Influence of mode of delivery on satisfaction with hospitalization for childbirth in the study Birth in Brazil

Controversial results have been reported on the association between mode of delivery and patient satisfaction. This study investigates which mode of delivery leads to greater satisfaction with hospital admission for childbirth. A cohort study was conducted with data from the Birth in Brazil study, which began in 2011. A total of 23,046 postpartum women were included from a random sample of hospitals, selected by conglomerates with a three level stratification. At the first follow-up, 15,582 women were re-interviewed. Mode of delivery, dichotomized into vaginal or cesarean section, and confounders were collected before hospital discharge. The outcome maternal satisfaction, investigated as a 10-item unidimensional construct, was measured by the Hospital Birth Satisfaction Scale up to six months after discharge. We used a directed acyclic graph to define minimal adjustment variables for confounding. The effect of mode of delivery on satisfaction was estimated using a structural equation model with weighting by the inverse of the probability of selection, considering the complex sampling design. The weight was estimated considering the different sample selection probabilities, the losses to follow-up, and the propensity score, which was estimated in a logistic regression model. The analysis revealed no significant difference in satisfaction with hospitalization for childbirth between respondents who had vaginal delivery and cesarean section in the adjusted analysis (standardized coefficient = 0.089; p-value = 0.056). Therefore, women who had vaginal delivery and cesarean section were equally satisfied with their hospitalization for childbirth.


Introduction
Although delivery is a physiological process typically without complications 1 , political, socioeconomic, cultural, institutional, technical-scientific changes, as well as changes in health and medical work models (which have become more technical and interventionist), have ended up characterizing delivery as an act of medical-hospital responsibility since the mid-21st century 2,3,4,5,6 .
While this medicalization of delivery has presented a positive impact on decreasing maternal and neonatal mortality rates, it has also resulted in medical interventions without scientific indications and loss of women's autonomy over the labor process 1,5 .Consequently, cesarean section rates have increased globally in the first two decades of the 21st century 4,7,8,9,10 , often exceeding the 10% to 15% recommended by the World Health Organization (WHO) 4,11 , and even surpassing vaginal childbirth rates in Brazil, Cyprus, Dominican Republic, and Egypt 12 .
Contrary to this increase in cesarean section rates 12 , studies carried out in countries at different levels of development have shown that, generally, women prefer vaginal childbirth 5,13,14,15,16,17,18,19,20 .Cesarean section, in turn, is preferred by multiparous women undergoing this mode of delivery 13,14,15,17,19,20 and among those with private health insurance 14,20 .The preference for cesarean section was associated with fear of pain 7,13,16,18,20 , perception that it is safer than vaginal delivery 13,18,20 , negative previous experiences in delivery 13 , influence of health professionals, friends, and family members 13 , and limited access to information about the characteristics of these two modes of deliveries 13,17 .
Delivering according to plan 21,22,23,24,25 contributes to maternal satisfaction with hospitalization for delivery.Participating in the choice of mode of delivery; ease of access to the hospital of delivery; facility with appropriate physical structure; availability of medicines and equipment; receiving dignified, respectful, and courteous treatment; privacy and confidentiality in care; availability of technically competent physicians and nurses (especially during emergencies); provision of cognitive and emotional support; and good delivery outcomes were associated with greater satisfaction with the delivery healthcare service in a systematic review 23 .Conversely, the perception of intense pain, especially during labor induction, instrumental vaginal delivery, emergency cesarean section, and prolonged labor, were associated with a negative experience 26 .
Maternal satisfaction with childbirth hospitalization has been more commonly investigated based on theories of satisfaction with care received in health services 23,24 .These theories are based on previous expectations or experiences that occurred during delivery or even on attributes of health services 24,27 .Its measurement instruments in quantitative surveys often assess one or more determinants of maternal satisfaction with labor/birth and vary in the number of items (6 to 30 questions), response options (dichotomous or Likert scale), and variable construction (single score and uni-or multidimensional constructs) 22,23,24,28 .
Based on the above, this study presents the following hypothesis: there should be no differences in maternal satisfaction with hospitalization for vaginal delivery and cesarean section if they are performed as idealized and planned by pregnant women.

Type of study
A cohort study was conducted with data from the first two stages of the Birth in Brazil study -a hospital-based population survey, which aimed to study the incidence, associated factors, and consequences of cesarean section in Brazil.The first stage was carried out from February 2011 to October 2012, and the second stage from March 2011 to February 2013 29,30,31 .

Sampling and data collection
From a list of 1,403 hospitals with more than 500 deliveries per year registered in the Brazilian Information System on Live Births (SINASC), 266 were selected.The sample was stratified by Brazilian Cad.Saúde Pública 2023; 39(3):e00138922 macro-regions (North, Northeast, Central-West, Southeast and South), type of municipality (state capital or not) and hospital administration (public, private or mixed).Each stratum presented at least 450 postpartum women selected from five or more hospitals.Inverse sampling was used to select the number of research days (minimum of seven) to reach 90 interviews per hospital.Information on data collection and sample design of the Birth in Brazil study is detailed in three articles 29,30,31 .
In the first stage, 24,200 puerperal women were interviewed during childbirth hospitalization 19,29,30,31 .In the first follow-up, 16,255 mothers were contacted and re-interviewed by telephone, on average 90 days (from 45 days to less than six months) after delivery 29,30,31 .
The basic sample weights were estimated as the inverse of the product of the inclusion probabilities at each stage and calibrated so that the estimates of the total live births of the strata corresponded to the total live births obtained from SINASC.For the telephone follow-up, conducted six months later, the probability of puerperal women's response was modeled by the variables available at the baseline survey to correct the sample weights for non-response in the second phase.The probability of response was estimated as a function of the three variables that define the stratum (macro-region, capital or not, and hospital administration) and the Brazilian Economic Classification Criteria 32 : age group, paid employment, satisfaction with the pregnancy at the beginning and stillbirth or neonatal death 31 .

Scale of satisfaction with hospitalization for delivery
The scale of satisfaction with hospital for delivery, with a total of eleven items, was part of Block III (Satisfaction with Hospital Care) of the Birth in Brazil survey.It is an instrument that mainly assesses the work of health professionals during childbirth hospitalization 28 .
The first seven questions of this scale were extracted from the World Health Survey and adapted to childbirth hospitalization (items 1 to 7) 33 .As satisfaction with hospital care during delivery involves aspects not covered in these seven World Health Survey questions, one question on verbal, psychological, and physical violence practiced by care professionals (item 8), and three questions on general satisfaction with delivery, postpartum, and neonatal (items 9 to 11) were added.Psychometric analyses of the scale showed that it was a unidimensional construct of ten items with 0.91 composite reliability 28 .

Theoretical model and variable
The theoretical model (Figure 1) was presented as a directed acyclic graph (DAG).The exposure variable mode of delivery was dichotomized into vaginal delivery and cesarean section.The outcome satisfaction with childbirth hospitalization was tested as a 10-item unidimensional latent variable.Mode of delivery and its predictor variables were collected in the after birth and the outcome satisfaction in the first stage of the follow-up.
The predictive variables of the mode of delivery were the following: Brazilian Economic Classification Criteria 32 , categorized as D/E, C and A/B, according to possession of goods and the education level of the head of the family, with categories A and B having greater purchasing power; maternal schooling, measured in years of study (0-4, 5-8, 9-11 or ≥ 12 years); marital status, categorized as: single, consensual union, and married; maternal age at birth, with categories < 20 years old, 20-34 years old, and ≥ 35 years old; type of hospital according to funding, with the Brazilian Unified National Health System (SUS) and private categories; previous cesarean section, with the primiparous categories, with and without cesarean section; preference for the mode of delivery, categorized in "had not decided", "normal delivery", and "cesarean section"; and, also, categorized as "no" and "yes"; "pregnancy and/or delivery complications" 19 ; "private health insurance coverage of delivery"; "link to the hospital"; and "same professional at prenatal and childbirth".

Figure 1
Directed acyclic graph for estimating the causal effect of cesarean section on women's satisfaction with childbirth hospital admission.Brazil, 2011-2013.
tion, hepatitis B and C, complications that influence the negative outcome of the newborn, complications that influence the mode of delivery, and previous uterine surgery and seizure.

Statistical analysis
Counterfactual approach 34 was used to estimate the effect of mode of delivery on maternal satisfaction with childbirth hospitalization.Initially, the propensity score was estimated with logistic regression 35 in the birth database.Via the back door criterion, a method proposed by Pearl 36 and Pearl et al. 37 , the minimum set of adjustment for confounding was selected with the help of the public domain program DAGitty (http://www.dagitty.net/) 38.
The balance of pre-exposure variables was verified via standardized absolute differences in means and variance ratios between vaginal childbirth or cesarean section groups using the tebalance command after the teffects ipw routine command in Stata, version 14 (https://www.stata.com) 39.Balance was reached as the standardized absolute differences in the means ranged from -0.10 to 0.10 and the variance ratios from 0.8 to 1.2 40,41 .Observations were weighted as follows: for the group born by cesarean section, weight was the inverse of the probability of cesarean section and it was the inverse of the probability of vaginal delivery for the group born by vaginal delivery, i.e., the inverse of one minus the probability of being born by cesarean section.
Cad. Saúde Pública 2023; 39(3):e00138922 The effect of the mode of delivery on satisfaction was estimated in a structural equation model with mode of delivery as exposure and the satisfaction construct as response variable, weighted by the inverse of the selection probability, considering the complex sampling design.The weight was estimated considering the different probabilities of sample selection, the losses to follow-up, and the weight obtained from the propensity score 42 .Standardization was performed only for the response variable.Comparisons were made between weighted percentages considering only the complex sampling design to those also considering losses to follow-up to assess to what extent inverse probability weighting was able to reduce selection bias.
Mplus version 8.5 software program (https://www.statmodel.com/) was used and, as the variables were ordinal categorical, the weighted least square mean and variance adjusted (WLSMV) 43 was used.
The model fit was tested using the following indicators: (a) p-value < 0.05 and upper limit of the confidence interval < 0.08 for the root mean square error of approximation (RMSEA) 44 index; (b) values > 0.95 for the comparative fit index (CFI) and Tucker Lewis index (TLI); and (c) standardized root mean square residual (SRMR) < 0.05 42,45 .

Ethical aspects
The project complied with the principles of Resolution n. 196/1996, which deals with research involving human beings, and the Brazilian National Health Council and its complementary rules.It was approved by the Ethics Research Committee of the Sergio Arouca National School of Public Health, Oswaldo Cruz Foundation (ENSP/Fiocruz, CAAE 0096.0.031.000-10).Those responsible in each institution and all postpartum women signed an informed consent form at the face-to-face interview.

Results
For the purposes of this study, from the total of 24,200 deliveries in the hospital sample, 488 multiple births, 64 neonatal deaths, as well as missing data for mode of delivery (n = 304), maternal education (n = 104), marital status (n = 8), economic class (n = 154), maternal age at delivery (n = 4), and the same professional attending prenatal care and childbirth (n = 28) were excluded, totaling 23,046 cases in the first stage.In the first follow-up, after losses and exclusions, 15,582 complete cases remained.
In the follow-up, there was a lower proportion of mothers from classes D/E, < 9 years of maternal schooling, single, younger than 20 years, users of SUS, without previous cesarean section, with preference for normal delivery, without pregnancy and labor complications, without health insurance, without link to the hospital, and not attended by the same professional at prenatal and childbirth.Inverse probability weighting considering losses to follow-up reduced selection bias -weighted percentages that considered losses to follow-up were closer to those at birth than to those at follow-up uncorrected (Table 1).
More than half of the interviewees in the final sample of this study (52.2%) underwent cesarean section (result not shown).Almost all predictor variables tested were independently associated with cesarean section, except for the Brazilian economic classification, having health insurance at delivery, and link to the hospital.The predictors strongly associated with cesarean section were same professional at prenatal and childbirth (OR = 5.53; 95%CI: 4.34-7.05),pregnancy and/or labor complications (OR = 4.58; 95%CI: 4.04-5.19),childbirth hospitalization in a private hospital (OR = 4.40; 95%CI: 2.48-7.79),and previous cesarean section (OR = 3.65; 95%CI: 3.06-4.34)(Table 2).
The balance was obtained for all variables, with the exception of preference for the mode of delivery and pregnancy and/or labor complications; as for these variables, the standardized absolute difference between the means of both groups was either slightly below or slightly above -0.10,more stringent cutoff, but still ranging from -0.20 to 0.20, i.e., a less stringent but still acceptable cutoff (Table 3).

Table 1
Comparisons between weighted percentages of the sample at birth and the sample at the follow-up after delivery.Birth in Brazil, 2011-2013.* Due to the huge sample size all differences were statistically significant; ** Weighted percentage considered the complex sampling design; *** Weighted percentage considered the complex sampling design, losses to follow-up and the weight obtained from the propensity score.

Discussion
The results of this study showed that users of Brazilian hospital services were equally satisfied with hospital admission for both vaginal and cesarean delivery.The limitations of this study are related to selection and memory biases, losses to follow-up, and characteristics of the instrument that measured satisfaction.Regarding memory bias, the authors consider that the data collection time from 45 days to less than six months after delivery was adequate to measure maternal satisfaction with hospital admission for delivery.This is because a systematic review on instruments to measure satisfaction with care during labor and childbirth suggested that  interviews should be conducted after hospital discharge so that problems that occurred during hospitalization were not overshadowed by the birth of a healthy baby 24 .Concerning the selection bias, the weighting technique by the inverse of the selection probability was performed.The final weight was estimated considering the different probabilities of sample selection, the losses to follow-up, and the weight obtained from the propensity score.The joint use of these techniques tends to reduce the likelihood of selection bias and confounding that arise due to the variables used to derive the weights.However, selection bias and confounding due to other variables not included in the weight calculation may still be present.
The scale used in this study to measure maternal satisfaction with hospital admission for delivery showed good evidence of validity and excellent psychometric properties 28 .As for the instrument to assess the work of health professionals in delivery care, a systematic review drew attention to the fact that the interpersonal relationship between the user and the health team was the main determinant of satisfaction with childbirth hospitalization 23 .
This study advances compared to other studies due to the set of methodological strategies used: (a) a prospective cohort study with random and stratified sampling at three levels was conducted: macroregion, type of municipality (capital or not capital), type of hospital administration, which brought together women from different socioeconomic and cultural conditions in a large country such as Brazil; and (b) a set of methodological techniques was used to better study causality relationship: Cad. Saúde Pública 2023; 39(3):e00138922 directed acyclic graphs, counterfactual approach, maternal satisfaction with childbirth hospitalization outcome estimated as a latent variable, and estimation of the causal effect in a structural equation model with inverse weighting of selection probability, with weights derived from the propensity score, follow-up losses, and considering the complex sampling design.
No difference in satisfaction with hospital admission according to mode of delivery can be explained by the women's final decision between the vaginal childbirth and cesarean section since, at the end of pregnancy, they gave birth as planned.This finding is in line with analyses of qualitative, quantitative, and systematic review studies that found similar satisfaction when the mode of delivery planned was performed 21,22,23,24,25,46 .Our results are in agreement with those reported in a cross-sectional study with 355 Swedish postpartum women who were interviewed before hospital discharge, in which no differences were observed in maternal satisfaction according to vaginal and cesarean mode of delivery 47 .In another study, which recruited 335 German women who had fullterm newborns without congenital malformations, no differences were found in the satisfaction of women who had vaginal childbirth or cesarean section, including emergency cesarean section and surgical vaginal delivery.The authors concluded that the mode of delivery did not directly influence women's satisfaction with delivery.The factors that improved woman's experience with birth the most were decision-making power, support received, and effective analgesia 27 .
However, in a study of 204 U.S. primiparous women who planned and managed to have a cesarean section (n = 44) or vaginal childbirth (n = 160), maternal satisfaction was higher for those who planned to have a cesarean 48 .Conversely, another study evaluating 894 women shortly after delivery in Ethiopia, greater satisfaction was observed after vaginal delivery 49 .Higher satisfaction scores were also reported after vaginal and cesarean section performed before labor in another U.S. study, whereas a greater dissatisfaction was observed in cesarean births performed after the onset of labor 50 .In another study carried out in Sweden, a greater dissatisfaction was detected when the delivery was prolonged or performed by cesarean section 51 .
Another study from the Birth in Brazil survey also found no differences in overall maternal satisfaction according to mode of delivery in the adjusted analysis.However, in that survey, the outcome general satisfaction with delivery was investigated using the question: "In your opinion, was your delivery care...", with an instrument measuring maternal satisfaction with hospital admission for delivery, via a dichotomous variable (no and yes) 30 .In contrast, in our study, a more comprehensive measure of satisfaction with childbirth hospitalization was measured via a latent variable, with no measurement error.
In conclusion, no difference was observed in maternal satisfaction with delivery when comparing women who had vaginal childbirth or cesarean section after adjusting for confounding.However, this equal satisfaction among women who gave birth by vaginal delivery in the Birth in Brazil study cohort does not indicate that Brazilian obstetric services always occur with quality and in a humanized way.We must recognize that advances and improvements have been observed in recent years both in the public and private sectors 52 ; however, inequalities persist, the adoption of good practices is not as frequent as ideal, and humanized care during childbirth is still an objective to be achieved 53

Figure 2
Figure 2Structural equation model adjusted by the inverse probability of selection to estimate the effect of cesarean section on satisfaction with childbirth hospital admission.Brazil, 2011-2013.

Table 3
Standardized absolute differences in means and variance ratios between vaginal and cesarean delivery groups.Birth inBrazil, 2011-2013.