Case review of perinatal deaths at hospitals in Kigali, Rwanda: Perinatal audit with application of a three-delays analysis

Background: Perinatal audit and the three-delays model are increasingly being employed to analyse barriers to perinatal health, at both community and facility level. Using these approaches, our aim was to assess factors that could contribute to perinatal mortality and potentially avoidable deaths at Rwandan hospitals. Methods: Perinatal audits were carried out at two main urban hospitals, one at district level and the other at tertiary level, in Kigali, Rwanda, from July 2012 to May 2013. Stillbirths and early neonatal deaths occurring after 22 completed weeks of gestation or more, or weighing at least 500 g, were included in the study. Factors contributing to mortality and potentially avoidable deaths, considering the local resources and feasibility, were identified using a three-delays model. Results: Out of 8424 births, there were 269 perinatal deaths (106 macerated stillbirths, 63 fresh stillbirths, 100 early neonatal deaths) corresponding to a stillbirth rate of 20/1000 births and a perinatal mortality rate of 32/1000 births. In total, 250 perinatal deaths were available for audit. Factors contributing to mortality were ascertained for 79% of deaths. Delay in care-seeking was identified in 39% of deaths, delay in arriving at the health facility in 10%, and provision of suboptimal care at the health facility in 37%. Delay in seeking adequate care was commonly characterized by difficulties in recognising or reporting pregnancy-related danger signs. Lack of money was the major cause of delay in reaching a health facility. Delay in referrals, diagnosis and management of emergency obstetric cases were the most prominent contributors affecting the provision of appropriate and timely care by healthcare providers. Half of the perinatal deaths were judged to be potentially avoidable and 70% of these were fresh stillbirths and early neonatal deaths. Conclusions: Factors contributing to delays underlying perinatal mortality were identified in more than three-quarters of deaths. Half of the perinatal deaths were considered likely to be preventable and mainly related to modifiable maternal inadequate health-seeking behaviours and intrapartum suboptimal care. Strengthening the current roadmap strategy for accelerating the reduction of maternal and neonatal morbidity and mortality is needed for improved perinatal survival. The study adopted a perinatal audit design [4]. The audit was conducted prospectively and was related to the cases of perinatal deaths that had occurred among babies born in two urban hospitals. This study was part of a programme that was aimed at analysing the causes and determinants of perinatal mortality at a district hospital (DH) and a tertiary referral hospital (TRH) from 18 July 2012 to 8 May 2013. These two hospitals, located in Kigali, Rwanda, assisted in almost 7900 and 2200 deliveries in 2012, respectively. The DH conducts the most deliveries in the country, whereas the TRH serves as the largest referral and University teaching hospital. During the study period, the midwifery, nursing and medical teams working in the maternity and neonatal units included 79 midwives and nurses, 14 general practitioners, 3 obstetricians, and 2 paediatricians in the DH, whereas there were 78 midwives and nurses, 4 obstetricians, 2 paediatricians, 5 residents in paediatrics and 15 in obstetrics and gynaecology in the TRH. Most patients received in the two hospitals lived in the capital, which had approximately 1 200 000 inhabitants in 2012. Other patients came from the surrounding areas of the city or were referred from other urban or rural health facilities. Patients attending both hospitals were mainly covered by the community health insurance system, as shown in our previous study [17]. Patients usually followed referral steps starting from public health centres or posts, then to district hospitals, and finally, to tertiary level hospitals. Those who bypassed these referral pathways were not covered by the community health insurance system, except in emergencies. Other existing health insurance systems cover civil servants, the military, and people working in the private sector. These insurance systems frequently covered patients who attended the TRH. Patients who had no insurance cover needed to pay all hospital charges themselves. Stillbirths and early neonatal deaths, which together constitute perinatal deaths, among births at the two hospitals were consecutively included in the study. A stillbirth was defined as a fetal death after at least 22 completed weeks of gestation or weighing 500 g or more at birth. An early neonatal death was defined as a live birth dying within the first seven days of life, the early neonatal period, and born after at least 22 completed weeks of gestation or weighing 500 g or more at birth [18]. Deaths that occurred among babies born outside the study sites were excluded from the study. In each hospital, one nurse, one midwife, and one doctor were recruited and trained over two days on the study’s aim and data collection procedures, and how to conduct interviews with bereaved mothers and healthcare providers involved in the study using a questionnaire containing both closed and open-ended questions. The first part of the questionnaire included information on the socioeconomic characteristics of the mothers and their households, such as household assets, maternal education and place of residence, and health insurance coverage, as well as the demographic characteristics of the mothers and their babies, such as parity, maternal and gestational age, birth weight, and sex. The second part included data on routine antenatal screening and care, pregnancy complications and their management, characteristics of labour and delivery, condition of the baby at birth (alive or stillbirth) and afterwards, Apgar score, management of the newborn’s illness that led to a fatal outcome, and time and causes of death. A fetus that died during labour or delivery was defined as a fresh stillbirth. These stillbirths were also labelled intrapartum-related stillbirths as most of them were due to intrapartum-related insults or injuries [19]. When a death occurred before the onset of labour or delivery with signs of a degenerative process, the fetus was classified as a macerated stillbirth [18]. The time of fetal death in relation to admission was also recorded. The third part of the questionnaire comprised information obtained from interviews with mothers and healthcare providers. Mothers were encouraged to describe the chain of events leading up to the death, including both clinical and social elements, such as reasons for seeking care at a health facility, the onset and development of symptoms, actions taken, barriers encountered when seeking care, and the clinical attention received at the health facility. Important dates and time-related events were recorded. When mothers were not available, information was sought from their partners or closest relatives. Healthcare providers also provided additional narratives about the circumstances that led to the death. Data collected through interviews were systematically cross-checked against the information extracted from clinical records. When discrepancies were noted, interviews were repeated or other information sources were re-checked. Random repeat interviews were regularly performed to reinforce the quality of the data. Narratives summarizing the information collated from the different sources were prepared for each case of death by the main investigator. The actual data collection was preceded by one month of piloting of the study procedures. The main investigator rigorously supervised the fieldwork, assisted by one doctor from each hospital. Members of the audit committee were recruited from the hospitals and included two obstetricians, two paediatricians, and the main investigator, who is also a paediatrician. The main investigator introduced the study objectives and procedures at the first meeting. He also presented all cases of perinatal deaths that were discussed in each session, which was scheduled according to the availability of the members. These presentations were based on the individual narratives and were followed by open discussions to identify the causes of the deaths, the underlying factors, and any potentially avoidable deaths. A single cause of death was assigned to each case. The avoidable factors were classified based on the three-delays model [8]. In some instances, several delays were identified for an individual case. The discussions were aimed at evaluating whether the delay had possibly or likely contributed to the fatal outcome. This process was consistent with previous evaluation studies completed elsewhere [4]. The quality of care provided to mothers and neonates was judged against evidence-based practices, as expressed in local guidelines and protocols or in the scientific literature [20]. The adequacy of care was also related to feasibility and the availability of resources and opportunities, such as any investigations in making a diagnosis, and the availability of drugs, consumables, equipment, and infrastructure. A death was considered avoidable if improved management using available resources and opportunities would have altered the outcome. This approach has been applied in other reports on perinatal audit in low- and middle-income countries [21, 22]. In addition, the timing of death in relation to the mother’s admission to the hospital provided important orientation regarding the prevention of death. Audible fetal heartbeat on admission increased the likelihood that death would have been avoided. The auditors discussed each case to reach a consensus. In case of disagreement, all sources of information were rechecked and conclusions were made during the subsequent meeting. The senior obstetrician had the final word in matters of pregnancy, labour, and delivery-related events, while the paediatrician made the final decision regarding neonatal care and mortality [23]. The socio-demographic and clinical characteristics of the mothers and their babies included in the study were summarised in descriptive tables. Based on the individual narrative stories for each case of death, the auditors assessed the contributing factors that could be attributed to the three delays as well as to potentially avoidable deaths. Factors contributing to the same phase of delay were grouped together and listed in descriptive tables. Each phase of delay was illustrated by a case study that described barriers to access adequate maternal and neonatal care. Potentially avoidable deaths and their frequencies were described. IBM SPSS Statistics version 20 (IBM Corporation, Armonk, NY) was used for descriptive statistics. The Rwanda National Ethics Committee granted ethical approval for this study (ethics approval no. 086 RNEC/2012, Kigali, Rwanda).