Improving Practice of Institutional Delivery in Southern Ethiopia CURRENT

Background Institutional delivery service utilization is one of the key and proven interventions to reduce maternal death. It ensures safe birth, reduces both actual and potential complications, and decreases maternal and newborn death. However, a significant proportion of deliveries in developing countries including Ethiopia occurs at home and is not attended by skilled birth attendants. This study aimed at determining the prevalence of home delivery and associated factors in three districts in Sidama Zone. Methods A cross sectional survey was conducted from 15th- 20th October 2018. A multi-stage sampling design was employed to select 507 women who gave birth 12 months preceding the survey. Quantitative data were collected by using structured, interviewer administered questionnaires. Univariate and multivariate logistic regression models were run to assess factors associated with home delivery. Measures of association between factors and the outcome variable were reported using 95% confidence intervals (CIs) and adjusted odds ratios (aORs). Results The response rate was 495(97.6%). The overall prevalence of home delivery was 113 (28%) with 95%CI (19%, 27%). Maternal rural residence, aOR=7.45(95%CI: 2.23-24.83); illiteracy of mothers, aOR=8.78 (95% CI: 2.33-33.01); those who completed grades 1-4, aOR =3.81(95% CI: 1.16-12.49); mothers who did not know the expected date of delivery, aOR=2.12 (95% CI: 1.21-3.71); mother being merchant, aOR=3.01(95%CI:1.44-6.3) and paternal illiteracy, aOR=3.27, (95% CI: 1.20-8.88) were predictors of home birth. Conclusion

Interventions targeting rural and uneducated mothers might help to increase skilled birth attendance in the region.

Background
Maternal mortality is a major public health problem in developing countries particularly, in sub Saharan Africa (sSA) [1]. Every year, nearly half a million women and girls needlessly die as a result of complications during pregnancy or childbirth, and 99% of these deaths occur in developing countries of which 66% occur in sSA [1]. According to the joint WHO/UNICEF 2015 estimate, the global maternal death rate was 216/100,000 live births (LBs), while 436/100,000 LBs and 546/100,000 LBs maternal deaths were reported for least developed countries and sSA, respectively [2]. Based on a recent report, Ethiopia has one of the highest maternal, neonatal and infant mortalities with 412/100,000; 29 /1000 and 48/1000 LBs, respectively [3].
Most maternal deaths which occur in developing countries are due to complications during labor, delivery and the immediate postpartum period. WHO recommends for every delivery to be attended by skilled personnel. However, a significant proportion of women do not have this access during child birth [1,4].
The global skilled birth attendance estimate by 2015 revealed that 22% of deliveries did not get access to the service. In Eastern and Southern Africa, about 38% deliveries occurred at home [2]. In Ethiopia, community based surveys conducted in various regions of the country reported the magnitude of home deliveries to range from 31% to 96% [5][6][7][8][9][10][11][12]. Similarly, according to the Ethiopian Demographic Health Survey (EDHS) 2016, the prevalence of home delivery was reported to be 73% [3]. The consistently high prevalence of home deliveries in different parts of the country indicate that large proportion of pregnant mothers and their babies are at risk of complications including death related to child birth [1,2].
Available evidence revealed that several factors such as socio-economic status, maternal education, infrastructure, place of residence, women's experiences with healthcare providers and access to primary health care services significantly influence mothers' choice of place of delivery [13][14][15][16]. For example, a study from Nigeria and Ethiopia showed that women in the rural areas were more likely to regard facility delivery as unimportant and also complained about distance and inability to pay for the services [17].
According to the EDHS data -covering 2011 to 2015 -the prevalence of home birth in Southern Nations Nationalities and Peoples Region (SNNPR) was 74% [3]. However, as part of an ongoing implementation research of Kangaroo Mother Care (KMC) for low birth weight babies (weight less than 2000 grams) in Sidama zone, SNNPR, Ethiopia, the KMC Implementation Team (KIT) observed that there were very few home births which were reported by the health extension workers (HEWs), who are community health workers, -only 64 home births out of 31,000 (0.2%) (data not shown). While the low home birth rate which was observed by KIT could signal an improving health system in the region, it could also indicate poor tracking and reporting systems of home births by HEWs. To assess the real burden of home births in the area, we conducted a home survey in three districts of Sidama zone -Hawassa City Administration, Dalle and Shebedino districts. The aims of the survey were assessing the prevalence and determinants of home birth rate in three districts of Sidama zone, SNNPR.  [18]. It is estimated that there are over 10,000 deliveries taking place in Hawassa every year. There are 3 public hospitals and 12 health centers in the city.

Study setting
Shebedino district, the second study area is located 30km south of Hawassa city and has 32 Kebeles.
Leku town is a capital of Shebedino district. There is one primary hospital, 9 health centers and 32 health posts in the district. An estimate of 121 deliveries is attended per month in Leku Primary Hospital.
The total population of Dalle district in 2017 was 317,246 with 11104 expected deliveries per year (18). Yirgalem town is the capital of the district and is located 45kms south of Hawassa. There is one General hospital, 10 health centers and 36 health posts in the district.

Study design and Population
A community based cross-sectional survey was conducted during 15 th -20 th October, 2018. Randomly selected mothers who gave birth in the last one year and residing at least 6 months in the area were included in the study.

Sample size
The sample size was calculated using Epi info 7 Statistical software for population survey. The assumptions considered were: proportion of 72.5% home delivery in SNNPR based on EDHS 2016, confidence level of 95%, margin of error 5%, design effect of 1.5 with a cluster of 10 and 10% nonresponse rate. The final sample was calculated as 506 [3].

Sampling procedures
A multistage sampling technique was used to reach to the study participants. There are 32 kebeles in Hawassa city, 35 in Dalle and 32 in Shebedino districts. We selected 11 kebeles [4 kebeles from Hawassa City, representing urban households (36%); 4 from Dalle and 3 from Shebedino Woredas, both representing rural households (64%)] using simple random sampling techniques. Then, households with mothers who gave birth during the last 12 months preceding the study were identified and listed with the help of family folder available at the health posts of the selected 11 kebeles. Finally, the calculated sample size was proportionally allocated to the kebeles based on identified number of eligible mothers. Mothers in each of the selected kebeles were randomly selected by simple random sampling technique using the list as a sampling frame.

Method of data collection
The questionnaire was primarily prepared in English and then translated to local languages: "Sidamu Afoo" for rural residents and "Amharic" for urban residents. Six data collectors who completed at least first degree in Public Health were recruited and one-day training was given. Face to face interview with structured questionnaire was used to collected data. Consent was obtained from the selected mothers. Evaluation team of KMC implementation project supervised the data collection.

Data analysis
The data were coded, entered and cleaned using Epi Info version 7.2 software package and exported to SPSS version 25 software for analysis. Descriptive, bivariate and multivariate analyses were done to assess for association between independent factors and place of delivery. Odds ratios and 95% CIs were computed. In multivariate analyses, variables having with p.value < 0.2 in bivariate analysis were re-entered in the model, such as place of residence, age, education and occupation of mothers, paternal education, distance of health center from home, family size, knowing the due date and birth order of the child (parity). Statistical significant association was declared with a p-value <0.05 in the multivariate model.

Socio-demographic characteristics of the respondents
We interviewed 495(97.6%) out of the selected 507 mothers who had given birth 12 months preceding the survey. The mean (±SD) age of the participants was 25.85(±4.95) years; 339 (68.5%) and 156 (31.5%) were rural and urban residents, respectively. Two hundred and seven (42%) respondents completed grades 5-8 while 61(12.3%) did not attended school. Majority, 377(76.2%) of mothers were housewives and less than half (45%) of them had at least five family members (Table   1).

Characteristics of deliveries
The overall prevalence of home delivery was 113 (22.8%, 95%CI: 19%, 27%). Home delivery rate among mothers from rural residence was 108(32%, 95%CI: 27.6-36%) while 5(3.2%) of mothers residing in urban settings delivered at home (  times higher than those who completed at least grade 9, respectively. The odds of home birth among mothers who did not know the due date of their index child was two folds higher compared to those who knew it, aOR=2.12(95%CI: 1.21-3.71). Furthermore, mothers whose husbands did not go to school were about 3 times more likely to deliver at home compared to their counter parts, aOR=3.27, 95% CI: 1.20-8.88)( Table 3).

Discussion
In the current study, the proportion of mothers who gave birth at home was 22.8% (32% and 3.2% among mothers from rural and urban areas, respectively). Educated male partners are well informed risks related to home birth and thus, are more likely to encourage and financially support their wives to deliver at health facilities.
One unique observation of the current study is that merchant mothers were found to be more likely to deliver at home compared to housewives. Merchant mothers are considered to be more mobile because of the nature of their work, more likely to use contraceptives and have better awareness compared to housewives. The finding of this study is contrary to this reality requiring further investigation. Lack of early plan of place of delivery and lack of preparedness for institutional delivery are among the factors that increase the probability that mother could give birth at home.

Limitations Of The Study
Being a cross-sectional, this study may not give picture of trends in home delivery status among the study population. However, this survey is the first of its type for the site to reveal reasons for home births and assists policy makers and health system leadership to design ways of handling the case.

Conclusions And Recommendations 13
The prevalence of home birth has significantly dropped in the study setting compared to figures reported in 2016. Parental and maternal education, mother's occupation, and place of residence and planned place of delivery were found to be predictors of home delivery.
We recommend interventions targeting rural and uneducated parents to further reduce home births.