Antenatal care and uptake of HIV testing among pregnant women in sub-Saharan Africa: a cross-sectional study

Introduction Current guidelines recommend inclusion of HIV testing in routine screening tests for all pregnant women. For this reason, antenatal care (ANC) represents a vital component of efforts to prevent mother-to-child transmission (PMTCT) of HIV. To elucidate the relationship between ANC services and HIV testing among pregnant women in sub-Saharan Africa, we undertook an analysis of data from four countries. Methods Four countries (Congo, Mozambique, Nigeria and Uganda) were purposively selected to represent unique geographical regions of sub-Saharan Africa. Using Demographic and Health Survey datasets, weighted crude and adjusted logistic regression models were used to explore factors that influenced HIV testing as part of ANC services. The study was approved by the Institutional Review Board of the University of Arizona. Results Pooled results showed that 60.7% of women received HIV testing as part of ANC. Ugandan women had the highest rate of HIV testing as part of ANC (81.5%) compared with women in Mozambique (69.4%), Nigeria (54.4%) and Congo (45.4%). Difficulty reaching a health facility was a barrier in Congo and Mozambique but not Nigeria or Uganda. HIV testing rates were lower in rural areas, among the poorest women, the least educated and those with limited knowledge of HIV. In every country, crude regression analyses showed higher odds of being tested for HIV if women received their ANC services from a skilled attendant compared with an unskilled attendant. After adjusting for confounders, women in the total sample had 1.78 (99% CI: 1.45–2.18) times the odds of having an HIV test as part of their ANC if they went to a skilled attendant compared with an unskilled attendant. Conclusions There is a need for integration of HIV testing into routine ANC service to increase opportunities for PMTCT programmes to reach HIV-positive pregnant women. Attention should be paid to the expansion of outreach services for women in rural settings, and to the training, supervision and integration of unskilled attendants into formal maternal and child health programmes. Education of pregnant women and their communities is needed to increase HIV knowledge and reduce HIV stigma.


Introduction
Following the implementation of the 2003 ''United States Leadership Against AIDS, Tuberculosis, and Malaria Act (P.L. 108-25)'' that authorized the President's Emergency Plan for AIDS Relief (PEPFAR) [1], the use of antiretroviral therapy (ART) in high HIV burden countries was scaled-up and HIVrelated deaths have significantly declined over the past decade. The Joint United Nations Programme on HIV/AIDS (UNAIDS) estimates that 9 million life-years have been saved in sub-Saharan Africa alone as a result of wider availability of ART [2,3]. Nonetheless, sub-Saharan Africa remains the region with the largest proportion of pregnant women living with HIV, accounting for nearly 85% of the global burden [4]. Historically, HIV testing was primarily delivered via voluntary counselling and testing that emphasized client-initiated testing. Numerous studies reported the inadequacies of this model [5,6], leading the World Health Organization (WHO) to establish guidelines in 2004 that recommended routine HIV testing by healthcare providers for all pregnancy-related visits, especially in high HIV transmission areas [7]. As a result of the implementation of this approach in sub-Saharan Africa, HIV diagnoses in women are most often made during pregnancy [8].
Knowledge of HIV status allows pregnant women access to prevention of mother-to-child transmission (PMTCT) services [9]. PMTCT services are designed to reduce the risk of HIV transmission from a HIV-positive mother to her child [9]. Mother-to-child transmission (MTCT) Á also known as vertical transmission Á is the most common cause of paediatric HIV infection, often occurring during pregnancy, birth or breastfeeding [10]. In 2013, the WHO recommended two approaches to PMTCT for HIV-positive pregnant women: Option B' or, if this is not possible, Option B [4]. In Option B', all HIV-positive pregnant women are placed on ART for life, regardless of their CD4 count or WHO clinical stage [4]. In Option B, pregnant women with HIV who are eligible for ART (CD4 counts under 500 or clinical stage 3 or 4) should be treated with ART for life; however, those who are not eligible for ART should stop ART after delivery or a week after completion of breastfeeding [4]. The PMTCT guidelines also recommend optimal breastfeeding techniques for HIV-positive mothers living in developing countries: exclusive breastfeeding for the first 6 months of life, followed by continued breastfeeding combined with nutritionally adequate complementary feeding up to 24 months of age. Furthermore, since 2010, the PMTCTguidelines have recommended exclusive formula feeding for women for whom it is acceptable, feasible, affordable, sustainable and safe [11].
While uptake of PMTCT services in sub-Saharan Africa has increased from 3% in 2003 to 73% in 2014, there remains a substantial need to increase PMTCT interventions [5, 12,13] due to their direct effects on reduction of vertical transmission of HIV [14,15]. Antenatal care (ANC) provides an opportunity for integration of routine maternal and child health (MCH) services with HIV screening and thus represents a critical entry point into the PMTCT care cascade [16Á18]. Service integration has the potential to improve care and reduce missed opportunities for key interventions such as HIV testing, provision of ART, education in optimal breastfeeding and adherence support [19]. In fact, studies conducted in sub-Saharan Africa have demonstrated an association between service integration Á delivery of services or multiple interventions together at the same patient visit by the same health worker or clinical team Á and increased maternal HIV care enrolment and uptake of ART [16Á18]. However, in sub-Saharan Africa, uptake of both overall medical care and ANC is sub-optimal. Thus, it is important that women who do access ANC are provided all of the recommended medical tests (including HIV tests) and linked to appropriate care through effective service integration.
Missed In brief, DHS used a cross-sectional survey design to obtain nationally representative data. The surveys were typically designed to collect information on demographics, family planning, HIV and other aspects of health. Data included a standard set of questions; however, some survey items reflect country-specific issues. Only women who received ANC during their last pregnancy Á which was within 5 years preceding the survey Á were retained for the analyses in this paper.

Measures
In the DHS, an indicator for comprehensive knowledge of HIV was created using UNICEF's definition of comprehensive knowledge [22]. Women were classified as having comprehensive HIV knowledge if they correctly: 1) identified two methods of preventing sexual transmission of HIV; 2) acknowledged that a healthy looking person can have HIV; and 3) rejected two common misconceptions about HIV transmission, that is, HIV can be transmitted through mosquito bites or by sharing food with a HIV-positive person [22]. A skilled ANC attendant was defined using the WHO's standard as an accredited health professional including a doctor, nurse or midwife [23]. Otherwise, the attendant was classified as nonskilled. Stigma towards HIV was defined as unwillingness to care for a relative with AIDS, purchase vegetables from someone known to be HIV-positive or allow a HIV-positive female teacher to continue teaching [24]. The DHS variable on difficulty in getting to a healthcare facility measured distance as a barrier to accessing a healthcare facility. Data were elicited using the item: ''Getting medical help for self: distance to health facility'' with corresponding response options as: ''Big problem'' versus ''Not a big problem.''

Statistical analysis
Due to oversampling of certain populations, individual weights were used as recommended by DHS [25]. Using weights allows for adjustment for non-response to questions and makes the data more representative of the study population on a national level. Weighted descriptive statistics were generated to demonstrate the sampling weight of each variable. Weighted chi-square tests were used to assess the relationship between pregnant women's characteristics and uptake of HIV testing. Weights were applied in all regression models. Using weighted logistic regression, the relationship between ANC provider (skilled or non-skilled attendant) and uptake of HIV testing was adjusted by participants' characteristics (age, education, area of residence, difficulty accessing a healthcare facility, wealth index, stigma towards someone with HIV and comprehensive HIV knowledge). Statistical significance was set at p B0.01. All data cleaning, variable manipulation and analyses were conducted using SAS 9.4 (Cary, North Carolina) using the ''PROC SURVEY'' command. DHS surveys were conducted under the scientific and administrative oversight of the local country, including ethical review by the corresponding local ethics review board. Data collection procedures were also approved by the ORC Macro institutional review board. In addition, this secondary data analysis was evaluated by the Mel and Enid Zuckerman College of Public Health Research Office and was considered exempt from human subjects review.

Results
Of the 25,201 pregnant women included in the analyses, 14.6% were Congolese, 24.8% were Mozambicans, 46.2% were Nigerians and 14.3% were Ugandans. Table 1 presents a full description of characteristics of the Congolese participants. Briefly, 53.1% of the sampled women were aged 20Á29, 62.8% had secondary school education, 67.6% lived in an urban environment and 39.2% of the women reported difficulty getting to a health care facility. The women were generally evenly distributed along the wealth index: poorest (16.5%), poorer (22.4%), middle (22.0%), richer (20.8%) and richest (18.3%). Over half of the sample (52.1%) had a stigma towards someone with HIV, whereas less than half (38.9%) had comprehensive HIV knowledge. Most women in Congo (96.9%) saw a skilled attendant for ANC; however, less than half (45.5%) had been tested for HIV during that time. Table 2 shows the relationship between HIV testing during ANC and participants' characteristics. There was a significant association between being tested for HIV as part of ANC and education (pB0.0001); those with a tertiary education had the greatest uptake of HIV testing (71.9%) compared with those with secondary, primary or no education (50.1, 34.7 and 23.7%, respectively). There was also a statistically significant association between area of residence and receipt of HIV test as part of ANC (pB0.0001), with more women who reside in urban areas (51.6%) being tested for HIV during ANC compared with women who lived in rural communities (32.8%). Difficulty accessing a healthcare facility was also significantly associated with uptake of HIV testing (p00.004); only 40.2% of women who reported difficulty accessing a healthcare facility had an HIV test during ANC compared with 49.0% of women who had no such difficulty. Wealth index was also significantly associated with uptake of HIV during ANC testing (p B0.0001). As women's wealth quintile increased, so too did the percentage of women who partook in HIV testing as part of ANC (poorest, 24.4%; poorer, 34.5%; middle, 44.5%; richer, 57.8%; and richest, 65.4%). Women who expressed stigma towards someone with HIV were less likely to have an HIV test as part of ANC (37.0%) than those who did not express stigma (54.8%, p B0.0001). Comprehensive HIV knowledge was associated with uptake of HIV testing as part of ANC (p 00.0016); 50.5% of women with comprehensive HIV knowledge were tested for HIV as part of their ANC compared with 42.3% of women who lacked comprehensive HIV knowledge. Finally, more women who received ANC from a skilled attendant were tested for HIV test as part of ANC (46.1%) compared with those who received ANC from an unskilled attendant (25.8%, p 00.0003). No relationship was observed between participants' age and uptake of HIV test as part of ANC (p 00.1591). Table 3 shows outcome of crude and adjusted logistic regression models for the relationship between having an HIV test and whether the ANC attendant was skilled or unskilled. The crude models showed an increase in the odds of women who saw a skilled ANC attendant having an HIV test as part of ANC compared with those who saw an unskilled attendant for their ANC (OR: 2.46; 99% CI: 1.26Á4.81).

Congo
However, when adjusting for potential confounders, no association between skilled or unskilled attendant and uptake of HIV testing was observed (Adjusted OR [aOR]: 1.43; 99% CI: 0.79Á2.59). Table 1 presents the characteristics of respondents from Mozambique. Briefly, most women were aged 20Á29 years (49.3%), had a primary education (52.6%), lived in a rural setting (69.6%) and reported difficulty getting to a healthcare facility (55.5%). The women from Mozambique were also generally evenly distributed along the wealth index: poorest (20.6%), poorer (20.7%), middle (19.8%), richer (21.2%) and richest (17.7%). Less than half of the sample (36.9%) reported stigma towards women with HIV; however, only 32.1% had comprehensive HIV knowledge. More women (65.3%) received ANC from skilled attendants than from unskilled attendants (34.7%). Finally, more than two-thirds of women (69.4%) were tested for HIV as part of their ANC. Table 2 shows the relationship between participants' characteristics and receipt of HIV testing as part of ANC. There was a significant association between age and HIV testing as part of ANC (p B0.0001), with more women aged 20Á29 years having an HIV test as part of ANC (72.5%) compared with those 15Á19 years (68.4%), 30Á39 years (67.5%) and 40Á49 years (58.6%). Education was significantly associated with having an HIV test as part of ANC (p B0.0001). As education increased so too did the per cent of women who partook in HIV testing as part of ANC (no education, 58.2%; primary, 69.9%; secondary, 94.1%; and tertiary education, 98.7%). Type of living environment was also significantly associated with having an HIV test as part of ANC (p B0.0001), with more women in urban locations having an HIV test as part of ANC (82.9%) compared with those in rural locations (63.6%). Women who reported difficulty accessing a healthcare facility were less likely to be tested for HIV as part of ANC compared with those who did not report having difficulty accessing a healthcare facility (62.0 and 78.8%, respectively; chi-square pB0.0001). Wealth index was also significantly related to having an HIV test as part of ANC (pB0.0001). As women's wealth quintile increased so too did rates of having an HIV test (poorest, 53.1%; poorer, 58.0%; middle, 66.6%; richer, 79.7%; and richest, 93.0%). Fewer women had an HIV test as part of ANC if they displayed stigma towards someone with HIV (59.4%) compared with those who did not display stigma towards someone with HIV (75.3%; chi-square pB0.0001). Having comprehensive HIV knowledge was significantly associated with having an HIV test as part of ANC (pB0.0001); more women who displayed comprehensive HIV knowledge had an HIV test (75.3%) compared with those who lacked comprehensive HIV knowledge (66.7%). Finally, women who received ANC from a skilled attendant were more likely to have HIV tests during ANC (75.7%) compared with those who received ANC from an unskilled attendant (57.8%; chi-square p B0.0001). As shown in Table 3 for both crude and adjusted logistic regression models, an increase in the odds of having an HIV test during ANC was observed if the participant saw a skilled attendant for ANC compared with if they saw an unskilled attendant (aOR:1.58; 99% CI: 1.22Á2.05). Table 1 presents a description of characteristics of Nigerian participants. Briefly, 45.5% of the women were aged 20Á29 years, 40.2% had a secondary education, 52.1% lived in an urban setting and 20.7% reported difficulty accessing a healthcare facility. Women's wealth quintile was as follows: poorest (8.7%), poorer (15.9%), middle (20.8%), richer (25.8%) and richest (28.8%). Over half of the sample (62.1%) reported stigma towards someone with HIV, and few (37.6%) had comprehensive HIV knowledge. Almost all of the women (94.3%) received their ANC from a skilled attendant, but only 54.4% were tested for HIV during that time. Table 2 shows the relationship between being tested for HIV and participants' characteristics. There was a significant relationship between uptake of HIV testing and age (p B 0.0001). More women aged 30Á39 years (58.4%) were tested for HIV as part of ANC compared with those aged 15Á19 (37.0%), 20Á29 (53.6%) and 40Á49 (50.7%). Education was significantly associated with having an HIV test as part of ANC (pB0.0001). As education increased, so did uptake of HIV testing as part of ANC (no education, 30.0%; primary, 44.4%; secondary, 67.0%; and tertiary education, 89.1%). Women who lived in urban areas were more likely to be tested for HIV as part of ANC compared with those who lived in rural areas (66.9 and 40.7%, respectively; chi-square p B0.0001). Women were less likely to be tested for HIV as part of ANC if they had difficulty accessing a healthcare facility (40.7%) compared with those who did not have difficulty accessing a  healthcare facility during pregnancy (57.9%; chi-square pB0.0001). Wealth index was significantly associated with having an HIV test as part of ANC (pB0.0001). As women's wealth quintile increased, so did the per cent of women who received an HIV test during ANC (poorest, 30.0%; poorer, 32.1%; middle, 46.5%; richer, 59.6%; and richest, 77.6%). Women who displayed stigma towards someone with HIV were less likely to be tested for HIV during ANC compared with women who did not display stigma towards someone with HIV (47.7 and 65.3%, respectively; chi-square p B0.0001). Women who displayed comprehensive HIV knowledge were more likely to test for HIV as part of ANC compared with those who lacked comprehensive HIV knowledge (65.6 and 47.6%, respectively; chi-square pB0.0001). Finally, women who received ANC from a skilled attendant were more likely to have an HIV test (55.7%), compared with those who received ANC from an unskilled attendant (31.5%; chi-square p B0.0001).

Nigeria
As shown in Table 3, for both crude and adjusted logistic regression models, an increase in the odds of having an HIV test was observed if the participant saw a skilled attendant for ANC compared with if they saw an unskilled attendant (aOR: 1.95; 99% CI: 1.30Á2.93).

Uganda
Characteristics of participants from Ugandan respondents are shown in Table 1. Over half of the women sample (56.0%) were aged of 20Á29 years and had a primary school education (62.9%). A majority of the women (85%) lived in a rural setting and less than half (45.4%) indicated that they had difficulty accessing a healthcare facility. Women were spread fairly evenly across the wealth index: poorest (22.0%), poorer (21.0%), middle (19.6%), richer (18.1%) and richest (19.3%). Less than half of the sample (44.2%) displayed stigma towards people with HIV; however, only 35.5% had comprehensive HIV knowledge. Almost all of the women received their ANC from a skilled attendant (97.5%) and were tested for HIV as part of their ANC (81.5%). Table 2 shows the relationship between uptake of HIV testing as part of ANC and participants' characteristics. There was a significant association between receiving HIV testing as part of ANC and age (p 00.0002) with more women aged 15Á19 and 20Á29 having a HIV test as part of ANC (83.8 and 83.4%, respectively) compared with those aged 30Á39 (79.3%) and 40Á49 (71.0%). Education was significantly associated with having an HIV test as part of ANC (pB0.0001). As education increased, so did the per cent of women who received HIV testing as part of ANC (no education, 68.3%; primary, 81.5%; secondary, 88.1%; and tertiary education, 93.3%). Area of residence was also significantly associated with having an HIV test as part of ANC (p B0.0001); more women in urban locations had an HIV test as part of ANC (92.7%), compared with those in rural locations (79.6%). Women who experienced difficulty accessing a healthcare facility were less likely to be tested for HIV as part of ANC compared with those who did not have such a difficulty (79.4 and 83.3%, respectively; chi-square p 00.0061). Wealth index was also significantly related to having an HIV test as part of ANC ( B0.0001). As women's wealth quintile increased, so did the likelihood of having an HIV test (poorest, 75.8%; poorer, 77.5%; middle, 82.1%; richer, 83.4%; and richest, 90.1%). Fewer women received HIV tests if they displayed stigma towards someone with HIV (76.2%) compared with those who did not display stigma towards someone with HIV (85.8%; chi-square pB0.0001). Having comprehensive HIV knowledge was significantly associated with having an HIV test as part of ANC (p B0.0001), with more women having an HIV test if they displayed comprehensive knowledge (85.7%) compared with those who did not have comprehensive HIV knowledge (79.3%). If a skilled attendant was used for ANC, women were more likely to receive HIV tests (82.4%) compared with those who used unskilled attendants (49.4%; chisquare pB0.0001). Table 3 shows the results of the logistic regression. For both the crude and the adjusted models, an increase in the odds of having an HIV test was observed if the participant saw a skilled attendant for ANC compared with if they saw an unskilled attendant (aOR: 4.48; 99% CI: 2.48Á8.08).

All countries
Descriptive characteristics of the sample are summarized in Table 1. In brief, there were more women aged 20Á29 (49.0%) than any other age group: 36.8% had a primary school education, 56.3% lived in a rural setting and 35.6% stated that they had difficulty accessing a healthcare facility. Women's wealth index were as follows: poorest (14.7%),  Table 2 shows the relationship between HIV testing and the participants' characteristics. There was a significant relationship between uptake of HIV testing and age (p B0.0001). More women aged 20Á29 years (62.2%) received HIV testing as part of ANC compared with women aged 15Á19 years (56.7%), 30Á39 years (61.3%) and 40Á49 years (53.4%). Only in Uganda was uptake of HIV testing highest among the youngest group (15Á19 years, Table 2). For all countries overall, there was also a significant association between education and uptake of HIV testing during ANC (p B0.0001); those with a tertiary education (88.0%) had the greatest uptake of HIV testing compared with secondary, primary and no education (66.9, 61.5 and 43.1%, respectively). There was also a statistically significant association between living in a rural or urban location (p B0.0001), with more women who reside in urban areas (67.5%) being tested for HIV during ANC compared with women who lived in rural communities (55.5%). Difficulty accessing a healthcare facility was significantly associated with uptake of HIV testing (p B0.0001); fewer women reported receiving an HIV test during ANC if they had difficulty accessing a healthcare facility than if they indicated no such difficulty (55.9 and 63.3%, respectively). Wealth index was also significantly associated with uptake of HIV during ANC testing (pB0.0001). As women's wealth quintile increased, so did the percentage of women who received HIV testing as part of ANC (poorest, 44.5%; poorer, 46.9%; middle, 55.9%; richer, 66.7%; and richest, 80.6%). Having stigma towards people with HIV was also associated with uptake of HIV testing as part of ANC (p B0.0001) with significantly fewer women having an HIV test if they showed stigma towards someone with HIV (51.7%) than if they did not display stigma (70.4%). Lack of comprehensive HIV knowledge was also significantly associated with uptake of HIV as part of ANC (pB0.0001), with 68.2% of women who had comprehensive HIV knowledge being tested for HIV compared with 56.5% of women who lacked such knowledge. Whether the ANC provider was skilled or unskilled was also significantly associated with having an HIV test as part of ANC (pB0.0001), with more women having an HIV test if they saw a skilled attendant for ANC (62.1%) compared with those who saw an unskilled attendant (50.6%).
As shown in Table 3, in the adjusted logistic regression model, an increase in the odds of having an HIV test was observed if the participant saw a skilled attendant for ANC compared with if they saw an unskilled attendant (OR: 1.78; 99% CI: 1.45Á2.18).

Discussion
While mother-to-child transmission of HIV has almost been eliminated in many high-income countries, it remains an important source of new HIV infections in sub-Saharan African countries. According to the 2014 report of the Joint United Nations Programme on HIV/AIDS (UNAIDS), sub-Saharan Africa accounted for 87% of the 1.5 million pregnant women living with HIV and 91% of children living with HIV worldwide in 2013 [26]. In spite of continued effort and the availability of simple, relatively inexpensive and highly effective ART for PMTCT of HIV, 32% of the women in the aforementioned UNAIDS report did not receive ART for PMTCT resulting in an estimated 240,000 new infections in children [26]. The enormous burden that paediatric HIV infections places on families and the financial strain HIV exerts on health systems warrant careful attention to efforts to eliminate MTCT of HIV in resource-poor settings. In many sub-Saharan African countries, less than 40% of deliveries occur in hospitals [27]. Thus, there is a need to optimize opportunities for HIV testing of the few pregnant women who do access antenatal services. Identification of HIV-positive pregnant women through routine HIV screening is a critical step to initiate PMTCT interventions, which is why current guidelines recommend HIV screening be part of the routine panel of screening tests for all pregnant women [7]. ANC facilities represent a critical entry point into the PMTCT care cascade. It is hoped that this paper will help to inform policy and practice related to integration of HIV testing in routine ANC services.
Overall, results of our analyses indicate that although a majority of the women (88.0%) received ANC from a skilled attendant, only 60.7% received HIV testing as part of ANC. Ugandan women had the highest rates of HIV testing as part of ANC with 81.5% of women tested compared with only 69.4% of Mozambican women, 54.4% of Nigerian women and 45.5% of Congolese women. For the overall sample, and for each country excluding Nigeria, women aged 40Á49 years were least likely to receive HIV testing as part of ANC. In Nigeria, women aged 20Á29 years were least likely to receive HIV testing as part of ANC. These age differences may reflect a need for educating ANC providers that women of all ages and relationship statuses can contract HIV and should receive HIV testing. Difficulty accessing HIV testing remains a problem in each country Á except Uganda Á with much lower HIV testing rates in rural areas, among the poorest and the least educated. The high rates of HIV testing among women in Uganda where a significant proportion of the population resides in rural areas may be due to local policies that ensure higher HIV testing rates such as the opt-out testing during ANC and mandatory testing for people receiving care in public health centres [6,7]. HIV testing rates were higher among women who did not express stigmatizing attitudes towards persons living with HIV/ AIDS and among those with comprehensive HIV knowledge. In every country, crude regression analyses showed higher odds of being tested for HIV if the women saw a skilled attendant compared with an unskilled attendant for ANC. In the total sample, women had higher odds of having an HIV test as part of their ANC if they went to a skilled attendant compared with if they went to an unskilled attendant (aOR: 1  (45.5%). This is particularly troublesome as this is a missed opportunity for HIV education and entry into PMTCT care cascade.
Overall, the findings of our study are consistent with the literature on access to ANC services and uptake of HIV testing among pregnant women in low-income countries. For example, in a community-based, cross-sectional survey of mothers in Western Kenya, Kohler et al. [28] found that stigma and lack of HIV knowledge were significantly associated with a decreased likelihood of maternal HIV testing. Another study in rural Kenya [29] also established that women who demonstrated higher HIV-related stigma (i.e. those who held negative attitudes about persons living with HIV) were less likely to deliver in a health facility with a skilled attendant. As the authors explained based on their qualitative data, childbirth at a health facility was commonly viewed as most appropriate for women with pregnancy complications (e.g. HIV); therefore, women who deliver at health facilities may be labelled as HIVpositive in the community [29]. Age, education and residence have also been associated with being offered HIV counselling or testing, with younger and more educated women and those residing in urban settings more likely to be offered HIV counselling and testing than older, less-educated and rural women [30]. It was not surprising that women who receive their ANC services from unskilled health workers were less likely to receive HIV testing compared with those who received their service from skilled workers, because unskilled workers may not have access to HIV testing resources.
The findings from this study have implications for policy and practice regarding PMTCT of HIV in sub-Saharan Africa. They underscore the need for PMTCT and MCH programmes to pay particular attention to socioeconomic determinants of access to HIV testing as part of ANC services. Evidence from this study and several others [31Á34] has revealed that women who lack comprehensive knowledge about HIV demonstrate stigmatizing attitudes towards persons living with HIV, have low educational levels, live in rural settings with difficult access to reproductive health services and receive their ANC services from unskilled workers are less likely to receive HIV testing as part of their ANC. This underscores the need for public health programmes in high HIV burden lowincome countries to consider focusing their meagre resources on targeted efforts to reach pregnant women of these socioeconomic backgrounds. MCH programmes in sub-Saharan Africa should invest in efforts to reduce HIV stigma among pregnant women since stigma remains an important barrier to the uptake of HIV testing. Interventions that use peer-educators [35,36] as well culturally appropriate channels, such as faith-based institutions, can be effective not only in reducing stigma-related barriers but also geographical barriers [36]. Similarly, integrated testing approaches where HIV testing is considered as one of many tests related to maternal health, instead of HIV-only testing, can be effective in reducing the stigma and anxiety that are often associated with HIV testing [37,38]. Although it has been decades since the discovery of HIV/AIDS, there remain significant misconceptions about the disease. With the availability of ART at a relatively low direct cost to recipients in many low-income countries, the notion of HIV/AIDS as a death sentence should be dispelled by MCH programmes. The transmission of HIV from mother-to-child during pregnancy, labour, delivery or breastfeeding can be as high as 45% in the absence of interventions, but with early diagnosis, prompt initiation of care and adherence to PMTCT services, the rate can be reduced to less than 5% [39,40]. Early diagnosis and prompt enrolment into care also helps to preserve the health of the pregnant woman while reducing the chances of HIV transmission to a male partner who is not infected [41,42].
Since it is known that provider-initiated modes of testing make testing accessible to women from lower socioeconomic groups, ANC services should promote HIV counselling and testing as an essential component of ANC service for all pregnant women. Since many women, especially low-income women in rural settings, still rely mostly on unskilled workers for their antenatal services, MCH programmes should invest in efforts to train and integrate such unskilled workers into the formal ANC delivery system [43,44] and to provide them with access to HIV testing resources. With proper training and supervision, they can also be effective channels for dissemination of factual information about HIV and the benefits of testing and treatment and for referral to available treatment services [45,46].
The DHS data used for this study are not without limitations. The data from Congo, Mozambique and Uganda were collected in 2011Á2012 and that from Nigeria in 2013. It is not known how well these data represent the current relationship between HIV testing and ANC care in their corresponding countries or in sub-Saharan Africa. It is also not known if there were any policy changes or events that may have affected the cohort in Nigeria in the two years after data were collected in the other three countries. Furthermore, there are several important questions that the DHS does not assess, including whether women knew their HIV status or were already on ART prior to ANC, how many ANC visits women had overall or in the first or second trimesters or how early HIV testing was performed. While these questions are beyond the scope of the DHS, they are important to know when considering the implementation and success of PMTCT services. Answering these questions would be an important future direction for empirical inquiry.

Conclusions
This paper reveals individual and health system factors that influence women's uptake of HIV counselling and testing as part of their ANC services in sub-Saharan Africa. At the individual level, age, poverty, lack of HIV knowledge, stigma towards persons living with HIV/AIDS and low educational levels were associated with reduced uptake of HIV testing as part of ANC services. At the health system level, difficulty in accessing a health facility, rural residence and receipt of ANC services from unskilled attendants were associated with low HIV testing as part of ANC services. These findings have clear implications for the organization and delivery of ANC and PMTCT services in low-and middle-income countries (LMICs). As is often the case with vertical programmes in LMICs, ANC services may not be seen as routine but only necessary when something is amiss. Thus, health education may be necessary to change cultural norms and reduce missed opportunities for HIV testing of pregnant women. Outreach services may also be necessary as a strategy for reaching women in rural and difficult-to-reach settings. The value of such outreach services in MCH programming has been amply demonstrated in the literature [47,48]. Typically, in order to reduce adverse MCH outcomes, it is recommended that pregnant women receive their ANC services under the supervision of skilled attendants [49]. However, for cultural, geographical and economic reasons, the reality remains that a significant proportion of women in LMICs have limited access to either health facilities or skilled attendants. This underscores the need for ANC outreach programmes in these settings to identify unskilled attendants who provide ANC services to women in their jurisdiction so that they can be trained, integrated into the formal ANC delivery system, and regularly supervised [50]. Their propensity to provide pregnant women with HIV/AIDS education, refer them for testing and support those found to be HIV-positive through the PMTCT cascade of care could be enhanced through an incentive mechanism that rewards them per client referred to ANC or PMTCT HIV screening centres [51]. An alternative possibility is to provide unskilled ANC workers with access to HIV testing materials along with appropriate training (e.g. on testing procedures, result interpretation, maintenance of client confidentiality) and supervision. Given the association between knowledge of HIV, AIDS-related stigma and HIV testing, efforts should be directed at community education and sensitization about HIV and the importance of knowing one's status. The notion that HIV is a death sentence still prevails, especially in rural settings, and continues to create a barrier against the utilization of available services [28]. Education about HIV/AIDS in communities should focus on providing basic, factual information and on reducing prevailing misconceptions. The fact that treatment is available to help prevent transmission of HIV to children born to HIV-positive women should also be emphasized.