This study is a secondary analysis of data collected by the Healthy Birth Study, a cross-sectional evaluative investigation carried out in 2017, 18 months after the beginning of the implementation of the Adequate Childbirth Project (ACP).
The Healthy Birth Study selected a convenience sample of 12 hospitals from the 23 private hospitals that joined the project to improve prenatal and childbirth care. For the selection of these hospitals, three criteria were considered: location of the hospital by geographic macro-region (at least one hospital from the northeast, southeast and south regions), type of hospital (owned or not by a health insurance company) and performance of the hospital in the prenatal and childbirth care improvement project (hospitals that reported good and bad results in achieving the ACP goals, according to administrative data provided by the project coordination board, were selected) [18]. Among the hospitals participating in the study, one was situated in the northeast region, nine in the southeast region and two in the south of Brazil.
The Healthy Birth Study’s sample was calculated to detect a 10% reduction in the proportion of cesarean sections, using a 50% cesarean rate as a reference, with 80% power and a 5% significance level. Overall, the sample size of 4800 women (12 hospitals times 400 women) had an accuracy of 80% in detecting a 2.5% reduction in the prevalence of cesarean sections. Losses and refusals accounted for about 5% of women and were replenished to complete 400 women in each hospital.
Electronic forms using the REDCap [19] application were developed. A pilot study was carried out in one of the maternity hospitals participating in the prenatal and childbirth care improvement project not included in this evaluative research. The questionnaires were tested and the logistical aspects of the fieldwork were refined.
Trained interviewers, external from the hospitals, conducted data collection, addressing all women admitted to the hospital who met the eligibility criteria to participate in the study, until 400 participants were included in each hospital. Women who did not speak Portuguese, with hearing loss, whose delivery occurred outside the hospital and who were hospitalized for judicial termination of pregnancy were not eligible for the study. The women were interviewed face to face at least 6 h after vaginal delivery and 12 h after cesarean section, after reading and signing an informed consent form.
Three instruments were used: 1. questionnaire applied to the puerperal woman with questions related to maternal characteristics, pregnancy and prenatal care, childbirth care, newborn and infant feeding; 2. form for extracting data from the medical record of the puerperal woman and the newborn about the type of pregnancy, type of delivery, condition of the newborn and feeding of the newborn during hospitalization; 3. form for extracting data from the prenatal card [18].
In prolonged hospitalizations, data were collected from medical records until the 28th day of infant hospitalization and on the woman's 42nd day of hospitalization. In the case of hospital transfer, data were obtained from the medical records of the hospital from which the puerperal woman and/or the newborn were discharged. Data collection was conducted from March to August 2017. Due to the variation in the size of hospitals, the time required for data collection ranged from 1 to 4 months, depending on the total number of births per month in each participating hospital [18].
The present study used as inclusion criteria living newborns with gestational age ≥ 34 weeks (stillborns: n = 19). Newborns or women who met one or more of the following criteria were excluded: newborns with malformations that could disturb or prevent breastfeeding (malformations: n = 47), sons of HIV-infected mothers (n = 6), postpartum women with severe maternal morbidity or transferred to the Intensive Care Unit (ICU) due to complications during delivery (n = 244) and babies transferred to the ICU at birth (n = 527).
The excluded malformations were severe congenital heart disease, Down syndrome, craniofrontonasal syndrome, hydrocephalus, cleft lip or soft palate cleft, pulmonary cystic adenomatosis, corpus callosum dysgenesis, gastroschisis, diaphragmatic hernia, vermin hipoplasia, omphalocele and Moebius and Robin sequence.
A total of 785 women were excluded, and data from 4,093 binomials were analyzed. Post-hoc calculations showed that, considering a prevalence of exclusive breastfeeding of 60% and a significance level of 5%, the sample after the exclusions used in the present analysis had 90% power to detect 5% differences in the proportion of this outcome.
The outcome was breastfeeding in the first hour of life (yes/no), obtained from two questions to the puerperal woman: "After birth, did you breastfeed in the delivery room?" and "How long, more or less, did it take you to breastfeed for the first time?" Breastfeeding in the first hour of life was considered when the answer to the first question was “yes” and/or the second covered up to 1 h. A hierarchical theoretical model was adopted, with maternal and household characteristics as distal exposure variables; pregnancy and prenatal care characteristics as intermediary exposure variables and hospital, delivery, and newborn characteristics as proximal exposure variables (Fig. 1).
The selected maternal characteristics were age (< 20; 20–34; ≥ 35), self-reported skin color (white; black; brown/yellow/indigenous), having a partner (single/separated/widowed; married/living with a partner), education (up to 14 years; ≥ 15 years). Household characteristics were analyzed based on economic conditions (class A—above average; B—average; C + D + E—below average) [20]. Characteristics of pregnancy and prenatal care were parity (primiparous; multiparous), being target of the Adequate Childbirth Project (yes; no), maternal disease—hypertension, diabetes, or other chronic diseases (yes; no), number of prenatal consultations (≤ 7; 8 or more), beginning of prenatal care (up to 12 weeks; ≥ 13 weeks), guidance on breastfeeding during prenatal care (yes; no). The characteristics of childbirth care selected were presence of a companion during delivery (yes; no/did not want), type of delivery (vaginal; cesarean section), skin-to-skin contact at birth (yes; no) and rooming-in (measured by the question: “did the baby go to the room with the mother?”: yes; no). Features of the newborn selected were birth weight (< 2500 g; ≥ 2500 g), gestational age (34–36 weeks; ≥ 37 weeks) and sex (male; female).
Different groups of women were target of the Adequate Childbirth Project, as defined by the hospital manager, such as: 1. primiparous women; 2. pregnant women belonging to Robson Groups 1 to 4 (nulliparous or multiparous without previous cesarean section, single fetus, term, cephalic, according to WHO [20]) because these women had greater chances of vaginal delivery; 3. women served by the on-call team; 4. pregnant women without previous uterine scarring whose delivery was performed by the on-call team [18].
All statistical analyses were performed using SPSS software version 17 (https://www.Ibm.com/), using data weighting and incorporating the design effect, considering the complex sampling process [18]. Considering the sample weight, the final number analyzed was 4060 binomials.
Initially, univariate analysis was conducted to identify the distribution of exposure variables and the outcome under investigation. Then, bivariate analysis was applied between each exposure variable and the outcome, using Pearson's chi-square test and crude odds ratios (OR) were obtained with their respective 95% confidence intervals (95% CI). The variables that obtained p ≤ 0.20 in the bivariate analysis were included in the statistical modeling. Finally, multiple logistic regression was conducted, following the hierarchical conceptual theoretical model applied, according to the temporal proximity of each variable with the outcome [21].