect the quality of access, and likelihood of ART initiation, adherence, and retention.	High Seven studies reported that health worker attitudes influenced whether women initiated and adhered to ART. Qualitative studies with strong designs reported that women’s perspective of beneficial interactions encouraging adherence. One quantitative paper proposed a second order interpretation that adherence is due to nurses taking opportunities to engage more effectively with patients. Positive interactions with doctors and traditional birth attendants were noted as enablers of ART initiation, adherence, and retention in two strong qualitative papers.	Moderate This finding was reported qualitatively across a range of contexts, suggesting strong context validity, and reasonable generalizability.