This study is an under sample of the larger study “Birth in Brazil”, a national hospital-based research with countrywide representation consisting of 23,894 mothers and their newborns, conducted between February 2011 and October 2012 in Brazil. All the authors were part of the team that conducted this survey and research. The sampling was carried out considering three stages of selection: all hospitals which had 500 or more births per year in 2007 were selected, classified according to Brazil’s five macro-regions (North, Northeast, Southeast, South and Mid-west), municipality (capital or interior), and type of hospital (private, public and mixed). Subsequently the number of days needed to reach the fixed sample of 90 women who had recently given birth in each hospital was calculated. Finally, these 90 women were selected from each hospital remaining in the sample. A total of 1356 (5.7 %) postnatal women selected were replaced, 203 owing to early hospital discharge and 1153 owing to refusal to participate. A detailed description of the “Birth in Brazil” methodology is given elsewhere [12].
Sample subjects
To assess the outcome of interest (i.e., assessment by women of the care received during labor and birth), only postpartum women classified as low risk during pregnancy who had experienced either spontaneous or induced labor and whose birth had occurred in the Southeast region of Brazil were included. This geographical delimitation has been chosen because the Southeast has the highest prevalence of adoption of the good practices in care during normal labor and birth recommended by WHO [13]. Women were defined as low-risk according to the following criteria used by Dahlen et al. [14]: absence of pre-existing or pregnancy-related hypertension or diabetes; body mass index <30 (above which the person is considered obese); HIV negative; gestational age between 37 and 41 weeks; singleton pregnancy with cephalic presentation and birth weight between 2500 and 4499 g (between the 5th and 95th centiles of birth weight for gestational age). This resulted in a sample of 4102 mothers, representing 64 % of the total sample in the region [15].
Data collection
A structured electronic questionnaire was administered face-to-face to women within the first 24 h after birth in the maternity ward querying their sociodemographic characteristics, obstetric history, prenatal care, and data related to labor and birth. In addition, medical record data of the mother and newborn were collected, and a photocopy made of the women’s prenatal care cards. Electronic forms were developed and validated to collect data and all interviews were conducted by interviewers previously trained by the investigation coordinators. Field research supervisors reapplied the questionnaire to a random sample of 5 % in the interviews with the women. Manuals were prepared with descriptions of procedures for data collection in order to ensure the quality of data and thereby minimize systematic or random errors.
Two telephone contacts were made with the mothers on average 45 days and 6 months after birth respectively, and structured questionnaires were applied at these moments. At the first telephone contact the women were asked about the presence of some WHO’s good practices recommended in care during normal labor and birth. At the second telephone contact they were asked about their assessment of the care received during labor and birth.
As it was not possible to contact all the women during the follow-up (68 % response rate in the first interview and 49.4 % in the second), a statistical model was adjusted to estimate the probability that each woman who took part at baseline would answer the telephone questionnaire, using a set of variables which differentiated the groups of respondents and non-respondents. Non-response adjustment factors attempt to compensate for the tendency of women to have certain characteristics (such as being unmarried or of lower education background) to respond at lower rates. On the basis of this model, specific sample weights were calculated for the analysis of the telephone interviews. The rationale for applying non-response weights is the assumption that non-respondents would have provided similar answers, on average, to respondents’ answers. More information about the sample design, data collection, and processing of lost segments is described elsewhere [12].
Study variables
The dependent variable of this study was the assessment by the women about the care received during labor and birth measured in the second telephone interview when they were asked: “In your opinion, how was the care that you received during labor and birth?” The answers were: 1) Excellent, 2) Good, 3) Regular 4) Poor, and 5) Very poor. Because of low frequencies for the categories poor and very poor, these were grouped into a single category called “Poor”.
The independent variables analysed were the good practices in care during normal labor and birth recommended by WHO (category A). They were obtained from the questionnaire administered to postpartum women in the hospital, medical record data, and the first telephone interview. Using these instruments, only some good practices could be analyzed: respecting the right of women to privacy in the birthing place, empathic support from caregivers during labor and birth, respecting women’s choice of companions during labor and birth, presence of companion throughout labor and birth, giving women as much information and explanation as they desired (time to ask questions and receive information), clarity of the information and explanation received, offering oral fluids and food during labor and birth (free nutrition), nonpharmacological pain relief during labor, freedom of position and movement throughout labor, early skin-to-skin contact between mother and child, and support for the initiation of breastfeeding in the birthing place [16].
The control variables used were parity (primiparous or multiparous), type of birth (vaginal, vaginal with use of forceps or vacuum extractor, and cesarean section), type of payment (public or private with payment by the patient or by health insurance), educational level (0–7, 8–10, 11–14, and 15 or more years) and economic level. According to the Brazilian Association of Research Companies (ABEP), the definition of economic level used in this study was based on the ownership of assets and education level of the head of household [15]. The categories of economic level were divided into five groups, ranging from A (highest) to E (lowest). Because of the low proportion of women in classes A and E, the categories were regrouped into three levels: A and B (high), C (mid level), D and E (low).
Data analysis
For this study, exploratory and descriptive data analysis were conducted first. After this, bivariate and multivariate analyses using the generalized linear modeling technique of multinomial logistic regression were conducted and Odds ratios (OR), crude and adjusted for potential confounding variables, and 95 % confidence intervals (CI) were obtained. These measures were used to assess the associations between the dependent and independent variables. For data analysis, R version 3.0 software (The R Foundation, Vienna, Austria) and IBM SPSS version 19.0 (IBM Corp., Armonk, NY, USA) were used.