Vitória , ES HIV prevalence in pregnant women and vertical transmission in according to socioeconomic status , Southeastern Brazil

MÉTODOS: Estudo ecológico exploratório utilizando a base de dados do Sistema de Informação de Agravos de Notifi cação de gestantes HIV-positivas e aids em crianças notifi cadas entre 2000 e 2006 em Vitória, ES. Para análise das informações socioeconômicas foi utilizado o Índice de Qualidade Urbana. A prevalência de HIV em gestantes e a taxa de transmissão vertical foram calculadas. A distribuição espacial dos casos foi realizada no programa Terraview 3.2.0. Para verifi car a associação entre a qualidade urbana e a prevalência de HIV em gestantes utilizou-se o modelo de regressão de Poisson.


INTRODUCTION
The AIDS epidemic appears to have stabilized in Brazil, although at a high level.Incidence rates were 18.7 cases per 100,000 residents in 1998 and 10.0 cases in 2008.a Since people of reproductive age are most affected by the infection, the number of children infected by vertical transmission (VT) is high.b a Ministério da Saúde.Secretaria de Vigilância em Saúde.Programa Nacional de DST e Aids.Boletim Epidemiológico Aids e DST.Brasília, DF; 2010.b Ministério da Saúde.Secretaria de Vigilância em Saúde.Programa Nacional de DST e Aids.Recomendações para profi laxia da transmissão vertical do HIV e terapia anti-retroviral em gestantes.Brasília, DF; 2007.
b A study in 2009, with a representative sample of pregnant Brazilian women between ages 15 and 49 years, reported an HIV prevalence of 0.42% at delivery, which corresponds to 13 thousand infected women in 3 million annual pregnancies. 17In Espírito Santo State (Southeastern Brazil), from 1985 to 2008, 329 AIDS cases were reported among children age less than 13 years, with VT responsible for 310 (94.2%) cases.c In the capital Vitória, a 2005 study reported a VT rate of 3.1%. 15derstanding the distribution of HIV-infected pregnant women in the community is necessary to estimate risk of VT and guide preventive measures.2][3] The study objective was to describe the relationship between HIV infection in pregnant women and VT, according to socioeconomic profi le of the mother's residential neighborhood.

METHODS
An exploratory ecologic study was performed in Vitória, between 2000 and 2006.The city has a land area of 93,381 km 2 and is divided in 79 neighborhoods, which were the unit of analysis.d Epidemiologic data was obtained from forms reporting HIV-positive pregnant women.Data of children less than 13 with an AIDS diagnosis was obtained from the respective reports.These cases were linked to cases of mothers, by searching reports of HIV-positive pregnant women in the period from 2000 to 2006.These data were made available by SEMUS, using notifi cation/ investigation forms from the Sistema de Informações de Agravos de Notifi cação (SINAN -Information System of Notifi able Conditions).Data was entered in an electronic worksheet on Excel®, and prevalence rates were calculated with Stata version 9.0.
The prevalence of infected pregnant women was calculated.The number of reported cases in each neighborhood was divided by the respective population of pregnant women, estimated by the number of live births 6 in the study years and multiplied by 100.To analyze the spatial distribution of HIV cases among pregnant women by neighborhood, average prevalence rates were calculated.
To calculate the VT rate, the numerator was the number of reported AIDS cases in the municipality among children less than 13 years.The denominator was the number of HIV-infected pregnant women over the period multiplied by 100.Due to the discrepancy in the denominator of pregnant women identifi ed in the database of HIV-positive pregnant women and the database of AIDS in children age less than 13 years, it was necessary to utilize two formulas to analyze the VT rate in Vitória.The fi rst formula used the 14 cases of children less than 13 years reported as HIV-positive in the respective SINAN database, with an increase in the denominator of seven mothers not yet reported as HIV-positive pregnant women during the period.The second formula utilized seven reported children in the numerator and the respective mothers as the denominator.
Spatial distribution of events was performed with geoprocessing software TerraView 3.2.0,and the unit of analysis corresponded to neighborhoods in the municipality, whose limits are well-defi ned.e To evaluate the association between IQU and prevalence of HIV-infection in pregnant women, Poisson regression was performed with Stata 9.0.To relate IQU of residential neighborhood with reports of AIDS in children less than 13 years, it was necessary to perform two analysis: one considering 14 reports and the other considering seven.

RESULTS
Between 2000 and 2006, of a total 30,854 pregnant women, 137 (0.44%) were infected with HIV.In general, prevalence varied little (0.41% to 0.50%) each year.The exceptions were 2000 (0.02%) and 2003 (0.75%), when prevalence was far higher than the average over the study period (Figure 1).
During the study period, 14 cases of HIV infection from VT were reported in children, of which only seven had mothers reported as HIV-positive.The two VT rates calculated were 9.7% and 5.0%.The fi rst rate considered the 14 children against 144 (137 + 7 unreported mothers); the second rate used the seven reported children and 137 mothers as the denominator.
Between 2000 and 2006, 137 cases of HIV-positive pregnant women were reported among residents of Vitória.Some cases were missing information about skin color (26.3%) and education (37.2%).Age varied between 13 and 43 years, with an average of 24.3 years and SD of 5.7 years.For the total 101 women with data for race/color, 39 (33.7%) were brown, 34 (38.6%)black and 28 (27.7%)white.Education data was missing for 51 cases (37.2%), and among the others (86), 41 women (47.6%) studied between 4 and 7 years, 26 (30.3%) studied between 8 and 11 years and 8 (9.3%) were illiterate (Table 1).Average age of children at time of notifi cation was 18 months (SD 14.2).Only nine (64.3%) had information for race/color (two white, three black and four brown).
Of the 79 neighborhoods in Vitória Municipality, ten reported VT in the study period.Seven neighborhoods presented one case of VT; two cases were reported in the city center and the Maruípe neighborhood; and São Pedro neighborhood presented three cases.
IQU distribution by neighborhoods in Vitória was heterogeneous, with low and high IQU neighborhoods next to each other (Figure 2).
The spatial distribution of children infected by VT (Figure 4) reveals greater frequencies in neighborhoods   Considering the 14 notifi cations of VT (9.7% VT rate), 64.3% of cases corresponded to low urban quality neighborhoods, 14.3% medium quality and 21.4% high quality.In analysis of the seven notifi cations (5.0%VT rate), 85.7% were from low IQU neighborhoods and 14.3% from high IQU neighborhoods.Underreporting is greater for residents of neighborhoods with low urban quality.
Poisson regression shows an inverse association between the IQU variable and prevalence rate of HIV-infection in pregnant women (Table 2): the lower the IQU level, the greater the infection prevalence among pregnant women.

DISCUSSION
An HIV prevalence of 0.44% was found among pregnant women in Vitória, similar to the rate in Brazil, which estimate among pregnant women was 0.41% in 2004. 17Studies in other states in Brazil and in Vitória report rates from 0.42% to 0.50% 6,8,12,15 Nonetheless, prevalence in the current study is less than the 0.6% found in public maternity hospitals of Vitória by Miranda et al. 13 The difference may be to the fact that these maternities attend pregnant women from other municipalities in the state.
The HIV VT rate found in this study (between 5.0% and 9.7%) is within the range reported by various other studies, from 0.1% to 27.2%, 8,13,15,16,18,19,21 and does not differ from the 6.8% estimated for Brazil and 7.6% for the Southeast Region in 2004. 17Nonetheless, since the highest rates occur in pregnant women that do not undergo preventive interventions, which can reduce VT to less than 0.5%, 7 these levels are far from ideal.
Although Vitória has a prenatal care rate of approximately 95%, a previous study 14 reported that the quality of prenatal care in municipal public maternities was unsatisfactory, with diffi culty to reduce the proportion of caesarian sections even in maternities that principally care for low risk women, with a low enrollment rate in the low income population and less than six prenatal visits recommended by the Ministry of Health.In addition, non-performance of prenatal care was associated with low education, single marital status and premature birth. 14The low quality of care may refl ect diffi culty in early detection of HIV-positive mothers and introduction of prophylaxis to prevent VT, which may be responsible for the high VT rate identifi ed by the present study.The epidemic dynamic in Vitória is similar to Brazil, with similar rates of VT and HIV-infection among pregnant women.In addition, in neighborhoods with low IQU, based on level of education among pregnant women, prevalence was higher, indicating a pauperization of AIDS. 4,10,11Use of neighborhood as the unit of analysis emphasized the internal differences in the city and its social and epidemiologic processes, with a tendency for an increased epidemic among women located in poorer neighborhoods, as in a study performed in Rio de Janeiro State (Southeastern Brazil). 19e main study limitation was the quality of data available in databases of HIV in pregnant women and of AIDS in children less than 13 years, which were shown to be incomplete and inconsistent.The greatest obstacle was the discrepancy between the databases, with seven mothers unreported in Sinan due to under-notifi cation of HIV-positive pregnant women and a consequent under-estimation of the VT rate.Due to the importance of these indicators, it is urgent to improve notifi cation by linking each child infected by VT with the respective notifi cation of their HIV-positive mother.In addition, the small number of events may have impacted data analysis due to the substantial amount of neighborhoods with no events.
Neighborhood socioeconomic condition, summarized by the IQU, was associated with HIV prevalence among pregnant women in the current study.Geoprocessing improved visualization of cases, revealing concentrations in neighborhoods with lower urban quality, allowing municipal managers and health professionals to act aggressively in the most vulnerable areas.

Figure 4 .
Figure 4. Neighborhood distribution of AIDS cases from vertical transmission in children less than 13 years.Vitória, Southeastern Brazil, 2000-2006.
The study was approved in May 2008 by the Ethics Committee of the Health Sciences Center of the Universidade Federal de Espírito Santo, and access to the database was authorized by SEMUS (process No 490/2008).In data worksheets, cases were only identifi ed by a registry code, to preserve confi dentiality and privacy.

Table 2 .
Poisson model considering Urban Quality Index and prevalence of HIV in pregnant women in Vitória, SoutheasternBrazil, 2000-2006.