Socio-demographic associations of HIV among antenatal care attendees in selected rural primary care facilities in South Africa's Eastern Cape province

of prevalence full A Cross-sectional study of women attending antenatal care in four Primary Care facilities was conducted using an interviewer-administered questionnaire which collected information on socio-demographic characteristics and medical history. Binomial logistic regression analyses were used to determine factors associated with HIV and to estimate the prevalence ratio (PR). The 95% condence interval (95%CI) is used for precision of estimates; p ≤ 0.05 for statistical signicance. This population showed a very low uptake of spousal attendance in antenatal care. In addition to dealing with a high teenage pregnancy rate and myths around condom usage, health promotion practitioners have a huge role to play in assisting health systems to address the new HIV challenge of this century. This cadre of professionals can have an impact on HIV prevention as they will plan, and conduct awareness campaigns related to HIV/AIDS issues, reproductive health and family planning.


Abstract Background
To effectively reduce vertical HIV transmission requires a reduction of HIV prevalence and incidence among pregnant women and a full understanding of its epidemiology. There is, however, literature paucity of HIV studies focusing on women attending antenatal care in rural areas in South Africa.

Methods
A Cross-sectional study of women attending antenatal care in four Primary Care facilities was conducted using an interviewer-administered questionnaire which collected information on socio-demographic characteristics and medical history. Binomial logistic regression analyses were used to determine factors associated with HIV and to estimate the prevalence ratio (PR). The 95% con dence interval (95%CI) is used for precision of estimates; p ≤ 0.05 for statistical signi cance.

Results
A total of 343 participants were recruited. The antenatal HIV prevalence was 38.2% (95%CI: 33.2-43.9). Participants older than 40 years were 4.6 times signi cantly more likely to be HIV positive compared to teenagers (p-value = 0.019). Furthermore, multiparous, unemployed and employed participants, were signi cantly associated with 40%, 60% and 70% higher antenatal HIV prevalences respectively when compared to primigravidas or tertiary students (p-value < 0.0001).

Conclusion
Despite a 100% antenatal HIV testing rate, the antenatal HIV prevalence remains high and is increasing in this population, coupled with no spousal attendance in antenatal care and poor condom compliance. It is therefore important to remain vigilant and monitor mother-to-child transmission that could be associated with this increased prevalence.

Background
It is widely accepted that South Africa is facing a quadruple burden of diseases [1], where the main causes of illnesses and deaths are due to four groups, namely; communicable diseases including HIV/AIDS and TB; Other infectious disease, maternal mortality, child mortality, and malnutrition; non-Communicable diseases (NCDs) and; violence and injuries [1][2][3]. Vertical transmission of HIV from mother to child can take place during pregnancy, delivery and lactation [4]. The World Health Organization (WHO) European regions especially the Eastern and Central sub-regions reported an increase of HIV infection in infants and children during the perinatal period and lactation from 347 to 494 per 1000 live births between the years 2004 and 2011 [5,6]. It is not clear what the rates are currently.
A 2009 to 2011 Kenyan study on the effectiveness of the Prevention of Mother-to-Child Transmission (PMTCT) programme showed improved immunity and viral load suppression of HIV positive pregnant women who were initiated on antiretrovirals (ARVs) [7]. In addition, 90% of their children were HIV negative [7]. UNAIDS previously estimated 330000 pregnant women to be HIV positive in South Africa in 2009, and 95% of them received ARVs through the PMTCT programme [8].
South Africa's PMTCT programme has gone through several phases from 2001, when the South African government was ordered by the country's apex court (constitutional court) to introduce effective PMTCT guidelines [9,10]. There have been several versions of the guidelines since then, starting with the 2003 version which placed emphasis on the provision of nevirapine, the extension of treatment to all HIV-infected pregnant women and their children and the expansion of other services like Voluntary Counselling and Testing (VCT) [9,10]. In addition, the Highly Active Antiretroviral Therapy (HAART) policy for individuals with clinical stage 4 HIV and/or a CD4 cell count of less than 200 cells/mm 3 was introduced [9, 10].
Despite guidelines being updated in 2008, a signi cant shift was in 2010 [11] where lifelong HAART could be initiated in HIV positive pregnant women with a CD4 cell count of less than or equal to 350 cells/mm 3 and/or in clinical stage 3 or higher [11]. All other pregnant women were to be placed on dual antiretroviral therapy from fourtheen weeks of pregnancy [11]. Whereas only doctors could initiate HIV positive patients on HAART before 2010, the 2010 guidelines also introduced Nurse Initiated Management of Antiretroviral Therapy (NIMART) [11]. Nyasulu et al. [12] noted in 2013 that the rolling out of NIMART had reduced the number of patients requiring initiation on ART in the main referral facilities; referral sites could focus more on complicated cases [12].
The 2015 guidelines [13] recommended for all HIV positive pregnant women to receive HAART from their rst antenatal care visit regardless of gestational age, clinical HIV stage or CD4 cell count; and all unbooked women who tested positive during labour would be given prophylactic treatment and initiated on lifelong ART before being discharged [13]. In 2016, the guidelines were updated further, to allow all HIV positive individuals to be enrolled on HAART regardless of CD4 count or HIV stage [14].
To effectively reduce the HIV incidence in newborns, requires an increased HIV testing uptake among pregnant women, a reduction of HIV prevalence amongst pregnant women, a suppressed viral load and a full understanding of the HIV epidemiology among pregnant women. With all the policy changes, assumptions are that the epidemiology of HIV would have also been affected, e.g. reduction in incidence, improved testing, reduced stigma, etc. There is, however, a literature gap of recent HIV studies focusing on women attending antenatal care in rural areas in South Africa.
The research aimed to determine the HIV antenatal prevalence, socio-demographic associations and the extent of the male partner involvement with pregnancy and antenatal care. Study ndings will provide valuable information to health providers and will assist in prioritising, planning and strengthening of the PMTCT programme. Health care users from similar settings in low-and middle-income countries (LMIC) will also gain information through community awareness and will be able to help minimise risks of HIV infection from mother-to-child.

Study setting
The Eastern Cape Province is one of nine of South Africa's Provinces. This is the Province with the second biggest surface area, the third most populous and the most rural with eight health districts [15]. Two of these (OR Tambo and Chris Hani) were purposefully selected for this study. The province was chosen for this study due to its rurality and size.

Study design, Population and Sampling
This quantitative cross-sectional study included all pregnant women with a gestational age of at-least 24-weeks, who used antenatal care services at the OR Tambo and Chris Hani Districts in the Eastern Cape Province, South Africa between March and November 2016. This gestational age was chosen to ensure that all women had been offered at least one HIV test and had results available in their antenatal care records. Study Participants were recruited from four community health centres. Participants under the age of 18 were included after providing written consent from both themselves and their parents or guardians. All other pregnant participants issued a voluntary informed, written consent to participate. Participants with missing date of birth and HIV status were excluded in the nal analysis.
A two stage-cluster sampling was used to recruit participants from four weighted facilities. Two Eastern Cape Health Districts were considered as the rst cluster level, (OR Tambo and Chris Hani district). The second cluster level of sampling was the convenient selection of two large community health centres (CHCs) in each of the two Districts. Thirdly, a convenience sampling technique was used to access pregnant women who met the inclusion criteria in each of the selected facilities.
A sample size of 352 was initially planned, based on using a one-sided 95% con dence interval and a 5% signi cance level (z = 1.96), an estimated antenatal HIV prevalence of 29.5% [16], a desired precision (d) of 5% and an addition of 10% to give allowance for non-responses.

Measurements
An interviewer administered questionnaire that was adapted from three instruments that have previously been used to measure HIV prevalence, PMTCT effectiveness and HIV stigma in developing countries [17][18][19]. It obtained information on socio-demographic characteristics and the medical history including HIV status (main outcome). The HIV status, the gestational age and antenatal care history were con rmed from the antenatal care card. Social desirability bias and language bias were mitigated through the use of a private interviewing space, training of researchers on professional conduct during interviews (e.g. phrasing of questions, avoidance of gestures, etc.) and the translation of the questionnaire into the local language (isiXhosa) respectively. The instrument was piloted among 12 pregnant women in the four study sites.

Statistical Analysis
Stata version 14.1 (STATA Corp, Colleage Station, Texas, USA) was used to analyse data. Missing data are treated as if they had never been offered to the participants. Numerical variables were explored using the Shapiro Wilk test. Numerical data were not normally distributed and thus reported on using non-parametric statistics (median and Interquartile Range (IQR)). The Wilcoxon Sum rank test (Mann-Whitney U test) was used to test for the equality of two medians, e.g. age in years by HIV status. Categorical variables are presented using, frequency tables, percentages and graphs.
Binomial logistics regression was used to determine the associations of an HIV positive status and to estimate the Prevalence Ratio (PR). The univariable models and the multivariable model selected through the forward selection process of model building are presented. The model with the lowest Bayesian Information Criterion (BIC) was selected as the better model. The 95% Con dence Interval (95%CI) was used to estimate the precision of estimates. The level of signi cance was set at 5% (p-value ≤ 0.05) for statistical signi cance.
The Walter Sisulu University Human Ethics and Biosafety Committee granted ethical clearance with ethics approval number (052/2016). The Eastern Cape Provincial Health Research Committee granted research access approval (EC_2016RP27_272).

Results
Socio-demographic and medical characteristics of participants are presented in Table 1. A total of 343/352 participants (97.4%; 95%CI: 95.2-98.8) were included in the nal analysis, of whom 38.2% (n = 131) were HIV infected. There was no signi cant difference in the HIV status of participants in the four health facilities. HIV positive participants (median age = 30) were signi cantly older than HIV negative participants (median = 25; p-value < 0.0001). The youngest participants were 15 years old, with 35 teenagers (10.2%) and the oldest were 43 years. High school learners and tertiary students comprised 33 (9.6%) and 32 (9.3%) of participants respectively. All participants knew their HIV status and the main reasons for having an HIV test included the fact that it was mandatory in antenatal care (44.0%); for health reasons or tests as a routine (47.2%); for sake of unborn baby (5.3%) and 3.5% reported to have tested due to being medically unwell.
The median gestational age at antenatal care booking for the 339 (98.8%) respondents who had complete information was 15.6 weeks (IQR 11.3-20.3) and there was no signi cant difference between HIV positive and HIV negative individuals. Whilst one patient had been on HAART for 13-years, at-least 80.3% (n = 102) of the participants had their HIV status diagnosed in the index pregnancy, with a median duration of diagnosis and initiation on HAART of 77 days (IQR 42-141).
Only 73 (21.3%) participants' partners were reported to accompany them to antenatal care. Most participants' partners (92.4%) were, however, reported to show an interest on the pregnancy. Partners did not attend antenatal care due to no longer being together (n = 1); they stayed far apart (37.4%); other commitments (41.9%); had never discussed it (8.9%); partner refused to come (5.9%) and 5.6% believed that males were not welcome in antenatal care.
In univariable analysis, the antenatal HIV prevalence was signi cantly higher in participants between the ages of 30-39 and 40-43 compared to teenagers (Table 2). In addition, multiparous, unemployed and employed participants, were signi cantly associated with 40%, 60% and 70% higher antenatal HIV prevalences respectively when compared to primigravidas or tertiary students. Other factors associated with a signi cantly higher antenatal prevalence included the perceived health status and spousal occupation.
Multivariable analysis had three variables in the best tting binomial logistic regression model (Table 2). Gravidity was not a statistically signi cant association of antenatal HIV prevalence (p-value = 0.116). Those older than 30 or unemployed, in employment or a high school leaner were signi cantly more likely to be HIV positive when compared to younger participants or tertiary students respectively.

Discussion
This study will hopefully add evidence to the already existing body of knowledge on South Africa's HIV epidemiology, especially among pregnant women. It is a unique study that does not only provide an update of the antenatal HIV prevalence but also seeks to use epidemiological data to inform health promotion practices in a rural South African environment. Lessons from this high HIV burden country will hopefully also be applicable to other LMIC and their planners. The major difference between this study and the South African antenatal surveys is that this study includes all pregnant women and not exclusively primigravidas [16,20]. This study has also been able to show that the HIV epidemic is maturing, characterised by an almost 40% antenatal HIV prevalence; a 100% proportion of women who know their HIV status; a higher HIV prevalence in older women, those who are unemployed, those who have had a previous pregnancy and those whose partners are locally employed. This study is also unique in that it has assessed the support provided by the male partner during antenatal care visits in public health facilities.
Of the 10.2% of teenagers interviewed in the study, 94.3% were high school students. This teenage pregnancy rate is lower than that described by Mchunu et al. [21] in a similar South African population wherein 19.2% of women reported to have fallen pregnant during their teenage years [21]. In that study, 74.1% of teenage pregnancies were attributed to lack of knowledge, and 55% claimed that they did not fully understand the risks involved with sexual intercourse [21]. Teenage pregnancy is known to be highly associated with a poor socio-economic status, risky sexual behaviours, violent behaviours and substance use [22]. Regardless of the percentage of teenagers who were pregnant, it cannot be ideal for school children who are themselves dependent on adults to be pregnant as this often has an impact on their long-term progress [21,22].
This is a poverty-stricken community with more than half of the women interviewed being unemployed (57.7%), never married (71.4%) and multiparous (63.9%). Almost 45% of the women's partners were employed locally, suggesting that there were economic opportunities locally that favour males. These compare to other antenatal care survey results such as that in another South African province (Limpopo), where 808 pregnant women were recruited, 51% from rural areas and 28% from peri-urban areas [23]. In that study both rural and peri-urban pregnant women had a high rate of being unemployed and being unmarried [23].
Encouraging is the fact that 50% of women had their rst antenatal care visit at 16 weeks. This is good as it allows adequate time for identi cation of congenital abnormalities, maternal or foetal risks and the suppression of the viral load if HIV positive, thus reducing the probability of Mother-to-Child-Transmission [5,6,[24][25][26][27]. This compares to antenatal care survey results of a Cameroonian study [28], of 293 pregnant participants, where 34% had started antenatal care in the rst trimester [28]. The explanations most commonly offered for a late antenatal presentation were nancial di culties and living a long way from the health facility [28].
The fact that all participants knew their HIV status and were open to disclose is positive and moves us closer to the goals of the 90-90-90 strategy [14,29]. This compares favourably with other similar studies where the HIV testing rate in antenatal care was was up to 99.0% [29]. Disclosure of HIV test results to sexual partners in a group of Ugandan women was relatively high and the ndings suggest that having a sexual partner who had also tested for HIV probably made it easier for women to disclose their HIV status [30]. The study also suggested the following as key factors in promoting disclosure: the need for promotion of sexual partner HIV testing; furthering of knowledge about HIV in women; and encouraging women to attend antenatal care [30]. The high testing rate is also a sign that health professionals were complying with HIV policies to encourage testing of all pregnant women so that they can be initiated on HAART without delays [13].
The data shows that antenatal HIV prevalence is increasing and higher than that presented in previous studies for same area [20]. The crude antenatal HIV prevalence of 38.2% is higher than the 31.9% (95%CI: 27.4-36.8) and the 33.3% (95%CI: 30.4-36.4) antenatal prevalences previously reported for Chris Hani and OR Tambo Districts respectively [20]. The differences could be attributed to the inclusion of multigravida women in this study [16,20]. The high prevalence could be a result of an increasing incidence most probably related to poor condom compliance and the concurrent reduction of HIV related mortality due to an improved antiretroviral programme [31]. There was no statistical difference in HIV prevalence between health facilities.
The 2015 South African National Antenatal Sentinel HIV and Syphilis survey reported a declining HIV prevalence for the Eastern Cape Province [20]. This contrasts with ndings from this study. The prevalence in the Eastern Cape province has stabilised since 2005, having increased by only 0.7% from 1990-2015 [20]. Over the 5-year period (2011-2015), the point prevalence estimate reached a peak in 2013 and 2014 at 31.4 (95%CI: 29.4% -33.5%); and declined by 1.2% in 2015 to 30.2% [20].
In women, the HIV risk has always been known to be decreasing with increasing age [1-3, 6, 16, 20, 31]. According to Stats SA [31], approximately 20% of South African women in their reproductive years (15-49 years) are HIV positive; however, HIV prevalence among those aged 15-24 has declined over time from 7.3% in 2002 to 4.6% in 2017 [31].
Several studies [1-3, 6, 16, 20, 31] have previously described a higher HIV incidence among teenage women since they had sex with both their peers and much older men [1-3, 6, 16, 20, 31]. Another South African study [32] previously reported a 'recent partnership' in 32% of young women aged 15-24 years to involve a partner ve or more years older than them [32]. Another 42% reported at least one age-disparate partner either in any of their three most recent partnerships or in their rst-ever partnership [32]. HIV prevalence was 29% among these women [32].
Pregnancy in early adolescence has been found to be associated with an increased incidence of HIV infection among South African women [33]. The higher risk is associated with sexual risk behaviour such as multiple partners and a greater age difference with partners [33]. This study, however, found a different phenomenon: the prevalence is higher amongst older women which suggests changes in the epidemiological characteristics possibly since the HIV infected women were infected many years previously but only knew their HIV positive status in the index pregnancy or it could well be a mark of an increasing incidence among older women. The aim of PMTCT programs is to improve the wellbeing of expectant mothers and to reduce the incidence of HIV among newborns [9,10]. It is therefore of interest to quantify the HIV MTCT trends among newborns in the same study population, especially since multigravidas were associated with a higher prevalence than primigravidas.
It is of little surprise that unemployed women had a signi cantly higher risk of being HIV positive than self-employed women. HIV is a disease of poverty [34], which further explains the increasing HIV prevalence in this community where more than 50% of the participants were unemployed [34]. Poverty may drive some women into risky sexual behaviours such as transactional sex and an inability to negotiate safer sexual practices with their partner [34]. Participants who reported to always use condoms during intercourse were possibly not being honest. After all, they were pregnant at the time of interview, making it unlikely that they always used condoms.
One participant reported her sexual debut to have consensual, without a condom and is the one that resulted in the index pregnancy which was unplanned. This kind of behaviour suggests an opportunity for the South African Department of Health to have more health promoters who can assist such women. The results further suggest that health promoters could help dispel myths about condoms. They should also advocate for latex free condoms for people who are allergic to latex condoms [35]. Plastic or polyurethane condoms are an effective barrier against HIV as well as bacterial sexually transmitted diseases while at the same time allowing better transfer of heat [35]. They are thinner than latex condoms and increase sensitivity [35]. However, plastic condoms are more expensive than latex condoms and less exible so more lubrication may be needed [35].
Women need to be empowered to be the ones who decide on the choice of condom use [36]. As evidenced in this study, 45.8% of women reported not using a condom because of the partner's preference. Women should therefore not blindly trust their partners but should instead take control of their sexual decisions and be empowered through interventions that encourage them to negotiate safe sex. They need to understand and acknowledge their own bodies, wishes and desires, in addition to being stimulated to participate in social movements to combat gender inequalities [36]. All these roles can be ful lled by quali ed health promoters [36].
Most women did not have an expectation for their partners to accompany them during their antenatal care visits. Spousal support during antenatal care can help improve acceptance and utilisation of preventive strategies in general and to an increased uptake of interventions to prevent vertical and sexual transmission of HIV [37]. Partner/couple counselling in the antenatal setting may have further bene ts to individual VCT [37]. In a Kenyan study, male antenatal care attendance was found to be associated with improved infant HIV-free survival [38]. Promotion of HIV testing in men and engagement in antenatal care services may improve outcomes in infants [38].
The non-involvement of partners in antenatal care services could discourage women in their ability to disclose their HIV-positive status due to fear of rejection, stigma and discrimination. It could also serve as a barrier to women beginning treatment and adhering to it and may disrupt HIV prevention services which could in turn result in poor HIV outcomes.
Even though attempts were made to reduce limitations the study encountered a few. Firstly, ndings from this study are not representative of the Eastern Cape Province as participants were recruited from only four health facilities. Findings from this study do, however, give a reasonable idea of the changing epidemiology of HIV in a rural environment amongst women attending antenatal care.
Secondly, the limited privacy during the interviews that occurred as a result of infrastructure challenges could have resulted in a social desirability bias, especially in questions pertaining to sexual behaviour and the use of condoms. Where this bias was noted results are reported truthfully. Lastly, only a few participants' CD4 cell counts and viral loads were included in the clinical records, as such these measures are not reportd altogether in this article. It is however unlikely that these limitations could have distorted the ndings on the epidemiology of HIV in this population especially since medical information was triangulated from clinical records.

Conclusion
It is in no doubt that a successful PMTCT program is bolstered by an early antenatal care attendance and high HIV testing rates. This was evident in this study with a median antenatal booking attendance of 15.6 weeks of gestation and an HIV testing rate of 100%. Whilst an e cient antiretroviral programme means that HIV infected individuals will live longer and will initially result in an increasing HIV prevalence it is important to reduce HIV incidence of both newborns and adults. The antenatal HIV prevalence of 38.2% is therefore high by any standards. It is therefore important to remain vigilant and monitor mother-to-child transmission that could be associated with this increased prevalence. Of further concern is the higher prevalence among the older participants who will soon also be at a higher risk of comorbid non-communicable diseases. Health systems must therefore be designed to address this burden using a comprehensive public health approach of disease prevention.
This population showed a very low uptake of spousal attendance in antenatal care. In addition to dealing with a high teenage pregnancy rate and myths around condom usage, health promotion practitioners have a huge role to play in assisting health systems to address the new HIV challenge of this century. This cadre of professionals can have an impact on HIV prevention as they will plan, and conduct awareness campaigns related to HIV/AIDS issues, reproductive health and family planning. Availability of data and material

Abbreviations
The datasets used and/or analysed during the current study are available on request from the corresponding author on request.
Ethics approval and consent to participate The Walter Sisulu University Human Ethics and Biosafety Committee granted ethical clearance and approval for the study to be conducted with ethics approval number (052/2016). The Eastern Cape Provincial Health Research Committee granted permission for the study to be conducted (EC_2016RP27_272). District managers of Chris Hani and OR Tambo Districts as well as the operational managers of the facilities concerned also granted permission for the study to be conducted. In addition to the fact that each participant gave informed written consent; con dentiality was maintained abiding by the four ethical principles of autonomy, bene cence, non-male cence, and justice. Participants under the age of 18 were included after providing written consent from both themselves and their parents or guardians.

Consent for publication
No personal identi ers or participants' images are contained in this study. The consent did however mention that ndings would be disseminated to scienti c audience, lay audience and policy makers.
SAM and KS conceptualised, designed, executed and did the main drafting of the manuscript and its revisions. SAM analysed the data and signed off on the nal version. WC sought funding and reviewed manuscript. TA and SN provided critical review and approved the manuscript.

Figure 2
Reported Condom Use

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. BMCInfectiousDiseaseresponseonPUBHD1900728Comments.docx