This was a prospective cohort study as part of the Breastfeeding Attitude and Volume Optimization (BRAVO) study (NCT01566812) in Jakarta, Indonesia [21].The participant enrollment was performed from June 2012 to January 2017. From 718 participants enrolled in BRAVO study, initially, 282 participants were excluded due to incomplete data on exposure. Afterwards, 152 participants were excluded due to incomplete infant’s growth measurement, thus we included 284 pairs of mother-infant with complete data on exposure and growth measurement of their infants for analysis. Both populations were comparable on important prognostic factors (supplemental data). Pregnant women in three centers, i.e., one private maternal and children hospital (Budi Kemuliaan Hospital) and two primary care centers in Senen and Jatinegara districts, Jakarta, Indonesia were recruited for participation in the BRAVO study during their 3rd trimester antenatal care visit. BRAVO was ethically approved by the Institutional Review Board of the Faculty of Medicine Universitas Indonesia/Cipto Mangunkusumo Hospital, Jakarta, Indonesia (reference number: 913/UN2.F1/ETIK/X/2012) [21].Informed consent was obtained from all study participants prior to study enrolment [21].
Maternal characteristics including maternal age, parity, history of abortion, family income, level of education, weight gain during pregnancy (expressed as ΔBMI: the difference between BMI at labor and BMI prior to pregnancy), and alcohol or illicit drug use in pregnancy were obtained from self-report questionnaires filled out by pregnant women at recruitment [21]. Gestational age at delivery, Apgar score, maternal hypertension and diabetes (pre-existing or gestational), certain neonatal morbidities (special nursery care requirements, including sepsis, respiratory distress, and hyperbilirubinemia) were obtained from medical records [21].
Measurement of exposure
Exposure to household pesticides was obtained by validated self-report questionnaires filled out by the pregnant women at recruitment [21]. Mothers were considered as exposed to household pesticides when they use mosquito insecticides, or other pesticides to eliminate cockroaches or other insects during their pregnancy. Due to the all year high incidence of dengue in Jakarta, we assume that the use of pesticides is a routine manner. We decided to classify the pesticides exposure into three groups as exposed to mosquito vs non mosquito vs combination of both type considering that e.g. pesticides used against cockroaches might be more potent than pesticides used against mosquitos. No data on the type or brand of household pesticides was recorded [21]. Exposure to other possible EDCs including water sources, fuel usage at home, and garbage burning were recorded [21]. Traveling using open vehicles (motorcycle or bajaj) without using a face mask during travelling was recorded as proxy to exposure to traffic air pollution [21]. Data on active and passive smoking in mothers during pregnancy were also recorded [21].
Birth sizes and post natal growth measurement
Birth and length were measured at birth, while head circumference (HC) was measured at day 7 after birth. Afterwards, the weight, height, and HC were measured at 7 days, 1, 2, 4, and 6 months of age. All the measurements were performed twice according to standardized procedures and then averaged. Infants with at least two measurements (other than at birth and day 7) available within the first 6 months of life were included in the analyses [22, 23]. Subsequently, linear mixed modeling was performed with extraction of estimated random slopes per child for weight, length, and HC. Linear regression was performed to calculate the predicted values per child, giving the estimated length gain rate, weight gain rate and HC increment per child. Weight gain rate was expressed in grams per day, while the length gain rate and HC increment were expressed in cm per months [22, 23]. Weight gain rate adjusted for length gain rate (WLG), reflecting excess weight gain, was assessed for each child by deriving internal Z-score in our study population and calculating the standardized residuals from the linear regression model with weight gain as the dependent variable and length gain as the independent variable [22, 23].
Covariates
Socio-economic status (SES) i.e. household income and level of education, mother’s age at pregnancy, BMI increment during pregnancy were considered as covariates in this study. Other explored confounders were active smoking and passive smoking during pregnancy, household water sources, fuel sources, exposure to traffic air pollution, and garbage burning. All of the possible covariates were treated as categorical, excluding mother’s age and BMI increment. Breastfeeding was considered as a potential effect modifier [24]. Information on breastfeeding was obtained by questionnaire at 1, 2, 4 and 6 months of infant’s age.
Other infant health outcomes
Episode of fever, respiratory symptoms, gastrointrointestinal, skin and eye infection symptoms were collected by use of an infection diary filled by mothers at 1, 2, 4 and 6 months of infant’s age. Fever was subjectively defined as increased temperature that needed antipyretic drug or axilla temperature > 37.5 °C. Gastrointestinal symptoms were recorded as any episodes of vomiting, diarrhea or loose stool. Any cough episodes, runny nose, or dyspnea were recorded as respiratory symptoms. Skin symptoms were recorded as eczema or acute urticaria, while eye symptoms defined as any discharge in one of both eyes or conjunctival injection (redness in the white portions).
Statistical analysis
Baseline characteristics were tabulated by exposure (yes/no) to household pesticides. Continuous variables were expressed as mean and standard deviation or median and interquartile range if distributions were skewed. Group differences were estimated and tested by independent groups t-test, chi-square test, or Fisher’s exact test where appropriate and p values were provided.
Multivariable linear regression adjusted for confounders was used to assess the associations between household pesticides exposure and birth sizes (weight, length, HC) and between household pesticides exposure and growth rate (weight, length, HC, and WLG). To that end, dummy variables were created for categories of household pesticides exposures (mosquito pesticides, non-mosquito pesticides, and combined pesticides groups) and simultaneously entered into the models as independent variables. Possible interaction between household pesticides categories and breastfeeding at 6 months of age on growth rates was tested by adding a product term of (dummy variable of house hold categories * breast feeding at 6 months) to the models. The difference in the incidence of infections in early life (fever, respiratory, gastrointestinal, skin, and eye infection) between exposed and non-exposed to household pesticides group were explored using the chi square test.
Statistical significance was assumed if 95% confidence intervals did not include the estimation null values, corresponding to a two-sided p value of < 0.05. Statistical analyses were conducted using IBM SPSS version 24 for Mac.