Global spatial autocorrelation was measured with Arc-GIS using the Global Moran’s-I statistics to determine whether the pattern expressed is clustered, scattered or random throughout the study areas. Moran's I is a spatial statistic that measures spatial autocorrelation by yielding a single output value between -1 and +1. Moran’s, I Values close to -1 indicate disease/intervention is dispersed, whereas moron’s I close to +1 indicate disease clustered and disease distributed randomly if I value is zero.
The presence of spatial autocorrelation is indicated by a statistically significant Moran's I (p < 0.05), which leads to the rejection of the null hypothesis (skilled birth attendance is randomly distributed). Local Anselin Moran's In terms of positively connected (high-high and low-low) or negatively correlated (high-low and low-high) clusters, used to look into the local level cluster locations of skilled birth attendance. A positive value for ‘I’ indicated that a case with adjacent cases that had similar values. A negative value for "I" indicated that a case was surrounded by cases that had values that were different from its own.
2.10. Host spot analysis (Getis-Ord Gi* statistic)
The Getis-Ord Gi* statistic was calculated to quantify how spatial autocorrelation varied over the research location. The statistical significance of the clustering was assessed using the Z-score and the p-value. Spatial clusters with high values (hot spots) and low values (cold spots) were distinguished by the Getis-Ord Gi* statistic. Gi* is a measure of local autocorrelation, i.e. it measures how spatial autocorrelation varies locally over an area and provide statistic for each data points. If z-score is higher, the intensity of the clustering is stronger. A z-score that is close to zero denotes no clustering, a positive z-score suggests high value clustering and a negative z-score suggests low value clustering.
2.11. Spatial interpolation
Spatial interpolation is the process of estimating values (spatially continuous variables) for spatial locations which have not value from spatial location using known values. It is the method of determining the unknown value for any given set of points with known values. Spatial interpolation technique was used to predict SBA delivery on the un-sampled areas in the country based on sampled EAs. Numerous deterministic and geo-statistical interpolation techniques exist. Ordinary Kriging and empirical Bayesian Kriging are regarded as the best techniques out of all of them since they statistically optimise the weight and take into account spatial autocorrelation. For this study, the Ordinary Kriging spatial interpolation approach was employed to forecast SBA delivery in Ethiopian regions that were not sampled.
2.12. Spatial scan statistical analysis
Statistical analysis using Sat Scan version 9.6, a Bernoulli-based model was used to determine whether statistically significant spatial clusters of SBA delivery were present or not. The spatial scan statistic uses a circular scanning window that moves across the study area. To suit the Bernoulli model, women who gave delivery with a skilled birth attendant were classified as cases and those who did not were classified as controls. Each location's case count followed a Bernoulli distribution and the model needed information on cases, controls and geographic coordinates. The number of observed SBA inside each probable cluster was compared to the expected number using the likelihood ratio test statistic and p-value to see if there was a significant difference. By comparing the rank of the maximum likelihood from the real data with the maximum likelihood from the random datasets, Monte Carlo hypothesis testing was used to assign a p-value to each cluster. The scanning window with the largest likelihood was the most likely performing cluster. Based on Monte Carlo replications, the primary and secondary clusters were found, given p-values and ranked according to their likelihood ratio tests31.
2.13. Multi-level analysis
The multi-level mixed effect logistic regression model was employed for the proper determinant estimate due to the hierarchical nature of the DHS data. Two-level multilevel Multivariable logistic regression (mixed effect model) was used to analyse factors associated with SBA delivery at two levels, at individual and community (cluster) levels. Four models were built. The first model was an empty model without any explanatory variables, to calculate the degree of variance within the cluster on SBA delivery. The second model was adjusted with individual level variables; the third model was adjusted for community level variables while the fourth was fitted with both individual and community level variables simultaneously. ICC (Intra-class correlation), MOR (median odds ratio) and the difference across clusters were measured using PCV (proportional change in variance). The ICC is a measure of within cluster variation (i.e., the variation between individuals within the same cluster)32. MOR refers the median value of the odds ratio between the cluster at high risk and the cluster at reduced risk when two clusters are randomly selected33. In comparison to the null model, PCV calculates the overall variation in the multilevel model that is attributable to individual and community level influences. The formulas for these measurements are as follows;
The model with the highest LLR and with lowest deviance was considered as a best fitted model. Variables in the multivariable multilevel logistic regression analysis were deemed statistically significant if their p-value was less than 0.05. Finding the Adjusted Odds Ratio (AOR) and its corresponding 95% confidence interval was used for identifying factors associated to SBA delivery. The variance inflation factor (VIF) was also used to test for multi-colinearity. There is multi-co linearity if variable have VIF>10 and tolerance< 0.1.
2.14. Dissemination of the finding
As part of a Master of Public Health (MPH) thesis, the findings of the study will be present and disseminated to the Bahir Dar University, College of Medicine and Health Sciences, School public health, department of health system management and health economics. The findings will also be shared with the regional health bureaus and other relevant governmental and nongovernmental organizations. It will be submitted to scientific Journals for publication and possibly presented to other research conferences and seminars for concerned governmental and non-governmental organizations and stakeholders.
3. Results3.1. Descriptive characteristics of the study population
A weighted sample of 11,023 from EDHS 2016 and 5,527 from EMDHS 2019 reproductive age women were incorporated into this research. However, the data of 21 clusters from EDHS 2016 were not included in the spatial analysis as geographical information was missed. 66.1% of study participants in EDHS 2016 and 53.6% in EMDHSB2019 had no formal education. About 96.5% of the study participant’s health expenditure was not protected by health insurance in EDHS 2016. Regarding media exposure, about 81.9% of participants had no media exposure in EDHS 2016. The median age of participants upon first birth, was 18 years in both survey years. Large of study participants 88.9% and 75.3% were lived in rural area in EDHS 2016 and EMDHS 2019 respectively. From the total participants more than half, 56.6% in EDHS 2016 and 57.1% in EMDHS 2019 were lived in communities with low poverty level (Table 1). The percentage of women delivery with the assistance skilled professional was 27.7% and 49.6% in 2016 and 2019 respectively.
3.2. Spatial and temporal distribution of skilled birth attendant delivery
The proportion of skilled birth attendant delivery across regions of Ethiopia was increased over survey periods in all regions, except in Addis Ababa. The overall Skilled birth attendant delivery showed increasing pattern in the country between survey years (2016 to 2019).
3.3. Hot spot analysis
The geographical distribution of SBA delivery was different in both survey periods. The Getis Ord Gi* statistical analysis identified the significant geographical distribution of hotspots and significant clustering of cold spots of SBA delivery. In EDHS 2016, hotspot of skilled birth attendant delivery was discovered in Dire Dawa, Harari, Addis Ababa, Northern part of SNNPR (Gurage) and Southern part of Amhara (north Shewa) regions. On the contrary, Afar, most part of Amhara, Benshangul-Gumuz, Western Gambela (Nuer), Central Oromia, SNNPR and Somali were regions with cold spots.
In EMDHS 2019, hotspot of SBA delivery was clustered in Dire Dawa, Addis Ababa, Harari and South west part of Benshangul-Gumuze (Asosa) regions; whereas afar, Somali, Amhara and Eastern part of SNNPR were cold spot areas of SBA delivery. The overall spatial pattern of skilled birth attendant delivery was different across regions of Ethiopia over survey periods except Addis Ababa, Dire Dawa and Harari showed similar spatial pattern over the two survey years.
The spatial interpolation methods allow estimating values for locations where no samples have been taken and also used to assess the uncertainty of these estimates. We have used geo-statistical spatial interpolation with Ordinary Kriging technique in ArcGIS 10.7 software for estimating values for locations where no samples have been taken. From EDHS 2016, geo-statistical ordinary kriging analysis predicted that the highest frequency of skilled birth attendant delivery (77.7% to 98.5%) was discovered in Dire Dawa, Harari, Addis Ababa and south east part of Tigray. In contrast, area with relatively low prevalence (5.1%-15.5%) was predicted in Eastern part of Somali (Doolo), Southern Oromia (Borena and Guji), Western Gambela (Nuer), Benshagul (Metekel), Central Amhara (South Gondar) and Afar regions. Based on EDHS-2019 data ordinary kriging analysis predicted that the prevalence of skilled birth attendant delivery was the highest percent about (79.09% to 98%) predicted in Addis Ababa, Dire Dawa, Harari and Western Benshangul-Gumuz (Asosa). In contrast, area with relatively low prevalence (13%-22.4%) was predicted in most part of Somali, southern Afar and south part of Oromia (Borena, Guji and Liben).
For detection of purely spatial clusters of SBA delivery, spatial analysis was carried out with the Culldorff spatial sat scan analysis. The circular window with highest likelihood ratio and containing more cases than expected was identified as the most likely (primary) cluster. A significance level of P < 0.05 was used to test whether the cluster was significant or not. A total of 13 significant clusters were found in the EDHS 2016. Of which, 1 was most likely (primary) cluster and 12 of them were secondary clusters. The primary cluster spatial window was found in Addis Ababa which was located at (8.977567 N, 38.738050 E) of geographic location with 14.56 km radius, as well as the Log-Likelihood ratio (LLR) of 441.312466 which was detected as the most likely cluster with maximum Likelihood (Table 2). It demonstrated that women within this spatial window had 3.12 times more probable to deliver with SBAs than the women outside that area of the spatial window. The scanning window for the secondary clusters was found in Dire Dawa and Tigray, Northern SNNPR, Southern Gambela, Western Amhara and Western border of Oromia, with a LLR range from 9.7-441.3. In EMDHS 2019, A total of 9 significant cluster with p-value <0.05 were detected. 28 locations with overall sampled population of 337 were found as primary clusters. The spatial window of primary cluster was situated in Addis Ababa. The spatial window of primary cluster was located at (8.651588 N, 39.118340 E) / 68.57 km in radius, with a log-likelihood ratio (LLR) of 181.62 and a relative risk (RR) of 2.02. It demonstrated that women within this spatial window had 2.02 times more probable to get SBA than the women outside areas of the spatial window. The location of the secondary cluster spatial window was in Dire Dawa, Benshangul, Tigray, South East and central Amhara, Eastern Afar, SNNPR and Eastern Gambela regions, with a log-likelihood ratio (LLR) of 9.6-181.6.
3.6. Determinant factors of skilled birth attendant delivery
Intra-cluster correlation coefficient (ICC) in the empty model indicated that 16.8% and 36.7% of the overall fluctuation for skilled birth attendant delivery was caused by variations amongst clusters in EDHS 2016 and EMDHS 2019 respectively. The remaining unexplained (83.2% in EDHS 2016 and 63.3% in EMDHS 2019) was due to individual variations. The null model's median odds ratio (MOR) was 2.8 and 3.8 in EDHS 2016 and EMDHS 2019 respectively. Which indicates in the event that two women are chosen at random from two distinct clusters, those in the cluster with the highest skilled birth attendant delivery had 2.8 and 3.8 times more likely to experience skilled birth attendant delivery as compared with women from the cluster that has a lower skilled birth attendant delivery utilization. Bi-variable logistic regression analysis was performed to find the variables for the multivariable multilevel logistic analysis. Variables with p-value <0.05 were taken into account for the multivariate study. Multi-collinearity was also checked using VIF, which was found with maximum value of 2.2 in EDHS 2016 and 2.1 in EMDHS 2019. It indicates that there was no multi-collinearity among independent variables. With the highest log-likelihood and the lowest deviation, the final model (model-4) was the best suited model.
3.7. Individual-level predictors
In multivariable multilevel mixed-effect logistic regression analysis; wealth index, maternal education level and parity were significantly linked to the delivery by skilled birth attendants in both survey periods, However, health insurance coverage, media exposure, ever use of family planning and maternal age at first birth was strongly correlated with the delivery by a skilled birth attendant in EDHS 2016 (Table 1).
For variables showing significant association in both surveys, by taking the recent survey (EMDHS 2019), The likelihood of using SBA delivery for those women residing with household wealth index of poorer, middle, richer and richest increased with 1.46 times (AOR=1.46, 95%CI =1.16-1.85), 1.51 times (AOR=1.51, 95%CI =1.17-1.96), 2.44 times (AOR=2.44, 95%CI =1.83-3.26) and 3.67 times (AOR=3.67, 95%CI = 2.48-5.43) respectively in contrast with women in poorest wealth index.
The likelihood of having SBA delivery for those women with primary education and those with secondary and higher were 2.18 times (AOR = 2.18, 95% CI = 1.82-2.61) and 4.43 times (AOR = 4.43, 95%CI = 3.18-6.17) higher as compared with no formal education respectively.
Women having 2-4 children and those who had 5 or above were 46.2% (AOR = 0.538, 95% CI =0.42-0.69) and 54.1% (AOR = 0.459, 95%CI = 0.33-0.64) less likely to experience SBA delivery in contrast to those who had only one child respectively.
The likelihood of experiencing SBA delivery for those women reside in rural areas were lower by 75% (AOR=0.25, 95%CI = 0.15-0.43) in contrast to those who live in urban areas.
The odds of experiencing SBA delivery for those women whose medical expenditure was covered by health insurance were 1.61 times (AOR =1.61, 95% CI=1.13-2.29) higher in contrast to those who don’t have health insurance.
The likelihood of having SBA delivery for those women who had media exposure were 1.34 times (AOR = 1.34, 95% CI =1.13-2.29) higher compared to those who had no media exposure.
The likelihood of experiencing SBA delivery for those women who had ever used family planning were 1.72 times (AOR=1.72, 95% CI=1.49-1.99) higher compared to those who never used family planning.
3.8. Community-level predictors
Multivariable multilevel mixed-effect logistic regression analysis showed that; region, type of location of the residence and community poverty level in both survey periods (2016 and 2019) and distance from health facility and community childcare burden only in EDHS 2016 were strongly linked to the delivery by a skilled birth attendant in EDHS 2016 (Table 1 and 2).
For variables showing significant association in both surveys, by taking the recent survey (2019), The probability of women residing in Afar 84.9% (AOR=0.151, 95% CI= 0.06-0.37) oromia 73.3% (AOR=0.267, 95% CI=0.12-.58), Somalia 88.7% (AOR=0.113, 95% CI= 0.045-0.29), SNNPR 59.9% (AOR=0.401, 95% CI= 0.18-0.87), Gambela 57.4% (AOR=0.426, 95% CI= 0.19-0.98) and Harari 69.2% (AOR=0.308, 95% CI=0.12-0.78) less likely to experience SBA delivery in contrast to those who live in Tigray region. Whereas, Addis Ababa, Dire Dawa, Amhara and Benshangul-Gumuz region had not changed all that much in proportion of SBA delivery from the region of reference Tigray.
The likelihood of having SBA delivery for those women reside in rural areas were lower by 75% (AOR=0.25, 95%CI = 0.15-0.43) in contrast to those who live in urban areas.
Women who reside in areas with high rates of poverty in the community were 57.6% (AOR=0.424, 95%CI =0.28-0.64) less likely experiencing of SBA delivery in contrast t with those women live in a region with low rates of poverty.
The odds of experiencing SBA delivery for those women who perceive distance from health facility as not a big problem were 1.34 times (AOR =1.34, 95% CI =1.2-1.56) greater than those who perceive distance from health facility as a significant problem.
The probability of receiving SBA delivery for those women live with a community of high childcare burden were lower by 31.1% (AOR=0.689, 95%CI =0.54-0.88) as compared with low community childcare burden.
Table 1: Multilevel logistic regression analysis result of individual level and community level factors associated with SBA delivery in Ethiopia, EDHS 2016. Characteristics |
Category | Null model AOR (95% CI) | Model 2 AOR (95% CI) | Model 3 AOR (95% CI) | Model 4 AOR (95% CI) |
Maternal age | 15-24 | -- | 1 | -- | 1 |
25-34 | -- | 0.901(0.754-1.077) | -- | 0.853[.713-1.021] |
35-49 | -- | 1 | -- | .909[.709- 1.167] |
Maternal education | No formal education | -- | 1.84[1.60 2.13] * | -- | 1 |
Primary | -- | | -- | 1.69[1.46- 1.95]** |
Secondary and higher | -- | 4.25[3.29- 5.51] * | -- | 3.377[2.604- 4.379] ** |
Parity | Primiparous (1) | -- | 1 | -- | 1 |
Multiparous (2-4) | -- | 0.43[0.36- 0.524] * | -- | .435[.357- .529]** |
Grand multiparous (>=5) | -- | 0.348[0.269-0.448]* | -- | .379[.294- .490]** |
Religion | Orthodox | -- | 1 | -- | 1 |
Muslim | -- | 0.741[0.593-0.926]* | -- | 1.120[.860- 1.459] |
Protestant | -- | 0.607[0.472-0.781]* | -- | .760[.571- 1.011] |
Other | -- | 0.497[0.298-0.827]* | -- | .560[.357- 1.01] |
Ever use of FP | Never used | -- | 1 | -- | 1 |
Ever used | -- | 1.89[1.65- 2.18]* | -- | 1.722[1.494- 1.985]** |
Media exposure | Not exposed | -- | 1 | -- | 1 |
Exposed | -- | 1.42[1.198-1.69]* | -- | 1.339[1.126- 1.592]** |
Health insurance coverage | No | -- | 1 | -- | 1 |
Yes | -- | 1.62[1.145-2.301]* | -- | 1.605[1.127-2.285]** |
Wealth index | Poorest | -- | 1 | -- | 1 |
Poorer | -- | 1.77[1.456- 2.119]* | -- | 1.576[1.304-1.905]** |
Middle | -- | 1.81[1.478- 2.218]* | -- | 1.585[1.288- 1.951]** |
Richer | -- | 2.15[1.732- 2.669]* | -- | 1.785[1.427- 2.233]** |
Richest | -- | 7.396[5.764-9.491]* | -- | 2.531[1.904- 3.364]** |
Maternal occupation | Have no work | -- | 1 | -- | 1 |
Professional/technical/managerial | -- | 1.463[0.821- 2.604] | -- | 1.314[.742- 2.327] |
Agricultural | -- | 0.762[.64- 0.906]* | -- | .819[.688- .976]* |
Other | -- | 0.976[0.833-1.142] | -- | .940[.802-1.102] |
Maternal age at first birth | <=19 | -- | 1 | -- | 1 |
20-24 | -- | 1.158[1.009-1.328]* | -- | 1.119[.975-1.284] |
>=25 | -- | 1.741[1.358-2.232]* | -- | 1.66[1.293- 2.141]** |
Community childcare burden | Low | -- | -- | 1 | 1 |
High | -- | -- | 0.555[0.427-0.722]* | .689[.538-.882]** |
Community poverty level | Low | -- | -- | 1 | 1 |
High | -- | -- | 0.397[0.304- 0.52]* | .640[.491-.834]** |
Region | Tigray | -- | -- | 1 | 1 |
Afar | -- | -- | 0.074[0.044- 0.124]* | .119[.069-.206]** |
Amhara | -- | -- | .143[.09- .227]* | .162[.104-.251]** |
Oromia | -- | -- | .136[.086- .215]* | .152[.095-.242]** |
Somali | -- | -- | .108[.068- .172]* | .201[.121-.336]** |
Benshangul | -- | -- | .268[.164- .437]* | .331[.203-.538]** |
SNNPR | -- | -- | .23[.147-.360]* | .286[.179-.455]** |
Gambela | -- | -- | .236[.143- .39]* | .342[.202-.580]** |
Harari | -- | -- | .428[.245- .746]* | .426[.239-.757]** |
Addis Ababa | -- | -- | 1.69[.797- 3.57] | 0.967[.464- 2.015] |
Dire Dawa | -- | -- | .485[.272- .864]* | 0.556[.306-1.011] |
Distance from HF | Big problem | -- | -- | 1 | 1 |
Not big problem | -- | -- | 1.54[1.349- 1.749]* | 1.367[1.195- 1.563]** |
Place of residence | Urban | -- | -- | 1 | 1 |
Rural | -- | -- | .104[.074- .146]* | .248[.173-.356]** |
Key: 1= reference group; *=significant with p-value 0.01-0.05; **=significant with p-value<0.01; -- Not applied.
Table 2: Multilevel logistic regression analysis result of individual level and community level factors associated with SBA delivery in Ethiopia, EMDHS 2019.
Individual level variable | Category | Null model AOR(95% CI) | Model 2 AOR(95% CI) | Model 3 AOR(95% CI) | Model 4 AOR(95% CI) |
Maternal age | 15-24 | -- | 1 | -- | 1 |
25-34 | -- | 1.229[.981-1.53] | -- | 1.165[.930-1.459] |
35-49 | -- | 1.407[1.027-1.929]* | -- | 1.258[.917-1.726] |
Maternal education | No formal education | -- | 1 | -- | 1 |
Primary | -- | 2.255[1.881-2.704]* | -- | 2.176[1.815- 2.609]** |
Secondary and higher | -- | 4.696[3.375-6.534]* | -- | 4.428[3.181-6.164]** |
Parity | Primiparous(1) | -- | 1 | -- | 1 |
Multiparous(2-4) | -- | .523[.405- .676]* | -- | .538[.416-.696]** |
Grand multiparous(>=5) | -- | .423[.305- .586]* | -- | .459[.330-.636]** |
Religion | Orthodox | -- | 1 | -- | 1 |
Muslim | -- | .743[.539-1.022] | -- | 1.282[.873-1.88] |
Protestant | -- | .661[.474- .922]* | -- | .789[.553-1.125] |
Other | -- | .498[.273-.907]* | -- | .593[.323-1.087] |
Wealth index | Poorest | -- | 1 | -- | 1 |
Poorer | -- | 1.664[1.321-2.096]* | -- | 1.464[1.161- 1.845]** |
Middle | -- | 1.879[1.454-2.428]* | -- | 1.514[1.167- 1.963]** |
Richer | -- | 3.541[2.67-4.696]* | -- | 2.445[1.832- 3.263]** |
Richest | -- | 9.208[6.441-13.165]* | -- | 3.673[2.483- 5.433]** |
Maternal age at first birth | <=19 | -- | 1 | -- | 1 |
20-24 | -- | 1.106[.923-1.325] | -- | 1.107[.923-1.327] |
>=25 | -- | 1.138[.809-1.601] | -- | 1.109[.786-1.565] |
Community childcare burden | Low | -- | -- | 1 | 1 |
High | -- | -- | .624[.396962 .979]* | .829[.544-1.262] |
Community poverty level | Low | -- | -- | 1 | 1 |
High | -- | -- | .243[.159-.370]* | .424[.283-.635]** |
Region | Tigray | -- | -- | 1 | 1 |
Afar | -- | -- | .131[.0549-.310]* | .151[.062-.365]** |
Amhara | -- | -- | .449[.202-.999]* | .495[.234- 1.047] |
Oromia | -- | -- | .274[.123-.608]* | .267[.122-.583]** |
Somali | -- | -- | .096[.038-.244]* | .113[.0445-.287]** |
Benshangul | -- | -- | 1.224[.513-2.924]* | 1.169[.507-2.697] |
SNNPR | -- | -- | .353[.161-.776]* | .401[.184-.872]* |
Gambela | -- | -- | .480[.205- 1.127] | .426[.186-.979]* |
Harari | -- | -- | .498[.199-1.248] | .308[.122-.777]* |
Addis Ababa | -- | -- | 1.105[.351-3.477] | .577[.189-1.760] |
Dire Dawa | -- | -- | .736[.284-1.910] | .506[.194-1.317] |
Place of residence | Urban | -- | -- | 1 | 1 |
Rural | -- | -- | .133[.077-.229]* | .25[.147- .426]** |
Key: 1= reference; *=significant with p-value 0.01-0.05; **=significant with p-value <0.01; -- Not applicable.
4. Discussion
This study was amid to explore spatio-temporal variation and determinants of SBA delivery among reproductive age women in Ethiopia over 3 years. The finding of this study revealed that, the recent prevalence of SBA delivery was 49.6% (in EMDHS 2019). This finding is in consistent with previous study done in Kembata tambaro , Zone in Ethiopia 50.9%(28), whereas higher than study done in North west Ethiopia 18.8%34, another study in Ethiopia 15.6%35 and Galkacyo district in Somalia 27%36. This inequality may be because of the deference in study setting, sample size and period. In addition, it might be because of socio-demographic and cultural differences across regions. But this finding is lower than the result from Ghana 60.5%37, Pooled prevalence in 12 East African Countries 67.18%, Rwanda 90.68%, Malawi 89.8%, Burundi 85.13%, Comoros 83.78, Zimbabwe 78.14%, Uganda 75.19%, Zambia 64.21%, Kenya 61.81% and Mozambique 53.65%38. This may be as a result of the difference in accessing the service as well as due to the fact that women in those countries had better economic and educational status. This study showed that the delivery attended by skilled birth attendant increased from 27.7% to 49.6% between survey year 2016 and 2019. This improvement might be due to the focus of government since the inclusion of SBA as a key outcome indicator in the MDGs39. In addition, some socio-demographic improvement in the country might contribute for the improvement of SBA delivery.
This finding revealed that the proportion of SBA delivery varies across the regions of the country. Studies conducted in developing countries also pointed out the significant regional variations in the use of SBA delivery29,40. This might be due to the socio-cultural and socioeconomic differences among regions of the country. The highest prevalence was observed in Addis Ababa (95.5%) and Tigray (73%), whereas the lowest prevalence was obtained in Afar (30.2%) and Somali (25.7%). This might be because of inequalities in the distribution of resources like skilled birth attendants. Moreover, women from pastoral regions might have limited access to information regarding maternal health services than those in agrarian and urban regions. Maternal health service utilization by pastoralists is extremely low because of lack of awareness, cultural beliefs, seasonal mobility and limited availability of health facilities and health staff41.
Global Moran’s I value showed that spatial distribution of SBA delivery was non- random across the country and there was statistically significant clustering of SBA delivery. In this study significant cold spot areas of SBA delivery were observed in Afar, Amhara, Benshangul Gumuz, Gambelia oromia and Somali regions. One explanation could be the discrepancy in the provision of maternal health services, as well as the poor accessibility of infrastructure such as roads for transportation in those regions. Furthermore, the communities in these regions were more pastoral; as a consequence, relative to the rest, health facilities are not nearly available41. This finding shows that public health planners and programmers should develop successful public health actions to improve SBA delivery in these substantial cold spot regions.
Women from households of better economic status were more likely to use
SBA during delivery in contrast to women raised from poorest household. This
finding has also been demonstrated in similar studies in Ethiopia29,35,40,
Kenya42, Nigeria26, Ghana23, India43 and
Bangladesh44,45. This might be due to the reason that women in a
higher wealth index had greater autonomy in decision-making on reproductive
health and more likely delivered under skilled birth attendance46.
Women with better economic status often have the financial empowerment to
access skilled attendance during delivery47. In contrast women in
the poorest households are less likely to get health access because of
economical constraint, which is backbone to get education and health services.
Even though maternal health services in Ethiopia are free, this could not give
a guaranty for the use of SBA delivery, because of the presence of other costs
such as time and transportation. Sometimes mothers may be asked to buy supplies
that are not available at the health facility at the time of delivery.
This study also indicated that women with increased education were more experiencing SBA delivery as compared with those not has formal education. This finding was consistent with studies conducted in Ethiopia28,29,34,40,48, Kenya42,49, Zambia24, Nigeria26, Bangladesh22, Ghana37 and east Africa38. This could be explained by the fact that women who are educated had knowledge about the delivery complication and their consequences on them and their child, which could push them to deliver with the help of SBAs. Educated women had more open and better communication with the husband, more decision-making power and better negotiating skills thus better ability to demand adequate services50. As a result, more emphasis is required in educating women specifically those with no formal education on the importance of skilled delivery services and its association with reduced maternal mortalities.
Those women whose health expenditure was covered by health insurance had higher odds of experiencing SBA delivery in contrast to those not have health insurance. This finding is in line with previous study done in Ethiopia29 and Ghana23. This might be because of women’s health expenditure is covered by health insurance, they sense free and seek any medical care frequently and it might be narrows the difference between rich households and poor households in using SBA delivery service. This is supported by this study in which wealth index is a significant factor associated with SBA delivery.
The finding of this study indicated that women who exposed to media had more likely to use SBA delivery service as compared to those women who were not exposed to media. This finding was line with a previous study done in Ethiopia35, Niger, Sierra Leone and Mali27. The possible reason is that health information may improve health-seeking habits, as information about what services are available, where and when to get them, as well as the benefits and risks of accessing specific services, can be communicated via different Medias.
Women with parity of more children were less likely to experience SBA delivery as compared with who had only one child. This finding was in line with previous studies in Ethiopia29,38, Nigeria26 and southern Ghana51.This might be because women with higher parity might not have any complications before and take childbirth as a natural process or might have bad contact with a health professional previously, so they may prefer delivering without SBA52. Another possible explanation for this is that women who are pregnant with their first child are usually more likely to have difficulties during delivery than women of more parity. This may result in low parity women being more motivated to deliver with assistance of SBAs than women with more parity.
Women who had used any family planning services were more likely to deliver with the help of SBA in contrast to those women who had never used family planning. The finding is in agreement with studies conducted in Bangladesh22. Family planning service could empower women by exposing them to the health education about the benefit of skilled birth attendance. Moreover, women could understand the complication before, during and after delivery and early detection of complications arising during birth preparedness and complication planning which push them to visit skilled birth attendants.
This finding is consistent with a study conducted in Bangladesh22 and sub-Saharan Africa53. The possible reason for this finding could be the fear of stigmatization, devaluation and shaming young pregnant women to receive at maternal health services54.
The odds of experiencing SBA delivery were lower among women who lived in Afar oromia, Somali, SNNPR, Gambelia and Harari as compared to Tigray region. This finding indicated that SBA delivery varies across regions and in line with a study in Ethiopia29 and Nigeria26. Regional variation in the health infrastructure can cause significant health service disparity. Another potential reason might be women in pastoral regions have poor access to education and are not permanent residents and because of these, there is limited availability and accessibility of maternal health services.
The likelihood of having SBA delivery for those women reside in rural was decreased by 75% compared to those women live in urban areas. The result was comparable with studies conducted in Kambata Tembaro Zone28, North West Ethiopia34,55, Ethiopia29,40, Nigeria26, East Africa38 and Bangladesh22. Women living in urban areas have much easy access to skilled birth attendance compared to rural areas due to the proximity of the health facilities and better availability of transportation56,57. Another possible reason might be urban women in Ethiopia tend to benefit from increased knowledge and access to maternal health services compared with their rural counterparts, because, various health promotion programs uses urban-focused mass media work to the advantage of urban residents and explain the close connection between urban residence and use of maternal health services. Moreover, rural women are more readily influenced by traditional practices that are contrary to modern health care.
Women residing in communities with high poverty levels had lower odds to give birth by SBA as compared to women residing in communities with low poverty level. This finding is supported by a study done in Ethiopia29,35,40, Bangladesh22, India43 and East Africa38. This is due to communities with high poverty level even might not be able to pay for health insurance and health insurance coverage is associated with SBA delivery in this study58. Women who perceive distance from health facilities as big problem were less likely to receive SBA delivery compared to those perceive distance from nearest health facility was not a big problem. The finding is also in agreement with studies conducted in Ethiopia29,40. This is due to the fact that long distance from health facilities is important factors to prevent mothers from seeking and utilizing skilled maternity care services59. Moreover, some women at times deliver on the way to the health facility due to long distances60. Women who live in communities with high childcare burden had less likely to deliver with SBA as compared to those with low child care burden and this finding is in line with study in Ethiopia29.This might be due to a high childcare burden may need cost and time for carrying children, which may prevent mothers from seeking and utilizing maternal health services like SBA delivery61.
5. Conclusion
In spite of the above limitations, this study tries to explore spatio-temporal variation and determinants of SBA delivery among reproductive age women in Ethiopia. The magnitude of SBA delivery was lower in EDHS 2016 and it showed improvement in 2019. But if it continues with the current pace, it will be difficult to achieve the national Health Sector Transformation Plan II target. Despite the efforts that have been made in recent years to improve maternal health outcomes in Ethiopia, the proportion of women who receive assistance from SBAs is still unacceptably low.
The spatial pattern of SBA delivery in Ethiopia was clustered non-randomly across regions of Ethiopia. The most prominent low SBA delivery was detected in Afar, Amhara, SNNPR and Somali regions more or less consistently over survey periods. Benshangul-Gumuz, Gambela and Oromia showed improvements after survey year 2016. High proportion of SBA delivery was detected in Addis Ababa, Dire Dawa and Harari regions in both survey years. Household wealth index, maternal education level, health insurance coverage, media exposure, parity, ever use of family planning, maternal age at first birth, region, type of place of residence, community poverty level, distance from health facility and community child care burden were significant predictors of SBA delivery among reproductive age women.
5.1. Acronyms and abbreviations
ANC: Antenatal care, ARC-GIS: Aeronautical Reconnaissance Coverage Geographic Information System, CMHS: College of Medicine and Health Sciences, CSA: Central Statistical Agency, DHS: Demographic Health Survey, Dr: Doctor, EA: Enumeration Area, EDHS: Ethiopian Demographic and Health Survey, EMEDHS: Ethiopian Mini Demographic and Health Survey, EmOC: Emergency Obstetric Centre, EPHI: Ethiopia Public Health Institute, GIS: Geographic Information System, ICC: Intra-Class Correlation, ICF: Inner-City Fund, MDGs: Millennium Development Goals, MoH: Ministry of Health, MOR: Median Odds Ratio, MPH: Master of Public Health, Mr: Mister, NGOs: Non-Governmental Organizations, PCV: Proportional Change in Variance, PHC: Population and Housing Census, SBA: Skilled Birth Attendant, SDGs: Sustainable Development Goals, SNNP: Southern Nations and Nationality of People , SPSS: Statistical Package for Social Science , UN: United Nation, USAID: United States of America International Development, WHO: World Health Organization.
6. Acknowledgment
We would like to thanks Bahir Dar University, College of Medicine and Health Sciences for providing this opportunity and support to conduct this research. Thanks a lot, Ethiopian Central Statistical Agency (CSA) for providing Ethiopian Demographic and Health Survey data for this study.
6.1. Authors’ contributions
YAM was responsible for a significant contribution to the conceptualization, study selection, data extraction, investigation, methodology, formal analysis and original and final draft preparation. Project administration, resources, software, supervision, validation, visualization and reviewing are all handled by MAA, HAG and EMA and the final draft of the work was read, edited and approved by all authors.
6.2. Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
6.3. Competing interests
None declared.
6.4. Patient and public involvement
Patients and/or the public were not involved in the design or conduct or reporting or dissemination plans of this research.
6.5. Patient consent for publication
Not required.
6.6. Ethical approval
Ethical clearance was obtained from the ethical review committee of Bahir Dar University, CMHS. In addition a permission for data access was obtained from a measure Demographic and Health Survey through an online request at (http: //www. dhsprogram.com). The data used for this study were publicly available with no personal identifier. Our study was based on secondary data from Ethiopian Demographic and Health Survey and we have secured the permission letter from the Measure Demographic Health and Survey.
6.7. Data availability
The data used in this study are the third-party data which is Demographic and Health Survey available at (http://www.dhsprogram.com). So, to access the data, someone needs to follow the steps and protocol outlined under the methods section or the data is available upon reasonable request from the corresponding author.
7. References