Vertical transmission of HIV from mother to fetus during pregnancy, delivery and the postpartum period can be avoided in almost all cases through existing interventions, such as use of antiretroviral therapy, elective cesarean section and avoidance of breastfeeding [1]. Nonetheless, it has been estimated in 2009 that, worldwide, 370,000 children become infected with HIV every year [2].

Brazil has a concentrated AIDS epidemic with a stable prevalence estimated to be less than 1% [3]. Increased heterosexual transmission has resulted in an increased number of cases among women, while vertical HIV transmission from mother to child during pregnancy and delivery accounts for almost all of the cases reported among children [3]. Based on the last Sentinel Parturient Studies, conducted in 2006 [4] and 2010 [3], it was estimated that approximately 0.4% of pregnant women in Brazil have HIV infection. Given the number of cases of HIV infection reported among pregnant women in 2012, it was inferred that HIV surveillance among pregnant women only reached 58.3% of expected cases, [3] thus indicating that there were some difficulties in diagnosing and/or reporting cases.

In Brazil, several government initiatives have made tests, medications, infant formula feeding and care routines available for managing HIV during pregnancy, delivery and the postpartum period since the end of the 1990s [5]. Studies with national coverage in Brazil have shown that reductions in vertical transmission started to occur from 1997 onwards [6-9]. The estimated vertical transmission rate was 6.8% in 2004, with a range from 13.4% in the North region to 4.3% in the Midwest region [10]. Local studies conducted in the Southeast region [11-14] have reported vertical transmission rates closer to 3%, while rates of less than or equal to 2% have been found in the Midwest and South regions [15-17]. In contrast, in the North and Northeast regions and in the state of São Paulo, local studies have found vertical transmission rates of 9.9% [18], 9.2% [19] and 6.0% [20], respectively.

Thus, despite advances, flaws in the Brazilian control program for HIV during pregnancy have prevented reaching vertical transmission rates of less than 2%, like the rates found in some developed countries such as the United Kingdom [21]. The Pan-American Health Organization (PAHO) has a target of eliminating vertical transmission of HIV from the Latin American and Caribbean region by 2015, which translates into reaching a vertical transmission rate of less than or equal to 2% and an incidence of cases of vertical transmission of HIV of less than or equal to 0.3 per 1,000 live births [22].

Identification of HIV-infected pregnant women is the first stage towards adopting the protocols available for both prevention of vertical transmission and treatment of women. Local studies [13,15,18,19,23,24] have indicated that the coverage of HIV testing has increased in all Brazilian regions. However, these results were from studies with municipal or state coverage while no studies with national coverage that could adequately provide estimates for the country have been conducted recently.

The aims of the present study were to ascertain the coverage of HIV testing during prenatal care, to evaluate the factors associated with performing at least one HIV testing during pregnancy and to estimate the prevalence of HIV infection among pregnant women in Brazil through a survey with national coverage that was conducted among puerperal women who underwent delivery in hospitals in 2011–2012.

The study “Birth in Brazil: national survey into labour and birth” was a hospital-based study conducted between February 2011 and October 2012. Additional information on the methodology of the study “Birth in Brazil” can be obtained from Leal et al. [25].

The sample was selected in three stages. In the first stage, hospitals with more than 500 deliveries per year were stratified according to the five macroregions of the country (North, Northeast, South, Southeast and Midwest), according to location (state capital or elsewhere in the state) and type of hospital service (public, mixed or private), and 266 hospitals were selected with a probability of selection proportional to the number of deliveries in each of the strata in 2007. The minimum size of the sample in each of the strata was 450 puerperal women. In the second stage, the number of days needed to interview 90 puerperal women in each hospital (minimum of 7) was selected by using an inverse sampling method. In the third stage, the women eligible on each day of the fieldwork were selected [26]. Sample losses due to refusal to participate or hospital discharge were replaced by selecting new puerperal women at the same hospital. In total, interviews were conducted with 23,894 women.

All puerperal women with live births in a hospital in which the fetus had any gestational age or weight, or a stillbirth in which the fetus had a gestational age of more than 22 weeks recorded in the medical file or weight greater than 500 grams, were considered eligible for the study. Miscarriages were excluded as the aim of the study was to evaluate the conditions of prenatal, birth and delivery care and the results from the care provided.

Interviews with the puerperal women were conducted during their hospital stay, at least 6 hours after delivery. Data from the medical files of the puerperal woman and newborn were obtained at the time of hospital discharge. In case of prolonged hospitalization, data from medical files were obtained at the 42nd day of hospitalization in the case of the puerperal women or the 28th day of life in the case of the newborns. For the interview and extraction of data from the medical files, electronic forms that had been developed specifically for this study were used. When available, prenatal cards, which are official forms provided by health services to all pregnant women during prenatal care where tests, procedures, and prescription drugs are recorded, were photographed in digital media with subsequent extraction and input of data into an online platform. This was done by a team of students and healthcare professionals who had been trained and were supervised by the central coordinating team of the study.

To evaluate the prenatal care, the protocols of the Ministry of Health were used. These recommend that prenatal care should begin by the 12^th gestational week [27], with a minimum of six consultations during follow-up. Furthermore, there should be two HIV serological tests over the course of the pregnancy: the first serological test should be requested at the time of the first prenatal visit and the second serological test should be requested at the start of the third trimester of the pregnancy [28].

In the present study, the following prenatal care indicators were estimated: proportion of puerperal women who reported having at least one prenatal consultation; proportion of puerperal women who reported having received a prenatal card; proportion of puerperal women who had their prenatal card at the time of hospital admission for delivery; and proportion of puerperal women with records of the first and second HIV tests on their card. It was assumed that the tests had been carried out when there was a record of any serological result relating to HIV. Other data on prenatal care that were analyzed in the “Birth in Brazil” study are available in Viellas et al. [29].

For the diagnosis of HIV infection during pregnancy, one of the following were considered: 1) reactive results recorded on the prenatal card (two rapid tests or Elisa + immunofluorescence or Elisa + Western Blot^a); and/or 2) records in the hospital medical files of a) diagnosis of HIV infection or b) indication of cesarean due to HIV infection or c) use of zidovudine during delivery labor and/or delivery or d) use of zidovudine syrup for the newborn or e) suspension of breastfeeding due to maternal HIV infection or f) record of the diagnosis of “child exposed to HIV”.

The prenatal care indicators were analyzed according to the following maternal characteristics: “region of residence” (North, Northeast, Southeast, South or Midwest); “age” (12 to 19 years, 20 to 34 years or 35 years and over); “schooling level” (up to 7 years, 8 to 10 years or 11 or more years of school attendance); “self-reported skin color” (white, black, mixed, East Asian or indigenous); “conjugal situation” (with or without a partner); “type of prenatal care service” where most of the consultations were attended (public or private); and “type of delivery care service” (public, mixed or private; mixed services were taken to be those that attended to women with either public or private funding for delivery care). The chi-square statistical test was used to ascertain differences between proportions.

To evaluate the factors associated with performing at least one HIV testing during pregnancy, univariate and multivariable logistic regressions were performed. In addition to the maternal characteristics, the following variables relating to prenatal care were included: the time when prenatal care began measured in gestational weeks (up to week 12, between weeks 13 and 28, or week 29 and beyond); and the adequacy of the number of consultations up to the time of delivery, considering a minimum of six consultations for a term pregnancy, adjusted for the gestational age at birth (sufficient or insufficient), according to the criteria defined by the Ministry of Health.

To evaluate the factors associated with HIV infection among pregnant women, univariate and multivariable regressions were performed. In addition to the maternal characteristics described above, “diagnosis of syphilis infection during pregnancy” (yes or no) was also included, as some authors point to high rates of coinfection [30,31]. The women were considered to present with syphilis infection if they presented with one of the following criteria: record of a reactive serological test on the prenatal card or hospital medical file; record of syphilis diagnosis on the prenatal card or hospital medical file; or case of congenital syphilis as an outcome from the current pregnancy.

A univariate analysis estimated the unadjusted odds ratio (OR) and 95% confidence interval (95% CI). Variables with a significance level estimated as lower than 0.20 were included in an initial multivariable model and those with a significance level of <0.05 were retained in the final multivariable model. The results from the multivariable model were expressed as adjusted odds ratios with their corresponding 95% confidence intervals (95% CI).

In all the statistical analyses, the complex sampling design was taken into consideration. The data weighting was calculated according to the inverse of the probability of inclusion of each puerperal woman in the sample. To ensure that the distribution of the puerperal women interviewed was similar to that observed among the births in the population sampled in 2011, a calibration procedure was used in each selection stratum [26]. For the univariate and multivariable logistic regression, the women who self-reported as East Asian or indigenous were excluded because they accounted for a very small proportion of the sample. Analyses were performed using IBM SPSS software version 19.0 (IBM Corp., Armonk, USA).

This study was approved by the Research Ethics Committee of ENSP/Fiocruz, under report no. 92/2010. Every care was taken towards ensuring privacy and confidentiality of the information. Before each interview was conducted, the interviewee’s consent was obtained, after reading the free and informed consent statement.

Fewer records of results from the first HIV test were observed in the North and Northeast regions; among women with a lower schooling level; among women who self-reported as black, mixed and East Asian; among those under 20 years of age; among those without a companion; and among those with prenatal and delivery care provided in public clinics. A similar pattern was observed for the results from the second HIV test, except in relation to the location where prenatal consultations took place: there was no difference between public and private clinics (Table 2).

The results from the adjusted analysis (Table 4) showed that women living in the South region continued to have the greatest odds ratio of HIV infection (OR = 2.95; 95% CI 1.78-4.91). After adjustment for the other variables, the strength of association with the woman’s age increased, such that HIV infection among women aged 20 years or over was four to six times greater than that of adolescents. HIV infection was also associated with syphilis coinfection (OR = 4.75; 95% CI 2.01-11.21), with self-reported black skin color (OR = 2.89; 95% CI 1.57-5.31), with living without a partner (OR = 1.87; 95% CI 1.21-2.87), with less than 8 years of schooling (OR = 2.26; 95% CI 1.28-3.98) and with delivery care provided in public clinics (OR = 4.08; 95% CI 1.65-10.08). If the births that took place in the public and mixed clinics are analyzed together, the estimated prevalence was 0.45% (95% CI: 0.36-0.58%), a rate that was significantly higher than that observed in the private clinics (0.10%; 95% CI: 0.04-0.22), with an OR of 4.67 (95% CI: 2.02-10.80).

The results of this study demonstrate that coverage of prenatal care with at least one consultation is practically universal in Brazil, reaching levels greater than 90%, irrespective of the geographical region of the country or maternal characteristics. Nonetheless, a slightly lower coverage was observed in the North region and among women with less schooling, indigenous women, and women without a partner, probably as a result of geographical access, and cultural and social barriers. Provision of a prenatal card was a well-established practice, without regional differences, and this result was similar to that observed in a recent national study [32].

More than 80% of the women who presented with their prenatal card at the time of hospital admission for delivery had prenatal cards with the results from the first HIV serological test. This proportion was higher than that observed in the Sentinel Parturient Study of 2006 [4], when 62.8% of the pregnant women analyzed presented with an HIV serological test. Our results thus corroborate the findings from recent local studies [13,18,19,23,24], which indicated coverage close to 80%. Although a notable increase in this coverage was seen, regional differences were still observed. The South region was the only one to attain coverage greater than 90%, which is the target defined by World Health Organization (WHO) for vertical HIV transmission control programs [33], but this is still lower than the coverage of 95% proposed by PAHO for the Latin American and Caribbean region [22].

Younger women, with less schooling and who self-reported as black or mixed, who were living in the North and Northeast regions, and whose prenatal care was provided in public clinics, with a late start and an insufficient number of consultations, were the ones who presented with the lowest coverage of at least one HIV test during pregnancy. These results show the importance of an early start to prenatal care and an adequate number of consultations for undertaking routine examinations, as already demonstrated in other studies [23,34,35], thus enabling timely implementation of care protocols for preventing vertical transmission. The results also reveal inequalities in control measures that may not reach women with greater social vulnerability, i.e., younger women with less schooling and black and mixed skin color, who live in less developed regions of the country. It is important to note that this state of inequality has remained unchanged over the years, given that these results have already been observed in national studies conducted at the beginning of the 2000s [4,34-36]. Similar inequalities have been described in Africa [37].

The rate of performing a second test is still poor in the country: less than 30%, with marked regional differences. Although differences in coverage by characteristics of women have been observed, the coverage attained has been low, even among those in a better socioeconomic condition who were attended in private clinics. The Brazilian Ministry of Health introduced the second serological test for HIV in 2006 [38], as part of the prenatal care routine. The low coverage observed reveals that there have been problems in implementing this recommendation.

Brazilian Guidelines recommend that all pregnant women without any previous serological test or without serological results from the third trimester of pregnancy have a rapid HIV test performed at admission for delivery. The observed low coverage of the second serological test, both in public and in private clinics, gives rise to the need to perform testing on a greater number of women at the time of admission for delivery, thus overloading these services and making it difficult to diagnose the infection, hindering timely implementation of preventive measures against vertical transmission. Moreover, other aspects of care are also impacted, such as delaying the start of breastfeeding after delivery [39].

The prevalence of HIV infection among the pregnant women of this study (0.40%) was similar to that observed in the Sentinel Parturient Studies conducted in 2004 (0.41%) [40], 2006 (0.41%) [4], and 2010 (0.38%) [3]. However, because private clinics were not included in these Sentinel Studies, comparisons between these and our data are difficult. When the births that took place only in the public and mixed clinics of our study are analyzed separately (i.e. using the criteria of the Sentinel Studies), the estimated prevalence was significantly higher than that observed in the private clinics. This is a novel result for Brazil, and it demonstrates that there is a greater prevalence of infection among female users of clinics within the Brazilian National Health System (SUS), which is a population formed by women of poor socioeconomic condition.

The South region had the highest prevalence of HIV infection among pregnant women in this country, and the North region had the second highest prevalence, although with no significant difference in comparison to the Southeast region. Data from local studies have also shown a higher prevalence in the North (0.6%) [41] and South regions (1.7%) [42]. Studies conducted in the Southeast [24], Northeast [43] and Midwest [44,45] regions have found prevalence values not statistically different from those estimated in our study for these regions.

The results indicated a higher prevalence of HIV infection with increasing age. The decline in mortality due to AIDS in Brazil since the 2000s [3], with increasing survival among HIV-infected individuals [46,47], is one possible explanation. The high proportion of pregnant women with diagnoses of HIV infection prior to the current pregnancy, which has been reported in several national studies [11,13,14,18,19], suggests that women who have been infected for longer periods are becoming pregnant, which may also explain the increase in prevalence with age.

A higher prevalence was also observed among women with less schooling, who were single, who self-reported as black and had diagnoses of syphilis infection. Syphilis infection is a known risk factor for HIV infection [30,31], and a combined approach to these two diseases is recommended by WHO for better results in programs for preventing vertical transmission of both diseases [48].

The greater prevalence of cases among women with less than 8 years of schooling, who were single, who self-reported as black and as users of public clinics may reflect greater social vulnerability in these groups. However, the absence of other variables relating to the women’s behavior, such as sexual partnerships, drug use and access to condoms, makes it more complex to interpret the associations encountered. A study conducted in 13 Brazilian cities [49] found that there was greater use of drugs, an earlier start to sexual life, low adherence to condom use and a greater proportion of histories of STD and sexual violence among the women living with HIV/AIDS. It should be emphasized that black women, those with less schooling and those attended in public clinics, who had high odds ratios for HIV infection, were the women who presented with the lowest coverage of testing. This reveals missed opportunities for making diagnoses in the population most affected by the infection.

This study was conducted in institutions where more than 500 deliveries take place every year. It is likely that pregnant women who deliver at home or in smaller hospitals have different degrees of access to prenatal services and routine examinations. Nonetheless, given that more than 99% of deliveries in Brazil take place in hospitals, and approximately 80% in larger hospitals [25], significant changes to the results presented here would not be expected.

One limitation of the present study was that the results from the HIV tests were only ascertained among the women who presented with the prenatal card at the time of hospital admission for delivery (71.6%). However, the estimates presented here for the coverage of HIV testing during pregnancy are not thought to be overestimated, because the women who least often presented with a prenatal card were the ones with the most schooling and who were attended in private clinics, which also had higher test coverage. This less frequent presentation of the card probably resulted from the care model in the private sector, in which the same professional provides the prenatal and delivery care. Thus, it is unlikely that these women who did not present a card had done anti-HIV tests less often during pregnancy.

Another limitation of the present study was that the serological tests were not performed in a standardized manner, with use of secondary data to estimate the prevalence of HIV infection. It is possible that failures in recording the medical file data may have led to underestimation of this infection in the cases of women who did not present a prenatal card on admission for delivery.

Finally, our eligibility criteria excluded women with miscarriages. This exclusion may have introduced bias in the estimated prevalence of HIV infection in pregnancy if women with miscarriages tend to have an increased likelihood of HIV infection [50,51].

The present study found a 0.40% prevalence of HIV infection among pregnant women, an almost universal coverage of antenatal care, and coverage of prenatal HIV testing of over 80%. Nevertheless, regional and social inequalities in accessing healthcare services, the low implementation of the second anti-HIV serological test, and the lower coverage of testing among women with characteristics associated with greater prevalence of HIV infection indicate shortcomings in the HIV infection control programs during pregnancy. These missed opportunities probably contributed to the high vertical transmission of HIV infection in specific group populations, despite the available resources guaranteed by public policies since the 1990s. This result stresses the importance of designing and implementing interventions targeted to the people most affected by the epidemic that would most benefit from the interventions available to prevent mother-to-child transmission of HIV infection. The monitoring of the prevalence of HIV infection among pregnant women to identify the groups most affected by the epidemic, as well as the monitoring and evaluation of antenatal care and HIV program indicators, are important tools for the development of these specific interventions.