Causes of perinatal death at a tertiary care hospital in Northern Tanzania 2000–2010: a registry based study

This study is based on data collected at KCMC hospital in Northern Tanzania. The hospital is a tertiary care and zonal referral hospital which serves about 10 million people from mainly four regions in Northern Tanzania, namely Kilimanjaro, Arusha, Tanga and Manyara. Being a tertiary referral hospital the KCMC labour ward receives normal deliveries as well as high risk mothers with maternal or obstetric complications referred at various stages of pregnancy or labour from Moshi urban area or from other health facilities in the Northern zone.

The KCMC obstetrics and gynaecology department has a team on call which includes one specialist or consultant obstetrician, one obstetric resident and one intern doctor, two anaesthesiologists and 3 midwives who take care of the department outside regular working hours for comprehensive emergency obstetrics and gynaecological care. The department has two operative theatres in labour ward for emergency caesarean sections. In the Kilimanjaro region 70% of all births take place at health facilities [18]. Around 50% of the deliveries at KCMC are from Moshi urban area. The caesarean section rate at the institution is about 33% [19].

Based on records from the birth registry linked to the neonatal registry from July 2000 to October 2010 [20], we established a cohort of births with birth weight 500 g or more. A total of 34087 births were recorded of which 158 (0.4%) the birth weights were either missing or below 500 g (Figure 1). Therefore, our study population was 33929 births with birth weight 500 g or more, of which 1958 died perinatally.

Information on all mothers who delivered at KCMC was obtained through a structured questionnaire and the mothers being interviewed within the first 24 hours after delivery. Informed consent was obtained from mothers prior to the interview. Information was also extracted from the antenatal care record cards. Detailed description of the data collection procedure and data collected for the birth registry and neonatal registry have been previously published [20, 21].

For stillbirths, time of death was recorded as before labour, during labour, or unknown. The status of the fetus was also recorded, whether it was a macerated stillbirth or fresh one. The information was also sought whether the fetus died before or after admission to labour ward. Reporting of early neonatal deaths included date of death, time of death (died within first 24 hours, died within first week), and up to three diagnoses of cause of death [20].

Early neonatal deaths include newborns that die during first week of life. We define perinatal mortality as stillbirth or early neonatal death with birth weight 500 grams or more [22]. Perinatal mortality rate (PNMR), stillbirth rate (SBR) and early neonatal mortality rate (ENMR), were calculated as follows: PNMR = (stillbirths + early neonatal deaths/total births) × 1000, SBR = (stillbirths/total births) × 1000 and ENMR = (early neonatal deaths/live births) × 1000.

Outcome was perinatal death, overall and according to cause of death. Causes of death were classified on the basis of maternal, obstetric, fetal and neonatal characteristics identified in the linked registry data, according to the NICE classification [23], with a mild modification of the unexplained asphyxia category (Table 1) based on our previous modification [20]. In a strictly hierarchical order, each stillborn or early neonatal death was classified into one of the 13 specific, mutually exclusive causes of death. For the two causes of death categories maternal disease and obstetric complications, we also investigated co-morbidity.

Main results were stratified according to referral status (mother referred for delivery due to medical condition yes/no). The following conditions recorded in the birth registry were considered; obstructed labour, malpresentation, prolonged labour, retained twin, fetal distress, cord prolapse, premature/prolonged rupture of membrane, abruption placenta, placenta previa, antepartum haemorrhage, ruptured uterus, preeclampsia, eclampsia, gestational or diabetic mellitus, hypertension, and malaria. Referral due to previous caesarean section without any of the complications above was not regarded a medical referral.

Data were analyzed using Statistical Package for Social Science (SPSS) program for Windows Version 19.0 (SPSS 19.0 Chicago Inc. III, USA). Descriptive measures such as mean, standard deviation, rate per 1000 and relative risk were calculated.

The protocol for this study was approved by Kilimanjaro Christian Medical college (KCM-College) research ethics committee, with certificate no. 333 of 15^th July 2010.

Overall, major causes of perinatal death were unexplained asphyxia (n=425, 12.5/1000), obstetric complications (n=303, 8.9/1000), maternal disease (n=287, 8.5/1000), unexplained antepartum stillbirths after 37 weeks of gestation (n= 219, 6.5/1000), and unexplained antepartum stillbirths before 37 weeks of gestation (n=184, 5.4/1000), (Table 3). In the large group of unexplained asphyxia, 236 (55.5%) were early neonatal deaths and 189 (44.5%) were intrapartum stillbirths. A further analysis of co-morbidities showed that obstructed/prolonged labour was present in more than 80% of the deaths in the obstetric complications category, and that preeclampsia/eclampsia was present in nearly 90% of the deaths in the maternal disease category.

Births to mothers referred for delivery due to medical conditions accounted for 19.1% of all births and 36% of all deaths. Perinatal mortality was 45.6 per 1000 in the non-referred group and 109 per 1000 in the referred group (RR 2.4). In the group of non-referred, unexplained asphyxia still was the most common cause of death, while deaths from obstetric complications and maternal disease were largely reduced. High relative risks for referred vs. non-referred group were observed for obstetric complications (38/1000 vs. 2.1/1000, RR= 18.1), placental abruption (RR 8/1000 vs. 0.8/1000, RR=10.0), and maternal disease (21.8/1000 vs. 5.5/1000, RR= 4.0).

Nearly ninety percent of the deaths in the maternal disease category were affected by preeclampsia/eclampsia. Inadequate screening and women’s unawareness of danger signs in pregnancy means that serious maternal complications may go unrecognized leading to a delay in seeking and receiving necessary delivery care. In the Tanzanian demographic and health survey 2004–05, only 47% of the women who attended antenatal clinics recalled having been informed about any danger sign during pregnancy [18]. In a recent study in Rufiji district in Tanzania, 42% of the women who attended antenatal clinics were not informed of any danger signs, whereas only 8.7% were informed of all seven danger signs (vaginal bleeding, severe headache/blurred vision, severe abdominal pain, swollen hands and face, fever, baby stopped or reduced movement, and excessive tiredness or breathlessness) [24]. The risk of perinatal death due to maternal disease was in particular high in the referred group (21.8/1000 vs. 5.5/1000, RR=4.0). Also deaths due to placental abruption, which is associated with hypertension in pregnancy [25] were more frequent in the referred group (8.0/1000 vs. 0.8/1000, RR=10.0). Screening as well as raised awareness of signs of hypertensive disease in pregnancy among the women and their families could improve pregnancy outcome in this study setting. A study in rural Tanzania indicated that only 9% of all women screened for hypertension by health workers were informed about their blood pressure results [26]. Furthermore, health workers were able to detect high blood pressure level only in four out of the 12 patients with elevated blood pressure levels, of which only one received appropriate counseling. Lack of providers’ ability to screen, counsel and inform women on danger signs seen in rural Tanzania may be present on our area, too. Other threats to improved care are lack of transport, financial constraints, poor compliance with referral, and lack of birth preparedness [27, 28].

Unexplained antepartum stillbirths accounted for around half of all antepartum stillbirths. These are deaths largely linked to maternal health conditions before pregnancy, complications of pregnancy, such as preeclampsia, and placental dysfunction, without being able to establish the cause [1]. Antepartum stillbirths are also largely linked to maternal infection and fetal growth restriction [29]. Maternal and fetal infections are estimated to cause about 10-25% of stillbirths in high income countries, whereas, the rate is expected to be higher in low income countries [30]. Studies in low income countries found maternal infections such as syphilis, malaria and HIV are associated with high risk for perinatal morbidity and mortality [31‐34]; but infection is not included among the causes of death in the NICE classification. Furthermore, the available information may not be detailed enough to identify the underlying causes for the large fraction of unexplained deaths [13].

In consistence with previous studies from Tanzania [5, 7] and other African countries [34‐36], obstructed/prolonged labour was present in a majority of the deaths due to obstetric complications (80%). Worldwide, obstructed/prolonged labour is the most frequent obstetric complication leading to perinatal mortality [12]. A study in Western Tanzania found obstructed/prolonged labour accounted for 18.5% of perinatal deaths [5]. The risk of perinatal death associated with obstructed labour was 2-fold in Uganda [35], and 8-fold in Kenya [36]. Obstetric complication was the category with the largest relative difference between births to non-referred vs. referred women (38.0/1000 vs. 2.1/1000, RR=18.1). It is likely that delay in referral, transportation problems, and delay in taking action for emergency care are responsible for a large fraction of these deaths in the referred group [5, 9].

Unexplained asphyxia includes intrapartum related stillbirths and intrapartum related neonatal deaths, and was the largest category among births to non-referred mothers. These are deaths closely linked to obstetric and maternal complications and services given during labour and delivery [11, 12]; hence they are largely avoidable by appropriate care during time of labour and delivery. The majority of intrapartum stillbirths and early neonatal deaths with Apgar score below 7 at 5 minutes were babies of normal birth weight. These are viable fetuses and babies who would have survived with proper care during labour and delivery. We have previously shown that birth asphyxia was the leading cause of deaths in a neonatal care unit, and that most babies who died due to birth asphyxia had normal birth weight and were born at term [20]. Among interventions recommended to reduce intrapartum stillbirths and deaths related to asphyxia are; appropriate care at birth, the use of partograph for monitoring labour and proper resuscitation for asphyxiated newborns [37]. A qualitative audit study in Tanzania found lack of appropriate fetal heart monitoring during labour in more than 40% of the perinatal deaths [9].

A major strength of the study is that data is retrieved from a birth registry where the information is collected on a daily basis including holidays with use of a standardized questionnaire. The birth registry is linked to a neonatal registry at the neonatal intensive care unit at the same hospital, where causes of death are routinely registered in a specially designed form. By means of the NICE classification we were able to identify cause of death in more than 50% of the stillbirths. Furthermore, we analyzed a large number of births over a period of 11 years, reducing the possibility of chance findings.

Since the study is based on births in a tertiary care hospital, the results may not be representative for the population because some women are selected to give birth at this level of care due to complications. This might affect overall perinatal mortality as well as the distribution of causes of death. Another limitation is that perinatal mortality may be underestimated since deaths of newborn who were discharged from hospital before seven days were lost if the death occurred at home or in another health facility. Although the standardized birth registry form is designed to collect detailed information, maternal conditions may be underreported due to poor diagnostic procedures in pregnancy, and obstetric conditions may be underreported especially in emergency situations.