Effect of Epidural Analgesia on Labour, Neonatal and Maternal Outcomes.

Lumbar epidural analgesia (EA) is the recognized gold standard in labour pain control [1]. However, with data available to both support and refute a relationship between EA and a significant prolongation of the second stage of labour [1-4] (especially with low-dose anaesthetic protocols), its role is still controversial [5,6] . The second stage of labour is described as the period between complete cervical dilatation and the delivery of the baby. In 2014 the American College of Obstetricians and Gynecologists (ACOG) defined the normal duration of the second stage of labour as up to 2 hours in multiparous women and 3 hours in nulliparous ones [7]. However, as long as progress is being documented [7-9], newer recommendations propose longer durations based on individual factors [10,11] such as parity, maternal age [2] and body mass index (BMI) [12], hypertension [13], foetal weight and position [14], maternal position [15], oxytocin augmentation [2], and EA [16]. This study's attention on this phase of labour is strictly related to the potential impact of EA on foetal and maternal outcomes, and obstetric decision making [17-19]. In the literature, some papers report no detrimental foetal outcomes in cases involving a longer duration [3,20,21], while others show increased rates of maternal morbidity (third- or fourth-degree perineal lacerations, postpartum haemorrhage, and chorioamnionitis) [17,22] and Caesarean sections (CS), with labour dystocia as one of leading indications [19]. In order to investigate the effects of EA on labour effectively a propensity score matching seems appropriate. Propensity score matching is a statistical method for collecting data retrospectively that minimizes the selective biases that can arise from patients' backgrounds. Many studies have reported that propensity score matching produces results similar to RCTs despite its retrospective nature [4]. Based on five easily definable maternal characteristics, the Robson-10 classification system (RTGCS) introduced in 2001 and recognized by the WHO as the global standard for the analysis of pregnant patients [23], minimizes bias by comparing pregnant populations both within and across insitutions [24].

Using the propensity score matching method the present study aimed to analyse the impact of EA on the length of the second stage of labour and on foetal and maternal outcomes in the population of pregnant women referred to an Italian university hospital. These women have been stratified according to the RTGCS in order to settle the maternal characteristic confounders.

METHODS A retrospective cohort analysis was performed at a tertiary university hospital over an 11-year period (October 2008 to October 2019). This population was divided according to the RTGCS. Pregnant patients were enrolled according to the following RTGCS groups: R1 (nulliparous, single cephalic full-term pregnancy with spontaneous labour); R2a (nulliparous, single cephalic full-term pregnancy with induced labour); R3 (multiparous, single cephalic full-term pregnancy with spontaneous labour); and R4a (multiparous, single cephalic full-term pregnancy with induced labour). The exclusion criteria for this study were cases involving: multiple pregnancies, known major fetal or chromosomal abnormalities, pre-labour Caesarean deliveries, and elective Caesarean deliveries.

All maternal and obstetrical data were prospectively collected by labour and delivery unit personnel by entering cases into a perinatal database, which were then cross tabulated on an Excel file. The collected data included the demographic and obstetric parameters: maternal age and BMI, hypertension, diabetes, foetal weight and position, gestational age, labour induction (intravaginal or intracervical prostaglandin E2 gel, oxytocin), operative vaginal delivery (OVD) (only via the Kiwi OmniCup [produced by Clinical Innovations, Muray, Utah, USA] vacuum extractor), Caesarean section (CS), maternal morbidity (uterine atony, episiotomy, 3rd to 4th degree perineal laceration), and foetal morbidity (Apgar score< 7 at 1 and 5 minutes, neonatal resuscitation).

Access to EA is active on a 24-hour basis, with protocols reserving its administration for consenting women previously informed in an epidural outpatient clinic. All women who request analgesia for pain relief during labour are evaluated by an anaesthetist for suitability. Patients meeting absolute (i.e. uncorrected hypovolemia, coagulopathy, anticoagulant therapy) or relative (i.e. anatomical deformities, certain neurological disorders, sepsis) exclusion criteria are not qualified to receive EA. During labour, in the presence of a cervical dilatation of ≥3cm and in the active phase of the first stage (established by partograph), maternal status (blood pressure and temperature), and foetal well-being (20 minutes of normal cardiotocography) are also evaluated. In the absence of abnormalities, intravenous access by a 14-gauge (G) or 16G cannula is positioned and a crystalloid infusion is started. Using an aseptic technique while the patient is in sitting position, an epidural catheter is then placed at the L2-L3 or L3-L4 interspace. Finally, analgesia is established with the epidural administration of a low dose of local anaesthetic, plus a lipid-soluble opioid (ropivacaine 0.1% and sufentanyl 0.5%, 20 mL). Maternal blood pressure, foetal heart rate, pain scores, and the extent of sensory block are then assessed at five-minute intervals for the first 15 minutes, then at every half-hour. Analgesia is maintained with a top-up regimen, using intermittent manual epidural boluses of increasing concentrations of ropivacaine, with up to 0.15% at full dilation, according to specific needs of individual participants.

Statistical analysis As the use of observational data prevents the control of treatment assignment through randomization, systematic differences may be present between treated and untreated subjects in an randomized controlled trial. This can introduce confounding and thus prevent associations from being reliably estimated. However, the propensity score method is a statistical technique that aims to predict the probability of receiving treatment and/or having a condition based on identified covariates and background characteristics. Thus, the study population has been matched using a propensity score model with the optimal algorithm, including maternal age, BMI, gestational age, neonatal weight, parity, diabetes, and hypertension. The algorithm excluded patients with missing data in the matching variables. The quality of the matching was verified by considering an acceptable standard mean difference of <0.1 that was obtained in all participant groups. Data have been reported by mean ± standard deviation or by rate (i.e. percentage), unless indicated otherwise. All numerical variables with normal distributions were compared with t-tests, while categorical variables were compared with the chi-squared test or Fisher test when expected frequencies were less than 5. Statistical significance was considered for p-values <0.05. All analyses were performed in the R-statistical computing software 4.1.