Introduction

Women constitute the highest proportion of people living with HIV globally [1]. In South Africa, the HIV epidemic also disproportionately affects women. In 2017, approximately two thirds of people living with HIV, aged 15–49 years and on antiretroviral therapy (ART), were female [2]. In the same year, disparities in HIV prevalence by sex were most pronounced among young adults. HIV prevalence among 20–24 year olds was three times higher in females (15.6%) compared to males (4.8%) [2]. Potential reasons for gender disparities in the profile of the HIV epidemic are both economic and sociocultural [3]. Extensive geospatial heterogeneity in the distribution of the HIV epidemic has been reported previously [4]. In 2018, provincial HIV prevalence estimates among people living with HIV aged 15–49 years ranged from 6.7% in the Western Cape to 18.4% in KwaZulu-Natal [4]. While poorly understood, variations in provincial HIV prevalence rates have been linked to differences in the uptake of HIV prevention interventions and societal norms between provinces [4]. Thus, provincial variations in the profile of the HIV epidemic necessitate optimized allocation of resources and targeted interventions for HIV/AIDS programmes. This is particularly pertinent for the prevention of mother-to-child transmission of HIV (PMTCT) programme, which has prioritized the elimination of mother-to-child transmission of HIV (eMTCT). The national MTCT rate of 1.0% at birth in 2017 equalled just under 250 cases per 100 000 live births [5]. However, this figure was five times higher than the World Health Organization (WHO) eMTCT target, defined as ≤50 cases per 100 000 live births at completion of breastfeeding and an overall mother-to-child transmission (MTCT) rate of <5% in breastfeeding populations [5]. Sub-national level variations in MTCT rates also exist, with some districts reporting almost double the national MTCT rate at birth [6]. This suggests the need for contextualized and tailored responses to maternal and paediatric HIV in South Africa.

We describe the distribution of women of reproductive age (15–49 years) living with HIV (WLHIV) as well as those without HIV by province in South Africa, in order to identify areas that may require additional eMTCT interventions. These may include interventions aimed at reducing the risk of HIV acquisition by women of reproductive age and unplanned pregnancies among WRLHIV (pillars 1–2 of the South African PMTCT strategy) [7]. Women of reproductive age living with HIV and sub-groups (adolescent girls and young women (AGYW) aged 15–24 years) are priority populations for achieving eMTCT. These women can potentially fall pregnant and transmit HIV to their infants. While the risk of vertical transmission of HIV is low in planned pregnancies, where WRLHIV fully access and utilize preconception and post conception PMTCT services, the contrary is true for unplanned pregnancies [8]. Pregnancies are more likely to be unplanned in AGYW compared to older women [9]. In addition, AGYW represent a sub-population that will remain in the reproductive age group for the longest time; are associated with high incident infections and where HIV positive, AGYW have poorer PMTCT outcomes compared to older WRLHIV [9]. Therefore, the risk of MTCT is higher in this population. However, an increase in number of women conceiving on ART is anticipated with the scale up of the universal test and treat strategy [10]. Consequently, understanding where WRLHIV, including AGYW (regardless of their HIV status) and those who are pregnant, are located in South Africa is key for eMTCT. Scaling up family planning services and ART/PMTCT services (for example, stock planning of Efavirenz based regimens and Dolutegravir rollout in relation to new ART guidelines) in geographic areas with a high burden of these groups may be beneficial for maternal HIV care and eMTCT.

Obtaining accurate estimates of vertical transmission is challenging in South Africa because of the lack of a unique patient identifier in the health system. Monitoring of early infant MTCT rates and coverage of early infant diagnosis currently relies on routine data from the District Health Information System (DHIS) and the National Health Laboratory Service (NHLS) [11]. There are no accurate, national data sources for postnatal transmission rates, breastfeeding practices, maternal virological control during pregnancy and the postpartum period. An analysis of current data sources used for tracking eMTCT, their strengths and weakness, are described elsewhere by Sherman et al [11]. Using six different data sources, we describe the distribution of WRLHIV in South Africa during 2018 in relation to where the intra-uterine transmissions occur.

Materials and methods

Data for these analyses were extracted from six independent data sources for HIV-related and demographic indicators from the public health sector in South Africa. These were the Thembisa Model, DHIS, Statistics South Africa (Stats SA), the South African National HIV Prevalence, Incidence, Behaviour and Communication Survey (SABSSM), the National Antenatal Sentinel HIV & Syphilis Survey Report, (herein referred to as the ANC sero-prevalence survey) and the National Institute for Communicable Diseases’ Surveillance Data Warehouse (NICD SDW), a division of the NHLS. The Thembisa Model was used as the primary data source because it had most of the indicators required for the analyses. The other sources were used for triangulation purposes where applicable. The reporting period was the year 2018. Where data for 2018 was not available, available data from the most recent year were used. A brief description of each data source and indicators that were used for the analyses follows (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Indicators used in the analyses by data sources.

https://doi.org/10.1371/journal.pone.0231228.t001

Results

There were approximately 58 million people in South Africa in 2018 with half of the population (51%) being female (Fig 1). Women of reproductive age constituted 27% of the total population, while WRLHIV (3.8 million) accounted for approximately 7% of the total population. There was a total of 4.7 million (8.1%) AGYW of which 477 798 (10.1%) were living with HIV in South Africa. Gauteng province had the highest number of females (all ages) followed by KwaZulu-Natal (KZN), the Eastern Cape and the Western Cape (Fig 2). The Northern Cape had the lowest number of females during the analysis period. A similar trend was observed for women of reproductive age (Fig 3A) and AGYW (Fig 3B), where 69.5% and 67.9% occurred in the four most populous provinces, respectively. However, the analysis of the geographic distribution of WRLHIV (Fig 3C) and HIV-infected AGYW (Fig 3D) showed that 74.4% and 86.7% respectively, of these women were located in four provinces. The latter comprised three of the four most populous provinces but included Mpumalanga instead of the Western Cape. KwaZulu-Natal and Gauteng provinces recorded the highest numbers of WRLHIV at approximately 1 million each, contributing 53.7% of the total population of WRLHIV in the country followed by the Eastern Cape, Mpumalanga and Limpopo provinces.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Profile of females of reproductive age in South Africa in 2018.

Abbreviation: Stats SA Statistics South Africa.

https://doi.org/10.1371/journal.pone.0231228.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Population density of females by province in South Africa, 2018.

Abbreviations: EC Eastern Cape, FS Free State, GP Gauteng Province, KZN KwaZulu-Natal, LP Limpopo Province, NC Northern Cape, NW North West, WC Western Cape, Stats SA Statistics South Africa.

https://doi.org/10.1371/journal.pone.0231228.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

Provincial distribution of percentages of: 3A) females of reproductive age 3B) adolescent girls and young women 3C) women of reproductive age living with HIV 3D) adolescent girls and young women living with HIV in South Africa, 2018. Source: Thembisa Model 4.1. Abbreviations: EC Eastern Cape, FS Free State, GP Gauteng Province, KZN KwaZulu-Natal, LP Limpopo Province, NC Northern Cape, NW North West, WC Western Cape.

https://doi.org/10.1371/journal.pone.0231228.g003

Overall, out of approximately 1 million live births in the country; between 250 000–286 000 births occurred to women living with HIV in 2018 (Table 2). Limpopo province had the third highest number of live births in 2018 ahead of the Eastern Cape and Western Cape provinces (Table 2). This was in keeping with Limpopo province having the highest fertility rate throughout the country at (3.1%) (Table 2). KwaZulu-Natal province recorded the highest number of live births to women living with

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Number of live births (total), number of live births to women living with HIV and intra-uterine transmission rates by province in South Africa, 2018.

https://doi.org/10.1371/journal.pone.0231228.t002

HIV, followed by Gauteng and the Eastern Cape respectively (Table 2). Similarly, KZN had the highest HIV prevalence rate among pregnant women (S1 Table) followed by Mpumalanga province (S1 Table). The intra-uterine transmission rate for South Africa was 0.8%, translating into 241 cases of HIV-infected neonates per 100 000 live births (Table 2). Provincial intra-uterine transmission rates ranged from 0.6% in KZN to 1.1% in Limpopo province. Districts intra-uterine transmission rates ranged from 0.4% to 1.7% with KZN districts recording intra-uterine transmission rates lower than the national average of 0.8% (Table 3). Provincial intra-uterine case rates were fairly similar and ranged from 179–325 cases per 100 000 live births in the Northen Cape and Mpumalanga respectively. District level intra-uterine case rates had a greater spread, ranging from 87–415 cases per 100 00 live births (Table 3). Overall, the four provinces with the most WRLHIV yielded 1 417 (65.5%) of all intra-uterine infections in 2018. In addition, Limpopo’s intra-uterine infections contributed a further 264 (12.2%). Therefore focussing eMTCT efforts in five provinces has the potential to reduce 1 681 (77.8%) of all intra-uterine infections per annum, nationally.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Intra uterine transmission rates and intra uterine case rates per 100 000 live births by geographic location in South Africa, 2018.

https://doi.org/10.1371/journal.pone.0231228.t003

Discussion

In 2018, South Africa’s population of 58 million people was evenly distributed between males and females. Gauteng province had the highest number of females, accommodating a quarter of the total population of females (all ages). The majority of all women, 20 253 423 (68%), and women of reproductive age, 10 958 758 (70%), were located in the densely populated provinces; that is Gauteng, KwaZulu-Natal, Western Cape and Eastern Cape provinces (“the big four”). Although the distribution of WRLHIV followed a similar trend, Mpumalanga and Limpopo had the fourth [364 113 (10%)] and fifth, [266 705 (7%)] highest number of WRLHIV ahead of the Western Cape. Adolescents and young women were also mostly located in the big four provinces, 3 219 832 (68%). Ten percent (477 798) of all AGYW were living with HIV in South Africa in 2018. Gauteng, KwaZulu-Natal, Eastern Cape and Mpumalanga, had the highest percentage of AGYW living with HIV at 32%, 30%, 13% and 12% respectively. Although the national intra-uterine transmission rate was <1.0%, 241 per 100 000 neonates acquired HIV from their mothers in 2018 at birth. Mpumalanga, North West and KwaZulu-Natal had intra-uterine case rates that were above the national average at 325, 283 and 254 cases per 100 000 live births. Districts intra-uterine case rates varied greatly and ranged from as low as 87–415 cases per 100 000 live births. However, even districts with the lowest intra-uterine transmission rate at birth, still had case rates higher than the eMTCT target without taking infections after birth into account.

Based on the demographic profiles of provinces, findings of these analyses suggest disparities in the need for eMTCT interventions. In this last mile to eMTCT, it may be worthwhile strengthening interventions in provinces that have the largest numbers of WRLHIV (north eastern parts of the country) as well as those regions with the highest intra-uterine transmission rates and intra-uterine case rates while maintaining current levels of support elsewhere. Although early transmission rates are below the eMTCT target of <5%, South Africa is not positioned to achieve eMTCT case rate targets because of the extremely high maternal HIV seroprevalence [18]. However, the country has a well-established PMTCT programme in place. Closing PMTCT gaps [5] and scaling up sexual and reproductive health for preventing vertical transmission of HIV could represent low hanging fruit. Sexual and reproductive health services should include intensifying family planning and contraception services amongst all women of reproductive age regardless of HIV status and providing pre-exposure prophylaxis to AGYW. As expected, most women & AGYW were situated in the ‘big four’ provinces. However, we found that the burden of HIV in women and AGYW was also pronounced in Limpopo and Mpumalanga provinces. Limpopo province has the highest fertility rate in the country and a high proportion of AGYW without HIV who may benefit from family planning and pre-exposure prophylaxis initiatives. Mpumalanga also has a significant amount of WRLHIV who may be in need of family planning services. Cross border mobility of pregnant women seeking healthcare from neighbouring countries may be a contributing factor to high WRLHIV and fertility rates in these two provinces.

The findings from our analysis of intra-uterine transmission rates did not necessarily follow the underlying population densities. We observed lower transmission rates in KwaZulu-Natal which has the highest number of WRLHIV and live births to women living with HIV. We attribute the low intra-uterine rates in this province and its districts to the larger denominator (total live births, Table 2) and/or a more effective PMTCT programme. The converse probably explains the high intra-uterine transmission rate in the Northern Cape and its districts, where smaller denominators result in higher intra-uterine rates. Poor de-duplication of HIV PCR positive test data may also result in some provinces like the Western Cape having intra-uterine transmission rates that are above the national average. These results suggest that case rates per 100 000 may be a better indicator for monitoring programme performance at the various geographic levels [18] as it is directly linked to the performance of the PMTCT programme, for example coverage of services and attainment of viral suppression in pregnancy. Since most provincial level intra-uterine case rates were fairly similar, it is necessary to tackle the intra-uterine case rates at district level where there is considerable variation. Districts with high intra-uterine case rates deserve special attention with respect to improving PMTCT services.

Despite the distinct geographical differences in WRLHIV and AGYW, attainment of eMTCT will require more than understanding the geographic distribution of all WRLHIV. In order to fast-track eMTCT, it is critical to target PMTCT pillars of prevention of HIV infection [7] and family planning particularly in AGYW. The proposed School Health Programme which caters for sexual and reproductive health education and HIV testing in schools among children aged ≥12 years, has the potential to reduce maternal HIV prevalence in future and enable eMTCT. Among AGYW with HIV and other WRLHIV, the focus needs to shift towards virologic control, with specific attention to pregnant women, especially considering maternal viral load is the strongest predictor of MTCT risk [19]. Therefore, improving viral load monitoring during pregnancy, delivery and breastfeeding, and improving quality of patient management among WRLHIV and all HIV positive pregnant women may translate into improved viral load suppression rates. Achieving this in the big four provinces may fast-track eMTCT. There is urgent need to effectively track and monitor viral loads of all HIV positive pregnant women. A marker of pregnancy, delivery and the postpartum period within the NICD SDW as recommended in the 2019 National PMTCT guidelines, should allow for monitoring of viral load suppression rates at facility level in near real time to advance UNAIDS 90-90-90 [20] targets in women of reproductive age, particularly in high-risk areas. Such monitoring will provide information on the true picture of viral load burden during pregnancy and prompt rapid intervention. Currently this information is lacking.

In our analyses, the Thembisa Model was used as our primary data source because the model was specifically designed and is used for estimating HIV populations in SA. As a result it contained data for 10/12 indicators reviewed for these analyses. However, estimates may be inaccurate if underlying assumptions are incorrect. We therefore used other data sources to validate the data. Although there were differences in absolute numbers; we found that data from the other sources was generally within the upper and lower limits of the Thembisa model estimates (S2 Table). This may be because the Thembisa Model uses some of the data we triangulated with it, for example Statistics South Africa. Where there were significant differences, we attribute these to differences in (i) methodology or (ii) in reporting intervals or (iii) data quality issues for some of the data sources used for triangulation purposes. For example, the latest data for the ANC seroprevalence survey is 2017, while the reporting period for these analyses was 2018 (S1 Table). The laboratory based data lacks unique patient identifiers therefore de-duplication may be incomplete, resulting in over-estimation of transmission rates. Intra-partum transmission rates would be anticipated to be similar to intra-uterine transmission rates however; postnatal transmissions would depend on breastfeeding practices (e.g. duration of breastfeeding), postnatal virologic control and incident maternal infections which we were not able to measure.

Conclusions

We have attempted to quantify the need for eMTCT interventions based on the demographic profile of provinces in South Africa. Our results confirm what is mostly known. The need for HIV/eMTCT resources is greatest in Gauteng, KwaZulu-Natal and Eastern Cape. Limpopo and Mpumalanga provinces also warrant attention. While advocating for allocation of interventions for eMTCT on a needs-basis; it is important not to neglect those provinces with the least need. Routine monitoring and surveillance needs to continue until the country reaches eMTCT. Ending paediatric HIV will require robust viral load monitoring & intervention among WRLHIV in general, as well as pregnant and breastfeeding women. The NHLS laboratory data have the potential to provide viral load monitoring service near-real time until clinical databases can assume this role.

Supporting information

Acknowledgments

The authors thank the NHLS for provision of HIV PCR test data. The authors would also like to thank Lebohang Radebe and Jimmy Khosa for assistance with maps.