Background: Maternal mortality remains a major global public health concern despite many international efforts. Facility-based childbirth increases access to appropriate skilled attendance and emergency obstetric care services as the vast majority of obstetric complications occur during delivery. The purpose of the study was to determine the proportion of facility delivery and assess factors influencing utilization of health facility for childbirth. Methods: A cross-sectional study was conducted in two rural districts of Hadiya zone, southern Ethiopia. Participants who delivered within three years of the survey were selected by stratified random sampling. Trained interviewers administered a pre-tested semi-structured questionnaire. We employed bivariate analysis and logistic regression to identify determinants of facility-based delivery. Results: Data from 751 participants showed that 26.9% of deliveries were attended in health facilities. In bivariate analysis, maternal age, education, husband’s level of education, possession of radio, antenatal care, place of recent ANC attended, planned pregnancy, wealth quintile, parity, birth preparedness and complication readiness, being a model family and distance from the nearest health facility were associated with facility delivery. On multiple logistic regression, age, educational status, antenatal care, distance from the nearest health facility, wealth quintile, being a model family, planned pregnancy and place of recent ANC attended were the determinants of facility-based childbirth. Conclusion: Efforts to improve institutional deliveries in the region must strengthen initiatives that promote female education, opportunities for wealth creation, female empowerment and increased uptake of family planning among others. Service related barriers and cultural influences on the use of health facility for childbirth require further evaluation.
A community based cross-sectional study was conducted in two rural districts of Hadiya zone, southern Ethiopia. In the zone, there were 280 Health Posts (HPs), 60 rural Health Centres (HCs) and one hospital serving a population of nearly 1.2 million. Hadiya zone is divided into 11 districts for administrative purposes. The vast majority of the population are Hadiya in ethnic group and they earn their living through rain fed agriculture. The Lemo district has a population of 144, 244 with about 33,176 childbearing women whereas the population of Gombora district is 113, 004 with about 26,330 childbearing women. Lemo district has 7 HCs and 33 HPs while Gombora has 6 HCs and 23 HPs. In both districts each rural kebele (the lowest administrative unit in Ethiopia), has one HP. About 98% of Lemo district and 100% of Gombora district populations are rural dwellers [10, 11]. The study population comprised women who reside in the kebeles selected from the two rural districts (Lemo and Gombora) and had delivered in the last three years preceding the study. The sample size was calculated using single population proportion determination formula; by taking point estimates of institutional childbirth of southern region which was 6.2 [4] Other inputs considered for the sample size determination were: 95% confidence level, design effect (deff) of 2, precision of 2.5% and 5% non-response rate which makes a total of 756 respondents. Stratified random sampling was used to select the study units. Firstly the study areas were divided into two groups; six of which were far (>20 km) from and four that were close (within 5–20 km) to the zonal capital. In order to have a good representation, one district was selected from each group using lottery method. Then all kebeles in the selected districts were grouped into five based on geographical direction, afterwards one kebele was selected from each group. Then enumeration was done in the selected kebeles by going house to house in order to identify eligible mothers (women who had given birth in the last three years preceding the survey; regardless of the current pregnancy status or outcome of the previous pregnancy or place of delivery) and 2,474 women were found to be eligible and all houses were coded. Thereafter, 756 women were randomly selected for the interview using proportional allocation. Regarding weighting a simple balancing was done and each observation had a weight of 1 as all eligible mothers were identified by enumeration and sampling procedure was done carefully. A structured questionnaire was adopted after reviewing relevant studies done previously Additional file 1). The 50-item questionnaires had 5 sections. It was constructed in English language and then translated into Hadiyigna language. Before actual collection of data, a written informed consent was obtained from all respondents after explanation of the purpose, objective, risk and benefit of the study. Also a pre-test was done in similar district in 10% of the respondents which was 75 and necessary corrections were made. The interviews were conducted in a convenient, quiet and private place for the mother. It took 25–40 min to complete the questionnaire. Data were collected by ten trained female data collectors who were High School Teachers and had Bachelor degrees. The questionnaires were checked for consistency and completeness before being entered into EPI data version 3.1 software for cleaning and exploration, and analysed using SPSS version 20.0. The responses concerning the outcome variable, place of delivery were three: health facility, home and on their way to the health facility. We had only two (0.0027%) of respondents who gave birth on their way to the health facility, we found these to be very small to be analysed alone therefore we added them to the home deliveries. Finally, home births were coded zero and health facility deliveries were coded one. We used principal component analysis method to generate wealth quintiles. It was based on household assets, dwelling characteristics, any livestock, agricultural land and others. Planned (a pregnancy which the woman becomes pregnant after intending to be pregnant or a pregnancy which a woman and her partner had discussed and decided in advance) and unplanned pregnancies (a pregnancy which the woman becomes pregnant without intending to or a pregnancy a woman and her partner had not discussed and decided in advance) were among the factors. Summarization was done using percentage, tables, figures and summary statistics. Binary logistic regressions were used to calculate Crude Odds Ratios (COR) for birth in a health facility and 95% Confidence Intervals (CI). Variables which were statistically significant at p < 0.05 on bivariate analysis were entered into multiple logistic regression model. Variables having strong correlation were planned to be excluded from the final model. Parity and place ANC attended were not included in the final model as it had strong correlation with age and numbers of ANC attendances respectively. Odds ratio was adjusted for all other variables. The significance level was set at p-value less than 0.05. The Hosmer-Lemshow goodness-of-fit test was also checked. Ethical approval letter was obtained from South Regional Health Bureau Ethics Committee.