Research setting and study design
During 2004, all the 4,261 live births to mothers living in the urban area of the city of Pelotas (Brazil) were eligible to The Pelotas 2004 Birth Cohort Study. Of those, 4,231 (99.3%) mothers consented to their children to being enrolled in the study. Trained fieldworkers in the five local maternity hospitals interviewed all mothers within the first 24 hours after the delivery. Detailed information about socioeconomic, prenatal care, behavior, morbidity, and other maternal and newborn characteristics were gathered. Children in the cohort were followed-up at mean (standard deviation) ages of 3.0 (0.1), 11.9 (0.2), 23.9 (0.4), 49.5 (1.7), and 82.2 (4.0) months and at 6.7 (0.2) and 10.3 (0.5) years of life. A detailed description of the methodology of the cohort was given in previous publications. (12, 13)
Data from the hospital interview and the 6- and 11-year follow-ups were used in this study. A total of 95 children died from birth to 6 years and three more between 6 and 11 years old. Response rates at 6- and 11-year follow-ups were 90.2% and 86.6%, respectively. The current analyses included 3,182 children (75.2% of the original cohort) with full information on BP measures at 6 and 11 years. The follow-up visits were held at a clinic installed at the headquarters of the Federal University of Pelotas Epidemiologic Research Center. Several child characteristics were investigated, including detailed anthropometric and health conditions evaluation. (13) Trained interviewers applied standard and pre-codified questionnaires in the hospital and follow-up interviews.
A digital automatic OMRON sphygmomanometer© (model HEM 742) was used to measure BP in both follow-ups. BP was measured twice (2 minutes apart), on the children’s left arm, after remaining seated quietly for at least 5 minutes. Appropriate-sized cuffs were used for each child arm circumference (≤ 23 cm or >23 cm). The mean systolic and diastolic BP values of the two measures were calculated and transformed into Z-scores by sex, age (in complete years) and height, following international recommendations. (14, 15) The outcome (repeated high SBP, DBP and SDBP) were defined, respectively, as systolic and diastolic BP Z-scores on the ≥ 95th percentile at the two visits (at 6- and 11-year follow-ups).
Household monthly income and maternal and child characteristics obtained at the hospital interview were among the independent variables used. Family income in Brazilian Real at the month before the child’s birth was categorized into quintiles (the first quintile representing the poorest and the fifth quintile representing the wealthiest families). Medical diagnoses of maternal gestational diabetes mellitus and hypertension during pregnancy were reported by the mother (yes or no). Tobacco use during pregnancy (yes, no) as reported by the mother was defined as at least one daily cigarette during any trimester of the pregnancy. Maternal age was categorized as ≤ 20, 21-25, 26-30, and > 30 years.
The newborn birth weight was measured and recorded by the hospital staff with pediatric scales that were checked weekly by the research team using standard weights. Infants were categorized into low birth weight (< 2500 grams) or normal weight (≥ 2500 grams). Gestational age was calculated using the first day of the last normal menstrual period or estimated by obstetric ultrasound obtained before 20 gestational weeks when information about the last normal menstrual period was unreliable or not available. In the absence of both menstrual and ultrasound information, gestational age was estimated from the physical and neurological assessment of the newborn, employing the Dubowitz method. (16) Gestational age was categorized into < 34 weeks, 34-36 weeks, and ≥ 37 weeks. Weight-for-gestational age variable was generated as per the standard population curve proposed by Williams et al. (17) and classified as small (birth weight lower than the 10th centile), adequate (birthweight between the 10th and the 90th centile) or large for gestational age (birth weight above the 90th centile) for a specific completed gestational age and sex.
The child’s skin color (reported by the mother) and family history of hypertension were collected at the 6-year follow-up visit. Skin color was later categorized into white and non-white. Information on family history of hypertension in the parents' first-degree relatives (grand-parents of the child) was categorized as none, from mother, father, or both.
Information on the child physical activity pattern, BMI-for-age, daily salt intake, and weekly consumption of package chips was obtained at the 11-year follow-up. Children were considered as active if they performed structured physical activities (with adult/teacher/coach intervention) in private services after school period. The child weight was measured with a high precision scale (0.01 kg) that was part of the BODPOD® machine (Cosmed, Italy, http://goo.gl/ 7jzfLc). Height was measured by trained anthropometrists using a Harpenden metal stadiometer, with 1 mm precision (Holtain, Crymych, UK). Children were classified by BMI z-score into "normal weight" (< +1 SD), "overweight" (≥ +1 to < +2 SD) and "obese" (≥ +2 SD). (18) A validated semi-quantitative food frequency questionnaire (FFQ) was administered to the mother to estimate the child mean daily sodium intake. The children assisted with the FFQ report by answering questions on frequency and portion sizes of a list containing 88 different foods. The recall period was the past 12 months. We reconfigured the 12-month-food consumption to daily consumption, with all portions standardized at 100 g. The sodium content of each food was estimated based on The Brazilian Food Composition Table. (19) Daily sodium intake in milligrams was then calculated, which was later categorized in quintiles (the first quintile representing the lowest sodium intake and the fifth quintile the highest). Information on weekly consumption of package chips was investigated and classified in none, up to 1, 2, and ³ 3.
Firstly, the maternal and child characteristics of the sample located at the 11-year follow-up and of those included in the analyses were compared to the whole cohort. Then, we calculated the prevalence (95%CI) of repeated high SBP, DBP, and SDBP with a 95% confidence interval, by the independent variables. The associations were assessed by Fisher's exact chi-square test.
Crude and adjusted logistic regression (20) was used to obtain odds ratios with 95%CI for the outcome. The backward selection strategy was employed in the multivariable analyses. The selection started with all variables in the model, then the variables with the largest p-value were removed one-by-one, thus stopping when all remaining variables were associated with the outcome at a p-value ≤ 0.20. All analyses were performed with Stata software version 12.0 with a statistical significance level of p-value < 0.05.
The Medical Ethics Committee of Faculty of Medicine of the Federal University of Pelotas, affiliated with the Brazilian National Commission for Research Ethics (CONEP) approved the study protocol of all follow-ups of the Pelotas 2004 Birth Cohort. At each stage of the study, the mothers or legal guardians of all subjects gave written informed consent. At 11 years of age, the subjects also gave written informed consent.