Introduction: About sixty million of home deliveries occur worldwide every year. The vast
majority of them in Low and Middle Income Countries (LMIC) where most of all out of
health facility deliveries are attended by relatives and traditional births attendants . Poor
hygienic conditions, ignorance of clean birth practices and lack of skills to manage the
complications when they occur, make home deliveries unsafe for the mothers and their
newborns. Thus, getting pregnant women to give birth at health facility is critical in the
efforts to improve reproductive health outcomes in many LMIC.
In Ghana, the out-of-pocket fees were waived for pregnant women since 2003 and the primary
health care system was modified to station community health nurses in rural areas. These two
strategies have a potential to reduce the proportions of home deliveries by removing financial
and distance barriers to health facility utilization. The aim of this study is to document the
trends and the predictors of home deliveries after the implementation of the above
interventions in two districts that pioneered their implementation in the country.
Material and methods: The study was carried out in Kassena-Nankana East and West
districts, of the Upper East region in Ghana, where the Navrongo Health Research Centre
operates the Navrongo Health and Demographic Surveillance System (NHDSS). The study
design was a series of cross sectional analytical studies using secondary data. All the
deliveries that occurred between January 2003 and December 2009 in the area were included
in the analysis. We used time series analysis to describe the trends of home deliveries over
time and a two level logistic regression models to determine the predictors of home deliveries.
Results: In all, 25539 deliveries occurred, majority (58.11%) of them at home. The trend
analysis showed a consistent and significant decline in rates of home deliveries over time
v
(p<0.001). The stationary time series pattern followed by the rate of home deliveries is an
Autoregressive Moving average ARMA (1, 1) model. The rates of home deliveries were
halved during the study period; from 69% in 2003 to 36.54% in 2009. Although the decline
was consistently observed within all the subgroups, it was more marked in the poor, the rural
residents and the uneducated populations.
Non maternal education (OR: 0.28, 95% CI: 0.23-0.34; for secondary/tertiary education
compared to no education), traditional religion practice by the mothers (OR: 0.59, 95% CI:
0.53-0.64; for Christians compared to traditional religion), the rural residency (OR: 0.09, 95%
CI: 0.08-0.12; for the urban residents compared to rural residents) and poverty (OR: 0.16,
95% CI: 0.13-0.20; for the rich compared to the poor) were strong predictors for home
deliveries. Other significant predictors include the high number of parity (OR: 1.98, 95% CI:
1.72-2.28; for multipare mothers compared with nullipare), the high number of previous home
deliveries (OR: 2.96, 95% CI: 2.25-3.90; for two or more previous home deliveries compared
to zero previous home delivery), the marital status (OR: 0.88, 95% CI: 0.78-0.99; for non
married mothers compared to married mothers and, the pregnancy status (OR:2.21, 95% CI:
1.77-2.75; for single fetus compared to multiple fetuses).
Conclusion: The results showed that home deliveries are sharply declining since 2003 in the
study area and the gaps between the rich and the poor, between the rural and the urban
residents and, between educated and uneducated persons are also reducing. Even though we
did not investigate the health care system related factors, the findings in our study are
consistent with those broadly reported elsewhere about the predictors of home deliveries.
Further studies with the before-after design are needed to show that the observed results are
attributable to the two interventions.