Gaps were identified throughout the HIV treatment cascade for the three populations, with the largest breakpoint occurring at knowledge of HIV status; all key populations fell below the 90% target for every indicator although this was most stark among MSM where only 8.8% had knowledge of their HIV status prior to the survey. Furthermore, results show that of the MSM, FSW and PWID participants who were aware of their status, only 40.0%, 52.6%, and 47.2%, respectively reported being on treatment at the time of the survey. This translates to two in five HIV-positive MSM participants, and one in two for both HIV-infected FSW and PWID participants. However, once initiated on ART, both MSM and FSW were retained above the 90% target, while a laudable achievement this only represents a small fraction of the HIV-positive population. In general terms, use of services appears higher among PWID, when compared to the other KP groups; however, PWID are still well below global targets.
When compared to other KP groups in the region, HIV testing uptake among survey participants was consistent with previous studies (9, 10). However, MSM and FSW surveys in Mozambique reported lower awareness of their HIV status compared to those same groups in other Sub-Saharan African countries, although PWID participants reported higher awareness than other PWID in the region (11–15). ART engagement was also lower across all populations groups (11, 15–17). In the few studies with results of viral load suppression in Sub-Saharan Africa, there was a range of 11%-42 among MSM, 11.0%-49.5% among FSW and less than 5% in among PWID in Kenya. (12–15, 18). The differences in results may be attributed to different survey measures, recruitment methodology, policies and interventions. However, accurate and quality data examining the cascade in Sub-Saharan Africa across the three population groups is still scarce (15). Nevertheless, to our knowledge, no countries in Sub-Saharan Africa have been able to meet ambitious 90-90-90 Fast Track Targets for key populations.
Comparing engagement in HIV services of MSM, FSW and PWID survey participants to the general population from the most recent AIS in Mozambique, shows that HIV testing among all three key population groups was consistent with people living with HIV (PLHIV) in the general population 71.3% (76.9% females vs 59.4% males) (1). Although linkage to care was not assessed in the AIS, KPs are much less aware of their status than PLHIV in the general population: 40% (46% females vs 28% males), leading to major programmatic challenges to link these vulnerable populations to care and treatment services. All three KPs also report lower levels of current treatment than the general population: 35% (40% females vs 23% males) further highlighting the potential for these population groups to transmit HIV. Our results do not include viral load, however given that viral suppression among PLHIV in the general population was low at 23% (28% females vs 13% males), we can assume that the outcomes among KPs are considerably lower given their lower engagement in services compared to the general population as well as the evidence of social, legal and structural barriers to access health care services (4). The low evidence of engagement in health services among MSM is consistent with low health service access among men in the general population in Mozambique (1). However, the majority of PWID participants were male and reported higher engagement along the cascade than the other KP groups, which may be due to greater use of health services as a result of co-morbidities, such as viral hepatitis, which would have required accessing health care services.
This is the first analysis of engagement in HIV services and progress through the HIV treatment cascade among key populations in Mozambique. Although this analysis offers important insights into the use of HIV treatment services among key populations in Mozambique, it is important to acknowledge this analysis’ limitations.
While RDS is a robust methodology for sampling among hidden populations, there are some inherent limitations such as selection bias in chain referral sampling methods, recall bias and social desirability bias for self-reported risk behaviors; non-response bias also applies to individuals who did not consent to an HIV test – and were subsequently removed from the analysis –however the proportion was relatively low (3.70% MSM, 0.24% FSW, and 9.55% PWID). Additionally, some respondents may have been previously aware of their HIV positive status but refused to disclose due to stigma or social desirability given the common knowledge that HIV-positive individuals should be engaged in care and treatment. This potential selective underreporting of HIV status has been observed in previous studies, and can underestimate prevalence results and subsequent cascade assessments (12, 13). In the most recent AIS conducted in Mozambique, only 26% of PLHIV self-reported current ART use however, when biomarker testing was used, the number of current on ART increased to 35% indicating that individuals did not self-disclose their status (1).
Next, the computation of unweighted pooled estimates among survey participants means that results are not generalizable to the KPs in the geographic locations where the surveys were conducted nor to other cities in Mozambique, but instead only represents survey participants. Due to the low sample size, it was not possible to conduct meaningful weighted estimates of the cascade for the KPs in each survey city. However, the results provide the best available proxy indication of engagement in services and progress through the HIV treatment cascade and highlight the need for enhanced efforts targeted to these groups in each survey location.
Finally, the results must be understood within the context of the care and treatment landscape when the surveys were implemented. Mozambique’s National Acceleration Plan was introduced in 2013, which changed treatment guidelines and vastly scaled up the availability of ART in the country. The roll-out of increased ART provisions occurred after the BBS was implemented among MSM & FSW. This change in the landscape may explain why PWID reported greater engagement in the care compared to the other KP groups. Despite this policy change however, there is still evidence in the most recent AIDS Indicator Survey of low engagement of the general population in the HIV treatment cascade, where men are less engaged than women, and thus it can be posited that even with greater access to ART services, social and structural barriers continue to impact KPs engagement in health services.