Statistical analysis
Baseline characteristics of study population were described for preterm birth and term birth separately. Data were expressed as mean ± standard deviation (SD) for normally distributed variables or median (interquartile range, IQR) for skewed variable or frequency (percentage) for categorical variable. To test for difference between preterm birth and term birth groups, continuous variables were compared byt -test or Wilcoxon test and categorical variables were compared by Chi -square test. ApoB, ApoA-1 and ApoB/ApoA-1 were categorized into tertiles (ApoA-1: < 1.98, 1.98-2.18 or ≥ 2.18 g/L; ApoB: < 0.88, 0.88-1.06 or ≥ 1.06 g/L; ApoB/ApoA-1: < 0.42, 0.42-0.52 or ≥ 0.52) and the lowest tertile was used as the reference in the analysis. Logistic regression was used to estimate the odds ratio (ORs) and 95% confidence intervals (CIs) for the associations between ApoB, ApoA-1, ApoB/ApoA-1 ratio and the risk of PTB. ApoB, ApoA-1and their ratio were included in two ways: categorical or continuous variables. All multivariate models were adjusted for potential confounders including maternal age at delivery (continuous), pre-pregnancy BMI (continuous), parity (primiparous or multiparous), delivery mode (vaginal delivery or cesarean delivery), GDM (yes or no) and infant sex (male or female). Since smoking is rare among Chinese pregnant population and in our study only four women admitted smoking during pregnancy, maternal smoking status were not adjusted. Restricted cubic spline analysis with four knots (5th, 35th, 65th and 95th percentiles) were conducted to further explore non-linear dose-response relationship between mid-pregnancy apolipoprotein profile and risk of preterm birth.
Subgroup analyses were conducted according to maternal pre-pregnancy BMI (≤ 24 kg/m2, >24 kg/m2), maternal age at delivery (< 35 years, ≥ 35 years) and parity (primiparous, multiparous). The cut-off of pre-pregnancy BMI was decided according to the recommendation of the Working Group on Obesity in China27.
The presence of pregnancy complications may affect the association between apolipoprotein and preterm birth. Therefore, we performed a sensitivity analysis among a subgroup of 5,080 mother-newborn pairs after excluding mothers with GDM.
All statistical analyses were performed by R software (version 3.5.3) and SAS (version 9.4; SAS Institute Inc.). P < 0.05 was considered statistically significant.