Prospective cohort study comparing outcomes between vacuum extraction and second‐stage cesarean delivery at a Ugandan tertiary referral hospital

To compare maternal and perinatal outcomes between vacuum extraction and second‐stage cesarean delivery (SSCD).


| INTRODUCTION
With 275 288 maternal deaths, 2.1 million stillbirths, and 2.0 million early neonatal deaths recorded worldwide in 2015, maternal and perinatal mortality are global health priorities. 1,2 Most maternal and perinatal deaths occur in low-income and middle-income countries (LMICs). 1,2 Vacuum extraction is an evidence-based intervention that is used to shorten the second stage of labor. Indications for this approach include fetal distress, prolonged second stage of labor, maternal exhaustion, or the need to avoid expulsive efforts among women with conditions such as heart failure or severe anemia. 3,4 Although vacuum extraction can reduce maternal mortality from hemorrhage and sepsis-as well as perinatal mortality from birth asphyxia-use of this method has almost disappeared from obstetric practice in many LMICs. [5][6][7][8] One study 6 found that instrumental vaginal delivery was not used in almost half of 1728 Sub-Saharan African hospitals, with usage rates below 1% in the remaining centers. Reasons for this deficit include lack of functioning equipment, lack of trained personnel, staff perceptions regarding trauma to the fetus, and fear of mother-to-child transmission of HIV. [6][7][8] Consequently, many women in LMICs undergo avoidable cesarean delivery. 9 The use of cesarean delivery, especially when performed during the second stage of labor, increases the risks of hemorrhage and infection, which are two of the main drivers of global maternal mortality. [10][11][12] In addition, a scarred uterus is a risk factor for uterine rupture and abnormally invasive placenta in subsequent pregnancies. 13 These risks are particularly high in low-resource settings, where many deliveries happen outside healthcare facilities; access to safe surgery and anesthesia cannot be taken for granted; blood for transfusion is in short supply; and fertility rates are high. 14,15 Therefore, it is crucial that unnecessary cesarean delivery is avoided. [16][17][18] Published literature regarding outcomes of vacuum extraction among LMICs is scarce. Most studies lack follow-up, and vacuum extraction was not compared with alternative management options. [19][20][21] In 2012, vacuum extraction was reimplemented in the main tertiary hospital in Uganda (Mulago National Referral Hospital, Kampala). This initiative led to declines in intrapartum stillbirths and uterine ruptures of 24% and 26%, respectively. 7 The use of vacuum extraction was hypothesized to reduce maternal morbidity, perinatal morbidity, and the decision-to-delivery interval (DDI) when compared with second-stage cesarean delivery (SSCD). 12 The aim of the present study was to test this hypothesis among pregnant women attending Mulago National Referral Hospital.

| MATERIALS AND METHODS
The present prospective observational cohort study was conducted among women undergoing vacuum extraction or SSCD in the main labor ward of Mulago National Referral Hospital between November 25, 2014, and July 8, 2015. Women with a term singleton pregnancy in vertex presentation who delivered by vacuum extraction or SSCD at the study center were included, as were those who developed a ruptured uterus while in the second stage of labor and waiting for the intervention. Women who experienced a ruptured uterus before the decision for intervention (vacuum extraction or SSCD) were excluded. Women who experienced intrauterine fetal death (IUFD) before the decision for intervention were excluded from the analysis of perinatal outcomes. Ethical approval for the present study was obtained from the Mulago National Referral Hospital Research and Ethics Committee (MREC 489) and the Uganda National Council for Science and Technology, Kampala, Uganda (HS1752).
Women provided written informed consent for their participation.
Mulago National Referral Hospital is a university teaching and government hospital with 2700 beds and greater than 31 000 deliveries recorded annually. It is the main training center for midwives, medical doctors, and obstetricians in the country. The maternity unit has an operating theater, which is accessible 24 hours per day. Vacuum extraction and cesarean delivery are performed by residents (50 trainee obstetricians at the center, with 5-7 on labor wards) with or without supervision, depending on experience, and specialist obstetricians (40 at the center, with 1-3 on labor wards). All doctors are trained in performing vacuum extraction and cesarean delivery; however, cesarean delivery is undertaken more frequently than vacuum extraction (approximately 20 cesarean deliveries per day compared with one or two vacuum extractions per day at the study center). Although vacuum extraction is used regularly, and the hospital has a protocol with indications, the decision regarding mode of delivery depends not only on clinical factors but also on the doctor's personal preference and expertise, as well as the availability of theater and vacuum equipment. Many women undergoing cesarean delivery at Mulago National Referral Hospital could be eligible for vacuum extraction. 7,12 The vacuum equipment used at this center comprises Kiwi vacuum extractors (Clinical Innovations, South Murray, Utah, USA), Bird and silicone cups, with hand and foot pumps. Forceps are available, but rarely used, as is the case in many hospitals in LMICs. 6 Spinal anesthesia during cesarean delivery is provided by anesthetic nurses or anesthesiologists.
An obstetric high dependency unit is available where women are monitored and given oxygen. The hospital has a general intensive care unit, with mechanical ventilation. There is a blood bank; however, the availability of blood for transfusion is limited. Fetal monitoring occurs using a Pinard fetoscope or handheld Doppler machine. The neonatology ward has incubators, phototherapy, and continuous positive airway pressure, but no mechanical ventilation. Most women come from Kampala and the surrounding area, although some have travelled for a day to attend the hospital. Delivery is free of charge, except in the private ward.
Within 24 hours of delivery, a member of the research team identified women with vacuum extraction from the delivery book. Women who underwent cesarean delivery were identified from the theater register and their medical records examined to identify those who had a fully dilated cervix at the time of decision for cesarean delivery. Eligible women were asked to participate in the present study on the day after delivery.
Data were extracted from the participants' medical records.
Indications for SSCD and vacuum extraction were classified as "delay", "fetal distress", "maternal", and "other" (Table S1). The women were also interviewed using structured questionnaires (File S1). Data were extracted from medical records and the admission, discharge, and mortality registers for neonates admitted to the neonatology unit.
Follow-up consultations occurred at 6 weeks and 6 months after delivery. During these visits, women were interviewed using semistructured questionnaires (File S1). Neonates were weighed and assessed according to the neurodevelopmental scoring chart of Van Wiechen. 22 Verbal autopsy forms were used to determine the cause of any neonatal deaths that occurred after hospital discharge. 23 Women who missed the postnatal consultations were interviewed by telephone using the same questionnaire; however, questions about HIV status were omitted for reasons of privacy.
The primary maternal outcomes were death and a composite of severe maternal outcomes, defined as death, uterine rupture, hysterectomy, or relaparotomy (Table S1). Secondary maternal outcomes were postpartum hemorrhage (PPH), infection, genital tract injury, and duration of hospital admission.
Primary perinatal outcomes were death after the decision for a second stage intervention and a composite severe perinatal outcome, which was defined as death, severe birth trauma, convulsions, or a 5-minute Apgar score below four. Secondary outcomes were admission to the neonatology unit, duration of admission, and diagnosis.
Outcomes assessed during follow-up were neonatal or infant death after discharge, and neurodevelopment anomalies.
Sample size calculations are shown in Table S2. Failed vacuum extraction with subsequent cesarean (or forceps) delivery was analyzed as part of the outcome of vacuum extraction, as this was the intended mode of delivery.
Baseline characteristics were reported as numbers with percentages, with P values calculated using a two-sided χ 2 test. However, a two-sided Fisher exact test was used for outcomes recorded fewer than 10 times.
Outcome parameters were reported as numbers with percentages, P values, unadjusted (univariate) odds ratios (ORs) and, for primary outcomes, adjusted (multivariate) ORs with 95% confidence intervals (CIs). A multivariate logistic regression model to calculate adjusted ORs (aORs) was constructed to adjust for potential confounders. Factors were tested one by one, stratified for mode of delivery, and included in the multivariate model based on differences in distribution and the strongest potential for confounding. The number needed to treat (NNT) was calculated for maternal death and the composite severe maternal outcome. A P value of less than 0.05 was considered to be statistically significant.

| RESULTS
Among the 13 152 deliveries recorded during the present study period, 369 (2.8%) women with a term vertex singleton underwent (trial of) vacuum extraction and 429 (3.3%) women with a term vertex singleton underwent SSCD. The inclusion process is outlined in Figure 1. The vacuum extraction and SSCD groups used to analyze maternal outcomes comprised 358 and 425 women, respectively.
F I G U R E 1 Flowchart of the study participation. Abbreviations: CD, cesarean delivery; SSCD, second stage cesarean delivery; IUFD, intrauterine fetal death. a Uterine rupture occurred before the decision to perform SSCD. b IUFD occurred before the decision to perform vacuum extraction or second-stage cesarean delivery. In all, 36 (9.5%) women experienced a failed vacuum extraction: 35 of these women delivered by cesarean delivery and one by forceps.
These 36 women were analyzed in the vacuum extraction group.
Women who experienced IUFD before the decision to intervene were excluded from the analysis of perinatal outcomes. Therefore, the vacuum extraction and SSCD groups used to analyze perinatal outcomes comprised 347 and 410 women, respectively.
Baseline characteristics of the participants are shown in Table 1.
More women in the SSCD group had a previous cesarean delivery, delivered a neonate weighing more than 4000 g, were in second stage of labor on admission, and had indication delay, fetal distress or impending uterine rupture. There were non-significant trends toward greater numbers of nulliparous women and women with HIV in the vacuum extraction group. Baseline data with the missing data included as a proportion are presented in Table S3.
Maternal outcomes at hospital discharge are shown in Table 2. In all, 5 (1.2%) maternal deaths during the first 6 weeks after delivery were found for the SSCD group; however, no maternal deaths were recorded in the vacuum extraction group. The difference in maternal mortality between the groups was not significant (P=0.066). Deaths in the SSCD group occurred among women who underwent the procedure for prolonged labor. The causes of death were complete spinal block with cardiac arrest (n=4) and complete spinal block with hypoxic brain damage (n=1). Contributing factors were PPH, infection, and aspiration pneumonia (Table S4). One woman in the vacuum extraction group died 5 months after delivery following an episode of fever; however, this event was unlikely to be related to mode of delivery.
As shown in Table 2, the composite severe maternal outcome was recorded among 3 of 358 (0.8%) women after vacuum extraction and 18 of 425 (4.2%) women after SSCD (OR 0.19, 95% CI 0.06-0.65). The aOR was 0.24 (95% CI 0.07-0.84; P=0.026) ( Table S5). The NNT to prevent one severe maternal adverse event during or after SSCD was 28 (95% CI 17-69) patients. The NNT to prevent one maternal death was 85 (95% CI 45-661). Among women with relevant data available, blood loss of at least 500 mL was more frequent in SSCD group (P<0.001), blood loss of at least 1000 mL did not differ (P=0.098), and number of blood transfusions did not differ (P>0.99). Hospital stay was shortened after vacuum extraction, with a duration of 0-2 days more common in the vacuum extraction group (P<0.001) and a duration of longer than 5 days more common in the SSCD group (P=0.001) ( Table 2). Maternal follow-up rates at 6 weeks after delivery were 79% for the vacuum extraction group and 87% for the SSCD group (Fig. S1). Maternal infection and urogenital tract injuries that had occurred after vacuum extraction or SSCD and were reported at the 6 week follow-up consultation are shown in Table 3 Neonates were admitted to the neonatology unit more frequently following vacuum extraction than following SSCD (P=0.048) ( Table 4).
Admissions to the neonatology unit for longer than 2 days were comparable between the groups (P=0.737), indicating that the "extra admissions" after vacuum extraction were usually for a short period.
Severe neonatal trauma was infrequent and occurred after six vacuum extractions versus three in SSCD (Table S8).
At 6-month follow-up, two (out of six) infants that had experienced severe trauma after vacuum extraction had died, whereas four had developed normally (three according to examination during follow-up visit using the scoring chart of Van Wiechen 22 ) and one (who had dislocation of a leg) according to maternal report by telephone (Table S8). After SSCD, one (out of three) infant had died, one was lost to follow-up, and one showed developmental anomalies suggestive of brain damage (Table S8). Among the 74 perinatal deaths that occurred during admission (regardless of mode of delivery), 68 (91.9%) had birth asphyxia as the only identifiable cause of death (Table S9).
DDI data are outlined in More than one adverse event could apply. c ORs and 95% CIs were calculated using univariate logistic regression analysis. Calculations of adjusted OR's are shown in Table S5. d P values were calculated using a two-sided χ 2 test. However, a two-sided Fisher exact test was used for outcomes recorded fewer than 10 times. The cut-off for statistical significance was P<0.05. e Death, uterine rupture, hysterectomy, or relaparotomy. f Relaparotomy after cesarean delivery or laparotomy after vacuum extraction.
As shown in Figure S2, neonatal and infant follow-up rates were 82% after vacuum extraction and 89% after SSCD. The rates at 6 months were 79% and 83%, respectively.

| DISCUSSION
The present study found fewer maternal complications after vacuum extraction than after SSCD, whereas perinatal outcomes were comparable for the two groups. Severe neonatal trauma and brain damage were infrequent regardless of the mode of delivery. The risk of severe maternal complications-including death-during or after SSCD was one per 24 women.
The present findings from Uganda were consistent with those from high-income countries, indicating that vacuum extraction is a safe intervention and that SSCD carries an increased risk of maternal adverse events. 3,5,12 Indeed, one study found maternal and neonatal mortality to be higher following cesarean delivery compared with vaginal, especially in African countries. 24 The present study found no maternal deaths after vacuum extraction but five after SSCD. Although this observation did not reach statistical significance, it is suggested here that this is highly relevant and probably not random. Anesthetic adverse events played an important role in this study (Table S4). All five women who died were  Values are given as number (percentage) unless indicated otherwise. b More than one adverse event could apply. c ORs and 95% CIs were calculated using univariate logistic regression analysis. d P values calculated using a two-sided χ 2 test. However, a two-sided Fisher exact test was used for outcomes recorded fewer than 10 times. The cut-off for statistical significance was P<0.05. e Infection of the operation wound or perineum. f Obstetric fistula in the vacuum extraction group occurred following failed vacuum extraction and subsequent cesarean delivery. g These operations were performed on re-admission and are not included in Table 2 In the present study, nearly half of the women with a previous cesarean delivery arrived at the hospital during the second stage of labor. This observation suggests that many women with scarred uteri attempted to deliver outside of hospitals. Birth asphyxia, rather than trauma, was the main cause of perinatal mortality. This finding calls for action to improve the quality of monitoring during labor, to prevent birth asphyxia. In all, 33 women had both IUFD and SSCD. The development of IUFD had been diagnosed before SSCD was planned among 15 women. One of these 15 women died and two sustained uterine rupture during DDI. A timely vacuum extraction or destructive operation could possibly have prevented these adverse outcomes.
In conclusion, it is of utmost importance that unnecessary SSCD is prevented whenever possible, and particularly in areas where the risks associated with cesarean delivery are high. Reintroduction of vacuum extraction is an important strategy to limit unnecessary cesarean delivery, reduce DDI, and prevent maternal and perinatal mortality and morbidity.

AUTHOR CONTRIBUTIONS
BN contributed to the design of the study, collection, analysis, and interpretation of the data, and writing the manuscript. FN contributed to the design of the study, collection and interpretation of the data, and revising the manuscript. JL and JB contributed to the design of the study, interpretation of the data, and revising the manuscript.
TvdA contributed to the interpretation of the data and revising the manuscript. JvR contributed to the design of the study, analysis and interpretation of the data, and revising the manuscript. All authors approved publication of the article.

ACKNOWLEDGMENTS
The present work was funded by the Otto Kranendonk Foundation of the Netherlands Society of Tropical Medicine and International Health.

CONFLICTS OF INTEREST
BN serves on an advisory committee for the development of a new vacuum extractor product by Clinical Innovations (South Murray, UT, USA). This company had no role in either the present study or the content of the manuscript. The authors have no other conflicts of interest.

SUPPORTING INFORMATION
Additional Supporting Information may be found online in the supporting information tab for this article.            File S1. Questionnaires used at inclusion and during follow-up.