Maternal and community factors associated with postnatal care checkups in Ethiopia: A multi-level analysis

Background Maternal mortality remained a public health issue, especially in developing countries. Although the rst two days after birth was a critical time in which the highest maternal death recorded, it was the most neglected period of maternal health services. Therefore, this study aims to determine the maternal and community-level factors of postnatal check-ups in Ethiopia Methods Demographic and Health Survey (DHS) in 2016 was utilized. A total of 3,948 women aged 15-49 giving birth in the two years before the survey were included. The response variables were post-natal check-ups in the rst two days after birth, dichotomized into no check-ups in the rst two days and check-ups within two days after giving birth. The explanatory variables were individual maternal characteristics such as education, household wealth, perceived distance to a health facility to get medical help, employment status, antenatal care (ANC), delivery by cesarean section, birth order and place of delivery, and selected community characteristics such as community literacy, community socioeconomic status, community ANC coverage, community perception of distance to a health facility, and regional states. A multi-level mixed-effects logistic regression model was employed. Result Only 17% of the women had a postnatal check-up within 2 days of giving birth in Ethiopia. Institutional delivery AOR 2.14 [95% C.I 1.70, 2.0] and giving birth by cesarean section AOR 1.66 [95% CI 1.10, 2.50] were found to be maternal factors, whereas administrative regions(Oromia 69%, Somali 56%, Benishangul 55%, SNNPR 43%, Gambela 66%, Afar 50% and Dire Dawa 55% which less likely to utilize postnatal care utilization as compared to Addis Ababa city administrative area), Higher community-level wealth AOR 1.44 [95% C.I 1.08, 1.2], ANC coverage AOR 1.52 [95% C.I 1.19, 1.96] and perceived distance of the health facility as a big problem AOR 0.78[95% C.I 0.60, 0.99] were the community factors associated with postnatal check-ups in Ethiopia. Conclusion Both A community the distance to a health facility as a big were 22% less likely to postnatal care as compared to their counterparts. We believe that no other authors have found that postnatal care utilization is less likely in the community who perceived the distance of health facility as a big problem. There is a good probability that community telling can inuence any health care service utilization. This study provides strong evidence in utilizing community-based representative data of DHS and the use of multilevel mixed-effects analysis which bring disaggregated data on individual characteristics and community-level determinants for designing contextual interventions. aware that our research may have three limitations. The rst is ndings based quantitative cannot explore the detailed reasons in the community for low levels of postnatal care The second is excluding men’s and other community-level signicant others view, may not give fully address the community-level determinants. Lastly, the information on postnatal care provided by mothers was retrospective and those women who had given birth 2 years ago can't probably remember accurately the service they received. These limitations are evidence of the DHS women’s data not inclusive of the above problems.


Background
Globally, approximately about 810 women die from preventable causes related to pregnancy and childbirth a day; with 94% of the death occur in low and lower-middle-income (LMIC) countries [1].
Evidence revealed that most of the maternal death recorded in the early postnatal period. For instance more than 80% of the mortality in the rst fourteen days after birth, and more than 60% in the post-natal period in both developing countries and the United States [2]. Yet, this is the most neglected period in the provision of quality care [3].
According to the World Health Organization(WHO), the postnatal period begins immediately after childbirth and lasts six weeks [4]. It is an essential element across the continuum of care in the maternal and child health care services, and serve as a gateway for family planning services. Woman and her partner/family require more information than they usually receive on the care of the baby and mother within the rst week after childbirth, to stay safe from maternal and neonatal complications [5].
Although there is a good improvement in maternal and child health care services coverage recently, less than one in ve women received a postnatal check within the rst two days of birth in Ethiopia [6]. The experiences and expectations of women and their families and barriers to the uptake of services and/or access to services also should be considered for improving postnatal care services in the country [4].
Although this approach is interesting, it fails to take into account the community-level variables, postnatal care clients were living in a community with different social contexts. A multilevel approach is decisive to understand both maternal and community-level factors, however, no studies in Ethiopia considered such analysis for postnatal checkups. Therefore, the aim of this study is to estimate the extent of maternal and community factors associated with postnatal checkups in Ethiopia.

Study setting
Ethiopia is the second populous country in Africa next to Nigeria bordered by Eretria, South Sudan, Sudan, Djibouti, and Somalia with a total of more than one hundred ten million people [17]. The country has a decentralized three-tier health care system named as primary, secondary, and tertiary care. At the primary level of health care, there were district hospitals, health centers, and their satellite health posts; the secondary level health cares are general hospitals, and the tertiary level health care are specialized hospitals [18]. Administratively, Ethiopia is divided into nine geographical regions (Tigray, Afar, Amhara, Oromia, Somali, Benishangul-Gumuz, SNNPR, Gambella, and Harari) and two administrative cities, Addis Ababa and Diredawa. The 2016 EDHS was a population-based cross-sectional study conducted from January 18, 2016, to June 27, 2016, across the country [6].

Data Source
The dataset utilized in this study was obtained from the DHS program accessed from http://dhsprogram.com/data/. The DHS program o ce gave an authorization letter to access the 2016 EDHS, which is the fourth comprehensive survey. The 2016 EDHS sample was selected in two stages. In the rst stage, a total of 645 clusters (202 in urban and 443 in rural) were selected randomly proportional to the household size from the sampling strata, and in the second stage, 28 households per cluster were selected using systematic random sampling [6]. Representative samples of 18008 households were selected and 16, 650 households were interviewed in 2016 EDHS. For an individual interview, 16, 583 eligible women were identi ed from the interviewed household. Interviews were completed with 15,683 women aged 15-49 years [18]. In this study, among women aged 15-49 years, 3984 gave birth in the two years before the survey were included.

Dependent variable
The outcome variable was categorized as "use of postnatal care coded as 1" and "do not use of postnatal care coded as 0" within two days after birth.

Independent variables
The independent variables for postnatal care use were broadly classi ed into individual/household characteristics and community-level variables in line with a multilevel analytic approach.

Maternal variables
Educational level of women, household wealth index, perceived distance to a health facility to get medical help, employment status of women's, number of ANC visits during pregnancy, delivery by cesarean section, birth order, and place of delivery were included as individual-level variables.
Community-level variables Aggregated variables at the community level based on the individual information and then the aggregated values were classi ed as low and high if the median values or the proportions of the clusters were below and above the national value respectively. Based on this, community-level wealth index, community-level ANC coverage, community-level women's education, perceived distance to a health facility to get medical help at the community level were considered as community-level variables. These variables were selected based on their signi cance in previous studies [19][20][21].

Region
The EDHS sample was collected from nine regions and two administrative cities. We used these administrative boundaries as community-level factors.

Place of residence
It was classi ed as rural (coded as 1) and urban (coded as 0), considered as a community level factor.

Community-level education of women
The median value of educational attainment at the national level was 5 years. Thus, the median value of the aggregated clusters below 5 was classi ed as low education (coded as 0) of women and the median value of the aggregated clusters 5 and more were classi ed as high education (coded as 1) of women.

Community-level wealth
Similarly, the median value of the wealth index at the national level was 3. Then, the aggregated clusters were classi ed as low wealth(coded as 0) and high wealth (coded as 1) by considering the national value as a cut-off point.
Perceived distance to a health facility at the community level At the national level the proportion of perception of distance to a health facility to get medical help as a big problem was 0.45. So, clusters were classi ed as low perception (coded as 0) and high perception (coded as 1) using the national proportion as a cut-off point after aggregated of it.

Community-level ANC visit
The median value of ANC visit at the national level was 3. Hence, the median value of the cluster less than 3 ANC visits were classi ed as low (coded as 1) and greater than or equal to 3 ANC visits were classi ed as high (coded as 1) after aggregated the ANC visit.

Data analysis
Two-level mixed-effects logistic regression analyses were employed using STATA version 14. Since 2016 EDHS data was hierarchical, i.e., individuals (women) were nested in households, and households were nested in the cluster. The unit of analysis for the characteristics of community-level factors was the cluster. For this study, we included 645 clusters in which all the women whose most recent birth was within two years preceding the survey resides.
First, bi-variable two-level mixed-effects logistic regression analyses were done to assess the association between the independent variables and the dependent variable of the study. The overall categorical variables with a p-value of < 0.25 at the bivariate two-level mixed-effect logistic regression analysis were included in the nal model of the multivariable two-level mixed-effect logistic regression model in which odds ratio with 95% con dence intervals were estimated to identify the independent variables of postnatal checkups. P-values less than 0.05 were employed to declare statistical signi cance. Fixed effect and random effect were calculated to assess the individual and cluster variations respectively. Moreover, the frequency table was displayed for the individual and community level variables. All analysis was done on weighted data.
In this analysis four models are displayed, null model (model containing no factors), a model I (containing only individual factors), model II (containing only community factors), and model III (both individual and community-level factors). The tted model was: The intra-class correlation (ICC) was calculated as the proportion of the between cluster variation in the total variation: The variability on the odds of institutional delivery explained by successive models was calculated by Proportional Change in Variance (PCV) as:

Results
Maternal characteristics of the respondents The majority (61.09%) of the women of reproductive age in Ethiopia who gave birth in the two years preceding the survey did not attend formal education. More than seventy-ve percent of the women were unemployed. Still, there is a great problem of access to a health facility in Ethiopia as the majority (61.43%) of the women's perceived distance to a health facility to get medical help was a big problem. Regarding maternal health services in the last two years, only 32.62% percent of the women attended four and more ante-natal care and 35.6% percent gave birth in a health facility (Table 1). Only 17% of the respondents were utilized postnatal checkups with two days among the respondents in Ethiopia during the study period (Fig. 1).
Community-level characteristics of the respondents More than eighty-eight percent of the clusters were from rural areas of Ethiopia. While sixty percent of them were classi ed under higher community-level wealth status. There is also about 75% community level of antenatal care coverage and women's unemployment status revealed. One in three clusters perceived the distance to Health facility to get medical help at the community level is a big problem ( Table 2).

Multi-variable Multilevel Analyses Result
A two-level mixed-effects logistic regression model was used to analyze the effects of community characteristics and women's individual-level factors in postnatal care services utilization in Ethiopia.
As depicted in the empty model, 36.5% of the variations in postnatal checkups could be attributed to community characteristics. The value of the log-likelihood result consistently decreased as tted models progressed from the empty model to Model 1, Model 2, and Model 3 indicating that the tted models were a better t to the data ( Table 5). The higher the ICC, the more relevant were the community characteristics for understanding individual variation in postnatal checkups for mothers. Accordingly, the combined model of maternal, and community factors was selected for determining postnatal checkups in Ethiopia.

Maternal factors associated with postnatal checkups in Ethiopia
The details of the effect sizes of both individual and community-level factors on the odds of postnatal care service utilization are described in Tables 3 and 4. Delivery by cesarean section was independently and signi cantly associated with postnatal care utilization.
After adjusting for individual and community-level factors, the odds of using postnatal care was 1.66 times OR 1.66 (95% CI 1.10, 2.50) higher among women gave birth by cesarean section compared to their counterparts. Similarly, women who gave birth at health facilities were twice higher odds of using postnatal care OR 2.14 (95% CI 1.70, 2.70) as compared to those delivered at home (Table 3).

Community factors associated with postnatal checkups in Ethiopia
There is signi cant administrative regional variation in postnatal care use. The conspicuous observation to emerge from the data comparison was in Oromia 69%, Somali 56%, Benishangul 55%, SNNPR 43%, Gambela 66%, Afar 50% and Dire Dawa 55% which less likely to utilize postnatal care utilization as compared to Addis Ababa city administrative area. While there is no signi cant difference between residents of Harari city, Amhara regional state, and Tigray regional state as compared to Addis Ababa city. Community-level wealth and antenatal care coverage were also found to be signi cant determinants of postnatal care utilization. Community-level wealth was l.4 times OR 1.44 (95% C.I 1.08, 1.2) and community level antenatal care coverage was 1.5 times OR 1.52 (95% C.I 1.19, 1.96) more likely to use postnatal care utilization. Perceived community level distance of health facility as a big problem was found to be highly signi cant determinants of postnatal care. A community who perceived the distance to a health facility as a big problem were 22% less likely OR 0.78 (95% C.I 0.60, 0.99) to utilize postnatal care as compared to their counterparts (Table 4).

Discussion
The ndings of the study are decisive in achieving SDG, the global target of reducing the maternal mortality ratio to less than 70 per 100,000 births [22]. Also, it was vital for the global strategy, aimed at ending preventable maternal mortality through enlightening predictors of inequalities in access to and quality of maternal, and newborn health care services in different clusters of the Ethiopian community [1]. Therefore, identifying maternal characters and community-level determinants of postnatal care has a great contribution in designing different interventions for improving maternal and child health.
This study found different individual and community-level determinants of postnatal care service utilization were identi ed. In this study, 36.5% of the total variance in the odds of postnatal care utilization was accounted for the characteristics of between-cluster variation. While 7.1% at maternal and 6.8% in community-level variation were identi ed. We found much higher values than a cross-sectional study in remote and poorest rural communities of Zambia in 2012, which revealed 22% and study in Nigeria 10.35% [23], variance in the use of PNC within 48 hours were attributable to the variations across community clusters [24]. This implies the issues of equity in postnatal care services for availability and accessibility were still a great challenge for the community.
This study found that women who gave birth by cesarean section were 1.66 times higher odds of using postnatal care services within the rst two days compared to those given birth by another mode of delivery. This is in good agreement with the study in rural Tanzania in 2015 revealed cesarean section delivery was positively associated with postnatal care use [25]. This might be women who gave birth by cesarean section stay at the facility for about two days for which they receive a postnatal check-up.
In another way, women who gave birth at the health facilities were twice more likely to utilize postnatal care as compared to those delivered at home. This substantiates previous ndings in the study in different parts of Ethiopia like Tigray and SNNPR, [9][10][11]25] and Tanzania [25]. This might be clients who visit the health facility would have different health-seeking behaviors with those never attended health facility.
There is signi cant administrative regional variation in postnatal care use in Oromia, Somali, Benishangul, SNNPR, Gambela, Afar, and Dire Dawa which less likely to utilize postnatal care utilization as compared to Addis Ababa city administrative area. While there is no signi cant difference between residents of Harari city, Amhara regional state, and Tigray regional state as compared to Addis Ababa city. Variation in postnatal care service utilization varies was also observed in different parts of Ethiopia [8-10, 26, 27] and West African countries [20]. There are several possible explanations for this nding, rst the difference in geographic accessibility of the postnatal care services due to topography and unfavorable roads for the mothers in rural areas of Ethiopia; second, there was a difference in local cultures and beliefs in different areas of the country; third, the difference in urbanized geographical areas among different regions.
Being in the higher community level antenatal care coverage were 1.5 times more likely to utilize postnatal care services. Our ndings appear to be well supported by a multilevel analysis of DHS in sub-Saharan Africa in 2014, which found signi cant associations between four or more antenatal care visits and ever breastfed with both outcomes [28]. The results point to the likelihood of information diffusion for postnatal care utilization in the community.
Being a resident of higher community-level wealth was l.4 times more likely to utilize post-natal care services. This concurs well with the study in West Africa in 2018, which revealed community-level poverty was a signi cant determinant of postnatal care use [20]. This would appear to indicate that a wealthy community was more probability of getting health information and reside in urban areas.
A community who perceived the distance to a health facility as a big problem were 22% less likely to utilize postnatal care as compared to their counterparts. We believe that no other authors have found that postnatal care utilization is less likely in the community who perceived the distance of health facility as a big problem. There is a good probability that community telling can in uence any health care service utilization.
This study provides strong evidence in utilizing community-based representative data of DHS and the use of multilevel mixed-effects analysis which bring disaggregated data on individual characteristics and community-level determinants for designing contextual interventions.
We are aware that our research may have three limitations. The rst is ndings are based on quantitative data only which cannot explore the detailed reasons in the community for low levels of postnatal care use. The second is excluding men's and other community-level signi cant others view, may not give fully address the community-level determinants. Lastly, the information on postnatal care provided by mothers was retrospective and those women who had given birth 2 years ago can't probably remember accurately the service they received. These limitations are evidence of the DHS women's data not inclusive of the above problems.

Conclusion
Less than one in ve women utilized postnatal care in the rst 2 days after birth in Ethiopia. Both maternal factors and community factors are found to be a signi cant association with postnatal care services utilization, however, based on the ICC maternal factors prevail the community-level factors.
Increasing access to postnatal care services to the remote areas of Ethiopia was recommended. Prevalence of postnatal checkups within two days among the respondents in Ethiopia.