Study design
We analyzed data from a population-based birth cohort study designed to include pregnant women living in the urban area of Pelotas, Southern Brazil, with an expected delivery date from January 1st, 2015 to 31st December 2015. The Pelotas 2015 Birth Cohort Study recruited pregnant women from all health facilities offering antenatal care including public and private services. During antenatal phase, three types of questionnaires were applied according to the gestational age of the woman at enrolment. Those before 16 weeks of pregnancy answered the initial assessment questionnaire and were contacted again between 16 and 24 weeks to answer the main assessment questionnaire. Women who were enrolled after 16 weeks of gestation answered to a combined assessment questionnaire containing questions from the initial and the main assessments. Thereafter, mothers were interviewed at hospital up to 48 h after delivery (perinatal study). Details on the cohort design, recruitment and enrollment can be found elsewhere [10].
The study was approved by the School of Physical Education Ethics Committee of the Federal University of Pelotas in an official letter numbered 522/064. Written informed consent was obtained from all participants.
ART exposure
Utilization of an ART procedure was gathered at the antenatal and the perinatal interviews. Women who answered “yes” to the question:” Did you have an artificial fertilization in this pregnancy? were contacted later (during 2017) and were invited to take part in a sub-study. Women in the sub-study were interviewed by phone. After five failed attempts of a phone interview, women were searched and interviewed at home.
Utilization of an ART procedure was firstly confirmed and then other related characteristics, such as type of ART were investigated. ART procedures included in vitro fertilization (IVF), intracytoplasmic sperm injection (ICSI) and artificial insemination (AI). Conceptually, ART does not include assisted insemination (artificial insemination) using sperm from either the woman’s partner or a sperm donor. However, considering that women who performed this procedure did not have a natural conception, we decided to include artificial insemination as part of our exposure.
Outcomes
Maternal health outcomes included mode of delivery (cesarean/vaginal); pregnancy complications (arterial hypertension, pre-eclampsia, gestational diabetes, and hospitalization during pregnancy) reported by the mother or extracted from medical records at the mother’s prenatal care card; and number of offspring delivered (single or multiple pregnancy).
Gestational weight gain was analyzed in two ways: (I) weight at the end of pregnancy (as recorded at the maternal prenatal care card or reported by the mother) minus the reported pre-pregnancy weight (II) assessed following the 2009 Institute of Medicine recommendations based on pre-pregnancy BMI and total gestational weight gain reported by the mothers at the hospital after delivery. Recommended weight gains during pregnancy for underweight, normal weight, overweight, and obese women were 12.5 to 18 kg, 11.5 to 16 kg, 7 to 11.5 kg, and 5 to 9 kg, respectively. Maternal weight at 3 months postpartum (kg) was collected by the research team with a TANITA scale model UM-080 with precision of 100 g and capacity for up to 150 kg.
Child health outcomes included gestational age at delivery that was calculated based on information collected in antenatal and perinatal study as follows: data on the last menstrual period as recorded at the woman’s prenatal care card and/or by maternal self-report; and through the obstetric ultrasound performed before 16 weeks of pregnancy. The final variable of gestational age was estimated by an algorithm that considered all the information previously collected, as well as the plausibility of the estimates based on birth weight, length and head circumference, according to the Fetal and Neonatal Growth Curves for the twenty-first Century [8]. Newborns with gestational age < 37 weeks were categorized as preterm. Intrauterine growth restriction (weight < 10 percentile for gestational age) and health problems at birth were also investigated.
Birth weight and 1-min and 5-min Apgar were collected from medical records at the hospital. Length and head circumference at birth were measured at the hospital up to 48 h after delivery by the trained research team. Information on health problems at birth, breastfeeding duration and breastfeeding up to 12 months after birth was obtained at the 12-month follow-up.
Covariables
Family income (1–3; 3.1–10; > 10 minimum wages) (01 minimum wage was equivalent to US$ 300.00) and maternal age (< 30; 30–35; 36–40; > 40 years), parity (1; ≥ 2), schooling (0–8; 9–11; ≥ 12 completed years), and pre-pregnancy body mass index (BMI) were used as independent covariables. Pre-pregnancy BMI was calculated dividing weight by height squared. Pre-pregnancy maternal weight and height were self-reported. Women were classified as underweight (< 18.5 kg/m2), normal weight (18.5–24.9 kg/m2), overweight (25.0–30.0 kg/m2), and obese (> 30 kg/m2), according to cutoffs defined by the World Health Organization.
Statistical analysis
Between groups difference was analyzed using the Student’s t-test (mean, SD) for continuous variables or the chi-squared and Fisher’s Exact test for categorical variables (n, %). Normality of data were checked graphically using histograms and by the parameters mean, median, skewness, and kurtosis. All continuous outcomes presented symmetric distribution.
Unadjusted and adjusted analyses were performed using linear regression expressed in β coefficients or binomial logistic regression expressed in odds ratios (OR). Analyses were adjusted for family income, maternal age, parity, and pre-pregnancy BMI. Mediation analyses were performed for statistically significant associations using G-computation in order to verify the proportion of the association between ART and child-health outcomes captured by multiple pregnancies. This was done by estimating the natural direct effects (NDE) and natural indirect effects (NIE) of ART on child-health outcomes. The NDE represents the effect of the exposure (ART) on the outcomes (gestational age at delivery, birth weight, length at birth and head circumference at birth) that is not captured by the mediator (multiple pregnancies), while the NIE estimates the effect that is captured. Therefore, the sum of the NDE and NIE would represent the total effect, and the quotient of dividing the NIE by the total effect would represent the percentage of the effect that is captured by a mediator. In our mediation analyses maternal age, family income, maternal schooling, parity, and pre-pregnancy BMI were considered as base-confounders and hospitalization during pregnancy as a post-confounder.
Statistical significance was set at 5, and 95% confidence intervals were adopted. All the analyses were performed using the software Stata version 12.1 (StataCorp, College Station, Texas, US).