ABSTRACT

Objective

To validate the Brazilian National Health System Hospital Information System (SIH/SUS) for maternal morbidity surveillance.

Methods

This was a cross-sectional study conducted in 2021/2022, taking as its reference a national study on maternal morbidity (MMG) conducted in 50 public and 28 private hospitals; we compared SIH/SUS and MMG data for hospitalization frequency, reason and type of discharge and calculated sensitivity, specificity, positive and negative likelihood ratios for seven diagnoses and four procedures.

Results

Hospitalizations identified on SIH/SUS (32,212) corresponded to 95.1% of hospitalizations assessed by MMG (33,867), with lower recording on SIH/SUS (85.5%) for private hospitals [10,036 (SIH/SUS)]; 11,742 (MMG)]; compared to MMG, SIH/SUS had a lower proportion of hospitalizations due to “complications during pregnancy” (9.7% versus 16.5%) as well as under-recording of all diagnoses and procedures assessed, except “ectopic pregnancy”.

Conclusion

Better recording of diagnoses and procedures on SIH/SUS is essential for its use in maternal morbidity surveillance.

Keywords
Validation Study; Morbidity; Pregnancy; Postpartum Period; Hospital Information Systems; Statistical Databases

Study contributions

Main results

The SIH/SUS system had good recording coverage of obstetric hospitalizations, but a lower proportion of hospitalizations due to “complications during pregnancy”; and under-recording of the diagnoses and procedures assessed, except “ectopic pregnancy”.

Implications for services

The quality of hospital admission authorization (AIH) filling in needs to be improved, including recording diagnoses and procedures to identify hospitalization as obstetric hospitalization as well as complications in pregnancy and the postpartum period.

Perspectives

Improving the quality of AIH records is essential for the use of the SIH/SUS system in maternal morbidity surveillance, this being a strategy recommended for better obstetric care, being complementary to maternal mortality surveillance.

Keywords
Validation Study; Morbidity; Pregnancy; Postpartum Period; Hospital Information Systems; Statistical Databases

RESUMEN

Objetivo

Validar el Sistema de Información Hospitalaria del Sistema Único de Salud (SIH/SUS) para vigilancia de la morbilidad materna.

Métodos

Estudio transversal, 2021/2022, utilizando como referencia datos de estudio nacional de morbilidad materna (MMG) realizado en 50 hospitales públicos y 28 privados; comparando: frecuencia, motivo y tipo de alta de internaciones en SIH/SUS y MMG y calculando sensibilidad, especificidad y razones de probabilidad positivos y negativos para siete diagnósticos y cuatro procedimientos.

Resultados

Las internaciones identificadas en SIH/SUS (32.212) correspondieron al 95,1% de internaciones evaluadas en MMG (33.867), observándose menor registro en SIH/SUS (85,5%) en hospitales privados [10.036 (SIH/SUS)]; 11.742 (MMG)]; comparado con MMG, SIH/SUS tuvo menor proporción de internaciones por “complicaciones durante el embarazo” (9,7% vs 16,5%), así como subregistro de todos los diagnósticos y procedimientos evaluados, excepto “embarazo ectópico”.

Conclusión

Mejor registro de diagnósticos y procedimientos en SIH/SUS es fundamental para su uso en la vigilancia de la morbilidad materna.

Palabras clave
Estudio de Validación; Morbilidad; Embarazo; Período Posparto; Sistemas de Información en Hospital; Bases de Datos Estadísticos

INTRODUCTION

Maternal mortality is a serious public health problem in Brazil and worldwide.¹1 World Health Organization. Trends in maternal mortality 2000 to 2020: estimates by WHO, UNICEF, UNFPA, World Bank Group and UNDESA/Population Division [Internet]. Geneva: World Health Organization; 2023 [cited 2023 July 11]. 86 p. Available from: https://www.who.int/publications/i/item/9789240068759.
https://www.who.int/publications/i/item/... Although the Brazilian maternal mortality ratio is high, maternal death is an infrequent event, especially in places with a low number of births. Since 2011, the World Health Organization (WHO) has recommended the study of severe maternal morbidity and maternal near miss as complementary strategies to the study of maternal death, as they are more frequent events and share the same determining factors, thus enabling more robust analyses.²2 World Health Organization. Evaluating the quality of care for severe pregnancy complications: the WHO near-miss approach for maternal health [Internet]. Geneva: World Health Organization; 2011 [cited 2023 July 11]. 29 p. Available from: https://apps.who.int/iris/bitstream/handle/10665/44692/9789241502221_eng.pdf;jsessionid=EEBF2C5BAD3A80F2E66623549925154B?sequence=1.
https://apps.who.int/iris/bitstream/hand...

The objective of the Brazilian National Health System Hospital Information System (SIH/SUS) is payment of hospitalizations with public funding. Although health surveillance is not one its purposes, the SIH/SUS has been used to investigate hospital morbidity,³3 Bittencourt SA, Camacho LAB, Leal MC. O Sistema de Informação Hospitalar e sua aplicação na saúde coletiva. Cad Saúde Pública. 2006;22(1):19–30. including maternal morbidity.⁴4 Magalhães MC, Bustamante-Teixeira MT. Morbidade materna extremamente grave: uso do Sistema de Informação Hospitalar. Rev Saude Publica. 2012;46(3):472-8.

5 Nakamura-Pereira M, Mendes-Silva W, Dias MAB, Reichenheim ME, Lobato G. Sistema de Informações Hospitalares do Sistema Único de Saúde (SIH-SUS): uma avaliação do seu desempenho para a identificação do near miss materno. Cad Saude Publica 2013;29(7):1333-45. doi: 10.1590/s0102-311x2013000700008.
https://doi.org/10.1590/s0102-311x201300...

6 Silva TC, Varela PLR, Oliveira RR, Mathias TAF. Severe maternal morbidity identified in the Hospital Information System of the Brazilian National Health System in Paraná State, Brazil, 2010. Epidemiol Serv Saude. 2016;25(3):617-28. doi: 10.5123/s1679-49742016000300017.
https://doi.org/10.5123/s1679-4974201600...

7 Carvalho BAS, Andrade AGBF, Dantas AS, Figueiredo IM, Silva JA, Rosendo TS, et al. Temporal trends of maternal near miss in Brazil between 2000 and 2012. Rev Bras Saude Mater Infant. 2019;19(1):115–24. doi: 10.1590/1806-93042019000100007.
https://doi.org/10.1590/1806-93042019000... -⁸8 Herdt MCW, Magajewski FRL, Linzmeyer A, Tomazzoni RR, Domingues NP, Domingues MP. Temporal trend of Near Miss and its regional variations in Brazil from 2010 to 2018. Rev Bras Ginecol Obstet. 2021;43(2):97-106. doi: 10.1055/s-0040-1719144.
https://doi.org/10.1055/s-0040-1719144... Previous studies,⁹9 Veras CMT, Martins MS. A confiabilidade dos dados nos formulários de autorização de internação hospitalar (AIH). Cad Saude Publica.1994;10(3):339-55. doi: 10.1590/s0102-311x1994000300014.
https://doi.org/10.1590/s0102-311x199400... ,¹⁰10 Mathias TAF, Soboll MLMS. Confiabilidade de diagnósticos nos formulários de autorização de internação hospitalar. Rev Saude Publica.1998;32(6):526-32. doi: 10.1590/s0034-89101998000600005.
https://doi.org/10.1590/s0034-8910199800... conducted in the 1990s, evaluated SIH/SUS data reliability, including information on childbirth.¹¹11 Bitencourt SA, Camacho LAB, Leal MC. A qualidade da informação sobre o parto no Sistema de Informações Hospitalares no Município do Rio de Janeiro, Brasil,1999 a 2001. Cad Saude Publica. 2008;24(6):1344-54. doi: 10.1590/s0102-311x2008000600015.
https://doi.org/10.1590/s0102-311x200800...

However, no studies were identified that have evaluated obstetric hospitalizations nationally using the SIH/SUS, when compared to hospitalization data obtained from hospital records, despite the use of administrative data to study maternal and neonatal morbidity being reported in the international literature.¹²12 Glance LG, Hasley S, Glantz JC, Stevens TP, Faden E, Kreso MA, et al. Measuring childbirth outcomes using administrative and birth certificate data. Anesthesiology. 2019;131(2):238-53. doi: 10.1097/aln.0000000000002759.
https://doi.org/10.1097/aln.000000000000...

13 Huennekens K, Oot A, Lantos E, Yee LM, Feinglass J. Using electronic health record and administrative data to analyze maternal and neonatal delivery complications. Jt Comm J Qual Patient Saf. 2020;46(11):623-30. doi: 10.1016/j.jcjq.2020.08.007.
https://doi.org/10.1016/j.jcjq.2020.08.0... -¹⁴14 Carmichael SL, Girsen AI, Ma C, Main EK, Gibbs RS. Using longitudinally linked data to measure severe maternal morbidity beyond the birth hospitalization in California. Obstet Gynecol. 2022;140(3):450-2. doi: 10.1097/aog.0000000000004902.
https://doi.org/10.1097/aog.000000000000... Analyzing the information available on the SIH/SUS is, therefore, relevant for confirming its validity in the study of maternal morbidity, developing strategies to improve obstetric care and reducing maternal mortality.

In 2021-2022, the national study entitled “Perinatal mortality, severe maternal morbidity and maternal near miss” study (“Mortalidade perinatal, morbidade materna grave e near miss materno” – MMG study)¹⁵15 Domingues RMSM, Dias MAB, Nakamura-Pereira M, Pacagnella RC, Lansky S, Pereira APE, et al. Mortalidade perinatal, morbidade materna grave e near miss materno: protocolo de um estudo integrado à pesquisa “Nascer no Brasil II”. Cad Saude Publica. 2024; 40(4):e00248222 doi: 10.1590/0102-311XPT248222.
https://doi.org/10.1590/0102-311XPT24822... assessed severe maternal morbidity in Brazilian public and private hospitals using medical record data.

This study aimed to validate SIH/SUS usefulness for monitoring maternal morbidity (MM), using the MMG study as a reference standard, by comparing obstetric hospitalization frequency, reason and discharge type, and calculating sensitivity, specificity and positive and negative likelihood ratio of diagnoses and procedures recommended by WHO²2 World Health Organization. Evaluating the quality of care for severe pregnancy complications: the WHO near-miss approach for maternal health [Internet]. Geneva: World Health Organization; 2011 [cited 2023 July 11]. 29 p. Available from: https://apps.who.int/iris/bitstream/handle/10665/44692/9789241502221_eng.pdf;jsessionid=EEBF2C5BAD3A80F2E66623549925154B?sequence=1.
https://apps.who.int/iris/bitstream/hand... for such surveillance.

METHODS

Study design

This is a cross-sectional study, conducted in 2021/2022, to validate the SIH/SUS for MM surveillance using the MMG study as a reference standard.¹⁵15 Domingues RMSM, Dias MAB, Nakamura-Pereira M, Pacagnella RC, Lansky S, Pereira APE, et al. Mortalidade perinatal, morbidade materna grave e near miss materno: protocolo de um estudo integrado à pesquisa “Nascer no Brasil II”. Cad Saude Publica. 2024; 40(4):e00248222 doi: 10.1590/0102-311XPT248222.
https://doi.org/10.1590/0102-311XPT24822...

Background

The MMG is a hospital-based study with national coverage, carried out in an integrated manner with the study entitled Birth in Brazil II: National survey of abortion, delivery and birth (“Nascer no Brasil II: pesquisa nacional sobre aborto, parto e nascimento – NBII”). All health facilities – public and private – with more than 2,750 births/year and participating in the NBII study were included in the MMG study. A obstetric hospitalization census was conducted in those facilities for 30 consecutive days in 2021-2022.

The SIH/SUS is a nationwide health information system, composed of two databases: the reduced database, which contains hospitalization diagnoses and the primary medical procedure performed; and the professional services database, containing a record of all professional acts carried out during hospital stay. The Hospital Admission Authorization (AIH) number identifies hospitalization of the same individual in both databases.

According to SIH/SUS standards,¹⁶16 Ministério da Saúde (BR). Secretaria de Atenção à Saúde. Departamento de Avaliação, Regulação e Controle. Coordenação Geral dos Sistemas de Informação. SIH – Sistema de Informação Hospitalar do SUS: manual técnico operacional do sistema. 2017 [Internet]. Brasília: Ministério da Saúde; 2017 [citado 2023 Jul 17]. Disponível em http://sihd.datasus.gov.br/documentos/documentos_sihd2.php.
http://sihd.datasus.gov.br/documentos/do... it is possible that, in specific situations, a woman may have more than one AIH during an obstetric hospitalization. For example, when a woman admitted for an obstetric procedure requires a surgical intervention. In such situations,¹⁶16 Ministério da Saúde (BR). Secretaria de Atenção à Saúde. Departamento de Avaliação, Regulação e Controle. Coordenação Geral dos Sistemas de Informação. SIH – Sistema de Informação Hospitalar do SUS: manual técnico operacional do sistema. 2017 [Internet]. Brasília: Ministério da Saúde; 2017 [citado 2023 Jul 17]. Disponível em http://sihd.datasus.gov.br/documentos/documentos_sihd2.php.
http://sihd.datasus.gov.br/documentos/do... the initial AIH must be closed with the “continuing inpatient stay” billing reason and a new AIH must be issued.

Data sources

In the MMG study, all data were derived from medical record data collection, performed in a standardized way by trained professionals. The total number of hospitalizations, reason for hospitalization, discharge type and diagnoses and medical procedures indicating MM¹⁵15 Domingues RMSM, Dias MAB, Nakamura-Pereira M, Pacagnella RC, Lansky S, Pereira APE, et al. Mortalidade perinatal, morbidade materna grave e near miss materno: protocolo de um estudo integrado à pesquisa “Nascer no Brasil II”. Cad Saude Publica. 2024; 40(4):e00248222 doi: 10.1590/0102-311XPT248222.
https://doi.org/10.1590/0102-311XPT24822... were obtained from the triage form, a data collection instrument filled out for all women included in the study. For women with morbidity recorded on the triage form, additional data collection was performed, using a detailed instrument.

In the case of the SIH/SUS, the total number of hospitalizations, reason for hospitalization, discharge type and hospitalization diagnoses were obtained from the reduced database, while procedures performed were obtained from both the reduced database and the professional services database. Both databases were accessed on August 15, 2022.

Participants

The MMG study included 78 hospitals – 50 public and 28 private, that had concluded data collection by February 2022 –, spread over all the country’s Federative Units, except Amapá.

In the case of the SIH/SUS, all obstetric hospitalizations that occurred in the same period as the MMG study were included, in each of the participating hospitals, identified by their National Health Establishment Registry (CNES) number. The hospitalization selection process had four stages (Figure 1). Initially, type 1 AIHs of women aged 10 to 49 years old were selected (without identification) from the reduced database, available on the website of the SUS Information Technology Department and captured by the Microdatasus package,¹⁷17 Saldanha RF, Bastos RR, Barcellos C. Microdatasus: pacote para download e pré-processamento de microdados do Departamento de Informática do SUS (DATASUS). Cad Saude Publica. 2019;35(9):e00032419. doi: 10.1590/0102-311x00032419.
https://doi.org/10.1590/0102-311x0003241... using R statistical programming language.

In the second stage , we identified each woman’s hospital episode of care,¹⁸18 Sheehan KJ, Sobolev B, Guy P, Bohm E, Hellsten E, Sutherland JM, et al. Constructing an episode of care from acute hospitalization records for studying effects of timing of hip fracture surgery. J Orthop Res. 2016;34(2):197-204. doi: 10.1002/jor.22997.
https://doi.org/10.1002/jor.22997... defined as the set of all information relating to a given hospitalization, which may consist of one or more AIH. In order to identify an episode of care formed by multiple AIH records, an algorithm was used that allowed identification of AIHs subsequent to an AIH with the “continuing inpatient stay” billing reason. In short, AIHs with the same CNES number and for women who had the same date of birth were considered as being part of the same episode of care, if the interval between the discharge date of an AIH containing the “continuing inpatient stay” billing reason and the date of subsequent AIH hospitalization was less than or equal to one day.

Subsequently, in order to identify obstetric hospitalizations, we selected AIHs with hospitalization diagnoses [according to the International Statistical Classification of Diseases and Related Health Problems – Tenth Revision (ICD-10)] and/or procedures as per the SUS Sistema de Gerenciamento da Tabela de Procedimentos, Medicamentos e Órteses, Próteses e Materiais Especiais)¹⁹19 Ministério da Saúde (BR). SIGTAP - Sistema de Gerenciamento da Tabela de Procedimentos, Medicamentos e OPM do SUS [Internet]. Brasília: Minitério da Saúde; c2022 [citado 2022 Ago 15]. Disponível em http://sigtap.datasus.gov.br/tabela-unificada/app/sec/inicio.jsp.
http://sigtap.datasus.gov.br/tabela-unif... described in Box 1. In this selection process, we identified AIHs with criteria for obstetric hospitalization in episodes of care with single or multiple records, regardless of the position of the AIH in the episode. Finally, after identifying obstetric hospitalizations on the reduced database, AIHs with the same number were selected on the professional services database.

Variables

The variables analyzed were:

1. a) Hospital type (public; private)

2. b) Total number of obstetric hospitalizations

3. c) Reason for hospitalization (delivery; abortion; complications during pregnancy; complications during the puerperium)

4. d) Discharge type (routine discharge; administrative; continuing inpatient stay; death; transfer)

5. e) Obstetric diagnoses

   1. – Severe pre-eclampsia

   2. – Eclampsia

   3. – HELLP syndrome (hemolysis, elevated liver enzymes, low platelet count)

   4. – Abruptio placentae

   5. – Postpartum/post-abortion hemorrhage

   6. – Rupture of uterus

   7. – Ectopic pregnancy

6. f) Hospital procedures

   1. – Hysterectomy

   2. – Laparotomy

   3. – Blood product transfusion

   4. – Hospitalization in an intensive care unit (ICU)

The seven diagnoses and four procedures listed in items “e” and “f” are part of the 26 criteria recommended by the WHO for studying severe maternal morbidity.²2 World Health Organization. Evaluating the quality of care for severe pregnancy complications: the WHO near-miss approach for maternal health [Internet]. Geneva: World Health Organization; 2011 [cited 2023 July 11]. 29 p. Available from: https://apps.who.int/iris/bitstream/handle/10665/44692/9789241502221_eng.pdf;jsessionid=EEBF2C5BAD3A80F2E66623549925154B?sequence=1.
https://apps.who.int/iris/bitstream/hand...

Detailing of the variables

In the MMG study, hospitalizations due to “complications during pregnancy” were classified as all those due to clinical and/or obstetric complications occurring during pregnancy, without indication of termination of pregnancy at the time of hospital admission; and “complications during the puerperium” were classified as all hospitalizations resulting from clinical and/or obstetric complications diagnosed after the end of pregnancy. The “continuing inpatient stay” discharge type was considered to be a woman who remained hospitalized after the 42^nd day after the end of pregnancy. We used medical records to obtain data on diagnoses and procedures but did not classify them. If there were no records for these items, we considered this to be lack of diagnosis and absence of the procedure.

In the case of the SIH/SUS, we used the criteria described in Box 1 to classify the reason for hospitalization and the discharge type. An episode of care in which the last AIH showed discharge due to continuing inpatient stay, and for which it was not possible to identify the subsequent AIH in that episode of care, was classified as “continuing inpatient stay” discharge type. In the case of episodes of care with more than one AIH, the “reason for hospitalization” recorded on the first AIH of the episode of care and the “discharge type” recorded on the last AIH of the episode of care were used. We used the ICD codes and procedures described in Box 1 to identify diagnoses and procedures, considering the records of all AIHs in the episode of care. In the case of a specific diagnosis or procedure recorded on more than one AIH of the episode of care, the diagnosis and/or procedure was only counted once.

Statistical methods

All analyses were performed comparing SIH/SUS hospitalization data (source to be validated) with hospitalization data presented in the MMG study (reference standard), for all hospitals together and for public hospitals and private hospitals separately, using version 4.3.0 of the R statistical programming language.²⁰20 R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2021 [cited 2022 Ago 08]. Available from: https://www.r-project.org/.
https://www.r-project.org/...

Initially, we compared the frequency of obstetric hospitalizations, reasons for hospitalization and types of discharge identified by the MMG study and corresponding data recorded on the SIH/SUS. When comparing the total number of hospitalizations, record coverage equal to or greater than 90% was considered adequate, this being a percentage used by the Brazilian Ministry of Health as a parameter for record coverage in other information systems.²¹21 Brasil. Ministério da Saúde. Portaria nº 1.520, de 30 de maio de 2018. Altera os Anexos XCVIII e XCIX à Portaria de Consolidação nº 5/GM/MS, de 28 de setembro de 2017, com a inclusão de metas e indicadores do Programa de Qualificação das Ações de Vigilância em Saúde - PQA-VS, a partir de 2018 [Internet]. Diário Oficial da União, Brasília (DF), 2018 Jun 6 [citado 2023 Set 1], Seção 1:47. Disponível em https://bvsms.saude.gov.br/bvs/saudelegis/gm/2018/prt1520_06_06_2018.html.
https://bvsms.saude.gov.br/bvs/saudelegi... The “reason for hospitalization” variable was included on the MMG study triage form after the start of fieldwork, whereby hospitals with more than 10% of hospitalizations without this variable were excluded from this analysis.

Subsequently, specific diagnoses and procedures were compared in both databases. For this analysis, only women from the MMG study who presented morbidity were included. These hospitalizations were deterministically linked to the SIH/SUS database, using the information available in both databases [CNES number; date of birth; date of admission; discharge date; race/skin color; postcode of residence]. Once the obstetric hospitalizations were linked, we compared the frequency of specific diagnoses and procedures, as well as the sensitivity, specificity, and positive (positive LR) and negative (negative LR) likelihood ratio of each diagnosis and specific procedure held on the SIH/SUS. Positive LR (sensitivity/1 - specificity) refers to presence of diagnosis; and negative LR (1 - sensitivity/specificity), refers to absence of diagnosis. Values ​​greater than 1 increase the probability of diagnosis and values ​​between 0 and 1 reduce this probability. The following criteria were used for the purposes of interpretation:

1. a) Positive LR: > 10 = strong; 5-9.9 = moderate; 2-4.9 = weak; 1-2 = very weak.

2. a) Negative LR: < 0.1 = strong; 0.11-0.20 = moderate; 0.21-0.50 = weak; 0.51-1 = very weak.²²22 Simel DL, Rennie D. The rational clinical examination: evidence-based clinical diagnosis [Internet]. New York: McGraw Hill, 2009 [cited 2024 Jan 15]. 744 p. Available from: https://jamaevidence.mhmedical.com/content.aspx?bookid=845&sectionid=61356244.
   https://jamaevidence.mhmedical.com/conte...

All the codes used are available at: http://github.com/coelicm/SIH-SUS-validation

Ethical considerations

The MMG study project was approved by the Research Ethics Committee of the Sergio Arouca National School of Public Health/Oswaldo Cruz Foundation, as per Opinion No. 4.230.028, issued on August 21, 2020, later amended as approved by Opinion No. 4.473.968, issued on December 18, 2020. As this was a retrospective study with data collection from medical records, waiver of signing of the Free and Informed Consent Form was requested, with access to records being authorized by the hospitals. All precautions were taken to guarantee information secrecy and confidentiality. Publicly accessible databases were used for analyzing SIH/SUS data, whereby there was no identification of the health service users in question.

RESULTS

The MMG study recorded 33,867 obstetric hospitalizations, of which 5,379 had a record of some MM, with detailed data being collected from 5,303 medical records. On the SIH/SUS system, 32,212 obstetric hospitalizations were identified in the same hospitals and same study period, 4,652 of which were related to hospitalizations identified by the MMG study that had diagnosis of morbidity (Figure 2).

The number of obstetric hospitalizations captured on the SIH/SUS corresponded to 95.1% of the number of hospitalizations assessed by the MMG study, with this proportion being 100.2% for public hospitals (22,176/22,125) and 85.5% for private hospitals (10,036/11,742). However, recording frequency varied between hospitals: when SIH/SUS data were compared with the MMG study data, the proportion of recording was below 90% for 8% of public hospitals and 50% of private hospitals, while the proportion of hospitalization recording was above 110% for 14% of public hospitals and 10.7% of private hospitals.

There was a higher proportion of hospitalizations due to “complications during pregnancy” in the MMG study, both in public hospitals (16.1% versus 9.9%) and in private hospitals (17.4% versus 9.1%), and a higher proportion of admissions for delivery on the SIH/SUS (78.7% versus 72.2% in public hospitals; 81.2% versus 73.0% in private hospitals). The reasons for discharge in the MMG study and on the SIH/SUS were similar (Table 1), highlighting that “continuing inpatient stay” had different interpretations in the two databases.

Analysis of specific diagnoses and procedures (Table 2) showed that the most frequent diagnoses in the MMG study were “severe pre-eclampsia” (17.7%) and “hemorrhage” (13.8%), with “hospitalization in ICU” being the most frequent procedure (7.1%). On The SIH/SUS, the most frequent diagnoses were “severe pre-eclampsia” (8.9%) and “ectopic pregnancy” (4.2%); and among the procedures, the most frequent were “hospitalization in ICU” (5.3%) and “blood product transfusion” (5.2%). We highlight the absence of HELLP syndrome cases on the SIH/SUS and the identification of procedures related to “blood product transfusion” and “hospitalization in ICU” only on the professional services database.

Two diagnoses (“eclampsia” and “abruptio placentae”) and three procedures (“laparotomy”, “blood product transfusion” and “hospitalization in ICU”) showed a “strong” positive LR in public and private hospitals, while “severe pre-eclampsia”, “rupture of uterus” and “ectopic pregnancy” also showed a “strong” positive LR in public hospitals. “Postpartum or post-abortion hemorrhage” showed a “moderate” positive LR in public hospitals and “very weak” positive LR in private hospitals. All diagnoses and procedures showed “very weak” negative LR in public and private hospitals, except “ectopic pregnancy” (“strong” in public and private hospitals), “blood product transfusion” (“weak” in private hospitals) and “hospitalization in ICU” (“weak” in public hospitals and “moderate” in private hospitals) (Table 2).

DISCUSSION

The results we found showed high capture of obstetric hospitalizations on the SIH/SUS, mainly for public hospitals, with similar proportions in terms of discharge type and reason for hospitalization, in the comparison between SIH/SUS records and MMG study records; with the exception of hospitalizations due to “complications during pregnancy”, which had a 40% lower proportion on the SIH/SUS. Eight diagnoses and procedures showed a “strong” positive LR in public hospitals and five in private hospitals, demonstrating the high probability of these diagnoses and procedures when registered on the SIH/SUS. However, all diagnoses and procedures, with the exception of “ectopic pregnancy”, showed evidence of under-reporting, limiting the use of the SIH/SUS for monitoring maternal morbidity.

The lower number of hospitalizations found for private hospitals is a result to be expected, since not all beds in SUS outsourced private hospitals provide hospitalization paid for with public funding. Possible explanations for the variation in the proportion of records found in the hospitals assessed include (i) non-issuance of AIH, (ii) non-authorization of AIH issued or even, (iii) issuance of AIH without group “O” diagnoses or procedures used to identify obstetric hospitalizations.

In hospitalizations with primary diagnosis and procedures for non-obstetric causes, as may have occurred in hospitalizations due to COVID-19, it is essential that group “O” codes/ICDs related to complications during pregnancy or in the postpartum period be recorded in secondary diagnoses, so that such hospitalizations can be identified as having occurred during pregnancy or in the postpartum period.

Regarding the greater number of AIHs identified on the SIH/SUS, the explanatory hypotheses include (i) non-identification of an AIH as part of an episode of care, being counted as hospitalization of another woman, (ii) failure to capture hospitalization of pregnant and postpartum women through the MMG study, if women diagnosed with COVID-19 were hospitalized outside the maternity ward (for example, in respiratory isolation beds), and (iii) issuance of AIH for hospitalizations that did not take place.

The findings as to a lower proportion of hospitalizations due to “complications during pregnancy” on the SIH/SUS can be attributed to the way in which the reason for hospitalization was classified on both databases: on the SIH/SUS, women diagnosed with a pregnancy complication during hospitalization were classified as “delivery” if the procedure performed was one of those provided for natural birth assistance or cesarean section. It is important to note the very similar proportions of hospitalizations for abortion care, indicating that there is no under-reporting of hospitalizations due to abortion, which would be possible, considering the illegality and stigmatization of the topic in Brazil.

The majority of discharges were routine hospital discharges, with the proportion of deaths being low and similar between the two databases. A study evaluating maternal mortality in Brazil in 2019, estimated through data registered on the SIH/SUS, concluded that use of the SIH/SUS may be valid in studies on maternal mortality and morbidity, as an information system complementary to the Mortality Information System.²³23 Ranzani OT, Marinho MF, Bierrenbach AL. Usefulness of the Hospital Information System for maternal mortality surveillance in Brazil. Rev Bras Epidemiol. 2023;26:e230007. doi: 10.1590/1980-549720230007.2.
https://doi.org/10.1590/1980-54972023000...

Under-recording of diagnoses and procedures on the SIH/SUS database has implications for the study of maternal morbidity, especially hypertensive and hemorrhagic complications, these being the most frequent causes found both in the MMG study and in national⁴4 Magalhães MC, Bustamante-Teixeira MT. Morbidade materna extremamente grave: uso do Sistema de Informação Hospitalar. Rev Saude Publica. 2012;46(3):472-8.,⁶6 Silva TC, Varela PLR, Oliveira RR, Mathias TAF. Severe maternal morbidity identified in the Hospital Information System of the Brazilian National Health System in Paraná State, Brazil, 2010. Epidemiol Serv Saude. 2016;25(3):617-28. doi: 10.5123/s1679-49742016000300017.
https://doi.org/10.5123/s1679-4974201600... ,⁸8 Herdt MCW, Magajewski FRL, Linzmeyer A, Tomazzoni RR, Domingues NP, Domingues MP. Temporal trend of Near Miss and its regional variations in Brazil from 2010 to 2018. Rev Bras Ginecol Obstet. 2021;43(2):97-106. doi: 10.1055/s-0040-1719144.
https://doi.org/10.1055/s-0040-1719144... and international studies.²⁴24 Dzakpasu S, Deb-Rinker P, Arbour L, Darling EK, Kramer MS, Liu S, et al. Severe maternal morbidity surveillance: Monitoring pregnant women at high risk for prolonged hospitalisation and death. Paediatr Perinat Epidemiol. 2020;34(4):427-39. doi: 10.1111/ppe.12574.
https://doi.org/10.1111/ppe.12574... ,²⁵25 Wolfson C, Qian J, Chin P, Downey C, Mattingly KJ, Jones-Beatty K, et al. Findings from severe maternal morbidity surveillance and review in Maryland. JAMA Netw Open. 2022;5(11):e2244077. doi: 10.1001/jamanetworkopen.2022.44077.
https://doi.org/10.1001/jamanetworkopen.... Problems with recording specific diagnoses on the SIH/SUS have already been reported in previous studies.³3 Bittencourt SA, Camacho LAB, Leal MC. O Sistema de Informação Hospitalar e sua aplicação na saúde coletiva. Cad Saúde Pública. 2006;22(1):19–30.

In the case of “severe pre-eclampsia” diagnosis, it is possible that other ICD codes related to hypertensive complications may have been used, reflecting the difficulty of differentiating diagnosis between pregnancy-specific hypertension, chronic hypertension and chronic hypertension overlapping with gestational hypertension.²⁶26 Bartal MF, Lindheimer MD, Sibai BM. Proteinuria during pregnancy: definition, pathophysiology, methodology, and clinical significance. Am J Obstet Gynecol. 2022;226(2S):S819-S834. doi: 10.1016/j.ajog.2020.08.108.
https://doi.org/10.1016/j.ajog.2020.08.1... ,²⁷27 Hurrell A, Webster L, Chappell LC, Shennan AH. The assessment of blood pressure in pregnant women: pitfalls and novel approaches. Am J Obstet Gynecol. 2022;226(2S):S804-S818. doi: 10.1016/j.ajog.2020.10.026.
https://doi.org/10.1016/j.ajog.2020.10.0... Failure to record the “HELLP syndrome” code may result from its having been more recently included in the ICD-10 with effect from 2019.

Underreporting postpartum hemorrhage diagnosis is contrary to what is reported in the literature, whereby this diagnosis is considered to be overestimated when using ICD-10 records.²⁸28 Masterson JA, Adamestam I, Beatty M, Boardman JP, Johnston P, Joss J, et al. Severe maternal morbidity in Scotland. Anaesthesia. 2022;77(9):971-80. doi: 10.1111/anae.15798.
https://doi.org/10.1111/anae.15798... A possible explanation is that the complication that gave rise to hemorrhaging was recorded in its place (e.g., uterine inertia) or that the ICD code for hemorrhaging was not recorded as hemorrhaging had been resolved. The more frequent recording of the “blood product transfusion” procedure, compared to diagnosis of “hemorrhages”, supports this hypothesis.

It should be noted that ICD codes recorded in any of the 12 fields available for diagnosis recording were considered and not just the code in the primary diagnosis field, with the aim of achieving greater sensitivity in identifying obstetric hospitalizations and complications. In some hospital units, all diagnoses of “eclampsia” were recorded in the “ICD notification” field and not in the primary diagnosis field (data not shown), probably reflecting concern with surveillance of this condition. Using only the diagnosis recorded in the primary ICD field would result in even greater under-reporting of complications, as well as lower capture of obstetric hospitalizations.

With regard to procedures, under-reporting of major surgical interventions, such as “hysterectomy” and “laparotomy” is noteworthy, these being markers of serious complications. Regarding recording of hospitalizations in ICUs, there was an improvement in relation to a previous study, which assessed hospitalizations due to acute myocardial infarction,²⁹29 Escosteguy CC, Portela MC, Medronho RA, Vasconcellos MTL. O Sistema de Informações Hospitalares e a assistência ao infarto agudo do miocárdio. Rev Saude Publica. 2002;36(4):491–9. doi: 10.1590/s0034-89102002000400016.
https://doi.org/10.1590/s0034-8910200200... although the frequency found is still lower than that recorded in medical record data. Absence of accredited beds for intensive care, or even rejection of the AIH, are possible explanations for this result.

This study has limitations. Only hospitals with more than 2,750 live births per year were analyzed, making it impossible to estimate whether the same results would be found in smaller hospitals. Comparisons between the number of hospitalizations, reason for hospitalization and discharge type were made between frequencies estimated in the two databases, and it is not possible to state that the hospitalizations found relate to the same women. Finally, the MMG study, used as a reference standard, contains data obtained from hospital records that depend on record quality in each health facility.

Notwithstanding these limitations, this is the first study dedicated to assessing obstetric hospitalizations held on the SIH/SUS and comparing them with medical record data obtained in a national study, conducted in public and private hospitals. Episodes of care were also analyzed, a method already adopted in previous SIH/SUS studies³⁰30 Portela MC, Schramm JM de A, Pepe VLE, Noronha MF, Pinto CAM, Cianeli MP. Algoritmo para a composição de dados por internação a partir do sistema de informações hospitalares do sistema único de saúde (SIH/SUS) - Composição de dados por internação a partir do SIH/SUS. Cad Saude Publica. 1997;13(4):771–4. doi: 10.1590/s0102-311x1997000400020.
https://doi.org/10.1590/s0102-311x199700... but not in recent studies on obstetric morbidity, thus allowing a better estimate of the number of obstetric hospitalizations and complications recorded for such hospitalizations. Finally, simultaneous analysis of the reduced database and the professional services database allowed us to identify morbidities and procedures that can be more adequately assessed using both databases.

Future studies need to investigate SIH/SUS recording in smaller hospitals and the reasons for differences in recording between different hospitals, as well as assessing strategies for improving the quality of recording on the SIH/SUS.

The study of maternal morbidity is an important component of strategies to improve obstetric care and reduce maternal mortality. The SIH/SUS is a nationwide information system that contains morbidity data and the results of this study show a high capture of obstetric hospitalizations when using the proposed operational definitions. However, specific diagnoses, with the exception of “ectopic pregnancy”, were under-reported, as were procedures that are indicators of management of serious complications.

Continuous improvement of AIH records, especially diagnoses and risk procedures most relevant to maternal morbidity and mortality, such as hypertensive and hemorrhagic complications, is essential for the use of the SIH/SUS in maternal morbidity surveillance. Possible strategies to be recommended include training of health professionals and implementation of standardized practices, to be followed even in exceptional periods, such as pandemics, aiming to improve accuracy and consistency in recording diagnoses and procedures, including secondary diagnoses, in both public and private hospitals.

REFERENCES

Publication Dates

History