Factors associated to the type of childbirth in public and private hospitals in Brazil

Objectives: to estimate the prevalence of cesarean sections and factors associated to the type of childbirth in Brazil. Methods: data on childbirths were collected in Brazil in 2014. Demographic characteristics, related to pregnancy and birth hospital regime (public or private) were evaluation. For each hypothesis raised, the variables were modeled by the binary logistic regression, which the outcome was considered in the type of childbirth. Results: the prevalence of the cesarean sections in Brazil in 2014 was 52.8%; that is 38.1% at public hospitals and 92.8% at private ones. The association between cesarean section and the legal regime at the hospital was highlighted in the logistic model which presented a positive association and interaction between age groups (OR = 23.26; 95% CI= 13.39 41.79 for women between 20 and 24 years old and OR = 51.04; 95% CI 31.06 84.23 for women aged 35 and over). Conclusions: the performance of childbirth in Brazil meets the routines and recommendations regarding women's health and humanized childbirth established by the Brazilian National Health System policies.


Introduction
The excessive medicalized childbirth care model has been criticized worldwide, which culminated in adopting in maternal health as a priority in the international agenda in recent years, 1,2 as well as the national policies, 3,4 translated into the creation of a systematic assessment routine for obstetric practices, in which the World Health Organization (WHO) has adopted over the last decades. 5,6ormal childbirth is considered a physiological event that requires support, evaluation and surveillance.Evidences indicate that to intervene in this process it should be justified as a valid reason. 7A cesarean section, on the other hand, is a procedure introduced into obstetrical practice with the purpose of preserving maternal and children's lives that are put at risk by complications during the prenatal period and childbirth.The WHO recommends that cesarean section rates should be kept below 15%. 8owever, this practice has increased over the last three decades, with rates observed up to 50% in some countries. 9Specifically in Latin America, most countries have high rates of cesarean sections.Brazil particularly presents high cesarean section rates and it is still increasing from 38.9% in 2000 to 46.5% in 2007.Preliminary data of cesarean sections in 2014 indicated a rate of 56.64% and 62.66% in the South region of the country. 10This increase in cesarean sections in Brazil, has been observed predominantly since the 1970's, highlighting the importance of identifying and studying the factors associated to decide on the type of childbirth.
The diagnosis of overutilization of cesarean sections in Brazil has generated growing concerns about the unnecessary use, generating issues as the quality of obstetric care up to the meaning of parturition for the women. 11,12Thus, it is necessary to understand the factors that lead to the increase of this practice, so that the public policy actions could reflect over the specific population groups in order to increase their effectiveness. 13n this context, the aims of this present study are to describe the prevalence of cesarean sections and estimate the magnitude of the associations among the type of childbirth and demographic and pregnancy-related characteristics in both public and private hospitals in Brazil.

Methods
This study consisted of a cross-sectional study which the unit of analysis is livebirth.This information is available through the Declaração de Nascido Vivo (the Brazilian Live Birth Registration), which it is sent to the administrative registry, afterwards sent to the Municipal Health Secretary (and subsequently to the other Health Information levels) and sent to a specific Information System that contains data concerning all the births in Brazil.
The Information System on Live Births used microdata regarding births that occurred in hospital environments in 2014.The following variables were: age, categorized in age ranges: "up to 19 years old", "20 to 24 years old", "25 to 29 years old", "30 to 34 years old ", and "35 years old and over"; marital status, categorized as with no partner ("single", "widower" and "separated") and with a partner ("married" and "consensual union"); schooling, categorized as "up to 8 years of schooling" and above "8 years of schooling"; type of pregnancy, categorized as "single" or "multiple"; gestational age, categorized as "preterm and post-term" and "term"; primiparous, categorized as "yes" and "no"; type of hospital, categorized as "public" and "private"; number of prenatal consultations, 4 characterized as "adequate" (7 consultations or more) or "inadequate" (less than 7 consultations); 4 place of residence, considering whether the puerperal woman resided in the same location as where the childbirth occurred, classified as "yes" or "no"; and, finally, the type of childbirth categorized as "vaginal" and "cesarean section".
The dichotomous variable type of childbirth was considered as the outcome variable (dependent), while other variables were evaluated as variables of interest (independent).Hypotheses were elaborated based on the variables of interest.For each hypothesis raised, the variables were modeled by the binary logistic regression, which the outcome was "cesarean section".In order to evaluate the adjustment of the tested alternative model, it was established a deviance analysis of the model, in order to compare the difference between the deviances of the null model (only with the intercept) and the variable of choice.
Afterwards the univariate modeling proceeded the introduction of the variables in a multivariate model, from the strength of the association that each variable assumed in relation to the outcome, observed the differences between the deviances assessing the adjustment of the model.After verifying the model with the inclusion of all the explanatory variables that were statistically significant, proceeded an adequacy test on some interaction terms.The choice of the interaction terms was based on the underlying theoretical referential.For this, the null hypothesis was considered as the model in Guimarães RM et al.

Prevalence of cesarean sections and factors associated to the type of childbirth
which the statistically significant variables were included, obtaining the following previous described step.
To validate the established logistic regression model it was necessary to apply some validation tests on this model and also to verify if it was adequate The Hosmer-Lemeshow, Pearson and Deviance tests were used for this purpose to validate the model. 14inally, as this study used secondary, public origin and unidentifiable databases according to the Resolution 466/2012, this study is exempt from the approval of the Ethics Committee.

Results
In 2014, the Information System on Live Births (Sistema de Informações sobre Nascidos Vivos -SINASC) registered 2,979,259 births in hospitals in Brazil.The descriptive data of the study estimated the prevalence of cesarean sections at 58.2%, predominantly among young, single, high schooling level, multiparous women with singleton, at term gestations.Regarding prevalence of cesarean sections by type of hospital, a statistically significant difference (p <0.001) was observed, with the prevalence of 38.1% of cesarean sections performed at public hospitals and 92.8% at private ones.This profile is not the same when observing births according to the type of childbirth (Table 1).Generally, vaginal childbirths are more frequent among younger, single, high schooling level women with single, multiparous and at term pregnancies, while cesarean sections are more frequent in slightly older women.Among these, a higher frequency of married multiparous women with high schooling level, multiple pregnancies and premature birth rates were observed.
Since a statistically significant difference was observed for all variables in the bivariate analysis (Table 1), modeling was conducted by the logistic regression.Initially, univariate models were tested in order to compare their ajustment with the null model, without the presence of any variables and considering only the intercept.When testing the alternative models, as described in the methods, all variables contributed to the explanation of the phenomenon, and thus, after assessing the differences between the deviances of the null and alternative models, multivariate modeling was chosen and performed.
Subsequently, based on the multiple logistic regression, a model was constructed to estimate the probability of a woman undergoing a cesarean section at hospitals in Brazil.A reduced model with 12 parameters (9 independent variables and 3 terms of interaction) was obtained and all statistically significant at the 1% level.The explanatory power of this model was 42% (Naegelkerke's R 2 ).Table 2 presents more detailed information about this chosen model.
After defining the model, the Hosmer-Lemeshow and Deviance adjustment adequacy statistics were applied to verify the hypotheses regarding the acceptance of the model.The hypotheses were formulated as H 0 , where the adjustment of the data is good versus H 1 , where the adjustment of the data is not good.Analysis of the residues through Chi-square test for the deviances resulted in a value of 0.96, while th Hosmer-Lemeshow statistic resulted in 2.84 (10 gl), obtaining a value of 0.94.
In addition to the presented statistics, three other model discrimination indices o were also assessed.The C statistic assesses the discrimination of capacity model by calculating the ROC curve area, and ranged from 0.5 to 1, the closer the values are to 1 the more appropriate is the model.The statistic value for this performed model was 0.82, classified as excellent according to Hosmer and Lemeshow's criteria. 11The D xy statistics (Somers correlation) establishes the correlation between the estimated probabilities and the observed responses, and ranges from 0 and 1.Whereas the value of zero means that the prediction models are completely random and the value of 1 means that the model is perfectly discriminatory.In this present study, the value found was 0.69.Finally, the sensitivity and specificity of the model were assessed through contingency tables with values of 66% and 89%, respectively.This indicates that this is a more specific model that demonstrates with more reliability concerning cesarean section compared to vaginal childbirth.Thus, the model is considered adequate for this purpose.
After adjusting the model, it was then applied to estimate the probability of a pregnant woman undergoing a cesarean section.Considering the particular interest in observing the difference of this probability at public and private hospitals, the probabilities were chosen to be estimated and the odds ratios for the terms concerning the type of hospital (public or private).
All variables included in the estimated probability formulas are the dummy type, so that the calculation for the success probability (cesarean section) for each type of hospital will be calculated by setting the values of the other variables and assigning the value of 1 when the location is a public hospital and zero when it is private one.Thus, there   is for each age group the estimated probabilities and the odds ratio according to the values presented in Table 3.Finally, Figures 1 and 2, respectively, represent the curves regarding the estimated probability of a cesarean section and a vaginal childbirth per type of hospital according to age group, as well as the tendency of increasing the odds ratio for each age group.It should be noted that the estimated probability increases with increasing age, at a different level for each type of hospital (there is unequivocally, a greater probability in private hospitals).However, it is important to note that the two curves are not parallel, evidencing an interaction effect between age and the type of hospital (public or private).The evidence regarding the possibility of a cesarean section being higher with the increasing age is, thus, reinforced with a tendency for age strata.
From the information displayed in the tables and the added graphs, there is an advantage of women in private hospitals being submitted to cesarean sections compared to women in public hospitals.In addition, an interaction with age was observed, or, in other words, this advantage presents a different magnitude according to age.Finally, this advantage grows with increasing age.For example, women aged 35 or more display an advantage concerning a cesarean section 5000% higher compared to women under 19 in private hospitals.It is worth noting that the estimated probability for 35-year-old women in private hospitals is approximately 97%, or, in other words, almost all the childbirths performed in these institutions occured in this age group.

Discussion
Maternal factors associated to cesarean sections were observed in Brazil.The risk factors presented herein are consistent with those presented in the literature.A cross-sectional study conducted in Rio Grande do Sul evaluating 2591 live births observed a significant association for the age group and the choice of childbirth for the age groups of 20-24 years old (OR = 1.13), 25-29 (OR = 1, 36) and 30

Odds Ratio
years old or more (OR = 1.21); for marital status with a partner (OR = 1.26); for high schooling level (OR = 1.28); for multiple pregnancies (OR = 2.01); and for protective association in multiparous women (OR = 0.94). 15ome associations seem to maintain even among patients who perform prenatal care in public units.In a study carried out with 322 pregnant women performing prenatal care at a Basic Health Unit in Rio de Janeiro, found an association for the most advanced age group (OR = 4.82) and the married women (OR = 3.05). 16tudies carried out in maternities also corroborate the direction of the associations observed.For example, a case-control study at a public maternity hospital in Rio de Janeiro city inclded 231 cesarean sections (cases) and 230 vaginal childbirths (controls).Through multivariate logistic regression analyses, the authors found a positive association for cesarean sections in women older than 35 years old (OR = 7.3) and for primiparous women within the multiparous reference category (OR = 6.7). 17In addition, a sample of 15,336 women in a general hospital (therefore, not a maternity), Padua et al. 18 found a significant association for the more advanced age groups, stratified as 20 to 24 years old (OR = 1.26), 25 to 29 (OR = 1.54), 30 to 34 (OR = 1.82), and 35 years and more (OR = 2.05).The same study found a significant association for the married women (OR = 1.25) and for a greater number of prenatal consultations (OR = 1.24).
It is important to emphasize the importance of understanding the social representations of normal and cesarean sections childbirths for women, which is a qualitative aspect difficult to be measured.A qualitative study was conducted with 20 women in Santa Catarina city who experienced both childbirth types.The results reveal several representations of motherhood experience, such as the search for information, the experience of parturition alone versus accompanied, and the woman has no choice.Vaginal childbirth encompassed central themes such as feelings of ambivalence, positive perception and hospitalization.Cesarean sections were also related to feelings of ambivalence, the solution of a problem and the preference of the procedure.In other words, vaginal childbirth is considered a challenge for women, although positive feelings overcome the difficulties, while cesarean sections are associated to physical benefits related to its accomplishment. 19t is also noteworthy that lack of humanized attention and induction often results in women opting for a cesarean section.In addition, unpreparedness for vaginal childbirth interferes directly with the emotional system of the pregnant woman or parturient patient, reducing her confidence in the ability to be the protagonist of her own childbirth if she is not received by a providing health service.Thus, she cannot understand the advantages of vaginal childbirth and concludes that the cesarean section will bring more benefits for her and the baby. 20 difference was observed between public and private services regarding the type of childbirth.This data, the most consistent of all explanatory variables, is corroborated by the literature.Barros et al. 21onducted a cohort study made up of all the newborns from the urban area of Pelotas city in 2004 indicated a 45% prevalence of cesarean sections for the population.When stratified by the type of service, the rates were 36% among the patients from the National Health System and 81% in the private service.Similarly, another study 15 indicated a cesarean section rate of 43% for the public group and 86% for the private group.In addition, a recent study was conducted in Maringá city to assess the temporal tendency of childbirth according to funding source, 5 during 11 years of observation.77.1% of the childbirths were cesarean sections and only 22.9% were vaginal childbirths.In addition, an increasing tendency for cesarean sections and a decreasing tendency for vaginal childbirth in both types of funding (public and private) was evidenced.Cesarean section rates in private hospitals were always higher than 90% and more frequent than in public hospitals, even with a 36% increase in public hospitals during this studied period.
It is important to note that factors such as excessive intervention during the gestation, childbirth and puerperium processes are obstacles to the success of this policy, making it difficult to reach goals to decrease maternal mortality. 22This problem occurs even in women with low obstetric risks. 23This phenomenon (intense medicalization of childbirth process) associated to the maintenance of high maternal and perinatal mortality is known as the Brazilian perinatal paradox. 24This is, therefore, evidence that justifies the reorientation of the model in pregnancy, childbirth and the puerperium care.In this sense, there has been a progress in Brazil in organizing obstetric care in the National Health System (SUS) network.Particularly highlights the Rede Cegonha (Stork Network), standardized by an Administrative Rule Number 1459, with the aim to increase the access and improve the quality of prenatal, childbirth and puerperium care, as well as child care up to 24 months of age, 4 as a stimulus to decrease maternal mortality.In this context of Prevalence of cesarean sections and factors associated to the type of childbirth discussion of a childbirth model in Brazil, the, attempts to systematize the routines and the itinerary of pregnant women are being made, providing pregnant and puerperal women and newborns with a humanized and quality care to achieve links for pregnant women to go to a reference unit for childbirth and have safe transportation and to implement good practices on childbirth and birthcare. 4,25his study has limitations, especially the use of secondary data from SINASC, thus it is impossible to assess other variables that are not present in the Brazilian Live Births Registration.However, since this database considers births throughout Brazil, it displays excellent accuracy, 26 and considering the evidence produced from its analysis should be taken into account.
This study indicates that childbirth in Brazil meets the routines and recommendations established in the women's health and humanized childbirth policies stated by the National Health System (SUS).Therefore, it is important that there is a reflection on this theme, so that monitoring measurements on obstetric practices are implemented, complying with the international recommendations for better clinical management and humanization in childbirth process.

Figure 1 2
Figure 1 Estimated probability curves for cesarean section and vaginal childbirth per type of hospital according to age group.Brazil, 2014 (N=2,979,259).1

Table 1
Frequency type of childbirth according to demographic and clinical characteristics.Brazil, 2014 (N= 2,979,259).

Table 2
Logistic regression model with adjusted associations for statistically significant covariates and the type of childbirth with interaction terms.Rio de Janeiro, 2014.(N= 2,979,259).
aThe reference range is Public hospital"; b The reference range is "up to 19 years of age"; c

Table 3
Estimated probabilities, chances and odds ratio to perform or not a cesarean section on women according to age group.Brazil, 2014 (N=2,979,259).
OR 95% CI: Confidence interval of the odds ratio at a significance level of 95%.