Abstract

To evaluate the causal effect of hypertensive syndromes during pregnancy (HSP) on prematurity. Data were obtained from the nationwide study “Born in Brazil” (“Nascer no Brasil”), conducted with 23,894 women. Hypertensive syndromes comprised the synthesis of positive responses to any of the questions relating to increased blood pressure contained in the questionnaires completed with data from patients’ hospital records and prenatal cards. The outcome was early and late preterm birth. The confounding variables were indicated in the directed acyclic graph, and the causal effect was estimated by the propensity score. Of the 20,494 postpartum women evaluated in this study, 2,369 presented a diagnosis of hypertensive syndromes, and among these, 5.8% were early premature births and 13.5% were late premature births. After weighting, women with hypertensive syndromes had a 2.74-fold greater chance of having an early preterm birth (ORadj: 2.74; 95%CI: 2.12-3.54) and a 2.40-fold greater chance of having a late preterm birth (ORadj: 2.40; 95%CI: 1.86-3.08). The causal effect of hypertensive syndromes on prematurity reaffirmed their role in causing an increase in premature births, reinforcing the importance of controlling syndromes during prenatal care.

Key words:
Hypertension Pregnancy-Induced; Early prematurity; Late prematurity; Directed acyclic graph; Propensity score

Introduction

Hypertensive syndromes during pregnancy (HSP) are among the main causes of maternal and perinatal morbidity and mortality¹1 The American College of Obstetricians and Gynecologists (ACOG). Practice Bulletin No. 202: Gestational Hypertension and Preeclampsia. Obstetrics Gynecology 2019; 133(1):e1-e25.^,22 Wang W, Xie X, Yuan T, Wang Y, Zhao F, Zhou Z, Zhang H. Epidemiological trends of maternal hypertensive disorders of pregnancy at the global, regional, and national levels: a population-based study. BMC Pregnancy Childbirth 2021; 21(1):364.. Worldwide, HSPs occur in approximately 3% to 14% of all pregnancies³3 Say L, Chou D, Gemmill A, Tunçalp Ö, Moller AB, Daniels J, Gülmezoglu AM, Temmerman M, Alkema L. Global causes of maternal death: a WHO systematic analysis. Lancet Glob Health 2014; 2(6):e323-e333.. In Brazil, a cross-sectional study conducted by Oliveira et al.⁴4 Oliveira CA, Lins CP, Sá RAM, Chaves Netto H, Bornia RG, Silva NR, Amim Junior J. Síndromes hipertensivas da gestação e repercussões perinatais. Rev Bras Saude Materno Infantil 2006; 6(1):93-98. found that of the 12,272 pregnant women studied, 10.2% had HSP. Leal et al.⁵5 Leal LF, Grandi SM, Miranda VIA, Dal Pizzol TDS, Platt RW, Silveira MF Bertoldi AD. Hypertensive Disorders of Pregnancy and Medication Use in the 2015 Pelotas (Brazil) Birth Cohort Study. Int J Environ Res Public Health 2020; 17(22):8541., based on data from the Pelotas birth cohort in 2015, found that 31.3% of the women had HSP. In the epidemiological context, the frequency of HSP varies greatly due to the different characteristics of the population, definitions, and diagnostic criteria used⁶6 Franco EP, Rocha DM, Gama SGN, Moreira MEL, Saunders C. Divergências metodológicas entre os estudos que avaliaram a associação entre as síndromes hipertensivas da gestação e a prematuridade: uma revisão integrativa da literatura. Res Soc Develop 2021; 10(12):e591101220591..

The diagnostic criterion for HSP is the presence of hypertension greater than or equal to 140/90 mmHg on two occasions with a minimum interval of four hours⁷7 Report of the National High Blood Pressure Education Program Working Group on High Blood Pressure in Pregnancy. Am J Obstet Gynecol 2000; 183(1): s1-s22.. The most commonly accepted recommendations come from the American College of Obstetricians and Gynecologists¹1 The American College of Obstetricians and Gynecologists (ACOG). Practice Bulletin No. 202: Gestational Hypertension and Preeclampsia. Obstetrics Gynecology 2019; 133(1):e1-e25., which classifies HSP into four categories: chronic arterial hypertension, gestational hypertension, preeclampsia (PE)/eclampsia, and PE superimposed on chronic hypertension. More recently, the Guidelines of the Brazilian Network of Studies on Hypertension in Pregnancy⁸8 Peraçoli JC, Costa ML, Cavalli RC, Oliveira LG, Korkes HA, Ramos JGL, Martins-Costa SH, de Sousa FLP, Cunha Filho EV, Mesquita MRS, Corrêa Jr MD, Araujo ACPF, Zaconeta AM, Freire CHE, Poli-de-Figueiredo CE, Rocha Filho EAP, Sass N. Pré-eclâmpsia - Protocolo 2023. Rede Brasileira de Estudos sobre Hipertensão na Gravidez (RBEHG); 2023. classify HSP into five categories: chronic arterial hypertension, white coat syndrome, gestational hypertension, preeclampsia (PE)/eclampsia, and PE superimposed on chronic hypertension.

Prematurity is one of the neonatal outcomes most frequently associated with HSP and can occur because of the interruption of pregnancy due to maternal and/or fetal compromise or spontaneous labor due to increased uterine contractility⁹9 Brandi LDA, Rocha LR, Silva LS, Bretas LG, Rodrigues MA, Araújo STH. Fatores de risco materno-fetais para o nascimento pré-termo em hospital de referência de Minas Gerais. Rev Med Minas Gerais 2020; 30:41-47.. Premature birth represents a major challenge for public health services worldwide¹⁰10 Berger AZ, Zorzim VI, Pôrto EF, Alfieri FM. Parto prematuro: características das gestantes de uma população da zona sul de São Paulo. Rev Bras Saude Mater Infant 2016; 16(4):427-435., and is considered to be the main risk factor for infant morbidity and mortality¹¹11 Chawanpaiboon S, Vogel JP, Moller AB, Lumbiganon P, Petzold M, Hogan D, Landoulsi S, Jampathong N, Kongwattanakul K, Laopaiboon M, Lewis C, Rattanakanokchai S, Teng DN, Thinkhamrop J, Watananirun K, Zhang J, Zhou W, Gülmezoglu AM. Global, regional, and national estimates of levels of preterm birth in 2014: a systematic review and modelling analysis. Lancet Glob Health 2019; 7(1):e37-e46..

In the systematic review conducted by Chawanpaiboon et al.¹¹11 Chawanpaiboon S, Vogel JP, Moller AB, Lumbiganon P, Petzold M, Hogan D, Landoulsi S, Jampathong N, Kongwattanakul K, Laopaiboon M, Lewis C, Rattanakanokchai S, Teng DN, Thinkhamrop J, Watananirun K, Zhang J, Zhou W, Gülmezoglu AM. Global, regional, and national estimates of levels of preterm birth in 2014: a systematic review and modelling analysis. Lancet Glob Health 2019; 7(1):e37-e46., data from 139 million live births in 2014 were evaluated. The authors identified that the frequency of premature births ranged from 8.7% in Europe and 13.4% in North Africa, while in Brazil this finding was 11.2%, which ranked Brazil 9th among 10 countries with the highest frequencies of premature births.

Although previous studies have already evaluated the association between HSP and prematurity¹²12 Antunes MB, Demitt MO, Gravena AAF, Padovani C, Pelloso SM. Síndrome hipertensiva e resultados perinatais em gestação de alto risco. Rev Mineira Enferm 2017; 21:1-6.^,1313 Lu CQ, Lin J, Yuan L, Zhou JG, Liang K, Zhong QH, Huang JH, Xu LP, Wu H, Zheng Z, Ping LL, Sun Y, Li ZK, Liu L, Lyu Q, Chen C. Pregnancy induced hypertension and outcomes in early and moderate preterm infants. Pregnancy Hypertens 2018; 14:68-71., Franco et al.⁶6 Franco EP, Rocha DM, Gama SGN, Moreira MEL, Saunders C. Divergências metodológicas entre os estudos que avaliaram a associação entre as síndromes hipertensivas da gestação e a prematuridade: uma revisão integrativa da literatura. Res Soc Develop 2021; 10(12):e591101220591. in a recent integrative literature review highlighted the presence of methodological differences between these studies, especially with regard to the classification of HSP and prematurity, and the control of confounding factors.

The present study therefore aimed to evaluate the causal effect of HSP on early and late preterm birth based on national data. In addition, the present study is justified by evaluating the causal effect of HSP on early and late preterm birth using the directed acyclic graph (DAG), a graph tool used to identify the covariates that may or may not confound this causal relationship¹⁴14 Hernán MA, Herández-Diaz S, Werler MM, Mitchell AA. Causal Knowledge as a prerequisite for confounding evaluation: an application to birth defects epidemiology. Am J Epidemiol 2002; S155(2):176-184., and the propensity score, a statistical technique to evaluate the effects of treatment (exposure) on the outcome when quasi-experimental or observational data are used¹⁵15 Patino CM, Ferreira JC. Escores de propensão: uma ferramenta para ajudar a quantificar os efeitos de tratamento em estudos observacionais. J Bras Pneumol 2017; 43:86..

Methods

Study design and population

This article is part of a nationwide, hospital-based cross-sectional study entitled “Nascer no Brasil: Inquérito Nacional sobre Parto e Nascimento” (“Born in Brazil: National Survey on Childbirth and Birth”), conducted between 2011 and 2012. The STROBE¹⁶16 Von EE, Altman DG, Egger M, Pocok SJ, Gøzsche PC, Vandenbroucke, JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 2007; 370(9596):1453-1457. guideline was used to guide the writing of all sections of this study.

The sample for the larger study was selected in three stages. In the first stage, hospitals with more than 500 births per year were stratified according to the five macro-regions of the country, location (capital or interior) and type of service (public, mixed, or private), while 266 hospitals were selected with a probability of selection proportional to the number of births in each of the strata in 2007. In the second stage, the number of days needed to interview 90 postpartum women in each hospital (minimum of 7 days) was defined using an inverse sampling method. In the third stage, eligible women were selected. Ninety interviews were planned per hospital, and 23,894 women were interviewed. Details of the sampling design and selection of postpartum women are available in Vasconcellos et al.¹⁷17 Vasconcellos MTL, Silva PLN, Pereira APE, Schilithz AC, Souza Junior PRBS, Szwarcwald CL. Desenho da amostra Nascer no Brasil: Pesquisa Nacional sobre Parto e Nascimento. Cad Saude Publica0 2014; 30:S49-S58., and data collection can be found in Leal et al.¹⁸18 Leal MC, Silva AAM, Dias MAB, Gama, SGN, Rattner D, Moreira ME, Theme Filha MM, Domingues RM, Pereira AP, Torres JA, Bittencourt SD, D'orsi E, Cunha AJ, Leite AJ, Cavalcante RS, Lansky S, Diniz CS, Szwarcwald CL. Birth in Brazil: national survey into labour and birth. Reprod Health 2012; 9:15..

For the larger study, women who had a live birth, regardless of weight or gestational age, or a stillbirth weighing ≥500 g and/or gestational age ≥22 weeks of gestation were included, and those with severe mental disorders, who were deaf, or who did not understand Portuguese were excluded¹⁸18 Leal MC, Silva AAM, Dias MAB, Gama, SGN, Rattner D, Moreira ME, Theme Filha MM, Domingues RM, Pereira AP, Torres JA, Bittencourt SD, D'orsi E, Cunha AJ, Leite AJ, Cavalcante RS, Lansky S, Diniz CS, Szwarcwald CL. Birth in Brazil: national survey into labour and birth. Reprod Health 2012; 9:15..

Exclusion criteria

The sample of the present study consisted of 20,494 women (18,125 without HSP and 2,369 with HSP). Of the number of newborns in the larger study (n=24,200), 3,686 were excluded for the following reasons: 489 twins, 61 without information on the presence of HSP, 933 without information on the adequacy of prenatal care, two due to the absence of data on maternal age, one due to the absence of data on parity, and 2,200 newborns with gestational age ≥41 weeks. Of this number (20,514 newborns), 8,336 early term newborns (37 and 38 weeks) were excluded, totaling 12,178 preterm and full-term births. Of this total, 534 were classified as early preterm birth; 1,605 were late preterm births, and 10,039 were full-term newborns (between 39 and 40 weeks) (Figure 1).

Study variables

The exposure variable was the presence of HSP, which consisted of the summary of positive responses to any question contained in the questionnaires completed with data from hospital records and prenatal cards: “hypertension with continued treatment”, “hypertensive syndromes during the current pregnancy (chronic hypertension, preeclampsia, or HELLP syndrome (H: hemolysis; EL: elevated liver enzymes; LP: low platelet count)”, “eclampsia/seizure in the current pregnancy”, “diagnosis of eclampsia and seizure upon admission”, “diagnosis of hypertension during pregnancy upon admission (any type)”, and “high blood pressure outside of pregnancy”.

The outcome studied was early preterm birth (gestational age <34 weeks) and late preterm birth (gestational age between 34 and 36 weeks). The gestational age considered to be full-term (between 39 and 40 weeks and 6 days of gestation) was used as the reference category, regardless of the onset of labor. Gestational age was estimated using an algorithm based mainly on estimates from early ultrasound performed between 7 and 20 weeks of gestation. In the absence of this examination, gestational age was based on information reported by the postpartum women in the interview and, finally, on the date of the last menstrual period¹⁹19 Pereira APE, Leal MC, Gama SGN, Domingues RMSM, Schilithz AOC, Bastos MH. Determining gestational age based on information from the Birth in Brazil study. Cad Saude Publica 2014; 30:S59-S70..

Newborns with a gestational age between 37 and 38 weeks (early term) and newborns with a gestational age ≥41 weeks were excluded to ensure that the comparison group, newborns with a gestational age between 39 and 40 weeks of gestation, had a lower prevalence of factors related to early and late gestational age²⁰20 Leal MC, Esteves-Pereira AP, Nakamura-Pereira M, Domingues RMSM, Dias MAB, Moreira ME, Theme Filha M, Gama SGN. Burden of early-term birth on adverse infant outcomes: a population-based cohort study in Brazil. BMJ Open 2017; 7(12):e017789..

Data analysis

Data analysis was performed in five stages. Initially, a DAG was developed based on a broad bibliographic survey. This graph tool aimed to establish the relationship among exposure (HSP), covariates, and outcomes (early and late preterm birth).

The DAG of the present study was developed in the DAGitty program (public domain, available at http://www.dagitty.net/) developed to create, edit, and analyze causal models²¹21 Silva AAM. Introdução à Inferência Causal em Epidemiologia: uma abordagem gráfica e contrafactual. Rio de Janeiro: Editora Fiocruz; 2021.. This program follows strict DAG rules to identify the minimum set of variables to be adjusted for confounding factors in order to identify the causal effect²¹21 Silva AAM. Introdução à Inferência Causal em Epidemiologia: uma abordagem gráfica e contrafactual. Rio de Janeiro: Editora Fiocruz; 2021.. This DAG is presented in Figure 2.

The process of selecting variables for adjustment followed the graphical algorithm²²22 Pearl J. The art and science of cause and effect. In: Causality: models, reasoning and inference. Cambridge: University of Cambridge; 2000. p. 331-358., which comprises six criteria, until a minimum set of potentially confounding variables was selected²³23 Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol 2008; 8:70.. After applying the graphical criteria, from the 26 covariates inserted in the graphical model, the minimum set of 14 potentially confounding variables to be used in adjusting the causal relationship between GHS and prematurity was identified, namely: (1) maternal age (12 to 19; 20 to 34; ≥35 years); (2) maternal education (incomplete elementary school; complete elementary school; complete high school; higher education or more); (3) marital status (without a partner; with a partner); (4) pre-gestational body mass index (BMI): low weight (<18.5 kg/m²); eutrophic (≥18.5 - <24.9 kg/m²); overweight (>25.0 - ≤29.9 kg/m²) and obesity (≥30 kg/m²); (5) gestational weight gain (insufficient; adequate; excessive); (6) adequacy of prenatal care (inadequate; partially adequate; adequate; more than adequate); (7) parity (primiparous; 1 to 2 births; 3 or more births); (8) anemia (no; yes); (9) pregestational diabetes mellitus (no; yes); (10) gestational diabetes mellitus (no; yes); (11) autoimmune disease (systemic lupus erythematosus - SLE) (no; yes); (12) chronic kidney disease (no; yes); (13) urinary tract infection (no; yes). The variable “interpregnancy interval” was not collected in the original survey, therefore, it was not included for adjustment in the data analysis (Figure 2).

To assess the pre-gestational nutritional status of women, pre-gestational BMI was used based on the cutoff points recommended by the Institute of Medicine²⁴24 Institute of Medicine. Institute of Medicine, National Research Council (US) Committee to Reexamine IOM Pregnancy Weight Guidelines. In: Rasmussen KM, Yaktine AL, editors. Weight Gain during Pregnancy: Reexamining the Guidelines. Washington D.C.: The National Academis Press; 2009.. Total gestational weight gain was calculated by subtracting the weight of the last prenatal visit from the pre-gestational weight, both collected from the prenatal card or self-reported by the postpartum woman. The adequacy of weight gain was corrected for gestational age at birth. For each week less than 40 weeks (full-term gestation), the average weekly weight gain was discounted from the minimum and maximum values for each pre-gestational BMI range in the second and third gestational trimesters.

To classify the adequacy of prenatal care, the gestational trimester at the time of the start of prenatal care, the number of consultations attended corrected for gestational age at the time of delivery, routine exams performed, and the indication of the reference maternity hospital for childbirth care were considered, which was considered based on the mother’s report on the medical advice received. Prenatal care was considered adequate when prenatal care began up to 12 weeks of gestation and 100% of the minimum consultations scheduled for the gestational age at the time of delivery had been completed, according to the recommendation of the Rede Cegonha²⁵25 Brasil. Ministério da Saúde (MS). Portaria no 1.459, de 24 de junho de 2011. Institui, no âmbito do Sistema Único de Saúde - SUS - a Rede Cegonha. Diário Oficial da União; 2011. in effect during the broader study²⁶26 Domingues RMSM, Viellas EF, Dias MAB, Torres JA, Theme Filha MM, Gama SGND, Leal MC. Adequação da assistência pré-natal segundo as características maternas no Brasil. Rev Panam Salud Publica 2015; 37(3):140-147..

After the DAG had been performed, the second stage consisted of a descriptive analysis of maternal, prenatal and postpartum characteristics according to the presence or absence of HSP. and the analyzed outcomes (early and late preterm birth). In this stage, the Chi-square test with Rao-Scott adjustment²⁷27 Rao JNK; Scott AIJ. The analysis of categorical data from complex sample surveys: chi-squared tests for goodness of fit and independence in two-way tables. J Am Stat Assoc 1981; 374(6):221-230. was used to compare the proportions between the exposed and unexposed groups. The significance level adopted was 5%.

In the third stage, bivariate logistic regression was performed between the adjustment covariates contained in the minimum set of the DAG (maternal age, maternal education, marital status, pre-gestational BMI, gestational weight gain, adequacy of prenatal care, parity, anemia, pre-gestational diabetes mellitus, gestational diabetes mellitus, autoimmune disease (systemic lupus erythematosus), chronic kidney disease, urinary tract infection), and the outcomes (early and late preterm birth), using full-term newborns as the reference. The results were expressed as odds ratios (OR) with their respective 95% confidence intervals (95%CI).

The propensity score weighting method was then applied, which consists of assuming interchangeability, i.e., treated/exposed individuals are similar to untreated/control individuals in such a way that they could be in either group if the exposure was not different between them²⁸28 D'Agostino Jr RB. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med1998; 17(19):2265-2281.. Once the selection probabilities for each woman had been estimated, the next step was to weigh the estimate by the inverse of the selection probability²⁹29 Paes A. Uso de Escores de Propensão Para Corrigir Diferenças Entre Grupos. 2012; 10(3):103-104.. With this method, the aim is to compensate for differences by assigning greater weights to less common observations and lower weights to more frequent ones in an attempt to achieve balance in the study²⁹29 Paes A. Uso de Escores de Propensão Para Corrigir Diferenças Entre Grupos. 2012; 10(3):103-104..

After estimation and weighting with the propensity score, the fourth stage verified the balance of the groups in relation to the adjustment covariates that could interfere in the relationship between the exposure and the outcome using the absolute standardized difference of the means. The balance was verified before and after the implementation of the propensity score and was considered adequate when less than 0.10²¹21 Silva AAM. Introdução à Inferência Causal em Epidemiologia: uma abordagem gráfica e contrafactual. Rio de Janeiro: Editora Fiocruz; 2021..

In the fifth stage, the crude and adjusted odds ratios between HSP and early and late preterm birth were calculated with the respective 95%CI. The analyses were performed using the R software, version 3.4.2 (The R Foundation for Statistical Computing), using the twang library to estimate the propensity score.

In the statistical analysis, the complex sampling design was considered with the use of weighting and calibration of the data and the incorporation of the design effect of 1.3 in order to ensure that the distribution of the postpartum women in the sample was similar to that observed in the population for the year 2011.

Ethical aspects

The larger study was approved by the Research Ethics Committee (Comitê de Ética em Pesquisa - CEP) of the Sérgio Arouca National School of Public Health, Oswaldo Cruz Foundation (ENSP/Fiocruz), logged under Opinion No. 92/10. For the purpose of the present study, analysis and approval by the CEP of the Fernandes Figueira National Institute of Women, Children, and Adolescents, Oswaldo Cruz Foundation (IFF/Fiocruz) was waived.

Results

In this study, data from 20,494 postpartum women were analyzed (18,125 without HSP and 2,369 with HSP). Of this total, 10.4% had premature births, 2.6% of which were early preterm births and 7.8% were late preterm births. Among women with HSP, the frequency of premature births was 19.3%, 5.8% of which were early preterm births and 13.5% were late preterm births.

When comparing the sociodemographic and clinical characteristics between women with HSP and the others, it was observed that the frequency of HSP was higher among women aged 12 to 19 years (20.2%), with incomplete elementary education (27.6%), without a partner (18.3%), and without gestational diabetes mellitus (93.1%) (Table 1).

Early preterm births were more frequent among women aged 12 to 19 years (27.3%), with incomplete primary education (38.6%), with low pre-gestational weight (9.9%), with inadequate prenatal care (31.1%), and who had 3 or more births (16.9%), as compared to women who had births between 39 and 40 weeks of gestation (Table 2).

When analyzing women who had late premature births in relation to those who had births between 39 and 40 weeks of gestation, a higher frequency of this outcome was observed among women aged 12 to 19 years (24%), with incomplete elementary education (30.8%), without a partner (20.7%), with low pre-gestational (10.2%), with insufficient gestational weight gain (28.5%), with inadequate prenatal care (30.9%), and with 3 or more births (14.5%) (Table 2).

In the logistic regression analysis, it was observed that the variables that increased the chance of early preterm birth were ages between 12 and 19 years (OR: 1.90; 95%CI: 1.55-2.32), incomplete elementary education (OR: 3.02; 95%CI: 2.18-4.18), low pre-gestational weight (OR: 1.35; 95%CI: 1.10-1.82), inadequate prenatal care (OR: 2.19; 95%CI: 1.75-2.74), parity with 3 or more births (OR: 2.24; 95%CI: 1.73-2.89), and urinary tract infection (OR: 1.41; 95%CI: 1.14-1.75). The protective variables for early preterm birth were age over 35 years (OR: 0.58; 95%CI: 0.41-0.83), pre-gestational obesity (OR: 0.68; 95%CI: 0.48-0.95), more than adequate prenatal care (OR: 0.49; 95%CI: 0.37-0.65), and gestational diabetes mellitus (OR: 0.55; 95%CI: 0.38-0.80) (Table 3).

It was observed that the variables that increased the chance of late preterm birth were ages between 12 and 19 years (OR: 1.62; 95%CI: 1.42-1.84), incomplete elementary education (OR: 3.35; 95%CI: 2.67-4.19) and complete high school (OR: 1.72; 95%CI: 1.38-2.15), having a partner (OR: 1.36; 95%CI: 1.91-1.55), low pre-gestational weight (OR: 1.38; 95%CI: 1.15-1.65), inadequate prenatal care (OR: 2.36; 95%CI: 2.05-2.72), and parity with 3 or more births (OR: 1.88; 95%CI: 1.59-2.22). The protective variables for this outcome were age over 35 years (OR: 0.35; 95%CI: 0.55-0.81), pre-gestational obesity (OR: 0.61; 95%CI: 0.49-0.75), more than adequate prenatal care (OR: 0.56; 95%CI: 0.47-0.66), gestational diabetes mellitus (OR: 0.52; 95%CI: 0.41-0.65), and urinary tract infection (OR: 0.54; 95%CI: 0.45-0.64) (Table 3).

It was found that after balancing the information, all mean differences for the covariates used to weight women were below 0.10, indicating that balancing after adjustment by the propensity score was adequate (data presented in the Scielo Data repository: https://doi.org/10.48331/scielodata.JVLT2K).

Women with HSP had a 3.34-fold greater chance of early preterm birth (OR: 3.34; 95%CI: 2.72-4.10) and a 2.41-fold greater chance of late preterm birth (OR: 2.41; 95%CI: 2.09-2.77). After analysis by propensity score, it was found that these women had a 2.74-fold greater chance of early preterm birth (ORadj: 2.74; 95%CI: 2.12-3.54) and a 2.40-fold greater chance of late preterm birth (ORadj: 2.40; 95%CI: 1.86-3.08) (Table 4).

Discussion

This study evaluated the causal effect of HSP on early and late preterm birth. The prevalence of HSP was 11.5%, similar to that found in the cross-sectional study conducted by Bacelar et al.³⁰30 Bacelar EB, Costa MCO, Gama SGN, Amaral MTR, Almeida AHV. Fatores associados à Síndrome Hipertensiva Específica da Gestação em puérperas adolescentes e adultas jovens da Região Nordeste do Brasil: análise múltipla em modelos hierárquicos. Rev Bras Saude Mater Infant 2017; 17(4):673-681., conducted with 2,960 women (13.4%). Regarding preterm birth, it was observed that the frequency of this outcome (19.3%) was almost twice as high among women with HSP when compared to the others, with a predominance of late preterm births (13.5%) as compared to early preterm births (5.8%).

Using the propensity score weighting method, the causal effect of HG on early (ORadj: 2.74; 95%CI: 2.12-3.54) and late (ORadj: 2.40; 95%CI: 1.86-3.08) preterm birth was observed. It is important to mention that although some researchers have analyzed the association between HSP and prematurity¹²12 Antunes MB, Demitt MO, Gravena AAF, Padovani C, Pelloso SM. Síndrome hipertensiva e resultados perinatais em gestação de alto risco. Rev Mineira Enferm 2017; 21:1-6.^,1313 Lu CQ, Lin J, Yuan L, Zhou JG, Liang K, Zhong QH, Huang JH, Xu LP, Wu H, Zheng Z, Ping LL, Sun Y, Li ZK, Liu L, Lyu Q, Chen C. Pregnancy induced hypertension and outcomes in early and moderate preterm infants. Pregnancy Hypertens 2018; 14:68-71., there are still few studies that have focused on the use of the methods employed to analyze the data in this study.

In Brazil, the cross-sectional study by Rezende et al.³¹31 Love ER, Crum J, Bhattacharya S. Independent effects of pregnancy induced hypertension on childhood development: a retrospective cohort study. Eur J Obstet Gynecol Reprod Biol 2012; 165(2):219-224., conducted with 4,464 pregnant women, found an association between PE and early prematurity (PR: 11.01; 95%CI: 7.21-14.80). In a cohort of 28,967 British pregnant women, Love et al.³²32 Rezende KBC, Bornia RG, Esteves APVS, Cunha AJL, Amim Junior J. Preeclampsia: Prevalence and perinatal repercussions in a University Hospital in Rio de Janeiro, Brazil. Pregnancy Hypertens 2016; 6(4):253-255. analyzed the association between gestational hypertension, PE, and early (<32 weeks of gestation) and late (33 to 36 weeks of gestation) prematurity, and found that women with gestational hypertension had a higher chance of early prematurity (OR: 1.55; 95%CI: 1.29-1.80), while those with PE had a higher chance of both early (OR: 4.27; 95%CI: 3.46-5.38) and late (OR: 1.55; 95%CI: 1.29-1.87) prematurity. When adjustment for maternal socioeconomic conditions was implemented, the association remained between PE and early and late prematurity.

Similarly, Johnson et al.³³33 Johnson KM, Zash R, Haviland MJ, Hacker MR, Luckett R, Diseko M, Mayondi G, Shapiro R. Hypertensive disease in pregnancy in Botswana: prevalence and impact on perinatal outcomes. Pregnancy Hypertens 2016; 6(4):418-422., based on a cohort study of 14,170 women, found that after adjusting for maternal socioeconomic factors, pregnant women with HSP had a 40% risk of prematurity (RRadj: 1.4; 95%CI: 1.3-1.6). Furthermore, when individually evaluating the components of HSP, the authors observed that the risk for this outcome was higher among pregnant women with PE (RR: 2.5; 95%CI: 2.2-2.8), followed by women with chronic hypertension (RR: 2.3; 95%CI: 2.1-2.6) and with gestational hypertension (RR: 1.2; 95%CI: 1.1-1.3).

Regarding the methods used to assess the causal effect of HSP on early and late preterm birth, it is important to briefly discuss the DAG and the propensity score weighting method. It is important to highlight that the assessment of causality is an extremely important issue for epidemiology and has been the subject of intense study for at least three centuries³⁴34 Werneck GL. Diagramas causais: a epidemiologia brasileira de volta para o futuro. Cad Saude Publica 2016; 32(8):e00120416..

It is understood that the study of the causal relationship between HSP and prematurity is quite complex and involves a network of socioeconomic³⁵35 Jacob LMDS, Santos AP, Lopes M HB M, Shimo AKK. Socioeconomic, demographic and obstetric profile of pregnant women with Hypertensive Syndrome in a public maternity. Rev Gaucha Enferm 2020; 41:e20190180.^,3636 Vogel JP, Chawanpaiboon S, Moller AB, Watananirun K, Bonet M, Lumbiganon P. The global epidemiology of preterm birth. Best Pract Res Clin Obstet Gynaecol 2018; 52;3-12., clinical³⁷37 Kahsay HB, Gashe FE, Ayele WM. Risk factors for hypertensive disorders of pregnancy among mothers in Tigray region, Ethiopia: matched case-control study. BMC Pregnancy Childbirth 2018; 18:482.^,3838 Billionnet C, Mitanchez D, Weill A, Nizard J, Alla F, Hartemann A, Jacqueminet S. Gestational diabetes and adverse perinatal outcomes from 716,152 births in France in 2012. Diabetologia 2017; 60(4):636-644., genetic³⁹39 Cavalli RDC, Sandrim VC, Santos JETD, Duarte G. Predição de pré-eclâmpsia. Rev Bras Ginecol Obstetr 2009; 31:1-4.^,4040 Sbrana M, Grandi C, Brazan M, Junquera N, Nascimento MS, Barbieri MA, Bettiol H, Cardoso VC. Alcohol consumption during pregnancy and perinatal results: a cohort study. Sao Paulo Med J 2016; 134(2):146-152., healthcare⁴¹41 Hinkosa L, Tamene A, Gebeyehu N. Risk factors associated with hypertensive disorders in pregnancy in Nekemte referral hospital, from July 2015 to June 2017, Ethiopia: case-control study. BMC Pregnancy Childbirth 2020; 20(1):16.^,4242 Pervin J, Rahman SM, Rahman M, Aktar S, Rahman A. Association between antenatal care visit and preterm birth: a cohort study in rural Bangladesh. BMJ Open 2020; 10(7):e036699., and nutritional⁴³43 Shen M, Smith GN, Rodger M, White RR, Walker MC, Wen SW. Comparison of risk factors and outcomes of gestational hypertension and pre-eclampsia. PloS One 2017; 12(4):e0175914.^,4444 Lefizelier E, Misbert E, Brooks M, Le Thuaut A, Winer N, Ducarme G. Preterm birth and small-for-gestational age neonates among pre pregnancy underweight women: a case-controlled study. J Clin Med 2021; 10(24):5733. factors. Thus, to deal with these multiple risk factors, the use of graph models would be the appropriate strategy, given that they are performed based on a flexible structure in order to explore the multidimensional determinants and complex causal mechanisms¹⁴14 Hernán MA, Herández-Diaz S, Werler MM, Mitchell AA. Causal Knowledge as a prerequisite for confounding evaluation: an application to birth defects epidemiology. Am J Epidemiol 2002; S155(2):176-184.. In this case, the use of DAG proved to be an adequate tool for dealing with consistent assumptions and multiple risk factors, in addition to allowing researchers to use relatively simple and systematic graph criteria to identify a set of confounding variables that need to be controlled in the analysis²¹21 Silva AAM. Introdução à Inferência Causal em Epidemiologia: uma abordagem gráfica e contrafactual. Rio de Janeiro: Editora Fiocruz; 2021..

Several points deserve to be highlighted in this study, including the novelty of assessing the causal effect of HSP on prematurity in Brazil based on a nationwide sample of public and private hospitals. The use of DAG to identify the minimum set of confounding variables also stood out, and the analysis method applied was weighting by the propensity score. In this type of study, the use of the propensity score aims to reduce bias in the estimates of the effect of exposure¹⁵15 Patino CM, Ferreira JC. Escores de propensão: uma ferramenta para ajudar a quantificar os efeitos de tratamento em estudos observacionais. J Bras Pneumol 2017; 43:86., by recovering the interchangeability between the treated (exposed) and untreated (unexposed) groups¹⁵15 Patino CM, Ferreira JC. Escores de propensão: uma ferramenta para ajudar a quantificar os efeitos de tratamento em estudos observacionais. J Bras Pneumol 2017; 43:86.^,4545 Austin PC. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behav Res 2011; 46(3):399-424.^,4646 Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983; 70(1):41-55.. With this method, it is possible to find identical or very similar individuals among the comparison groups, especially when all possible confounding covariates can be observed in relation to the outcome of interest⁴⁷47 Leite W. Practical propensity score methods using R. Los Angeles: Sage; 2017..

It is important to note that prematurity in the present study was not assessed in a conventional manner, considering a gestational age of less than 37 weeks⁴⁸48 Arce-López, KL, Iglesias-Leboreiro J, Bernárdez-Zapata I, Rendón-Macías ME, Madrazo M. Hematological disorders in preterm newborns born to mothers with pregnancy-induced hypertension. Bolet Med Hospital Infantil Mexico 2022; 79(6):363-368.^,4949 Cruz SS, Costa VS, Santos JM, Gomes TAO, Brandão ITR, Fonseca CHA, Vilasboas SWSL, Silva RB, Gomes-Filho IS, Souza LM, Figueireido ACMG, Batista JT. Prematuridade ao nascer, hipertensão materna e outros fatores associados: um estudo de caso-controle na região do Vale do São Francisco. Rev Bras Pesqu Saude 2018; 20(4):113-119.. Given that gestational age at birth is a factor that impacts physiological maturity, the present study categorized prematurity into early and late. The gestational period between the 34^th and 36^th weeks is essential for the immunological, cerebral, and pulmonary development of the newborn⁵⁰50 Purisch SE, Gyamfi-Bannerman C. Epidemiology of preterm birth. Semin Perinatol 2017; 41(7):387-391.. Regarding complications for premature infants, signs of severity are greater for early premature infants, as they are more likely to receive active resuscitation at birth, respiratory support, and parental nutrition during the hospitalization period, which requires a longer hospital stay⁵¹51 Bajaj M, Natarajan G, Shankaran S, Wycokoff M, Laptoock AR, Bell EF, Stoll BJ, Carlo WA, Vohr BR, Saha S, Van Meurs KP, Sanchez PJ, D'Angio CT, Higgins RD, Das A, Newman N, Walsh MC; Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network. Delivery Room Resuscitation and Short-term Outcomes in Moderately Preterm Infants. J Pediatr 2018; 195:33-38.e2.. Moreover, late premature infants are still physiologically and metabolically immature, presenting an increased risk of neonatal complications (respiratory distress, hypothermia, hypoglycemia, hyperbilirubinemia, feeding difficulties, and infections)⁵¹51 Bajaj M, Natarajan G, Shankaran S, Wycokoff M, Laptoock AR, Bell EF, Stoll BJ, Carlo WA, Vohr BR, Saha S, Van Meurs KP, Sanchez PJ, D'Angio CT, Higgins RD, Das A, Newman N, Walsh MC; Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network. Delivery Room Resuscitation and Short-term Outcomes in Moderately Preterm Infants. J Pediatr 2018; 195:33-38.e2.

52 Boyle EM, Johnson S, Manktelow B, Seaton SE, Draper SE, Smith KL, Dorling J, Marlow N, Petrou S, Field DJ. Neonatal outcomes and delivery of care for infants born late preterm or moderately preterm: a prospective population-based study. Arch Dis Child Fetal Neonatal Ed 2015; 100(6):F479-F485.^-5353 Almeida AHV, Gama SGN, Costa MCO, Carmo CN, Pacheco VE, Martinelli KG, Leal MC. Prematuridade e gravidez na adolescência no Brasil, 2011-2012. Cad Saude Publica 2020; 36(12):e00145919..

This study does have its limitations. Although the minimum set of variables also includes the variable interpregnancy interval, this was not included in the regression analysis, as it was not collected in the larger study. Although an extensive literature review was carried out to construct the DAG for this study, it is important to note that every graph may contain errors, since the true causal structure is often unknown, due to the considerable limitation of scientific knowledge²¹21 Silva AAM. Introdução à Inferência Causal em Epidemiologia: uma abordagem gráfica e contrafactual. Rio de Janeiro: Editora Fiocruz; 2021..

However, it is important to highlight that not using this approach due to uncertainty about the veracity of the DAG simply demonstrates that chance, rather than rational deliberation, is allowed when choosing between the different causal diagrams. Therefore, causal inference in observational studies will nearly always be a challenging exercise and will depend on the validity of the chosen model²¹21 Silva AAM. Introdução à Inferência Causal em Epidemiologia: uma abordagem gráfica e contrafactual. Rio de Janeiro: Editora Fiocruz; 2021.. Therefore, even in view of this issue, it is worth noting that the diagram was constructed by experienced researchers in the maternal-child area and that it can be used by other observational studies that have objectives similar to those of the present study.

Considering that HSP have a causal effect on early and late prematurity, early diagnosis and treatment of these syndromes should be analyzed during prenatal monitoring, aiming at their reduction and possible short- and long-term consequences for the newborn. Thus, it is assumed that the findings of the present study have important clinical implications, in addition to being an aid in the planning and creation of public health policies aimed at preventing these conditions, as well as in further studies in search of more effective preventive and interventional strategies for this population during prenatal care.

References

Publication Dates

History