Addressing Critical Knowledge Gaps in Newborn Health

What is going on? Why is distance to care not associated with neonatal mortality in Malawi and Zambia!?

By Sabine Gabrysch on March 1, 2013
Malawi, Zambia
Africa

Access to adequate care at birth is needed to reduce newborn as well as maternal deaths. Interestingly, there have been hardly any studies on the influence distance to quality delivery care has on neonatal mortality, partly due to a lack of adequate data.

To address this gap, a group of researchers from Heidelberg University and the London School of Hygiene & Tropical Medicine, linked existing Demographic and Health (DHS) household data and national Health Facility Census data in a Geographic Information System to see whether babies who lived further from care and from better quality care were more likely to die in the early neonatal period in Zambia and Malawi.

What was seen was a big surprise: In Malawi, we found no difference in early neonatal mortality between babies who lived far or close from facilities, whereas in Zambia, babies who lived further from services were found less likely to die. Additionally, the level of care in the closest facility did not seem to have any influence on neonatal mortality in either country!

What is going on?

In order to better understand these unexpected results, we analysed the role of facility delivery as the mediating factor (see conceptual framework). As expected, women who lived closer to facilities were more likely to have their babies delivered in a facility rather than at home, compared to women who lived further away. The next step was to find out whether or not facility delivery was associated with lower neonatal mortality. This link is not as straightforward to analyse as one might think, because facilities tend to attract complicated births that have a higher risk of neonatal death. This can lead to facility births showing a higher mortality than home births.

To get around this problem, we studied the link between facility delivery and newborn mortality not for individuals, but for villages. Villages were split up in different groups depending on how common facility delivery was. In villages where facility delivery was rare, a high proportion of facility deliveries tended to be complicated cases, whereas in villages where facility delivery was common, health facilities also had many low-risk normal deliveries. In villages where most women delivered in facilities, there was a tendency of lower neonatal mortality among facility births compared to home births. Overall, however, and despite this link, there was no difference in early neonatal mortality between villages with high and low facility use.

One possible explanation of these surprising results could be that facility delivery cannot save newborn lives because of the low quality of care women and newborns receive at facilities. Alternatively, the results could be explained by differential underreporting of neonatal deaths in the DHS, if in the most remote locations early neonatal deaths are not reported or misreported as stillbirths. It would be interesting to compare to findings using alternative data, e.g. from demographic surveillance sites.

With some improvements in the data sources, our approach of linking existing national datasets could be used in other places to go beyond monitoring the percentage of births with a skilled attendant (health outcome indicator) and gather information upstream on the level of care at facilities (health system output indicator), to know where improvements are needed, and eventually downstream to know whether better service indicators upstream really translate into reduced mortality (health impact indicator).