Spatial distribution and determinants of newbornsnot receiving postnatal check-up withintwodays after birth in Ethiopia: a spatial and multilevel analysis of EDHS 2016

The study was done in Ethiopia using theEthiopian Demographic Health Survey 2016. Ethiopia is among the oldestcountry worldwide, which is located in the Horn of Africa at 3′ and 14.8″ latitude 33′ and 48′ longitude. The countryis bordered by Sudan in the west, Somalia and Djibouti in the east, Eritrea in the north, and Kenya in the south. The country has a surface area of1,112,000 km². It is a rugged, landlocked country split by the Great Rift Valley, with archaeological finds dating back more than 3 million years; it’s a place of ancient culture. Among the important sites in Lalibela with its rock-cut Christian churches from the 12th–13th centuries. Aksum is the ruins of an ancient city with obelisks, tombs, Our Lady Mary of Zion church, and Gondar Fasil castles. Ethiopia is the 10th largest and the 2nd-most populous country in Africa after Nigeria. Administratively Ethiopia has ten regions (Tigray, Afar, Amhara, Benshangul, Gambela, Harari, Oromia, Somali, Southern, Nations, Nationalities, and People’s Region (SNNPR), and Sidama (a recently added and two city administrations (Addis Ababa and Dire Dawa). Concerning resident, 79.2% of the Ethiopian population lives in rural, and 43.3% is under fifteen ages [19]. Ethiopia uses a three-tier healthcare system; 1) primary healthcare systemconsists of health posts, health centers, and primary hospitals, 2) secondary healthcare consists of zonal hospitals, and 3) tertiary healthcare consists of comprehensive specialized hospitals [20].

Across-sectional retrospective study design was conducted to assess the geospatial variation and determinants of newborns not receiving postnatal check-ups within 2 days after birth.

The data used in this article were obtained from the Ethiopia Demography and Health Survey (EDHS) 2016, which was accessed at the MEASURE DHS website after securing a formal request to the MEASURE DHS program. The survey was carried out by the central statistics agency of Ethiopia, and the Ethiopia Public Health institute Ethiopia with the technical assistance provided by ICF International. The authors requested the measure DHS trough briefly stating the objectives of this analysis and access was granted to use the data on the (http://​dhsprogram.​com) website [21].

The source population of this study was all mothers with newborns born in the last 2 years preceding the survey. Multi-stage stratified cluster sampling was used to select the study participants. In the first stage, 645 clusters or enumerations areas were selected randomly, and stratified into urban and rural. In the second stage, a fixed number of 28 households in each cluster was randomly selected [21]. Geographic coordinates of each survey cluster were also collected using Global Positioning System (GPS) [21]. Mothers aged 15–49 and children born within the 2 years preceding the survey in each selected household were subjected to our study.

The study was conducted among (3832 un-weighted and 4036 weighted frequency) newborns to assess the postnatal health checkups within 2 days after delivery (Fig. 1).

The independent variables used in this study were as nested into; 1) individual levelfactorssuch as marital status, maternal age, religion, maternal occupation, maternal education, residence, sex of newborn, wealth index, number of ANC, place of delivery, size of the newborn at birth and number of living children inthe family. 2) Community level factors such as region,distance to the health facility, community illiteracy level, health insurance, media exposure, access to electricity, access to safe water, and community poverty. Some of the community-level factors were aggregated from the individual-level factors.

We performed a secondary analysis of the EDHS 2016, using the Kids Records (KR) dataset. STATA version 14 and Microsoft Excel 16 were used for data cleaning and coding, and the spatial analysis, and mapping were done using ArcGIS version 10.8. Descriptive statistics such as frequency and percentage of different variables were computed and presented using texts, tables, and graphs.

After preparing the data we imported the data to ArcGIS version 10.8. Joining the outcome with the GPS data and projections of the Geographically coordinated data to the projected coordinatedata were conducted before the analysis. Spatial autocorrelation analysis was done to test whether there is spatial variation across the study area. The spatial autocorrelation test signifies whether there is clustering or dispersion of postnatal check-ups within 2 days after birth. The value of the Morans Index is standardized into Z-score. Positive Morans Index value with positive Z-score (> 1.96, P-value< 0.05) indicates clustering/ hot spots areas. Negative Morans Index value with a negative Z-score (<− 1.96, P-value1.96 (P < 0.05) was considered a cold spot area. GetisOrd Gi* statistic was applied to detect hotspot area or spatial clustering of newborns not receiving postnatal check-ups. Ordinary kriging interpolation was used to estimate/ predict the spatial distributions of not receiving a postnatal check within 2 days after delivery.

A multilevel binary logistic regression model was applied for each independent variable and p-value < 0.2 were entered into the multivariable multilevel logistic regressions model. The adjusted odds ratio was calculated and used as the measure of association between the dependant and independent variables, and variables having a p-value < 0.05, 95% CI were considered statistically significant.

In the EDHS data, the newborn is nested within a cluster, newborns within the same cluster were more similar to each other than within different clusters. Therefore, this violates the standard regression model assumptions, which are independence of observation and equal variance across the cluster assumptions. This implies they need to take into account between-cluster variables by using an advanced model. Therefore, a multilevel random intercept logistic regression model was fitted to estimate the association between individual-level and community-level factors and the likelihood of new-born not receiving postnatal care within 2 days after birth. Models were compared based on deviance (−2log likelihood) since the models were nested. Log-likelihood and intracellular correlation coefficient (ICC) was computed to measure the variation between clusters. The ICC indicates the degree of heterogeneity of new- born not receiving postnatal care within 2 days after birth. A multilevel binary logistic regression analysis was performed to examine the effects of individual and community level factors on newborns not receiving postnatal check-ups within 2 days after birth to identify individual and community level factors.

A total of 4036 weighted live birthswere included in this analysis, of whom nearly half (48.79%) of the newborns were males. The mean ages of the mother were 33+ SD 1.3 and near to one-third of the mothers (29.26%) were between the age group of 25 and 29 years. Near to 60 % of mothers, (58.97%) were unemployed, and more than 60 % (61.59%) were not educated. The majority (88.77%) of the study participants were residing in rural and nearly half (47.5%) of the mothers were in a poor wealth status (Table 1).

More than one-third (36.04%) of mothers hadnoANC visits, and 63.64% of the mother were delivered at home. Among mothers, 53.85% had 1–3 live children in their families. The majority (96.51%) of the study participants have not used health insurance, 61.28% of the mother had a big problem reaching the health facilities, and near to 50% of the community (47.39%) were illiterate. Around 40%, (38.98%) and (72.31%) of the mother have no access to media, and electricity (Table 2).

The overall prevalence of not receiving postnatal check-ups within 2 days after birth in Ethiopia was 84% (Fig. 2). The postnatal check-ups within 2 days were higher in Addis Ababa and Tigray regions, whereas, low postnatal check-ups were observed in Afar, Somali and Gambela regions (Fig. 3).

The spatial autocorrelation analysis result showed that newborns not receiving a postnatal check-up within 2 days after delivery significantly varied across Ethiopia, with a Global Moran’s I value of 0.47, p-value< 0.0001, and z-score 7.13. This indicates that the newborn not receiving postnatal check-upswithin the first 2 days after birth in Ethiopia has spatial dependence (Fig. 4).

In the hotspot map, significant spatial clustering of not receiving postanal check-ups within 2 days after birth was observed in Eastern (Somali), Northeastern (Afar), Southern (Oromia and SNNPR), and Western (Gembela and Benshangul) parts of Ethiopia. Whereas, spatial dispersion of newborns not receiving postnatal check-ups was observed in Northern, Central, and Southern parts of the country specifically in regions of Tigray, Amhara, and Addis Abeba (Fig. 5).

Purely Spatial analysis scanning for clusters with high rates was applied using the Bernoulli model. A total of 287 significant clusters of newborns not receiving postnatal check-ups were specified, of which 41 and 100 were primary (most likely clusters) and secondary clusters, respectively. The primary cluster, was found in the Eastern (Somali region) of Ethiopia at 5.589269 N, 44.175032 E geospatial locations, with a relative risk of 1.47 and log-likelihood ratio (LLR) of 48.1 at a p-value< 0.0001. Those shows mothers who lived in this spatial window is 1.47 time at higher risk of not receiving postnatal check-up than those outside the window. Whereas the secondary clusters were honored in the Western parts of the country(Gambela and Benishangul) at 7.417914 N; and 35.317365 E geospatial locations, with relative risk and LLR of 1.29 and 31.2, respectively (Table 3).

Spatial scan statistical analysis of hotspot areas of newborns not receiving postnatal check-ups in Ethiopia, produced using Arc GIS version 10.8 (Fig. 6).

Ordinary Kriging interpolation was used to predict the prevalence ofnot receiving postnatal check-ups within the first 2 days among newborns in unobserved areas. The high predicted prevalence of not receiving postnatal check-ups was observed in the Eastern, Southeastern, and Southern parts of Ethiopia. Whereas low prediction of not receiving postnatal check-upsamong newborns was observed in the Southern, Western, and Northern parts of Ethiopia, specifically in Amhara, Tigray, and Addis Ababa regions (Fig. 7).

The multilevel binary logistic regression model was the best-fitted model for our data; after LR and ICC, tests were checked. Thus, the two-level logistic regression model was fitted to obtain an unbiased result and to come up with a valid inference. Deviance was used to check the model’s fitness and a model with the lowest deviance value is the best fit model. Hence,the final model was the best-fitted model in our data, which gives the lowest deviance value. The ICC value was 0.13 (95% CI: 0.11–0.16) in the null model, which indicates that about13% of the overall variability of the newborns not receiving postnatal check-ups was presented between cluster variability, and the deviance of the null model was (−2 ×  − 1571922) = 3, 142.78. An ICC value of the individual model was 0.3442 = 34.42,95% CI: (0.17–0.71), which indicates that about 34.42% of the overall variability of the newborn, not reciving postnatal check-up was presented between cluster variability and the deviance of the individual model II was (−2 ×  − 1136.6712) = 2, 273.34, the ICC value of community level model III was 0.4582245 = 45.82, 95% CI: (0.27–0.79%) random effect whereas the deviance was (rand1448.7476) = 2, 8 97.50 and the multilevel model IV ICC value was 0.0902263 = 9.02, 95% CI:(0.01–0.76), and its deviance was (, an1101.4277) = 2, 202.86.

In this study a two level binary logistic regression model was fitted to determine the factors associated with newborns not receiving postnatal check-up. Among the individual level factors (model II); ANC visit AOR = 0.37,95%CI:(0.24–0.58), place of delivery AOR = 0.02, 95%CI: (0.01–0.03), and health insurance AOR = 0.49,95%CI:(0.30–0.80) were significantly associated with newborns not receiving postnatal check-ups; wheras, in the model III (community level factors) regions such as Afar AOR = 0.16, 95%CI:(0.08–0.34), Oromia AOR = 0.27,95%CI:(0.15–0.48), Somali AOR = 0.25, 95%CI:(0.14–0.46), Benishangul AOR = 0.37, 95%CI:(0.22–0.69), SNNPR AOR = 0.32,95%CI:(0.18–0.57), Gambela AOR = 0.24,95%CI:(0.13–0.47), Harari AOR = 0.40,95%CI:(0.22–0.72), and Dire Dawa AOR = 0.32,95%CI:(0.17–0.59), community wealth status AOR = 0.59,95%CI:(0.43–0.81),community illiteracy AOR = 0.62, 95%CI:(0.46–0.83), and media exposure AOR = 0.79,95%CI:(0.62–0.99) were significantly associated with not receiving postnatal check-up among newborns. However, in the final model (model IV) ANC visit AOR = 0.42, 95%CI: (0.27–0.66), place of delivery AOR = 0.02, 95%CI: (0.01–0.92), residence AOR = 1.90, 95% CI: (1.29–2.81) and regions such as Afar AOR = 0.34, 95%CI:(0.16–0.69), Oromia AOR = 0.53,95%CI:(0.30–0.91), Somali AOR = 0.40, 95%CI:(0.22–0.74), Benishangul AOR = 0.50, 95%CI:(0.27–0.92), SNNPR AOR = 0.33,95%CI:(0.19–0.57), Gambela AOR = 0.30, 95%CI:(0.16–0.56), Harari AOR = 0.44, 95%CI:(0.25–0.78), and Dire Dawa AOR = 0.30, 95%CI:(0.17–0.54) remains statsically significant.

In our study, the odds of not receiving postnatal check-ups within 2 days after birth was increased by 58% among mothers who have not attended ANC visits as compared to their counterparts. The odds of not receiving postnatal check-ups among mothers who delivered at home and others were increased by 80 and 25% higher than mothers who had delivered at the hospital (AOR = .0.02(0.01–0.0.29)) and (AOR = 0.76(0.59–0.99)), respectively. The majority of Ethiopian communities who have lived in different regions were significantly associated with new- borns not receiving postnatal check-ups. The odds of new borns not receiving postnatal check-up among mothers in Afar 66% (AOR = 0.34(0.16–0.69)), Oromia 47%(AOR = 0.53(0.30–0.91), Somali 60%(AOR = 0.40(0.22–0.74)),Benishangul 50%(AOR = 0.50(0.27–0.92), SNNPR 67%(AOR = 0.33(0.19–0.57)), Gambela 70% (AOR = 0.30(0.16–0.56), Harari 56% (AOR = 0.44(0.25–0.78)), and Dire Dawa 70% (AOR = 0.30(0.17–0.54) were higher than mothers in Addis Ababa, respectively. The odds of new born not receiving postnatal checks among mothers who had lived in rural were 1.90 times higher than in urban mothers (AOR = 1.90(1.29–2.81)) (Table 4).

The strength of our study is using large sample size and a country representative data. The other strength of this study is using advanced statistical models (geo-spacial and multilevel analysis), which account the cluster variabilities. However, our study has some limitations, first, since we used secondary data analysis, we fail to incorporate some clinical variables, which might effect on the outcome variable. The second limitation of this study was we coundn’t include the paternalinformations, which might infulunce our result.