Methods
The current study was conducted at the Unit of Obstetrics and Gynaecology of the San Paolo Hospital, Medical School in Milano, Italy. The Unit is a secondary referral centre, i.e. comprises a maternal intensive care unit (ICU) but not neonatal ICU. The audit system in our institution involves a weekly discussion of relevant clinical cases, the use of standardized and up-to-date protocols and the publication of an annual report about labour, delivery, maternal and neonatal outcomes of all women. The majority of the non-Italian women delivering in our Unit, are assisted by a non-profit social cooperative that deals with cultural mediation. Data were obtained from an electronic database that includes information on deliveries that occurred in our clinic from 1996 to 2019. Data were already available for the analysis for all women as part of the annual clinical report of the Unit. Obstetric outcomes, maternal and fetal characteristics included in the database are collected from medical records at the end of each month by residents in Obstetrics and Gynaecology. The database includes all the live births and stillbirths from 23 weeks of gestational age. The board of the San Paolo Department of Health Sciences approved the study protocol. When information on parity, number of fetuses, previous caesarean section, onset of labour, gestational age, and fetal presentation were all available, women were classified according to the modified TGCS; for the purpose of the analysis we selected women belonging to Group 1 (nulliparous, single cephalic, ≥37 weeks, in spontaneous labour), 2A (nulliparous, single cephalic, ≥37 weeks, induced), 3 (multiparous, excluding previous CS, single cephalic, ≥37 weeks, in spontaneous labour, 4A (multiparous, excluding previous CS, single cephalic, ≥37 weeks, induced) and 10 (all single cephalic, ≤36 weeks). We also included in the regression analysis women of Group 5 (previous CS, single cephalic, ≥37 weeks) only if they attempted a trial of labour after caesarean (TOLAC). Hereafter we will refer to these women as Group 5. We excluded from the analysis women of Groups 2B (nulliparous, single cephalic, ≥37 weeks, CS before labour), 4B (multiparous, excluding previous CS, single cephalic, ≥37 weeks, CS before labour), 6 (all nulliparous breeches), 7 (all multiparous breeches), and 9 (all abnormal presentations) because they underwent elective caesarean sections as for protocol. We also excluded multiple pregnancies (group 8) due to the presence of several potentially confounding factors and to the limited population The analysis was conducted inside each group and not between groups. Induction of labour was defined as the use of any mechanical method (intrauterine Foley catheter or Cervical Ripening Balloon), medication (prostaglandins or oxytocin) or amniotomy when not in labour.
The CS rate was calculated for each group as the number of CS out of the number of all the deliveries. We also calculated the CS rate per year in each group. The impact of advanced age (maternal age equal to or higher than 40 year), diabetes (including pre-pregnancy diabetes mellitus and gestational diabetes mellitus), hypertensive disease (including pregestational hypertension, gestational hypertension, preeclampsia or eclampsia), fetal macrosomia (neonatal birthweight higher than 4000 g), obesity (pre-pregnancy body mass index ≥30 kg/m2), obstetric analgesia and immigrant status was evaluated. Gestational diabetes mellitus and hypertensive disease of pregnancy were diagnosed according to the guidelines available at the time of the pregnancy. The dependent variable was the type of delivery, i.e. vaginal (both spontaneous or operative) or caesarean section. The variable macrosomia was not included in the analysis of group 10 because all the newborn weighed less than 4000 g. A binomial logistic regression was performed to ascertain the effects of variables on CS rates and odds ratio were calculated. For missing values, listwise deletion approach was applied. Linear regressions and Anova tests were performed to evaluate temporal trends in CS rates for considered Robson groups.
All calculations were performed using SPSS version 25.0 (SPSS Inc, Chicago, IL, USA). A p-value lower than 0.05 was considered statistically significant.