Background
Reviews of perinatal deaths are mostly facility based. Given the number of women who, globally, deliver outside of facilities, this data may be biased against total population data. We aimed to analyse population based perinatal mortality data from a LMIC setting (Mpumalanga, South Africa) to determine the causes of perinatal death and the rate of maternal complications in the setting of a perinatal death.

Methods
A secondary analysis of the South African Perinatal Problems Identification Program (PPIP) database for the Province of Mpumalanga was undertaken for the period October 2013 to January 2014, inclusive. Data on each individual late perinatal death was reviewed. We examined the frequencies of maternal and fetal or neonatal characteristics in late fetal deaths and analysed the relationships between maternal condition and fetal and/or neonatal outcomes. IBM SPSS Statistics 22.0 was used for data analysis.

Results
There were 23503 births and 687 late perinatal deaths (stillbirths of?=?1000gr or?=?28 weeks gestation and early neonatal deaths up to day 7 of neonatal life) in the study period. The rate of maternal complication in macerated stillbirths, fresh stillbirths and early neonatal deaths was 50.4%, 50.7% and 25.8% respectively. Mothers in the other late perinatal deaths were healthy. Maternal hypertension and obstetric haemorrhage were more likely in stillbirths (p?=?<0.01 for both conditions), whereas ENNDs were more likely to have a healthy mother (p?
Conclusion
A significant proportion of women have no recognisable obstetric or medical condition at the time of a late perinatal death; we may be limited in our ability to predict poor perinatal outcome if emphasis is put on detecting maternal complications prior to a perinatal death. Intrapartum care and hypertensive disease remain high priority areas for addressing perinatal mortality. Consideration needs to be given to novel ways of detecting growth restriction in a LMIC setting.