pregnant women work out

The interactive effect of per-pregnancy overweight and obesity and hypertensive disorders

Study participants

Data were extracted from two Chinese birth cohorts. The first birth cohort (Cohort A) was conducted in the cities of Shenyang, Wuhan, and Guangzhou from April 2009 to March 2010. The second cohort (Cohort B) was performed in Zhuhai city from May 2014 to December 2016. In Cohort A, maternal age, infant sex, birthdate, delivery mode, gestational weight gain (GWG), BP measurements during pregnancy and gestational condition (e.g. taking antihypertensive medications, present of diabetes mellitus, etc.) were obtained from the patient Perinatal Health Booklets; maternal height and infant weight and length were measured by trained nurses; and maternal pre-pregnancy weight, gestational age, household income, educational level, maternal secondhand smoke exposure during pregnancy, the feeding practices during the first month of life, and cities were obtained using self-reported questionnaires. In Cohort B, maternal age, infant sex, birthdate, delivery mode, BP measurements during pregnancy, and gestational conditions were obtained from the Maternal and Child Health Information System; infant weight and length were measured by trained nurses; and maternal pre-pregnancy weight and height, gestational age, household income, educational level, maternal secondhand smoke exposure during pregnancy, the feeding practices during the first month of life, and GWG were obtained from self-reported questionnaires.

A total of 4,036 mother-child pairs (2,066 in Cohort A and 1,970 in Cohort B) were originally recruited with information on maternal pre-pregnancy weight and height, BP measurements during pregnancy, infant sex, infant weight and length measured at least twice at birth, 1, 3, 6 and 12 months, and questionnaires were filled out during the study recruitment, before delivery, and 1 month after giving birth. Among all participants, 97 mother-child pairs in Cohort A and 174 mother-child pairs in Cohort B were excluded for mothers using antihypertensive medications, having diabetes mellitus, tuberculosis, thyroid or hepatobiliary diseases, or a non-singleton pregnancy, or missing data in infants’ weight and length at birth. A total of 3,765 mother-child pairs (1,969, 1,698, 1,574, and 1,605 participants in Cohort A, and 1,796, 1,601, 1,574 and 1,665 in Cohort B with available data at birth, 3, 6 and 12 months, respectively) were finally included in this analysis. No significant differences were observed in the main variables between the samples used for analysis and those originally included (Supplementary Table S1). Therefore, the possible withdrawal bias was eliminated in this study. A flowchart of the study design is presented in Fig. 1.

Figure 1 Flowchart of participants. Cohort A was conducted in Shenyang, Wuhan and Guangzhou cities (2009–2010). Cohort B was conducted in Zhuhai city (2014–2016). All protocol-required procedures in the four cities of Cohort A and B were carried out in accordance with the Declaration of Helsinki and data were collected by trained staff. The ethical approval was obtained from the Ethics Committees of the Tongji Medical College, Huazhong University of Science and Technology, and all participants signed informed consent before study participation.

Anthropometric assessments

In Cohort A, infant weight and length at birth, 3, 6, and 12 months were measured twice by trained health staff with an electronic scale (WHS-I, Wuhan Computer Software Development) to the nearest 0.05 kg and a length measurement instrument (WHS-I) to the nearest 0.1 cm, respectively. In Cohort B, infant weight and length at the corresponding months were measured twice by trained health staff with an electronic weighing scale (Hengxin HCS-20-YE) and a horizontal baby bed (Hengxin HX-II), respectively. For all measurements, average values were used. The weight and length at each visit were linearly adjusted by the measured intervals to the exact duration of the corresponding months based on World Health Organization criteria11. For instance, an infant that weighed 3.5 kg at birth and 11.0 kg at the 370-day measurement had an adjusted 12-month weight of (11.0–3.5 kg) × (365.25/370) + 3.5 = 10.9 kg (12-month exact days were: 12 × 30.4375 = 365.25days). Maternal height in Cohort B was measured by trained staff using a stadiometer (Leicester height measure; Invicta Plastic, Leicester, UK).

Covariates

Births were divided into vaginal and cesarean deliveries. Infant birth weight and length, gestational age, maternal age, and maternal GWG were continuous variables. GWG was defined as the maternal weight before delivery minus their weight before pregnancy. The maternal educational level was classified as primary and secondary school, high school, university/college, and postgraduate education. Secondhand smoke exposure during pregnancy was defined for non-smokers as being in the presence of cigarette smoking for at least 15 minutes per day or at least 1 day per week12. Feeding patterns at 1 month were classified as exclusive breastfeeding (WHO definition13), mixed feeding (both formula and breast milk), and formula feeding (exclusively formula without breast milk). Household income was categorized as <3000RMB ($426.6), 3001–5000RMB ($426.7–711.0) and 5001–8000RMB ($711.1–1137.6), or ≥8001RMB ($1137.7). The adjusted models were performed with the mentioned confounders and cohort variable adjusted. The missing data was compensated using a non-parametric missing value imputation (missForest based on random forest package in R3.5.2)14. The included data and the imputed datasets representing the main variables did not show significant differences (Supplementary Table S2).

Definition of maternal pre-pregnancy OWO and HDP

BMI was calculated by dividing weight in kilograms by height in meters squared. An expert panel of the World Health Organization appealed for additional research on the health implications of infants using a BMI indicator based on length15. It has been shown that the 85th percentile cutoff of BMI in infancy defined a highly-specific threshold for an increased risk of becoming overweight or obesity in early childhood16,17. In addition, compared with weight for length, BMI was recognized as a better indicator of body adiposity in infancy18. To make the data internationally comparable and able to be tracked over time, we defined a BMI ≥ 85th percentile as a high BMI status at 12 months of age, according to the BMI percentiles of the WHO Child-Growth Standards19. The maternal pre-pregnancy BMI status was classified as non-OWO (BMI < 24 kg/m2) and OWO (BMI ≥ 24 kg/m2), according to the Working Group on Obesity in China20. Gestational BP status was divided into the HDP (systolic BP ≥ 140 mmHg and/or diastolic BP ≥ 90 mmHg during pregnancy) and normal BP (NBP, systolic BP/diastolic BP < 140/90 mmHg) groups. We were unable to distinguish specific gestational hypertensive disorders, such as gestational hypertension and preeclampsia since urine protein data were not available. Therefore, the definition of HDP in this study represents a combination of gestational hypertensive disorders during pregnancy, which was similarly defined in a previous study10.

Statistical analysis

All data analyses were conducted using SPSS software, version 16.0 (SPSS, Inc., Chicago, IL, USA). Maternal age, GWG and gestational age (continuous variables) with non-normal distributions were expressed as the median and inter-quartile range (IQR). Maternal education level, feeding patterns at 1 month, mode of delivery, and maternal secondhand smoke exposure during pregnancy (category variables) were presented as N (%). The Kruskal-Wills test was used to compare continuous variables, and the Chi-square test was used to compare categorical variables between the groups. The generalized estimating equation models were used to assess the association of pre-pregnancy BMI and BP during pregnancy with the BMI-z at birth, 3, 6, and 12 months. Both multiplicative (statistical interactions) and additive (biological interaction) interactions21 were assessed in this study. A total of three logistic regression models were used to evaluate the effects of pre-pregnancy BMI status and gestational BP status and their multiplicative interactions on the high BMI status of the offspring at 12 months of age, adjusting for potential confounders. The three models are Model 1, the unadjusted model; Model 2, adjusted for birth-related variables (infant sex, gestational age, birth weight and length, and delivery mode), maternal variables (maternal age and GWG) and cohort variable and; Model 3, a model similar to Model 2 with further adjustments for family environment-related variables (maternal educational level, maternal secondhand smoke exposure during pregnancy, household income, and feeding patterns at 1 month). These potential variables were included because there were either well-established associations with the offspring BMI-z values or because there was at least a 10% variation in the interest estimates if removed22. P-values <0.05 were considered statistically significant.

The odds ratios (OR) of OR00, OR10, OR01, and OR11 were calculated for the Non-OWO + NBP (reference group), Non-OWO + HDP, OWO + NBP and OWO + HDP groups, respectively. The calculation of the relative excess risk of interaction (RERI = OR11 − OR10 − OR01 + 1), attributable proportion (AP = RERI/OR11) and the 95% confidence interval(CI) was performed using a Microsoft Excel spreadsheet generated by Andersson et al.23. RERI > 0 or AP > 0 indicates a biologically additive interaction, which had been widely used in previous studies24,25.

Leave a Comment

Your email address will not be published. Required fields are marked *