br Conflict of Interest Disclosures br Author contributions

Conflict of Interest Disclosures
Author contributions
Introduction Large for gestational age (LGA), defined as a birth weight≥90th percentile for gestational age, is the predominant adverse outcome associated with maternal hyperglycemia (Langer et al., 2005; Metzger et al., 2008). LGA is the main factor underlying birth trauma and preterm birth, as well as obstructed labor, that leads to cesarean delivery (Kc et al., 2015; Zhang et al., 2008). Long-term effects of LGA for the offspring include obesity, the metabolic syndrome, type 2 diabetes and insulin resistance (Damm et al., 2016). In clinical practice, the main aim of gestational Phenformin (GDM) treatment is to control glucose metabolism and thus reducing fetal macrosomia and obstetric complications as well as to prevent obesity in the offspring. Treatment of GDM is supposed to decrease the risk of fetal macrosomia (Crowther et al., 2005; Landon et al., 2009). In 2010, the International Association of the Diabetes and Pregnancy Study Groups (IADPSG) recommended new criteria for GDM diagnosis, based on odds ratios of abnormal birth weight, cord C-peptide and percent body fat observed in the Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study (Metzger et al., 2008), with the primary aim of prevention of obesity risk among high-risk offspring. The test criteria include relatively lower cut-off values and single abnormal of a fasting, 1h or 2h glucose measured by an universal, single stage screening of 2h 75-g oral glucose-tolerance test (OGTT) is adequate to make a diagnosis, which consequently increased the prevalence of GDM in many countries including China adopted this criteria (Cundy et al., 2014). Although OGTT has been well recognized as the “gold standard” for GDM diagnosis, it has many disadvantages of not reproducible, time-consuming and fairly demanding for both the pregnant women and the laboratory (Davidson, 2002; Hanna and Peters, 2002). Given the uncertainty of the utility of each glucose measurement in the prediction of fetal macrosomia and the significant resource implications of the IADPSG criteria, there remains controversy regarding this screening methodology for identification of macrosomia risk during pregnancy (Kalter-Leibovici et al., 2012). In the setting of developing countries with limited resource, the screening and management of high-risk population may be more important and cost-effective than GDM diagnosis. Fasting plasma glucose (FPG) provides a cheap, acceptable reliable, reproducible alternative GDM screening method to the OGTT for the last three decades (Mortensen et al., 1985; Zhu et al., 2013), with renewed attention following introduction of the IADPSG criteria. Early studies suggested that FPG had significantly higher predictive value for LGA in comparison to post-load glucose, independent of maternal BMI and 2h glucose value (Disse et al., 2013; Legardeur et al., 2014). A recent systematic review also found that fasting glucose concentration has stronger associations with LGA than post-load glucose concentration (Farrar et al., 2016). FPG performance has been largely determined by utility to GDM detection using specific criteria, with limited information regarding prediction of adverse pregnancy outcomes (Agarwal, 2016). However, the U.S. Preventive Services Task Force has suggested that the gold-standard for GDM screening tests, should include an acceptable, agreed set of relevant pregnancy outcomes (Donovan et al., 2013). There remains uncertainty regarding whether a single FPG measurement is sufficient for prediction of increased risk of adverse perinatal outcomes.
Methods
Results A total of 15,198 pregnant women recruited to the cohort study delivered between Feb 2012 and Jun 2016 (Fig. 1). We excluded 2309 women for the following reasons: withdrawal (n=623), diagnosis of pre-pregnancy hypertension (n=44), diabetes (n=28), multiple pregnancy (n=319), termination of pregnancy (n=143), stillbirth (n=27), missing delivery data (n=227), and missing OGTT (n=898). We included information obtained from 12,889 mother-newborn pairs in the analyses. There were no significant differences in birthweight or gestational age between those women included in the final analysis (n=12,889) and those who had missing OGTT results (n=898). The characteristics between women who dropped out or missing OGTT results and delivery data and women who were included in the analysis were only statistical different in parity (Appendix Table 1).