Study design and setting
Manicaland is the country’s second-most populous province with a population of 1,752,698 (2012 Census) constituting approximately 13.4% of the country population. Most Zimbabweans, 67%, live in rural areas. According to DHIS2 data January to December 2013, the province recorded 1540 perinatal deaths, against 44,610 (42,875 Institutional and 1735 home). In the same year, Mutare district recorded 534 (stillbirths + ENND) against 9673 (institutional live births) which translated to 55.2 deaths per 1000 live births [
17]. However, this could be an underestimate due to the poor vital registry system.
Mutare district population (35.7% of Manicaland) is served by a network of 48 primary, 1 secondary (Sakubva) and 1 tertiary (Mutare Provincial) health centres. For the period January to June 2013 (DHIS2 data): Manicaland province recorded 493 adverse pregnancy outcomes, against 20,869 deliveries while Mutare district recorded 4690 births (5497 expected births) and 197 adverse pregnancy outcomes (stillbirths and fresh neonatal deaths). Thus, Mutare district contributed close to 40% of the adverse pregnancy outcomes [
17].
Data were sourced from birth registers of Sakubva and Mutare Provincial Hospitals in Manicaland Province, Zimbabwe. Women resident in Mutare district, aged 18 years and above who had a singleton birth during the period January to June 2014 were included in the study. Records with more than 20% missing data were excluded.
Primary health facilities (PHF) manage non-complicated Antenatal (ANC), deliveries and postnatal cases. All para 0, under 18 pregnant women and those who have had previous caesarean sections, have ANC monitoring at clinic but are referred to district hospitals for delivery. For emergency referral, facility calls the district hospital to request an ambulance and if ambulance is not readily available, local transport is hired for transfer of patient. Physicians at Sakubva manage cases and refer complicated cases to Mutare provincial hospital for management by gynaecological specialists. All referrals from Sakubva were followed up at Mutare Provincial Hospital and considered for the study.
About 20% of women in Zimbabwe deliver at home or in the community (ZDHS 2015 report). Home deliveries have been attributed to perceived low economic, social and cultural opportunity costs, religious beliefs, culture, influence of family members, fear of hospital procedures, accessibility, and perceived high costs. Community maternal providers allure home delivery as they are perceived to be cheaper, have flexible payment terms and promote family cohesion thereby provide sensitivity to cultural and religious preferences [
18].
We conducted a descriptive study on data extracted from the “delivery register” from the period January 1st to June 30th, 2014. Data missing from registers but available from patient’s admission notes and antenatal (ANC) registers were also collected from these sources. All women who met the inclusion criteria were eligible to participate. For each birth record, socio-demographic information, maternal factors such as previous obstetric history (maternal age), neonatal information (sex, birth-weight) and delivery factors which included attendance by a skilled health worker, mode of delivery, and post-natal factors were extracted.
The main outcomes of this study were stillbirths and early neonatal deaths. Stillbirth was defined in accordance with World Health Organization definition of stillbirth for international comparison as death of a fetus weighing at least 500 g or after 22 completed weeks of gestation occurring before the complete expulsion or extraction from its mother [ICD-10]. Early neonatal death was defined as death that occurred within the first seven days of life. Data on late neonatal deaths could not be extracted as they are referred to local clinics for postnatal and therefore was not analysed.
Data from registers were double entered into Excel, cleaned and transferred to Stata 12.0 (Stata Corp, Texas, and USA) for analysis. Descriptive summaries were undertaken presenting frequencies, proportions, standard errors and associated 95% confidence interval for categorical characteristics while means, median, standard errors, p-values and associated 95% confidence interval for continuous characteristics.