The utility of customised growth charts for identifying macrosomia and the effect of intervention

Objective: Best management of suspected large for gestational age (LGA) fetuses is unclear. In some hospitals, women with an LGA fetus by customised growth charts are offered earlier induction. This study aimed to examine scan accuracy for this group and the outcome with intervention. Study design: This is a retrospective cohort study of pregnant women taken from 3 groups; women with a suspected LGA fetus, women with diabetes (DM) and a control group of women that underwent induction of labour on or after 280 days gestation. Scan accuracy using GROW and WHO charts in the LGA and DM cohorts was assessed using ROC curves and outcomes between the cohorts were compared. Results: Over a 12 month period 845 suitable cases were identified; LGA (128), DM (116) and control cases (601). Mean gestation at induction was 275.5 (5.8), 270.0 (7.6) and 287.1 (3.3) days. Mean birthweights were significantly different across groups (3818g (345), 3434g (416), 3653g (399)). No excess shoulder dystocia or neonatal morbidity was seen in the LGA group. Spontaneous vaginal delivery was significantly less likely in the LGA group vs the control group at 69/128, 53.9% vs 413/601, 68.7% RR 0.78 (95% CI 0.66-0.93 p0.02). Postpartum haemorrhage >1500mL was more likely in the LGA group, RR 2.28, vs the control group (95% CI 1.41 to 5.62 p<0.01). Mean scan error was -5.2% for the DM group and +15.6% for the LGA group. Positive predict value of scan estimated weight for birthweight >90th centile on GROW chart, >90th centile on WHO chart, and projected birthweight of >4000kg was 0.35-0.40. Of these three thresholds projected birthweight of >4000kg from scan significantly better predicted itself (AUROC 0.80, 0.79 and 0.87). Conclusion: Induction of suspected large for gestational age pregnancies at 39 weeks is an intervention based on a test known to have poor positive predict value. This may increase liabilities and costs without benefit.研究背景与目的：目前针对疑似大于胎龄儿（large for gestational age, LGA）胎儿的最优管理方案尚不明确。部分医院会根据定制生长曲线诊断为大于胎龄儿的孕妇，实施提前引产。本研究旨在评估该类人群的超声扫描诊断准确性，以及干预措施的临床结局。 研究设计：本研究为回顾性队列研究，纳入3组孕妇：疑似大于胎龄儿组、糖尿病（DM）组，以及孕周≥280天时接受引产的对照组。采用GROW生长曲线与世界卫生组织（WHO）生长曲线，通过受试者工作特征曲线（Receiver Operating Characteristic Curve, ROC）评估大于胎龄儿组与糖尿病组的超声扫描诊断准确性，并比较三组队列的临床结局。 研究结果：本研究在12个月内共纳入符合入组标准的病例845例，其中大于胎龄儿组128例、糖尿病组116例、对照组601例。三组孕妇的引产平均孕周分别为275.5(5.8)天、270.0(7.6)天与287.1(3.3)天；三组的平均出生体重分别为3818g(345)、3434g(416)与3653g(399)，组间差异具有统计学意义。大于胎龄儿组未出现肩难产发生率升高或新生儿并发症增加的情况。与对照组相比，大于胎龄儿组的自然阴道分娩率显著更低：大于胎龄儿组为69/128（53.9%），对照组为413/601（68.7%），相对危险度（RR）为0.78（95%置信区间CI：0.66~0.93，P=0.02）。大于胎龄儿组发生产后出血量>1500mL的风险显著高于对照组，RR为2.28（95%CI：1.41~5.62，P<0.01）。糖尿病组的平均超声扫描估算体重误差为-5.2%，大于胎龄儿组为+15.6%。针对GROW生长曲线出生体重>90百分位、WHO生长曲线出生体重>90百分位、超声估算出生体重>4000g这三项指标，其阳性预测值为0.35~0.40。在三项评估阈值中，超声估算出生体重>4000g的预测效能最佳，对应的受试者工作特征曲线下面积（AUROC）分别为0.80、0.79与0.87。 研究结论：针对疑似大于胎龄儿的孕妇在孕39周实施引产，是一项基于阳性预测值较差的检测手段的干预措施，该措施可能在无临床获益的情况下增加医疗纠纷风险与医疗成本。