Syphilis in pregnancy and congenital syphilis in Amazonas State , Brazil : an evaluation using database linkage

This study analyzes notification of syphilis in pregnancy and congenital syphilis in Amazonas State, Brazil, from 2007 to 2009 and verifies underreporting in databases in the National Information System on Diseases of Notification (SINAN) and the occurrence of perinatal deaths associated with congenital syphilis and not reported in the Mortality Information System (SIM). This was a cross-sectional study with probabilistic record linkage between the SINAN and SIM. There were 666 reports of syphilis in pregnant women, including 224 in 2007 (3.8/1,000), 244(4.5/1,000) in 2008, and 198(4.0/1,000) in 2009. The study found 486 cases of congenital syphilis, of which 153 in 2007 (2.1/1,000), 193 in 2008 (2.6/1,000), and 140 in 2009 (2.0/1,000). After linkage of the SINAN databases, 237 pregnant women (35.6%) had cases of congenital syphilis reported. The SIM recorded 4,905 perinatal deaths, of which 57.8% were stillbirths. Probabilistic record linkage between SIM and SINANCongenital Syphilis yielded 13 matched records. The use of SINAN and SIM may not reflect the total magnitude of syphilis, but provide the basis for monitoring and analyzing this health problem, with a view towards planning and management. Syphilis; Congenital Syphilis; Pregnancy; Statistical Databases Resumo O estudo descreve o perfil das notificações de sífilis em gestantes e sífilis congênita no Estado do Amazonas, Brasil, entre 2007 e 2009, e verifica a subnotificação em bases de dados do Sistema Nacional de Agravos de Notificação (SINAN) e a ocorrência de óbitos perinatais associados à sífilis congênita não notificados no Sistema de Informação de Mortalidade (SIM). Estudo seccional com relacionamento entre SINAN e SIM. Ocorreram 666 notificações de sífilis em gestantes, sendo 224 em 2007 (taxa detecção de 3,8/1.000 nascidos vivos); 244 (4,5/1.000) em 2008 e 198 (4,0/1.000) em 2009. Encontraram-se 486 casos de sífilis congênita, sendo 153 (incidência: 2,1/1.000) em 2007, 193 (2,6/1.000) em 2008 e 140 (2,0/1.000) em 2009. Após o relacionamento das bases-SINAN, 237 (35,6%) gestantes tinham casos de sífilis congênita notificados. No SIM, constavam 4.905 óbitos perinatais, sendo 57,8% fetais. No relacionamento do SIM/ SINAN-sífilis congênita foram encontrados 13 registros pareados. A utilização do SINAN e SIM podem não refletir a total magnitude da sífilis, mas formam base para o monitoramento e análise da situação de saúde, tendo em vista o planejamento e a gestão. Sífilis; Sífilis Congênita; Gravidez; Bases de Dados Estatísticos 715 ARTIGO ARTICLE http://dx.doi.org/10.1590/0102-311X00156312 Soeiro CMO et al. 716 Cad. Saúde Pública, Rio de Janeiro, 30(4):715-723, abr, 2014 Introduction Syphilis in pregnancy, when untreated, results in a high proportion of stillbirths and early neonatal deaths 1. Maternal syphilis poses a threat to the fetus, given the high likelihood of vertical (mother-to-child) transmission of the disease, especially in the primary and secondary stages, increasing the risk of perinatal deaths 2. In Brazil, considering 3,059,402 deliveries per year and a 1.6% prevalence rate for syphilis in pregnancy, there were an estimated 48,950 cases of pregnant women with syphilis in 2004 3. However, in 2005, the year after the above estimate, Brazil reported only 1,863 pregnant women with syphilis, increasing to 8,755 in 2009, but still far below the expected figure 3. Congenital syphilis can be prevented by adequate diagnosis and treatment during pregnancy 4. In Brazil, the conditions for such prevention are provided by the public healthcare system at the municipal level and can help eliminate the disease, since vertical transmission of syphilis can occur in up to 25% of cases of infected mothers 5,6. Congenital syphilis has been subject to compulsory notification in Brazil since 1986 7, with its own case definition, periodically revised 8. Information on miscarriage, stillbirth, and live birth with congenital syphilis must be reported to the Information System on Diseases of Notification (SINAN) 9. Syphilis in pregnancy has been subject to compulsory notification since 2005 according to the same principles as congenital syphilis 10. Epidemiological surveillance of both diseases is essential for eliminating congenital syphilis, providing the necessary information for planning and intervention 3,5,6. The Mortality Information System (SIM) was created in 1975 by the Brazilian Ministry of Health to gather regular mortality data in the country 11. The SIM can serve as an alternative source for measuring perinatal mortality (stillbirth plus early neonatal mortality) from congenital syphilis, particularly relevant when assessing the quality of prenatal care 12. The lack of a mechanism to integrate the large health information systems in Brazil makes database linkage an important tool for integrating the various bases. Database linkage has been applied in public health for the last 50 years, since publication of the study by Newcombe et al. 13. Recent decades have witnessed growing recognition of the importance of assessing the results of probabilistic or deterministic record linkage 14. The objectives of this study were to describe the epidemiological profile of reported cases of syphilis in pregnancy and congenital syphilis in Amazonas State from 2007 to 2009, verify underreporting of such cases in the SINAN databases, and verify perinatal deaths associated with congenital syphilis not recorded in the SIM using probabilistic record linkage.


Abstract
This study analyzes notification of syphilis in pregnancy and congenital syphilis in Amazonas State, Brazil, from 2007 to 2009 and verifies underreporting in databases in the National Information System on Diseases of Notification (SINAN) and the occurrence of perinatal deaths associated with congenital syphilis and not reported in the Mortality Information System (SIM).This was a cross-sectional study with probabilistic record linkage between the SINAN and SIM.There were 666 reports of syphilis in pregnant women, including 224 in 2007 (3.8

Introduction
Syphilis in pregnancy, when untreated, results in a high proportion of stillbirths and early neonatal deaths 1 .Maternal syphilis poses a threat to the fetus, given the high likelihood of vertical (mother-to-child) transmission of the disease, especially in the primary and secondary stages, increasing the risk of perinatal deaths 2 .In Brazil, considering 3,059,402 deliveries per year and a 1.6% prevalence rate for syphilis in pregnancy, there were an estimated 48,950 cases of pregnant women with syphilis in 2004 3 .However, in 2005, the year after the above estimate, Brazil reported only 1,863 pregnant women with syphilis, increasing to 8,755 in 2009, but still far below the expected figure 3 .Congenital syphilis can be prevented by adequate diagnosis and treatment during pregnancy 4 .In Brazil, the conditions for such prevention are provided by the public healthcare system at the municipal level and can help eliminate the disease, since vertical transmission of syphilis can occur in up to 25% of cases of infected mothers 5,6 .
Congenital syphilis has been subject to compulsory notification in Brazil since 1986 7 , with its own case definition, periodically revised 8 .Information on miscarriage, stillbirth, and live birth with congenital syphilis must be reported to the Information System on Diseases of Notification (SINAN) 9 .Syphilis in pregnancy has been subject to compulsory notification since 2005 according to the same principles as congenital syphilis 10 .Epidemiological surveillance of both diseases is essential for eliminating congenital syphilis, providing the necessary information for planning and intervention 3,5,6 .
The Mortality Information System (SIM) was created in 1975 by the Brazilian Ministry of Health to gather regular mortality data in the country 11 .The SIM can serve as an alternative source for measuring perinatal mortality (stillbirth plus early neonatal mortality) from congenital syphilis, particularly relevant when assessing the quality of prenatal care 12 .
The lack of a mechanism to integrate the large health information systems in Brazil makes database linkage an important tool for integrating the various bases.Database linkage has been applied in public health for the last 50 years, since publication of the study by Newcombe et al. 13 .Recent decades have witnessed growing recognition of the importance of assessing the results of probabilistic or deterministic record linkage 14 .
The objectives of this study were to describe the epidemiological profile of reported cases of syphilis in pregnancy and congenital syphilis in Amazonas State from 2007 to 2009, verify under-reporting of such cases in the SINAN databases, and verify perinatal deaths associated with congenital syphilis not recorded in the SIM using probabilistic record linkage.

Methodology
This was a descriptive cross-sectional study of cases of congenital syphilis and syphilis in pregnancy included in the SINAN (version SINAN-NET) from 2007 to 2009 in Amazonas State, Brazil.Data on stillbirths (fetal losses after the 22 nd week of pregnancy or greater than 500 grams) and early neonatal deaths (up to 7 days of life) related to congenital syphilis, coded in the International Classification of Diseases, 10 th Revision (ICD-10) as A50.0 to A50.9 were identified in SIM for 2007-2009 through a search for multiple causes of death, using Stata version 9 (Stata Corp., College Station, USA).
After selecting cases reported to SINAN, we conducted probabilistic linkage of the databases for syphilis in pregnancy and congenital syphilis and subsequently for the SINAN-Congenital Syphilis and SIM databases, using the variable "pregnant woman's name" in the SINAN-Syphilis in Pregnancy base and "mother's name" in the SINAN-Congenital Syphilis and SIM databases to identify cases, with Reclink (version 1) 15 supported by Stata (version 9) to analyze cases present in the 3 databases.
In the SINAN-Congenital Syphilis database, recording of cases of miscarriage was preceded by the word "miscarriage by" before the name of the mother in the patient's name variable.This term was used to identify cases of miscarriage in the SINAN.The search in SIM identified one such case as stillbirth rather than miscarriage.For cases of stillbirths, we searched for the words "stillbirth of" or "fetal death of" before the mother's name in the SIM and SINAN-Congenital Syphilis databases in the patient name variable.The first blocking step was performed with "pregnant women's name -mother's name".Given the small number of cases, we then proceeded to a manual search.Mother's age, mother's municipality of residence, and temporal correspondence of cases were used in the manual revision of the probabilistic linkage.
After probabilistic linkage of the databases, a final base was constructed in which the variables related to the pregnant woman (or mother of the case of congenital syphilis) were year of diagnosis, age, race, schooling, prenatal care, timing of syphilis diagnosis, clinical classification of the maternal syphilis, and treatment of the pregnant woman and her partner.For cases of congeni-tal syphilis, the study variables were sex, age at diagnosis of congenital syphilis, presence of symptoms at diagnosis of congenital syphilis in the child, long-bone x-rays, cerebrospinal fluid analysis, and final case classification.
We began with a descriptive analysis, including frequency distributions of categorical variables and measures of central tendency and dispersion for continuous variables.The detection rate for syphilis in pregnancy was calculated as the number of notifications of syphilis in pregnancy divided by the number of live births in the same year, and the incidence rate for congenital syphilis was calculated as the number of notifications divided by the live births in the same year, according to the Brazilian Ministry of Health's definitions for these indicators 16 .
Data were provided by the Amazonas State STD/AIDS Division and the Health Surveillance Foundation (FVS).The study was approved by the Ethics Research Committee of the Amazonas State University/Amazonas Foundation for Tropical Medicine, case n. 1993/2010.

Cases of syphilis in pregnancy
As for reporting of syphilis in pregnancy to the SINAN database, there were 666 notifications during the period, 224 of which in 2007, or a detection rate of 3.8/1,000 live births, 244 (4.5/1,000 live births) in 2008, and 198 (4.0/1,000 live births) in 2009.Table 1 shows the pregnant women's demographic and clinical data.Most were brown/ mixed-race (75.1%), less than 30 years old (80.6%), and had low schooling (66.2%).Sixtyfour percent of notifications were for women residing in Manaus, the State capital.Staging of the disease showed 56.8% classified as primary syphilis, 82.9% with a reactive VDRL test at notification, and 58.7% treated with single-dose benzathine penicillin G, 2.4 million units.

Cases of congenital syphilis
A total of 486 cases of congenital syphilis were notified to SINAN from 2007 to 2009 in the Amazonas State, of which 153 cases were in 2007, or an incidence rate of 2.1/1,000 live births, 193 (2.6/1,000) in 2008, and 140 (2.0/1,000) in 2009.Seventy-eight per cent of notifications referred to mothers residing in the State capital.In 252 reports of congenital syphilis (51.9%), the diagnosis of maternal syphilis occurred late (at childbirth or post-miscarriage curettage), while 144 (29.6%) cases were diagnosed during prenatal care.Only 17% of the women were adequately treated.Less than one-fourth of the mothers' partners were treated (Table 2).Only 25 infants were reported as symptomatic at notification, 56.6% presented reactive peripheral blood VDRL, 20.6% underwent long-bone x-rays, and 6% cerebrospinal fluid analysis for syphilis (Table 2).

Deaths related to congenital syphilis
The SIM listed 4,905 perinatal deaths from 2007 to 2009 in Amazonas State, of which 57.8% were stillbirths.Of the total, only six presented congenital syphilis coded as A50 (ICD-10), five as the underlying causes and one as an associated cause.The proportion of stillbirths from congenital syphilis was 0.04%.

Probabilistic record linkage
After probabilistic record linkage between the databases for syphilis in pregnancy and congenital syphilis, 237 pregnant women (35.6%) were found with resulting cases of congenital syphilis notified during the period, of which 3 women were recorded twice, in different years, and were thus considered as 2 cases of diagnosis of syphilis in pregnancy, resulting in 2 cases of congenital syphilis in distinct periods (Figure 1).
Probabilistic linkage of the SIM and SINAN-Congenital Syphilis databases yielded 13 cases in which the mother's name was recorded on both the Death Certificate and the notification of congenital syphilis.Nine were stillbirths, none of these made any mention of congenital syphilis among the causes leading to death, and only two cases were notified in the SINAN-Congenital Syphilis database.Two cases of early neonatal deaths were in the SINAN database, one of which with congenital syphilis as the underlying cause and another as associated cause; a third early neonatal death was not notified, although congenital syphilis was listed as the underlying cause of death (Figure 2).Meanwhile, of the 20 cases listed as stillbirths in the SINAN-Congenital Syphilis database, only two were located in the SIM.

Estimated underreporting of syphilis in pregnancy and congenital syphilis
Approximately 60% of the women diagnosed with syphilis in pregnancy were treated with single-dose penicillin G, considered inadequate for controlling congenital syphilis.One could thus expect 25% of these mothers to transmit the disease, with an additional 100 cases of congenital syphilis during the period.Meanwhile, some 50% of mothers of cases of congenital syphilis were not reported, which could have added 243 cases of syphilis in pregnancy during the period.

Discussion
The current study showed far fewer reported cases of pregnant women with syphilis than estimat- ed for the North of Brazil in 2004 3 , or some 1,200 pregnant women per year.Meanwhile, the study found an incidence rate of congenital syphilis ranging from 2.0 to 2.6 cases per 1,000 live births in the State of Amazonas, still above the target of elimination of the disease and indicating possible flaws in primary care 17 .In Brazil, congenital syphilis is still an important health problem, with rates around 2.0 per 1,000 live births in the study period 16 , thus highlighting the need for a highpriority policy for its control 18 .
The characteristics of the pregnant women reported with syphilis feature the high proportion of cases of primary syphilis (56.8%).The primary lesion (hard chancre) in women almost always presents with internal genital localization 2 , and most cases of syphilis in pregnancy are diagnosed in the latent phase, without signs or symptoms, only through serological screening with the nontreponemal test (VDRL) 19 .Logically, one would not expect such a diagnostic percentage of chancre in women submitted to serological  screening tests for diseases with risk of vertical transmission.In addition, since most of the women were probably treated inadequately, we were not able to assess treatment of the partners 20 .When verifying the quality of the mother's treatment in the report of congenital syphilis, the failure becomes evident since only 16.7% of the women received adequate treatment, even though 67.3% attended prenatal care.For the same years in Brazil as a whole, slightly more than 70% of mothers reported having attended prenatal care 16 .
As for cases of congenital syphilis reported in Amazonas State, 1.4% were classified as miscarriage due to syphilis and 4.2% as syphilitic stillbirth.In the year 2007, Brazil showed 1.7% as miscarriage and 12.3% as stillbirth 16 .In a study in the city of Nova Iguaçu, Rio de Janeiro State, 11.7% of recorded stillbirths were related to maternal syphilis 21 .In the current study, only 5.1% of cases were symptomatic at birth, a very low proportion, which should lead to immediate case investigation based on the Brazilian national protocol 5,22 .However, the data showed a low rate of long-bone x-rays (20.6%) and an even lower rate of cerebrospinal fluid analysis (6%) to assess involvement of the central nervous system.
Probabilistic record linkage of the databases found 237 women with syphilis in pregnancy whose newborns were also reported to the SINAN-Congenital Syphilis database during the same period.This corresponds to 35.6% of the pregnant women that were reported.One would not expect that all cases of syphilis in pregnancy would be followed by notification of congenital syphilis, since adequate treatment of the pregnant woman and partner should avoid the case 18 .Still, as discussed above, adequate treatment was only provided in a very low proportion of the pregnant women with syphilis, suggesting that there was underreporting of congenital syphilis, approximately 100 cases in the current study, as shown in another study in Rio de Janeiro 23 .Meanwhile, not all the mothers had been reported in the SINAN-Syphilis in Pregnancy database.
Considering the detection rate for syphilis in pregnancy of approximately 2 per 1,000, the proportion of stillbirths due to syphilis was extremely low (0.04%).The World Health Organization considers a proportion of less than 2% tolerable 24 .Of the 20 cases recorded as stillbirths in the SINAN-Congenital Syphilis database, only 2 were located in SIM.Many stillbirths are not claimed for burial by the families at the maternity hospitals and may not receive a Death Certificate.However, with the establishment of Committees for the Investigation of Infant and Fetal Mortality 25 , all such cases should be investigated, and when necessary an "epidemiological" death certificate should be issued to permit burial.
In this study, probabilistic record linkage of different databases containing data on congenital syphilis was a useful tool for evaluating the quality of information contained in these large national systems.In general, probabilistic record linkage is important for monitoring target events 26,27,28,29 or for upgrading information in different databases 13,30 .
The use of secondary databases such as SINAN and SIM may not reflect the total magnitude of the respective diseases, given the need for adequate completion of notification forms and/ or death certificates according to the prevailing national protocols and by responsible health professionals.Nevertheless, these health information systems can provide the basis for monitoring and analyzing the health situation in municipalities and states, with a view towards health planning and management.

Final remarks
Two issues are essential for a better understanding of the current study's findings and their extension to other locations.The first is adequate prenatal care provided in the municipalities and states, where diagnosis and treatment of syphilis in pregnancy can be the sentinel event for the quality of services, avoiding cases of congenital syphilis.The current study's findings corroborate this issue.The second issue is training for health professionals in epidemiological surveillance and for physicians in the completion of death certificates, in addition to the implementation of inquiries into stillbirths and infant deaths according to Ministry of Health criteria.With the development of more reliable information systems with broader coverage, the analysis of the health situation can contribute more robustly to the control of diseases such as syphilis.

Figure 1 Result
Figure 1 Result of probabilistic record linkage between the SINAN-Syphilis in Pregnancy and SINAN-Congenital Syphilis databases.Amazonas State, Brazil, 2007-2009.

Table 1
Distribution of cases of syphilis in pregnancy in SINAN (Information System on Diseases of Notification) according to demographic and clinical variables.Amazonas State, Brazil, 2007-2009 (n = 666).

Table 2
Distribution of cases of congenital syphilis notified to the SINAN (Information System on Diseases of Notification) according to infant and maternal variables.Amazonas State, Brazil, 2007-2009 (n = 486).