Design, study location and period
This is a retrospective cohort, developed with data from the survey “Born in Belo Horizonte: Survey on childbirth and birth”, carried out in 11 maternity hospitals in Belo Horizonte, Minas Gerais(MG), 7 with public care and 4 with private care [18].
The study “Born in Belo Horizonte: Survey on childbirth and birth” adopted the same criteria as the “National Survey: Survey on Childbirth and Birth in Brazil. The research "Born in Brazil: Survey on childbirth and birth" aimed to describe the incidence of caesarean section and examine the consequences on the health of women and new-borns (NB), investigate the relationship between excess caesarean sections, premature delivery and low birth weight and verify the relationship between excess caesarean sections and the use of technical procedures after birth [18].
This study included all women admitted to the maternity hospitals selected at the time of delivery, who had a single pregnancy, adults, residents of Belo Horizonte or Contagem (MG) and who had data on weight and height (necessary for calculating pre-pregnancy BMI). The final sample consisted of 506 mothers.
Study protocol
The information came from face-to-face interviews, performed by trained nurses, at least 6 h and maximal 24 h after delivery [18], from November 2011 to March 2013. Data from maternal records were also used. More information on the sample design is detailed in other publications [19, 20].
It is noteworthy that the difference in temporality between the years of data collection and the analysis of this proposal will not compromise the results, since it is believed that there was no temporal dissociation in the environmental variables (analyzed in relation to the temporal relationship of the time) and in the buffers design during the study period.
Outcome variable
The outcome variable of this study was the GWG calculated through the difference between the pre-pregnancy weight and the pre-delivery weight or that registered in the last prenatal care (PN) consultation.
Weight gain was classified based on the IOM's recommendations [2]. According to the pre-pregnancy BMI, pregnant women who had weight gain in the recommended interval were categorized as “adequate GWG”, those who presented weight gain below that recommended for pre-pregnancy were classified as “insufficient GWG” and those who had a gain above the recommended, considering the pre-pregnancy BMI, were categorized as “excessive GWG”.
Individual explanatory variables
The explanatory variables included in the study were socioeconomic and demographic, obstetric factors and related to childbirth factors (age, ethnicity/race[21], education, marital status, parity, gestational age, pre-pregnancy BMI, smoking in the first 5 months of pregnancy, consumption of alcoholic beverages during pregnancy, number of prenatal consultations, location where most of the prenatal and professional consultations were performed, which attended most of the prenatal consultations). GDM and non-gestational diabetes mellitus, arterial hypertension developed during pregnancy were included as adjustment variables. These variables were chosen as adjustment variables based on previous studies [3, 4].
Characterization of geographical data
From the home address provided by the participants at the time of data collection, a geographical coordinate (latitude and longitude) was assigned to each study participant. Based on the address and Postal Address Code (CEP) of the location, a geocoded database was developed, obtained through several commercial and government sources to assess the characteristics of the environment built in the buffers. In this way, it was possible to perform the georeferencing of participants and establishments selling the environment built in the Belo Horizonte and Contagem space, as well as the categorization regarding the proximity of their homes to points of sale and food and local stores for the practice of PA. For the process of geocoding the addresses, the GGMAP package was used in R, version 3.4.3.
The union of environmental data, including food outlets and PA practice with individuals, located through their homes and the definition of the buffer, took place through the QGIS program, version 2.18.14.
Contextual explanatory variables
The geocoding of the addresses of the environmental variables was performed with the GGMAP package in R, version 3.4.3. In this process, the geographical coordinates (latitude and longitude) of the points of sale of food, places where there are PA practices and the individuals' residence were located on a map.
To characterize the exposure of pregnant women to a specific physical and social environment, the concept of neighborhood was used through the delimitation of a buffer with a radius of 500 m. around the residence, used as a centroid. This radius was established based on the fact that the walking time can vary from 10 to 20 min [22].
To characterize the context of the neighborhood, information was obtained on points of sale of food registered according to the National Classification of Economic Activities (CNAE), a standard council that assigns codes of economic activity and defines criteria used by tax authorities in Brazil and the municipalities studied.
The establishments were classified into three categories: 1. Establishments with a predominance of healthy food offerings: where the acquisition of fresh or minimally processed foods represents more than 50% of the total acquisition; 2. Establishments with a predominance of unhealthy food supply: where the purchase of ultra-processed foods represents more than 50% of the total purchase and 3. Mixed food purchasing establishments: where there is a predominance of purchase of culinary preparations or processed foods or where there is no predominance of purchase of fresh / minimally processed foods or ultra-processed foods [23].
The food environment in this study was assessed using the number of healthy, unhealthy and mixed establishments available in the neighborhood environment assigned to each participant. It is representative of the timeframe that participants were pregnant (2011, 2012 and 2013). The establishments around the buffer were classified as mixed establishments: hypermarkets, restaurants, bakeries, dairy retailers, retailers of food products in general, supply of prepared food for home consumption, supermarkets, grocery stores, canteens and mobile food services [23].
The places used for the practice of PA were classified as public and private, such as: squares, gyms, bike paths and other places for this purpose, the data were obtained through government sources.
To characterize the social environment, the nominal income of the census sectors was calculated, which was divided by the number of people residing in the census sectors that made up the buffer, and this value was assigned to each study participant. This variable was categorized into terciles. Neighborhood income and population data were obtained from the demographic database of the Brazilian Institute of Geography and Statistics (IBGE) 2010, Belo Horizonte and Contagem, MG, Brazil.
Statistical analysis
For the descriptive analysis of the sample, the estimates were presented in proportions (%), with a 95% confidence interval (95% CI). For the quantitative variables, the asymmetry was verified by the Shapiro–Wilk test and the data were presented by means of median and interquartile range (IQ). The correlation between the variables of the buffer food environment was verified using the Spearman correlation test.
To assess the association between independent variables and insufficient GWG or excessive GWG, logistic regression was performed using Generalized Equation Estimation (GEE), and the pregnant women neighborhood was adopted as the aggregation unit, in order not to hurt the assumption of data independence, due to the possibility of sharing the context variables.
For the construction of the logistic regression model with the individual variables, the p ≤ 0.20 value obtained in the bivariate analysis was used as a criterion for the inclusion of the variables, in addition to theoretical criteria. Subsequently, the environmental variables elucidated by the literature associated with insufficient GWG or excessive GWG were included. For all analyzes, a significance level of 5% was considered. As an association measure, Odds Ratio (OR) and 95% CI were used. For data analysis, the statistical package Statistical Software for Professional (Stata), version 14.0 was used.
It is noteworthy that due to the high correlation between the variables of the community food environment (healthy, unhealthy and mixed establishments available around the buffer), individual models were built for each of them, thus avoiding collinearity in the models [24]. It is also worth noting that in the models built using Logistic Regression analysis with Generalized Equation Estimation (GEE), the insufficient GWG showed no statistical significance in the multivariate analysis.
The project “Born in Belo Horizonte: Survey on childbirth and birth” was approved by the Research Ethics Committee of the Federal University of Minas Gerais (UFMG), under Opinion CAAE-0246.0.203.000. All puerperal women and directors of each maternity hospital signed the Free and Informed Consent Term, according to the ethical guidelines described in Resolution No. 466, of December 12, 2012, of the National Health Council, which involve research with human beings. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.