Background: In low-resource settings, access to emergency cesarean section is associated with various delays leading to poor neonatal outcomes. In this study, we described the delays a mother faces when needing emergency cesarean delivery and assessed the effect of these delays on neonatal outcomes in Rwanda. Methods: This retrospective study included 441 neonates and their mothers who underwent emergency cesarean section in 2015 at three district hospitals in Rwanda. Four delays were measured: duration of labor prior to hospital admission, travel time from health center to district hospital, time from admission to surgical incision, and time from decision for emergency cesarean section to surgical incision. Neonatal outcomes were categorized as unfavorable (APGAR <7 at 5 min or death) and favorable (alive and APGAR ≥7 at 5 min). We assessed the relationship between each type of delay and neonatal outcomes using multivariate logistic regression. Results: In our study, 9.1% (40 out of 401) of neonates had an unfavorable outcome, 38.7% (108 out of 279) of neonates' mothers labored for 12-24 h before hospital admission, and 44.7% (159 of 356) of mothers were transferred from health centers that required 30-60 min of travel time to reach the district hospital. Furthermore, 48.1% (178 of 370) of cesarean sections started within 5 h after hospital admission and 85.2% (288 of 338) started more than 30 min after the decision for cesarean section was made. Neonatal outcomes were significantly worse among mothers with more than 90 min of travel time from the health center to the district hospital compared to mothers referred from health centers located on the same compound as the hospital (aOR = 5.12, p = 0.02). Neonates with cesarean deliveries starting more than 30 min after decision for cesarean section had better outcomes than those starting immediately (aOR = 0.32, p = 0.04). Conclusions: Longer travel time between health center and district hospital was associated with poor neonatal outcomes, highlighting a need to decrease barriers to accessing emergency maternal services. However, longer decision to incision interval posed less risk for adverse neonatal outcome. While this could indicate thorough pre-operative interventions including triage and resuscitation, this relationship should be studied prospectively in the future.
This retrospective cross sectional study was conducted at three rural district hospitals (Butaro, Kirehe and Rwinkwavu) in Rwanda. These district hospitals are under the management of the Rwandan Ministry of Health (MOH) with technical and operational support provided by an international non-governmental organization, Partners In Health, known locally as Inshuti Mu Buzima (PIH/IMB). These district hospitals receive patients referred from 46 health centers in their catchment area. At each of the district hospitals, there are two operating rooms where cesarean deliveries can be performed and most cesarean deliveries are conducted by general practitioners. In Rwanda, pregnant women and their families are responsible for their transport from home to the health center in their catchment area. Sometimes community health workers accompany laboring women to the health center. At the health center, registered nurses or midwives manage the woman. In the event of a potential complication for the fetus or the woman, the nurses/midwives will transfer the woman to the nearest district hospital via ambulance or with transport arranged by the woman’s family if no ambulance is available. Each district hospital has five ambulances for all emergencies, while only 11 out of the 46 health centers in the districts of study have ambulances. The rest of the health centers depend on ambulances from the district hospital for emergency transport. Most of the roads in these rural settings are unpaved and public transport has limited reliability. In cases where the health center is located in the same compound as the district hospital, an ambulance is not necessary for the transport. On rare occasions, laboring women present directly to the district hospital. At the district hospital, registered nurses or midwives perform initial obstetrical review and call general practitioners for further laboratory and ultrasound assessments, and decision for cesarean delivery. If cesarean delivery is decided, the pregnant woman is transferred to a delivery room where nurse-anesthetists administer anesthesia and the general practitioner, supported by nurses or midwives, performs cesarean section. Each hospital assists with 10–12 deliveries per day. During the day, one general practitioner is assigned to the delivery room; at night, one general practitioner covers the entirety of the maternity department. At all times, two nurses or midwives are assigned to the delivery room. After delivery, the woman and the neonate are transferred to the post-partum room where midwives closely monitor their progress with daily check-ups by the general practitioners and until discharge. Obstetricians are sometimes but not always available at the district hospitals. The decision to seek care at a health facility depends on the ability to pay or having a valid health insurance [16]. The majority of Rwandan households (79%) have at least one member with health insurance [16]. For women who have community-based health insurance (97% of households with insurance [16]), the insurance covers 90% of the total cost of care and the woman pays the remaining balance. Women without health insurance cover entirety of the cost out of pocket. The study included mothers and their neonates born via emergency cesarean section between 01 January 2015 and 31 December 2015 at the three hospitals. All emergency cesarean sections without intrauterine fetal death prior to the decision for cesarean section were eligible for inclusion. Because of the large population of cesarean deliveries at the district hospitals, we calculated a study sample size that could detect differences in neonatal outcomes by delays. Due to limited data a priori, we assumed that the overall negative neonatal outcome was 50% and that delay categories were binary to calculate the largest sample size possible. To have a 90% power to detect a 20% difference in neonatal outcome (i.e. one group had 40% negative neonatal outcome and the second group had 60% negative neonatal outcome), with α=0.05, we calculated a necessary sample of 140 women per group (delayed or not delayed). In anticipation of missing data, we increased the sample size by 30% to 200 per group or 400 women total. We further increased the sample size to 600 women, 200 per hospital, to account for any losses in statistical efficiency due to stratification by hospital and differences in group sizes. As the majority of cesarean deliveries at these hospitals were due to the referral of emergent and complicated cases from health centers, we originally assumed all cesarean deliveries were emergency and used segmented sampling to sample 200 women in each hospital. However, if a woman’s chart did not meet the inclusion criteria, that is if the cesarean delivery was indicated as elective or the infant died before the decision for cesarean section, we removed and replaced the chart with an eligible chart, randomly picked from the same month of admission. Among the sampled cesarean section deliveries, we performed in-depth review of indications for cesarean section. During the data cleaning stage, we excluded any deliveries whose indication for cesarean section was not emergent. These included women whose sole indication for cesarean delivery was previous cesarean delivery, women who were past their due date, or women who were having twins without any other emergent indication. Included emergency cesarean deliveries had at least one of the following indications: cord prolapse, uterine rupture with a previous cesarean section scar, fetal distress, eclampsia, obstructed labor, mal-presentation, cephalopelvic disproportion, prolonged rupture of membranes with a previous cesarean section scar, and pre-eclampsia [19, 20]. Finally, any neonate who did not have record of neonatal outcome, either death within the first 24 h or APGAR at 5 min after delivery was excluded from analysis. Trained data collectors extracted chart data on women’s demographic and clinical characteristics, individual and health facility delays, and neonatal outcomes. We categorized the indication for cesarean delivery as very severe or severe through consultation with a local obstetrician and based on complications to the fetus. Very severe indications included fetal distress, cord prolapse, uterine rupture, eclampsia, abruption placenta, and obstructed labor. Severe indications included prolonged rupture of membranes, pre-eclampsia, placenta previa, cephalopelvic disproportion, and mal-presentation, including breech presentation, transverse presentation, and occiput posterior presentation. If a woman had more than one indication, we prioritized the most severe indication. A neonatal outcome was considered favorable if the neonate was alive and had an APGAR score ≥ 7 at 5 min and unfavorable if there was death within 24 h after delivery or an APGAR score < 7 at 5 min. We captured four measures of delays in reaching and receiving care that were routinely collected in the charts (Fig. (Fig.1).1). We used two measures for delays in reaching care. First, we looked at the duration of labor prior to district hospital admission, as self-reported by the women. We also used the average ambulance travel time from the health center, where an emergency condition occurred or could have been detected, to the district hospital. The latter was a standardized time based on health center records and did not include the actual time from when the emergency was detected until the ambulance was called or the actual travel time for the mother in case an ambulance was not available. We used three cutoffs for this measure: less than 30 min (indicating that the health center was located on the same compound as the district hospital), 30–60 min, and more than 60 min. Delays in receiving care were measured by the time from admission to the district hospital to surgery start and the time from the decision for a cesarean section by a doctor to surgical incision (decision-to-incision interval). Possible delays for emergency cesarean section We assessed the relationship between neonatal outcome and potential confounders using Chi-squared tests. Types of delays were described using frequencies. We assessed the relationship between each of the delays and the neonatal outcome using multivariate logistic regression, controlling for potential confounders. We developed a separate model for each delay predictor. Potential confounders considered for this study were mother’s age, gestational age, number of fetuses, woman’s heart rate, fetal heart rate, district hospital, history of prior pregnancy, and severity of indication. To determine the appropriate confounders for each delay and neonatal outcome, i.e. the specific factors that were potentially related to the delay and the outcome but were not on the causal pathway, we constructed directed acyclic graphs (DAGs) (Fig. (Fig.2).2). For the duration of labor prior to admission at the district hospital, we controlled for district hospital, gestational age, number of fetuses, history of prior pregnancy, and woman’s age as potential confounders. For the travel time from health center to district hospital, we controlled for district hospital as a potential confounder. For the admission to surgical incision and decision to incision intervals, we controlled for district hospital, gestational age, number of fetuses, woman’s age, woman’s heart rate, fetal heart rate, and severity of cesarean section indication. We reported the resulting odds ratios (OR), adjusted odds ratios (aOR), 95% confidence intervals (95% CIs), and p-values. Directed acyclic graphs (DAGs) mapping the relationship between the four possible delays and neonatal outcomes, noting potential confounders For bivariate analyses, we excluded all observations with missing data. We also excluded observations if delay for that individual was not measured; however, if a confounder was included in a multivariate analysis, we created a missing category for that confounder if more than 15% of patients had missing data in order to avoid excluding these patients from the analysis. We used α = 0.05 significance level for all analyses and completed all analyses using Stata v14 (College Station, TX: StataCorp LP).
N/A