The 2017 South African national HIV household survey data analysis on virally unsuppressed people aged ≥ 15 years revealed three key findings. Firstly, against a backdrop of 18.7% HIV prevalence and 37.8% viral non-suppression, an overall HIVDR prevalence of 26.3% was identified. Secondly, rural farm residents had the highest estimated HIVDR prevalence at 35.0%, with the lowest (23.0%) among rural informal residents. On a provincial level, the Eastern Cape had the highest HIVDR prevalence (36.6%), and the North-West province had the lowest (21.6%). Thirdly, following adjustment for confounding factors, ARV-positive status, attainment of secondary schooling as highest level of education, and displaying HIV stigmatization/discrimination attitudes were found to significantly elevate the odds of developing HIV drug resistance (HIVDR). Conversely, individuals aged 25–34 years and those aged ≥ 45 years were significantly associated with reduced odds of HIVDR. Notably, sex, race, and HIV knowledge were not significant predictors.
These findings, drawn from a nationally representative household-based survey, underscore the importance of optimizing HIV management strategies to achieve UNAIDS 95-95-95 targets. The study also identifies key populations, factors, and regions with high HIVDR burdens, aiding resource prioritization. As per WHO threshold of > 10% for HIVDR levels, we have detected high HIVDR levels of 26.3% among adults without viral load suppression, particularly in the ARV-positive group at 43.3% for ADR, and the ARV-negative at 21.1% for PDR. On the contrary, pooled country estimates of 84% and 11.0% for the overall ADR and PDR prevalence, respectively, were reported for Uganda, Kenya, Tanzania and Nigeria [28].
Our findings also revealed elevated HIVDR levels for NNRTI-based regimens (17.3%), particularly EFV resistance (60.3%). In comparison, pooled country estimates for Uganda, Kenya, Tanzania and Nigeria were 21.5% for NNRTIs and 68.8% for EFV [28]. Much higher HIVDR levels, compared to our findings, were observed by the first South African national sentinel-based survey with estimates of 95.4% and 92.0% NNRTI HIVDR prevalence among the ARV-positive, and ARV-negative, between 2013 and 2014 [14, 29]. Prior to the South Africa’s transition to INSTI-based first- and second-line regimens whereby DTG replaced EFV and NVP from October 2019 [30], EFV was a widely used NNRTI drug in the first-line ART regimen in Southern Africa, including South Africa, at a rate of 76.0% versus NVP (23.0%, as second option) [4, 31–33]. Country surveys from 21 out of 30 LMICs from four regions, the Americas, South-East Asia, Western Pacific and nine sub-Saharan African countries including South Africa, showed PDR prevalence of 24% among the ARV-positive and 7% among the ARV-negative [9].
Our findings showed low HIVDR levels for NRTI (0.2%) and moderate levels for NNRTI + NRTI (8.2%) regimens. Additionally, an analysis of annual PDR prevalence across various LMICs demonstrated that NRTIs and PIs remained stable at a level of 5% over the period spanning from 2000 to 2016 [7, 9, 29, 32, 34–38]. South African studies showed that NRTI HIVDR has been increasing since the introduction of TDF as a component of the first- and second-line ART regimens in 2010, particularly among PLHIV infected with HIV-1 subtype C (> 95%) [2, 11, 13, 35, 39]. Moreover, other studies including in South Africa and the sub-Saharan Africa showed that preexisting NRTI + NNRTI HIVDR, including NNRTI HIVDR, increased the risk of DTG failure and resistance by compromising the virological efficacy of the NRTI backbone in DTG-based regimens during the transition, as the functionality of DTG relies on NRTIs, especially 3TC [7, 30, 40, 41].
Our study found no geographic or locality variations pertaining to the HIVDR prevalence across South Africa. Despite that, all provinces and locality types had HIVDR levels above 20%, ranging between 21.6% and 36.6% on provincial level and from 23.0–35.0% for localities. In contrast, Rossouw et al. found a substantial disparity in HIVDR prevalence between urban and rural settings in South Africa's sentinel sites [42]. This disparity was because of increased viral load monitoring, adherence support and ART switching in the urban group. In 2013, a sentinel HIVDR surveillance in the province of KZN found an estimated HIVDR prevalence of 89.0% among PLHIV failing ART in 2013 [38].
Our study found that after adjusting for potential confounders, the ARV-exposed had higher odds of HIVDR compared to the ARV-negative PLHIV. Consistent with our findings, a national survey in Rwanda also found that ART regimens were significantly associated with HIVDR [43]. In contrast, a Nigerian national survey observed a significant association between HIVDR and being ARV-negative [28]. In the current study, the adjusted analysis showed that those with secondary education level had higher odds of HIVDR compared to those with primary schooling as the highest level of education. In contrast, studies from Kenya, Uganda, Tanzania and Nigeria showed that primary, secondary or tertiary education level was significantly associated with HIVDR when compared to less than primary education. In addition, in a Chinese study, Zhou et al [44] found that compared to no schooling, secondary level was significantly associated with lower odds of HIVDR, while tertiary or primary education level had higher odds of developing HIVDR. This could be due to challenges such as HIV stigmatization/discrimination, inaccessible healthcare facilities and healthcare systems that are not tailored to specific age groups.
Our study also found that PLHIV who displayed HIV stigmatization/discrimination had higher odds of having HIVDR compared to those without this attitude. In addition, the 25–34 years old and ≥ 45 years old age groups had reduced odds of developing HIVDR compared to 15–24 years old age group. On the contrary, national surveys from Rwanda, Kenya, Tanzania, Uganda and Nigeria found no significant association between HIVDR and age [28, 43]. In a Chinese study by Zhou et al [44], being ≥ 15 years old significantly predisposed individuals to higher odds of HIVDR.
Our study found no significant association between HIVDR and sex or race. A Chinese study from a province with the highest burden of HIV also found non-significant association between HIVDR and sex or ethnicity [44]. In addition, national surveys from Kenya, Uganda, Tanzania, Nigeria and Rwanda also found no significant association between HIVDR and sex [28, 43]. On the contrary, the national Italian study found that male gender was significantly associated with HIVDR [45]. Furthermore, our study did not establish a significant link between HIVDR and employment or marital status and, whereas Zhou et al. reported lower odds of developing HIVDR in employed individuals and that the divorced/widowed were significantly associated with higher odds of HIVDR compared to the married [44].
Since a majority of our study participants were ARV-negative(75.6%) compared to the ARV-positive, this suggests that ARV-negative status was significantly associated with virological failure, as also observed by an Ethiopian study [46]. Our findings may suggest that NNRTI-based first-line regimen, compared to NRTI- and PI-based second-line regimens, was suboptimal to attain a sustainable population-level viral suppression among ART initiators or ongoing ART recipients. Furthermore, our results indicate delayed or absent ART switching in South Africa, coupled with a high prevalence of HIVDR for NNRTI-EFV (60.3%), insufficient viral load (VL) monitoring, and suboptimal adherence to ART.
Our findings of moderate levels of NNRTI + NRTI HIVDR suggest that NRTI HIVDR is driven by NNRTIs. Additionally, our low HIVDR levels for NRTI-only regimens and PI-based regimens might suggest that viral non-suppression among PLHIV exposed to NRTI-backbone based regimens was driven by non-adherence in the ARV-exposed. Consequently, our findings validate the country’s transition because INSTIs have a high genetic barrier to resistance development compared to the NNRTIs with a low genetic barrier [47].
LIMITATIONS
The limitations of this study, typical of cross-sectional designs, include the inability to establish causality due to the lack of temporal sequence between the exposure and outcome variables. Cross-sectional studies provide a snapshot of data at a single point in time, limiting the ability to capture changes over time or assess the directionality of relationships between variables. Additionally, reliance on self-reported data introduces potential for recall bias and social desirability bias, impacting the accuracy and reliability of information collected. However, the survey series used validated questionnaires. The representativeness of the sample might also be affected by selection bias, as individuals who were more likely to participate may differ systematically from non-participants. To correct this bias, survey weights were used across all analyses to adjust for unequal sampling probabilities and non-responses. Furthermore, the cross-sectional nature of the study prohibited the assessment of long-term trends or outcomes. Considering these limitations, it's advisable to interpret the study findings with caution and in context.
RECOMMENDATIONS
To effectively combat drug resistance, transitioning to INSTI-based regimens is crucial, particularly among individuals previously exposed to NNRTI-based regimens. Targeted intervention programs should be implemented to address the diverse needs of different demographic groups, focusing on enhancing ART adherence, promoting viral suppression, and reducing HIV transmission. Comprehensive education and awareness campaigns are needed to tackle HIV stigma and discrimination, fostering acceptance and support for people living with HIV within communities. Strengthening surveillance and monitoring systems are essential to track changes in HIVDR prevalence and assess the effectiveness of interventions, emphasizing the importance of longitudinal studies or repeat cross-sectional studies for evaluating treatment outcomes and identifying emerging challenges in HIV management. Additionally, policies supporting age-appropriate/tailored healthcare services, coupled with investments in capacity building for healthcare providers, are necessary to ensure accessible, confidential, and non-judgmental care, particularly for the younger and low literacy or illiterate populations.
The scope of sentinel HIVDR surveillance should be broadened to include other provinces such as the Eastern Cape, Mpumalanga, Western Cape, and/or Limpopo, given their elevated prevalence of HIVDR in contrast to KwaZulu Natal and Gauteng Province, which have been the primary focus of HIVDR monitoring due to their substantial HIV burden.