Improving maternal and perinatal health is a global health priority and forms part of the World Health Organization’s sustainable development goals.¹ The highest maternal and perinatal mortality rates occur in low-income and middle-income countries (LMICs), and mothers delivering in rural areas are at the highest risk for adverse outcomes.^1,2 Complications during labour contribute significantly towards adverse maternal and perinatal outcomes, with obstetric haemorrhage, pregnancy-related sepsis and anaesthetic complications causing more than 35% of maternal deaths with primary obstetric cause in South Africa in 2023.^3,4 Offering safe interventions during labour is an essential component of improving outcomes globally, but especially in LMICs like South Africa.

Vacuum extraction and caesarean section are obstetric interventions employed for maternal or foetal indications during the second stage of labour, which is defined as the period from complete cervical dilatation to the full delivery of the foetus.^5,6 Either intervention carries risks and benefits, and the decision of procedure should be individualised, depending on the indication for intervention, facility and healthcare provider ability, decision-to-delivery interval (DDI) and medico-legal considerations.⁵

South Africa and many other LMIC countries have seen a steady rise in caesarean delivery (CD) rates.^7,8 The reasons for this include medicolegal concerns, income generation, healthcare convenience, the perceived benefits of CDs over normal vaginal or instrumental deliveries, and a lack of equipment and training to perform instrumental deliveries.^3,8 There are safety concerns with the unnecessary use of CD as they are associated with increased risk of haemorrhage and infection, prolonged hospital stays, and need for blood transfusions.^3,4,9 Furthermore, long-term risks of a scarred uterus – such as hysterectomy, abnormal placentation, uterine rupture, haemorrhage, and postpartum infection in future pregnancies – are important considerations.^3,10 These risks are particularly high in rural settings where challenges in access to healthcare mean many births happen outside of healthcare facilities, and complicated deliveries are being performed by non-specialist medical practitioners with limited training and experience.¹¹ In resource-poor settings, access to vacuum-assisted deliveries (VADs) can potentially improve maternal and neonatal outcomes and prevent unnecessary CD. Despite these recommendations, the use of VAD has almost disappeared from obstetric practice in most LMICs, with remarkably low rates of use.^3,8 There is a paucity of literature on the use of VADs in South Africa and other LMICs, and the available literature is mostly from tertiary centres with specialist obstetricians overseeing care.^12,13,14

The aim of this study was to compare maternal and perinatal outcomes of VADs and second-stage caesarean deliveries (SSCDs) performed by medical doctors with no obstetric specialist support at a rural district hospital in South Africa.

As shown in Figure 1, there was a total of 7293 deliveries during the study period, which included 354 vacuum extractions (4.8%), 1605 caesarean deliveries (22%), and 5334 normal vaginal deliveries (73.1%). There were 72 cases of failed vacuum extraction (16.9% failure rate) included in the VAD attempt cohort (intention a priori). A total of 567 cases met the inclusion criteria and were included in the analysis, consisting of the two cohorts of VAD (417) and SSCD without vacuum attempt (150).

Almost a third of participants were adolescents, two-thirds were nulliparous, and around a quarter were HIV-positive. There was a significant difference for previous CD, with 47 (11%) of the VAD attempt group and 29 (19%) of the SSCD group having had a previous CD (p = 0.01). Level of clinician for decision-making differed across groups with community service medical officers more likely to opt for a SSCD (39%) compared with VAD (21%); medical officers (junior and senior) more likely to opt for a VAD attempt (17% and 21%) compared with SSCD (10% and 14%); and family medicine registrars and family physicians equally as likely to opt for a VAD attempt (24% and 15%) or SSCD (23% and 15%) (p = 0.01). Decision-to-delivery interval was shorter in the VAD attempt group, with a median delivery time of 8 min compared with 49 min in the SSCD without vacuum attempt group (p < 0.01). Prolonged second stage was more likely the indication for intervention in the SSCD group (62%) compared with the VAD attempt group (51%), while foetal distress was more likely in the VAD attempt group (40%) compared with the SSCD group (33%) (p = 0.01). Other clinical and sociodemographic characteristics are shown in Table 1.

Maternal outcomes between the two groups are shown in Table 2. Women who underwent a VAD attempt had a 33% decrease in odds of an adverse maternal outcome (OR 0.67, 95% CI 0.42–1.06; p = 0.09). There were statistically significant decreased odds of secondary outcomes of PPH, need for blood transfusion, need for transfer, need for further surgery, and prolonged hospital stay in the VAD attempt group. Major perineal injury only occurred in the VAD attempt group.

Perinatal outcomes between the two groups are shown in Table 3. There was no difference for the primary outcome of adverse perinatal outcome composite (OR 0.92, 95% CI 0.48–1.75, p = 0.8). There were also no statistically significant differences for secondary outcomes of need for admission, need for transfer, major birth trauma, neonatal seizures, low Apgar score, foetal death after decision to deliver made, and early neonatal death.

Table 4 presents the findings of the multivariable logistic regression analysis for the principal study outcomes. Significant covariates from the initial analysis were previous CD, maternal age, parity, number of antenatal care (ANC) visits, gestation age (GA) at booking and delivery, maternal body mass index (BMI), HIV status, maternal hypertensive disorder, level of clinician for decision-making, birth weight, and indication for second stage intervention. After adjusting for these covariates, women who underwent VAD attempt had a 70% decrease in odds of having a PPH (OR 0.30, 95% CI 0.15–0.60) and a 36% decrease in odds of a composite adverse maternal outcome (OR 0.64, 95% CI 0.39–1.05). There were no statistically significant differences for adverse perinatal outcome composite (OR 0.82, 95% CI 0.42–1.60) and low Apgar score (OR 1.22, 95% CI 0.43–3.50).

This study showed that for women who required intervention in the second stage of labour, there was a reduction in the odds of an adverse maternal outcome after VAD attempt as compared with SSCD. Although the composite adverse maternal outcome did not reach conventional statistical significance, several individual maternal outcomes were significantly less frequent in the VAD attempt group. This reflects the heterogeneity of events within the composite outcome, where outcomes of differing frequency and clinical severity (some favouring VAD attempts and others occurring only in that group) may dilute overall statistical significance despite clinically meaningful differences in key components. Specifically, major perineal injury (which is non-life-threatening) was disproportionately overrepresented as a severe adverse outcome in the VAD attempt group. In contrast, more severe life-threatening complications such as PPH, need for transfer to a tertiary centre and need for further surgery contributed more significantly towards adverse outcomes in the SSCD group. In addition, this study focused on short-term maternal and perinatal outcomes only. This is important to note, as existing literature has shown that CD (particularly when performed in the second stage of labour) is associated with clinically relevant long-term risks, including abnormal placentation, uterine rupture, hysterectomy and increased maternal morbidity in subsequent pregnancies.^3,10,20 These longer-term outcomes were not assessed in the present study and warrant further investigation, particularly in rural district hospital settings. Lastly, this study included cases of failed VAD and subsequent CD in the VAD attempt group, as well as cases of foetal distress, which differs from similar research conducted in Nigeria (which excluded interventions due to foetal distress).¹²

This study showed that perinatal outcomes of babies delivered after a VAD attempt were equivalent to those following SSCD. These findings are consistent with studies conducted between 2017 and 2020 in tertiary hospitals in Uganda and Nigeria, which similarly reported comparable neonatal outcomes under specialist obstetric care.^12,14 Although the overall use of VAD has declined over time, these studies were broadly contemporaneous with the current study and reflect clinical practice within the same era. In contrast, a 2017 study from Israel reported poorer neonatal outcomes following SSCD,¹³ highlighting potential contextual differences in practice and case selection.

The sample size of this study was similar to studies conducted in Nigeria and Uganda and larger than studies conducted in Israel and Turkey.^12,13,14,21 The CD rate of 22% in this study is comparable to studies conducted in Turkey and Nigeria, which reported CD rates of 18% and 20%, respectively.^12,21 This rate is higher than those observed in some LMICs, which range from 16% to 18%.⁷ However, it is lower than the rates reported in Uganda and Israel of 31% and 29%, respectively.^13,14 The management of many lower-risk and some higher-risk pregnancies at this rural facility likely contributed to some of the similarities and differences in sample size and CD rate. The VAD attempt rate of 4.8% was higher than similar research from Uganda, Nigeria, Turkey, and other LMICs, ranging from 0.5% to 3.3%.^7,12,14,21 No accurate rates for South Africa are available, but rates in this study are above the 0.5% – 1% estimates.⁸ The failure rate of 16.9% for VAD attempts was higher than research from Nigeria, Uganda, and global averages of 5.9%, 9.2%, and 13.9%, respectively.^7,12,14 This, combined with the relatively high VAD attempt rate, suggests the regular use of VAD attempts as a second-stage intervention in this group and the acceptance of failure as not enough of an adverse event in itself to discourage its use.

Baseline clinical characteristics were similar across the two groups and comparable to national averages, suggesting a representative and comparable sample.²⁰ Low rates of ANC attendance likely reflect broader structural barriers to healthcare access in rural South Africa, including socioeconomic disadvantage, lower educational attainment, rural residence, and transport limitations, all of which have been shown to reduce ANC utilisation.^4,22,23 Nulliparity is a recognised risk factor for complications in the second stage of labour, and the proportion of nulliparous women in this cohort is consistent with available literature.²⁴ Higher rates of previous CD in the SSCD group are consistent with previous literature identifying this as a risk factor for complications in subsequent pregnancies.⁷ It may have introduced confounding in increasing the odds of adverse outcomes in the SSCD group as well as impacted clinician decision-making between VAD attempt and SSCD. The DDI in this study was shorter for both the VAD attempt and SSCD groups as compared with other literature in Nigeria and Uganda.^12,14 Shorter DDI in the VAD attempt group likely contributed significantly towards the maternal and perinatal outcomes. This is because indications for intervention in the second stage often involve time-sensitive contributing factors, for example, maternal haemodynamic stability and foetal hypoxia.⁸ The intention-to-treat analysis of the VAD attempt group helps inform the performing clinician that the odds of a shorter expected DDI at this facility remain regardless of whether the VAD attempt was successful or not.

Decision-making in labour is complex and impacted by clinician experience and available resources. Recent research showed that junior doctors in South Africa are not sufficiently equipped with the knowledge and experience to perform instrumental deliveries safely.²⁵ As a result, they are more likely to opt for a CD. However, data consistently suggest that maternal and perinatal outcomes from CDs performed by junior doctors in rural settings are poor.^20,26 Community service medical officers and medical officers constitute a significant portion of the staffing in rural district hospitals.²⁵ Their limited knowledge and experience likely contributed towards the differences in decision-making as compared with family medicine registrars and family physicians in this study. The similar decision-making patterns observed among family physicians and family medicine registrars (compared with other clinician categories) suggest a potential stabilising role for family medicine expertise in second-stage obstetric care in rural district hospitals, although this requires further investigation.

Rising CD rates in many countries have not been associated with meaningful improvements in maternal or perinatal outcomes.^8,27 In fact, they have been associated with adverse outcomes in current and future pregnancies, particularly in rural hospitals.^7,8,20 In addition, several studies in LMICs and upper-middle-income countries have shown that the use of VAD remains a safe and effective method of intervening in the second stage of labour.^{12,13,14,21,27} This approach does not compromise neonatal outcomes and may even improve maternal outcomes and the overall birthing experience.^14,28,29,30 Despite this, there is an ongoing aversion to VAD use and decreasing rates of use.^5,8 This study adds to the literature that the use of VAD is a safe alternative to SSCD, especially in low-resource settings and those with no obstetric specialist support.^{8,12,13,14,21}

Despite the value of this study, it is not without limitations. The use of retrospective medical record reviews introduces the possibility of information and selection bias. As an observational cohort study, the findings are also subject to residual confounding, including confounding by indication. In this study, indications for second-stage intervention differed between groups, with foetal distress more commonly prompting VAD attempts and prolonged second stage more frequently resulting in SSCD. These differing clinical contexts are likely to influence both decision-making and outcomes and may have contributed to the observed associations, despite adjustment for indication and other relevant covariates. In addition, this study only assessed short-term outcomes, lacking long-term follow-up data that would provide a more comprehensive understanding of the overall implications. Additionally, variations in baseline characteristics among the two groups, along with data collection being restricted to a single site, further limit the generalisability of these results to broader patient populations and varying clinical contexts. There is a need for a larger multicentre study with a longer follow-up period that explores various aspects of second-stage interventions, including the impact of family physicians and other clinicians. Investigating the complexities of decision-making during the second stage of labour in different healthcare facilities and among a range of clinicians, utilising both qualitative and quantitative research methodologies, could provide valuable insights into the factors contributing to the aversion to VAD. Such understanding may inform strategies to enhance the adoption of VAD practices and promote their reintegration within clinical protocols.

Vacuum-assisted delivery attempts, performed in a rural district hospital setting in South Africa by medical doctors with no obstetric specialist support, are safe alternatives to SSCD with improved maternal and equivalent perinatal short-term outcomes. Considering the adverse long-term outcomes of CD, including the risk for future pregnancies, these findings support safe use of VAD as an alternative second-stage intervention in appropriately selected cases within rural district hospitals.