During the 4-year period 2005–2008, 27630 patients delivered in the two centres. Among them, 1 832 (6.6%) women had a newborn heavier than 4000 g, of which 268 (15%) had an elective cesarean section and 1564 (85%) attempted a vaginal delivery and were included in the study. The proportion of macrosomia remained steady over the 4-year study period. Most of the patients were European (71%), had a mean body mass index (BMI) before pregnancy of 23.9 (10% had a BMI ≥ 30 before pregnancy) and did not experience diabetes during or before pregnancy (92.7%) (Table 1). Mean weight gain during pregnancy was 14 kg (range: -6 kg; 42 kg). Median duration of pregnancy was 40.5 weeks (range 36.9 – 42 weeks). Labor was induced in 33%, mainly for post-term pregnancy (60%) or maternal - fetal reasons (30%) such as hypertension, diabetes or oligohydramnios. Median duration of labor was 6 hours with 34% patients delivering before 5 hours, 52% patients delivering between 5 and 9 hours, and 14% delivering after more than 10 hours. Two hundred and sixty two women (17%) experienced a cesarean delivery during labor with two main indications: non progressive labor/dystocia (65%) and non-reassuring fetal heart rate (34%). First degree perineal tears occurred in 26% of women, and 2nd, 3rd and 4th degree in 36%, 0.9% and 0.1% respectively. Two hundred and sixty one patients (17%) experienced postpartum hemorrhage with different treatment required: oxytocin (69%), prostaglandins (sulprostone) (27%), uterine artery embolization (2%), internal iliac artery ligation (2%), and haemostatic hysterectomy (0.5%). Overall, 95 patients (6%) presented maternal complications according to the MC composite criterion. Regarding neonatal complications, shoulder dystocia occurred in 5.1% of deliveries and 5 children presented with transient brachial plexus injury that completely resolved within 6 months without sequelae.

^§ Among vaginal deliveries.

*At least one of the maternal complications (3rd or 4th degree perineal tears or severe postpartum hemorrhage or transfusion).

**Clavicle: 11; Humerus: 3.

Mean birthweight was 4207 g (95% CI [4010; 4590]). Eight percent weighed more than 4500 g and 0.4% more than 5000 g. The gender was male in 67% of newborns and the 5 minute Apgar score was greater than 7 in 98.5% of cases. Seventy three newborns (4.7%) were admitted to neonatal intensive care unit, mainly for respiratory distress (43%) or hypoglycemia (30%), and for a median duration of 2 days [range 1–10].

Factors significantly associated with MC in univariate analysis were patient’s origin, history of a macrosomic infant vaginally delivered, prolonged duration of labor, cesarean delivery during labor and neonatal birth weight.

In multivariate analysis (Table 2), the adjusted odds ratios of MC were statistically significant for Asian women, prolonged labor (>10 hours), cesarean during labor and neonatal birthweight greater than 4500 kg. Multiparous women who had already vaginally delivered a macrosomic infant had a decreased risk of MC: ORa = 0.5 [0.2; 0.9] (p = 0.03). Attributable risk associated with birth weight greater than 4500 g was 10% whereas it was 17%, 12% and 3% for cesarean delivery during labor, duration of labor and Asian origin respectively. Sensitivity analyses for the classification of elective cesarean section (see methods section) gave similar results, i.e. same significant risk factors and same magnitude for attributable risks.

*Adjusted odds ratio and confidence interval.

Adjustment for all the variables in the Table and gestational age at delivery and hospital centre.

**Computed if OR significant and greater than 1.

***Composite criteria for maternal complications: at least one of the maternal complications (3rd or 4th degree perineal tears or severe postpartum hemorrhage or transfusion).

We report that the occurrence of maternal complications when attempting vaginal birth is 6% among women delivering infants with a birthweight ≥ 4000 g. The prevalence of macrosomic infants and the characteristics of our sample were very similar to those depicted in previous studies 4 5 11 19 20 21 . The risk factors for MC were not only fetal weight but also Asian origin, long duration of labor, and cesarean during labor. It should also be noted that multiparous women with a previous vaginal delivery of a macrosomic child had a decreased risk of MC.

It is well known that maternal and neonatal morbidity increases with birthweight and especially over 4500 g 5 9 11 12 22 . Zhang et al. 11 have shown a J-shaped birthweight-specific perinatal mortality and morbidity curves, with 2 important thresholds: 4500 g and 5000 g. Over those two thresholds, complications are dramatically increased. The 4500 g threshold was always present in our study, but to a lesser extent since the sample was limited to newborns heavier than 4000 g. In turn, this intended limitation allowed us to study other factors since it decreased the role of birthweight in MC. Moreover, the large and unselected sample (all women with a newborn heavier than 4000 g were included) provided a sufficient statistical power and results that are potentially generalizable.

Handa et al. 13 found that perineal tears risk increased 2-fold in case of macrosomia, but increased also in Asian women, Filipinas, and Indian women compared to white women with OR = 1.37, 1.63 and 2.50 respectively for anal sphincter laceration. Our findings confirmed these results. Ethnic origins may be associated with differences in body type and variations in perineal anatomy 23 24 . These particularities in body tissue may increase the risk of perineal tears and, as found in our study for Asian women, may favor maternal complications.

The potentially “protective” effect of having previously vaginally delivered a macrosomic infant has been reported by Mahony et al. 25 . Maternal perineal tissues have therefore already experienced macrosomic delivery and are more likely to experience an uneventful subsequent delivery.

Maternal diabetes as well as body mass index were not significantly associated with maternal complication in our study. However, it is well known that maternal diabetes is associated with shoulder dystocia and also perineal tears 26 . Data from the literature also support that maternal obesity is associated with increased postpartum hemorrhage 27 28 and prolonged labors 29 . This discrepancy in our study may be explained by the fact that the role of these factors in maternal complications is primarily through birth weight. Another explanation may also be that the proportion of maternal diabetes and obese patients remained consistent, albeit relatively low, in our sample (respectively 7.3% and 11.5%).

Cesarean delivery is a known risk factor of postpartum hemorrhage especially when it occurs during labor and after a long duration of labor 30 31 . Similarly, prolonged labor with the use of oxytocin is the main contributor of uterine atony that account for 79% of the cases of postpartum hemorrhage 30 . In fact, the occurrence of post-partum hemorrhage is more increased when one of those parameter is associated with macrosomia 14 32 . In our study, both parameters (cesarean and prolonged labor) appear to increase significantly the occurrence of MC, probably through their influence on post-partum hemorrhage. Those results are in agreement with the literature but quantification of their association after adjusting for birth weight has never been reported.

In addition, we computed the attributable risk as an alternative means to evaluate the role of each risk factor in maternal complications associated with fetal macrosomia , Although it can not ensure the causality of association the attributable risk quantifies the magnitude of the proportion of MC that could be avoided by removing the risk factor. Attributable risk estimation is particularly useful for factors that may be modified, but is useful to a lesser extent for unchangeable factors such as ethnic origin.

Estimation of attributable risks enabled us to precisely identify that only 10% of MC may be attributed to a birth weight greater than 4500 g, which is almost identical to the contribution of prolonged labor. Cesarean during labor appeared to be the most important factor with attributable risk equal to 17%. This could be explained by the fact it is likely responsible for the highest rates of post partum haemorrhage and blood transfusion. Finally, as the sum of attributable risk did not reach 100%, we should also mentioned that a fraction of the cases of MC were not associated with identifiable risk factors.