Study design and sample
This is a quantitative, exploratory, cross-sectional study approved by the Research Ethics Committee of Universidade Federal dos Vales do Jequitinhonha e Mucuri-UFVJM (Protocol: 2.773.418), with written informed head parent consent and participant assent and all protocols are carried out in accordance with relevant guidelines and regulations. Data collection took place from July to December 2019. Pre-school children, that is, children from 3 to 5 years old, from public schools in a Brazilian municipality, were eligible. The sample size was calculated using the GPOWER 3.1 statistical program. For this, we used linear multiple regression considering a partial determination coefficient of 0.50 for body fat mass as outcome. Thus, alpha error of 0.01, statistical power of 99%, considering 20% of possible sample losses, the sample size was estimated in 51 preschoolers.
Exclusion criteria were preterm and low birth weight infants; infants with pregnancy and delivery complications; infants with signs of malnutrition or illness that interfere with growth and development.
Instruments and procedures
Body fat mass was quantified using Dual Energy Radiological Absortometry (DEXA) (Paediatric medium scan mode software, Lunar Radiation Corporation, Madison, Wisconsin, USA, modelo DPX), known as a reliable quantification tool1.
The children were invited to the DEXA evaluation and, to encourage adherence, a video of another child performing the scan was made available before the measurement. The instrument was properly calibrated and the scans were analyzed by a trained technician. The body composition variable chosen for this study included a measure of total adiposity, that is, FM. To measure the weight, an analog scale (0.1 kg precision) was used. To measure the height, a portable, folding infant stadiometer was used. The children were instructed to remove their shoes and these measurements were performed by a properly trained examiner.
The sample was characterized according to BMI, as well as the z score, using WHO Anthro software version 3.2.2 (Geneva, Switzerland), developed by WHO38. Thus children with z-scores between -1 and +1 were classified as normal weight, > +1 as overweight; > +2 as obese There was a high correlation between BMI and FM (Spearman's correlation, r = 0.898, p < 0.001).
As possible independent outcomes, birth weight, presence of obese parents, birth order, sex, age, marital status, economic status, maternal education39, quality of the home and school environment36, PA15, caloric intake 19 were considered.
The biological and sociodemographic factors were collected using an specific questionnaire, containing information about the history of pregnancy, data on the vaccination card, such as weight and height at birth, presence or absence of siblings, self-report of maternal and / or paternal obesity. In addition, information about environmental opportunities for active and sedentary behavior, such as the time the child is exposed to screens, the presence of internal and external physical space in the house, the presence of a playground at school, and other outcomes were collected. The outcome ‘time of exposure to screens’ was collected considering the parents' report of the time in minutes that the child is exposed to the screens (television and cell phone).
Sociodemographic variables were collected using a specific questionnaire. To verify the economic level of families, the Brazil economic classification criterion, from the Brazilian Association of Research Companies was used. This is a questionnaire that stratifies the general economic classification resulting from this criterion from A1 (high economic class) to E (very low economic class)40, considering the assets owned by the family, the boss's education and housing conditions, such as running water and street paving.
The quality of the environment in which the child lives was assessed using the Early Childhood Home Observation for Measurement of the Environment (EC_HOME) 41. The EC_HOME is applied through observation and semi-structured interviews during home visits, standardized for children aged 3 to 5 years. The instrument contains 55 items divided into 8 scales: I-Learning materials, II-Language stimulation, III-Physical environment, IV-Responsiveness, V-Academic stimulation, VI-Modeling, VII-Variety, and VII-Acceptance For analysis, the sum of the raw scores of the subscales was used, after the environment is classified as High stimulation, Medium stimulation and risk environment.
The quality of the school environment was assessed using the Early Childhood Environment Rating Scales (ECERS) 42, which contain inclusive and culturally sensitive indicators for many items. The scale consists of 43 items organized into 7 subscales (1-Space and Furnishings, 2-Personal Care Routines, 3-Language and Literacy, 4-Learning activities, 5-Interactions, 6-Program Structure, 7- Parents and staff). Each quality indicator was marked, considering its presence or absence in each collective environment (classroom), with the items scored from 1 to 7. The final score of the scale is given by the mean of the seven subscales. It is an ordinal, increasing scale, from 1 to 7, the interpretation of quality being 1: inadequate; 3: minimal (basic); 5: good; 7: excellent.
The PA level was measured using an accelerometer (Actigraph®- Model GT9X); for a period of 3 days, without including the weekend43, for a minimum of 570 minutes a day15, which is considered suitable for preschoolers43. Accelerometers were initialized and analyzed using 5-second epochs. In all analyses, consecutive periods of ≥ 20 minutes of zero counts were defined as non-wear time44, with a sampling rate of 60 Hz. The acceleration units were expressed in triaxial vector magnitude (VM). The accelerometer was positioned on the right side of the hip to capture accelerations and decelerations of the body and determine objective measurements of gross acceleration, intensity of physical activity, heart rate intervals and total time of suspension of use44. Pediatric cutoff points validated for preschool children, with score values, classify as sedentary (0 to 819 counts / m), mild (820 to 3907), moderate (3908 to 6111) and vigorous (above 6612)45. For this study, the child's mean time at these intensities was used. The classification adopted for “active” or “insufficiently active” was established according to the WHO, which considers an active child to be one who has a PA of at least 180 minutes/day, with a minimum of 60 minutes/day in moderate to vigorous PA46.
For the assessment of food intake, the food diary was used to collect information about an individual's current intake. In this method, the responsible person writes down, in a specific form, all the food and drinks consumed over one or more days, and must also note the food consumed outside the home47. For helping the portion size the best estimates, of the portion size, we used counted on the help of traditionally used homemade measures, containing portion sizes and three-dimensional models of food48. The Average daily total energy values (Kcal) were calculated using the DietPro 5i software (A.S. Sistemas, Viçosa, Minas Gerais, Brazil). The first stage was carried out at the child's home with the completion of the survey questionnaires to assess socioeconomic data 40, quality of the home environment (EC-HOME)41, data on opportunities environmental aspects of active and sedentary behavior, clinical history of pregnancy, childbirth, child and parents, anthropometric assessment, in addition to guidance on the instrument (accelerometer) that the child used to measure the level of physical activity. The second stage was carried out in the school environment, where the daycare environment assessment (ECERS)42 was applied. In the third stage, the parents and the child were referred to the Exercise Physiology Laboratory (LAFIEX), on Campus 2 (UFVJM) for DEXA. All children were evaluated in the same places.
The researchers first went through training to apply the tests and measures to carry out the measures of weight, height, application of tests to assess body mass, as well as to apply the questionnaires. To ensure greater reliability, only 1 examiner per test and step was used, ensuring internal control for the measurements of the outcomes in a sequential study.
Data analysis
Statistical analysis was performed using SPSS 24.0. First, a descriptive analysis of the outcomes was performed to determine the data distribution. The Shapiro-Wilk test was used to determine the normality of the data. Afterwards, Spearman's or chi-square correlation was used. Simple linear regression analyzes were performed to determine the strength of the associations between the variables (child's age, sex, maternal age and education, marital status, weight acquired during pregnancy, birth weight, economic status, presence of siblings , father practices PA, son practices PA, presence of siblings, breastfeeding time, obese father or mother, ECERS score, EC_HOME classification, screen time for the week and weekend, sedentary PA, mild to vigorous PA, Classification PA in active or little active, measured calories) and the outcome (FM). All possible explanatory outcomes were inserted into the multiple linear regression model. The stepwise method was used to determine which variables remained associated with FM, with only explanatory variables with a p-value <0.05 remaining in the final model after adjustments. Given that outcomes that had no significance in simple linear regression can become significant in multiple linear regression when associating with other outcomes, as they can be considered confounding outcomes, multiple linear regression was performed including all the outcomes analyzed in simple regression.