Background
At the beginning of this century, 189 member states of United Nations endorsed a millennium declaration and committed themselves to eight Millennium Development Goals (MDGs) to improve the living standards of citizens worldwide [
1]. Two of these goals were to reduce child mortality (MDG 4) and maternal mortality (MDG 5) [
1]. By the end of 2015 some of these MDGs like poverty reduction and primary education for all were achieved. However achieving MDG 4 and MDG 5 were failed especially in sub-Saharan Africa [
2].
Globally, it was estimated that 289,000 maternal deaths occurred in 2013 [
3]. This indicates a 45% reduction from the baseline of 1990 [
3]. According to the 2013 World Health Organization (WHO) report, maternal deaths are highest in sub-Saharan Africa and Southern Asia. Sixty two (62%) percent of global maternal deaths took place in sub-Saharan Africa alone, followed by southern Asia (24%) [
3]. Ethiopia is among countries with highest maternal mortality ratio (MMR) in the world with an estimated MMR of 676/100,000 live births [
4].
In the past century, pastoral zones have been neglected by governments in terms of social services provision including health care. Geographical dispersion and mobility of pastoralists would create significant barriers to health service delivery, including institutional delivery. Governments often encourage settlement as a solution to the difficulty of providing health services for mobile pastoralists [
5]. Mobility of pastoralists, however, is a well-recognized survival strategy in arid and semi-arid lands of sub-Saharan Africa. Moreover mobility is very critical for the livestock survival and, the livelihoods and survival of the pastoralists [
6].
Many studies explored health institutional delivery utilization and its predictors in most socio-economic settings of Ethiopia. Few efforts, however, have been made to study this in a pastoral community. Therefore, this study aimed to assess the utilization of institutional delivery and associated factors among women of reproductive age in the mobile pastoral community of the Liban District in Guji Zone, Oromia.
Methods
Study setting
The study was conducted in the Liban District of Guji Zone, State of Oromia, Southern Ethiopia. The district is located in the lowlands of southern Ethiopia approximately 925 meters above sea level. The district’s administrative town, Nagelle Borana is located 595 km south of Addis Ababa, Ethiopia’s capital. The district is predominantly inhabited by the Borana agro-pastoral and pastoral community. Among the inhabitants are mobile pastoralists whose lifestyle is characterized by seasonal mobility for wise management of scarce water and pasture for a livestock.
Study design and population
A community based cross-sectional survey was conducted among women of reproductive age who had given birth within the last 2 years preceding the survey. The study was conducted in the area of the district considered to host mobile pastoralists.
Sample size and sampling procedures
The report of 2014 Ethiopian mini Demographic and Health Survey (EDHS) was used to estimate sample size for the first objective. According to the report 10.4% of rural women delivered in health institutions [
7]. Considering 95% confidence level and 3% absolute precision; the minimum required sample size was calculated as following:- n = (z
1-α/2)
2*p (1-p)/
d
2 ; were: z
1-α/2 = z-score corresponding to 95% confidence interval, which is 1.96; p- is anticipated sample proportion, which is 10.4% and d- is an absolute precision, which is 3%. Accordingly the minimum required sample size was 398. Sample size for the second objective was calculated by Epi-info7 statCalc software. Three independent variables were considered for the sample sizes calculation. The sample sizes were calculated with common assumptions of: Power of 80%; two sided confidence level of 95% and ratio of control to case of 1. Finally the largest sample size among four sample sizes was used for the study. Therefore sample size for the first objective (i.e. n = 398) was used, because it is the largest one. Because participants’ selection was not done by direct simple random sampling method, the design effect of 2 was considered. Additionally, 10% non-response rate was considered. Final sample size required was: 398 * 2 + 0.1(398*2) ≈ 876.
Sampling procedures
The area of the district which hosts majority of the mobile pastoralists is found to one side of the district. This side of the district was considered as one geographical area. The geographical area is formed by four different kebeles. The Sample size was allocated to the kebeles proportional to the approximate number of households. Each kebele was further subdivided into smaller administrative units called “Gare”. Finally, the sample size taken from each “Gare” was allocated proportional to the number of households in the “Gare”. In each “Gare” eligible women were enumerated and participants were selected by systematic random sampling. The whole process of approximating households, allocating sample sizes to each kebele and “Gare”, selecting the participants and data collection was accomplished within three weeks to reduce effects of mobility on the study.
Data collection
Data was collected by a pretested structured questionnaire. An English language questionnaire was developed by authors after reviewing relevant literatures. Finalized version of the questionnaire was translated into local language (i.e. Afan Oromo). The Afan Oromo questionnaire was pretested in nearby pastoral kebele not participating into the study. The Afan Oromo questionnaire was used to collect data. Informed verbal consent to participate was obtained from participants. Verbal consent was preferred over written consent, because majority of participants were non-literate women.
Data analysis
The data were entered into Epi-Info version 3.5.4, cleaned and analyzed with SPSS version 16. Frequency distributions were run and the prevalence of institutional delivery utilization was estimated. Bivariate analysis was done for all independent variables. P < 0.1 on bivariate analysis was used as a cutoff point to select the independent variables for multivariate analysis. Adjusted odds ratio with 95% confidence interval (CI) and P < 0.05 was used to declare statistical significance.
Discussion
The prevalence of institutional delivery in this study was 13.9% (95% CI: 11.5–16.3%). Women who had readily available cash when labor onset, delivered the birth preceding the most recent one in a health institution and faced birth related complications during the birth preceding the most recent one were more likely to have institutional delivery.
Prevalence of institutional delivery in this study is similar with a study in Banja District, Amhara regional state of Ethiopia. In Banja District, 15.7% of women delivered in health institutions [
8]. However, the prevalence in this study is much lower than the prevalence in Goba Woreda of Oromia regional state among urban and rural women where 47% (95% CI: 42.9–51.1%) delivered in health institutions [
9]. In Ethiopia, evidences show that prevalence of institutional delivery among rural women is lower than among urban women [
4,
7]. The current study shows lower prevalence of institutional delivery compared to the study in Goba Woreda, probably because of inclusion of urban women in the later. In addition to rural–urban difference, the difference in institutional delivery may be due to difference in socio-economic settings. Pastoral communities have limited access to social service infrastructures, including health care [
5]. The limited access to health care might influence utilization of institutional delivery. However, the prevalence in the current study is a bit higher than the national rural average of institutional delivery in Ethiopia [
7]. Mini Ethiopian Demographic and Health survey of 2014(EDHS 2014), reported that 10.4% (95% CI: 9.6–11.2%) of rural women gave their most recent birth in a health institution [
7].
Women who had readily available cash at the time of labor onset were almost three times more likely to deliver at a health institution compared to those who did not have cash. Though delivery services are free of charge at government health institutions; indirect health care costs may be a barrier. Ninety percent of women who gave birth at the health institutions delivered at health centers or hospitals. Almost all of these health institutions are found at more or less urban areas, whether it is a small rural or urban town. Moving to such health institutions may demands rural women and/or accompanying persons to have at least some money to cover essential expenses like food. In pastoral communities, however, money consists of livestock, which might be difficult to sell any time unlike other assets. First, market centers where livestock are to be sold might be located far away at urban areas, requiring a long distance drive with their livestock to the market centers. Second, despite the distant market centers, market days are limited within a given week. In the study area, for instance there are only 2 market days per week. Third, because livestock are highly valued and major income resources, its sell may require discussion among family members. All these three above conditions make livestock an unreliable source of money during emergencies of a woman in labor, unless livestock is sold prior to onset of labor. This may be further evidenced by the fact that livestock possessions did not affect institutional delivery. This study did not assess birth preparedness and complications readiness, which includes capital preparation. A study in Goba Woreda showed that women who were birth prepared and had complication readiness were more likely to deliver at health institutions.
Provided that delivery services are given free of charge for all women, the current findings support the importance of indirect health care costs on institutional delivery. Work done in five low income countries showed that even in the presence of fee waiver and exemption systems women continue to pay for maternal health services, a large proportion of which is informal payments [
10]. The same work demonstrated that fee waiver and exemption mechanisms will not alleviate the burden of out of pocket costs, because more than 80% of out of pocket costs for maternal health services are informal costs. Poor clients do not benefit from government fee subsidies because of poor awareness of fee waiver and exemption mechanisms [
10]. In the current study, majority (84.2%) of the women had good knowledge of delivery service including free ambulance service and free delivery services at government health institutions. However, institutional delivery was not significantly different between women who had unfavorable and favorable attitudes towards institutional delivery.
The women who delivered the birth preceding the most recent birth at health institutions were almost seven times more likely to deliver at health institutions than those who delivered such birth elsewhere. The finding is in line with a study in Banja District of Amhara regional state, Ethiopia [
8]. This may due to the fact that women who delivered at health institutions may better appreciate the advantages of institutional delivery, and this may encourage seeking health institutional delivery for subsequent deliveries.
The women who had complications during the birth preceding the most recent birth were more likely to deliver in health institutions. This finding is similar to a study in northern Ethiopia, which showed that women who had a history of obstetric complications tend to deliver at health institutions [
11]. Also in current study, the majority of the women who delivered at the health institutions did so because of the complications they faced during labor. Among women who gave the most recent birth at the health institutions, only about 27.3% of them had planned to give birth at a health institution. The rest delivered at the health institutions because of perceived complications during labor.
Socio-demographic characteristics of women like age, literacy, religion and family size did not seem to affect the place of delivery. Place of delivery was not associated with current maternal age, consistent with a study in northern Ethiopia [
11]. But it is inconsistent with many others which reported current maternal age as the factor affecting place of delivery [
4,
7,
8]. Contrary to many studies, the current study reported that women’s literacy status did not an have association with place of delivery [
7‐
9,
11,
12]. Consistent with many other studies, the current study reported that decision maker [
8,
9,
11], husbands’ education [
11] and habit of listening radio [
8,
9,
11]
, did not significantly influence the choice of place of childbirth. Even though pastoral mobility was claimed as one cause of the difficulty in providing health care to pastoralists, the current study did not show a significant association between extent of household mobility and the place of childbirth. Similarly, distance to the nearest health institution and functional transport road were not associated with the place of childbirth. Similar findings were reported in a study from Goba Woreda [
9]. However, 51.1% of the women who delivered outside health institutions mentioned lack of transportation and/or distance to health institution as one of the reasons for not delivering in a health institution.
Unlike many studies [
7,
9,
12], antenatal care utilization was not associated with place of delivery in the current study. Also marital status and parity were not associated with the place of birth. These findings are consistent with studies done in Gondar and Goba, Ethiopia [
9,
11]. In this study, the area of the district that considered hosting the mobile pastoralist and housing characteristics were criteria to distinguish mobile from settled pastoralists. But these criteria are not perfect in distinguishing the two from each other. So, chance of including some settled pastoral women in the study is possible. Therefore, this should be taken into account while interpreting the study.
Acknowledgements
We would like to express our deepest gratitude and appreciation to Hawassa University College of Medicine and Health Sciences for financially supporting the study. We are grateful to Guji Zone and Liban Woreda Health Office for the permission to undertake the study. Our honest gratitude also goes to all data collectors and supervisors for their commitment and hardship resiliency. Lastly, but not the least, we would like to extend our thanks to those study participants who volunteered to participate.