Background
Antenatal care (ANC) is a maternal healthcare service provided by skilled healthcare professionals to pregnant women and adolescent girls. It is provided throughout pregnancy to ensure the best health outcomes for both the mother and the newborn. The care service includes the following components: risk identification, prevention and management of pregnancy-related or concurrent diseases, and health education and health promotion [
1]. The World Health Organization (WHO) recommends Midwife-led continuity of care models throughout pregnancy, delivery and postnatal period. Antenatal home visits are also recommended to improve antenatal care utilization and perinatal health outcomes [
1].
Antenatal care has the potential to reduce maternal and child morbidity and/or mortality and to improve newborn health [
2‐
6]. For instance, in a systematic review and meta-analysis, it was found that ANC visit was significantly associated with a 34% reduction in neonatal mortality [
5]. In a cohort study carried out in Ethiopia, it was reported that having four or more ANC visits was significantly associated with 81.2, 61.3, 52.4 and 46.5% reduction in postpartum haemorrhage, early neonatal death, preterm labour and low-birth weight, respectively [
6]. Inadequate antenatal care visits, or late visits, or fewer than the recommended number of visits have been related to poor pregnancy outcomes [
7]. A lack of relevant and high quality antenatal care services is a major concern in sub-Saharan Africa [
8].
In the case of uncomplicated pregnancies, the 2002 Focused Antenatal Care model of the WHO recommended at least four antenatal care visits; the first visit to take place before 16 weeks of gestation [
9]. However, this model has now been superseded by the 2016 WHO ANC model; where a minimum of eight ANC contacts is recommended. The word “visit” in the previous model has now changed to “contact” to indicate an active interaction between a pregnant woman and a health-care provider. The first contact should take place in the first trimester; that is, within the first 12 weeks of gestation. The other recommended contacts are two contacts in the second trimester (at 20 and 26 weeks) and five contacts in the third trimester (at 30, 34, 36, 38 and 40 weeks of gestation) [
1].
In the developing regions of the world, according to the United Nations report, over the 25 years period (1990 to 2014), there was slow progress in the use of four or more antenatal care visits [
10]. In 2014, on average, only 52% of pregnant women in the developing regions had received a minimum of four antenatal care visits, which was a 17% increase from 1990. It was reported that 36% of pregnant women in Southern Asia and 49% in sub-Saharan Africa had received four or more antenatal care visits in 2014 [
10]. In Ethiopia, the rates of women using antenatal care at least once increased from 27 to 62% from 2000 to 2016 [
11‐
14]. In 2016, 31.8% of Ethiopian women received four or more ANC visits [
14] and the median timing of the first antenatal care visit was 5.2 months [
13]. About 63 and 89% of women from urban areas and Addis Ababa respectively had four or more antenatal care visits as compared to 27% from rural areas and those from the Somali region (11.8%).
Both the demand and supply side factors are important in determining health service use. However, the majority of previous studies were mainly focused on the demand side factors of health service use. For instance, most studies identified the demand side factors of antenatal care use [
15‐
20]. Amongst the demand side factors, women’s education [
16,
21,
22], husband’s occupation [
21], socioeconomic status [
16], and place of residence [
16,
17] were significantly associated with the use of antenatal care service. In most studies, the supply side factors of ANC use have been overlooked. Understaffed health facilities [
23] and distant ANC facilities [
23,
24] were negatively associated with the use of antenatal care services. The most important supply side factors, such as health facilities general service readiness, availability of ANC services and facilities readiness to provide ANC service [
25,
26] were not addressed. Due to the increasing availability of georeferenced health facility and population data, it is important to link these data sets and identify both the demand and supply side factors of ANC use.
So far, in Ethiopia, four Demographic and Health Surveys (DHS) were conducted, but the Service Provision Assessment (SPA) survey was the first to be carried out in the country. The DHS survey provides detail information about population characteristics including their health service use history [
27]. On the other hand, the SPA survey provides information about healthcare services available at each health facility and facility’s readiness to provide a particular service [
28]. Using the capability of geographic information system (GIS), a linked analysis of population and health facility data has enormous importance for investigating the links between population healthcare needs and uses, and the health service environment [
29]. Despite this importance, these two datasets have not been used in Ethiopia. Therefore, this study aimed to assess the spatial variations in the use of antenatal care services among women who gave birth in a 5 year period in Ethiopia. Furthermore, it aimed to identify the potential factors associated with the use of antenatal care services throughout the country using national population and health facility data.
Discussion
This study aimed to assess the spatial variations in the use of ANC in Ethiopia. Furthermore, it aimed to identify factors associated with ANC use throughout the country, using the national population and health facility data. This is the first study to provide a comprehensive assessment of ANC use in Ethiopia by region of living, service type and demographics. ANC visit was found to be a spatial problem in Ethiopia. It was also found that women’s ANC visit was significantly associated with different individual and region level factors.
In Ethiopia, the proportion of at least four ANC visits was 36.78%; the highest proportion was reported in urban areas (66.93% vs 28.41%). More than half of antenatal care visits were started during the second trimester of pregnancy, this was far below the WHO recommendation of having at least four ANC visits [
9] started during the first trimester of pregnancy [
1,
9]. Nevertheless, in 2016, more women had at least four ANC visits as compared to the results of the previous three DHS surveys [
11‐
13]. Similarly, the proportion of women who received ANC in the first trimester was higher than the 2000, 2005 and 2011 DHS survey findings [
11‐
13]. Despite these improvements, this figure is lower than the 2014 United Nations report where 52% of pregnant women in developing regions and 49% in sub-Saharan Africa had a minimum of four ANC visits [
10]. There are also variations in ANC visits across the different regions in the country. The highest proportions, more than 50% with at least four ANC visits, were reported in the Addis Ababa and Dire Dawa city administration and the Tigray region. The lowest, below 20% use of at least four ANC visits, were reported in the Somali and Afar regions. Despite the overall increase in ANC visits in the country, it was found that there was significant regional variation in undertaking four or more ANC visits.
Hot spots of at least four ANC visits were detected in the Southern Nations, Nationalities and Peoples region, and some parts of the Gambela and Oromia regions. The identified hot spots were located closest to teaching hospitals (Jimma, Hawassa, Wolaita and Arba Minch hospitals); they have a high number of service providers including students. These teaching hospitals also have antenatal care services available and are ready to serve the target population. Therefore, women who are living closest to these facilities are more likely to have frequent antenatal care visits, as the services could be more attractive to them.
The majority of cold spots were detected in the Addis Ababa city administration followed by some parts of the Oromia region. This was an unexpected finding, as Addis Ababa is where the majority of health facilities are concentrated. This highlights the need to specify how hot spot analysis works. In hot spot analysis, every feature has a neighborhood and that neighborhood is compared to the study area, and the feature is marked with the result of that comparison. If the neighborhood is significantly different from the study area, then that feature will be marked either a hot spot or a cold spot depending on whether there are high values or low values.
One important note here is that ‘where are the hot spots?’ is not necessarily the same as ‘where are the highest values?’ Hot spot analysis is a test of spatial randomness. In hot spot analysis, one can get a feature of low values, even zero, which is marked as a hot spot because its neighborhood was high enough to bring that local average to be significantly different from the global average. In our study finding, even though the overall prevalence of antenatal care use was high in Addis Ababa, the majority of clusters in the city had very low values. When every neighborhood in the city was compared to the study area, the neighborhood values were significantly lower than the study area. Thus, the spatial statistics marked every feature in Addis Ababa as a cold spot that is the local average is significantly lower than the global average. Due to this low prevalence clusters, Addis Ababa did not show hot spots of ANC4+ visits. Similarly, most of the clusters in Addis Ababa had the least first ANC visits during the first trimester. This could also be explained by the spatial variations of timing of first ANC visits as observed from the Moran eigenvector spatially varying coefficient (M-SVC) regression model. In Addis Ababa, it was found that first ANC visits during the first trimester were not strong predictors of at least four ANC visits as observed in the Amhara and Tigray regions. Even though the statistics gave this finding, further study is required to understand why low rates of at least four ANC visits are clustered in Addis Ababa.
Spatial regression models assume the potential between-neighborhood correlations due to spatial process [
44]. Standard multilevel models, however, do not assume spatial dependence; neighborhood observations are independent of one another [
44,
45]. This could lead to the overstatement of statistical significance of neighborhood effects [
44]. Our paper considered same variables for both spatial and multilevel analysis. However, only wealth quintiles and availability of ANC supplements were shared between the two. This doesn’t mean that they have the same interpretation on ANC visit. Spatial regression models enables to identify what is happening in a particular geographic location and why that is happening. In the spatial regression analysis, it was found that four or more ANC (ANC4+) visits were varied across geographic areas. These geographical variations were explained by different variables, such as first ANC visits, availability of ANC supplements, facilities readiness to provide skilled care and distance to ANC providing facilities. Furthermore, it helped us to explain cold spots of ANC4+ visits in Addis Ababa of which the standard multilevel analysis could not explain it. These enables for informed decision making like which communities and health facilities need especial attention and where should the government spend more money.
Different individual and regional level factors were significantly associated with the use of more ANC visits (ANC4+ visits). Amongst the regional level variables, it was found that a one-unit increase in the mean score of antenatal care service availability in a typical region was significantly associated with a five-fold increase in the odds of having more ANC visits. In Nigeria, it was found that health facilities staffed with fewer antenatal care providers were negatively associated with the use of antenatal care services [
23]. Availability and provision of antenatal care commodities at every antenatal care facility will help to reduce the costs associated with purchasing those drugs and thus improve women’s antenatal care visits. Furthermore, health facilities readiness, for instance, having the required number of physicians or services providers attending pregnant women would minimize the waiting time that a woman could spend at a health facility. This could make the service attractive and hence, get more women for antenatal care services very easily.
Every one-kilometre increase in distance to the nearest ANC providing facilities in a typical region was negatively associated with the odds of having more ANC visits. This finding was supported by a study carried out in Ethiopia where proximity to a health facility [
21] was significantly associated with the use of antenatal care service. Furthermore, a study carried out in Northern India found that living far from a health facility was negatively associated with maternal health service use [
17]. Similarly, distant health facilities were negatively associated with the use of antenatal care services [
23,
24]. However, those women who were living close to obstetric health facility were more likely to use antenatal care services [
46]. In Ethiopia, having access to obstetric care facilities within an-hour travel time was also significantly associated with the use of antenatal care services [
47,
48]. In rural Burkina Faso, pregnant women who had access to obstetric care facility within 5 km was significantly associated with the odds of having at least three antenatal care visits [
49]. This indicates that geographic accessibility, measured in either distance or travel time, has a greater influence on maternal health service utilization [
50].
Similarly, pregnant women who were living in rural areas were 47% less likely to have at least four ANC visits as compared to urban women. This finding was supported by another study carried out in Ethiopia where place of residence as well as administrative region were significantly associated with antenatal care use [
51,
52]. Furthermore, this was in agreement with previous studies carried out in Ghana [
19], Vietnam [
15], Nigeria [
16] and Northern India [
17]. Worldwide, the inequalities in the distribution of health facilities were reflected by the higher proportion of antenatal care use in urban centres as compared to rural areas [
18,
53]. These disparities could be due to the local inaccessibility of obstetric care services in rural areas as well as variations in some regional administrations. Therefore, the government and other service providers should work together for improving communities’ easy access to healthcare services.
Amongst the individual-level factors, women’s autonomy in their own healthcare decision was significantly associated with the odds of having more ANC visits. A woman whose husband/partner made decisions on her own healthcare was 24% less likely to have at least four ANC visits as compared to a woman who had autonomy to make decisions. This was consistent with other study findings where husbands’ approval had a greater effect on the use of antenatal care services [
48,
54]. Another study conducted in Ethiopia found that women’s autonomy on their own healthcare decision-making was significantly associated with the higher odds of using antenatal care service [
52]. Therefore, the empowerment and autonomy of women in all aspects of life, especially in their own healthcare decision is a highly important end in itself.
In Ethiopia, among the individual-level variables, husbands’ level of education was significantly associated with the odds of having more ANC visits. This study found that a woman whose husband had attained a primary level of education was 53% more likely to have at least four ANC visits as compared to those whose husband had no education. This finding adds to previous research conducted in Ethiopia which found that women’s education [
21,
22] and husband’s attitude [
22] were significantly associated with antenatal care use.
The odds of having at least four ANC visits was significantly associated with the increased in household wealth, in agreement with previous research conducted in Ghana [
19] and Nigeria [
24]. In some settings, service fees and socio-economic status were strong predictors of antenatal care use [
19,
20,
55]. In Ethiopia, even though antenatal services are free of charge at government health facilities, services fees at private health facilities as well as transportation costs are high. Moreover, most women end up spending the entire day at health facilities for their check-ups and on travelling to and from health facilities. This kind of indirect cost is associated with the women’s daily life of which they might go to a farm or market to make their daily living. Therefore, women in the highest wealth quintile will be more likely to make more antenatal care visits as compared to women in the lowest quintile.
In Ethiopia, having unwanted pregnancy was negatively associated with the likelihood of having more ANC visits as compared to wanted pregnancies. This finding was consistent with other studies carried out in Ethiopia, as those women who had a wanted pregnancy were more likely to have antenatal care visits [
48,
52,
56]. Women with wanted pregnancies could want to have a healthy pregnancy and childbirth, and thus they might give great attention for antenatal care services.
In this current study, it was found that a one-child increase in the number of living children a woman had was significantly associated with a 7 % decrease in the likelihood of having more ANC visits. This finding was supported by studies carried out in Ghana [
19] and India [
57] where a significant reduction in the use of ANC services was observed with increasing in the number of living children. This could be related to woman’s previous experience, as a woman might be reluctant to have ANC visits in a subsequent pregnancy if she had a negative previous experience or if she perceived the importance of ANC to be low with subsequent pregnancies. To avoid any complications and/or adverse pregnancy outcomes, more attention should be given to encouraging women to have more ANC visits. However, in another study, it was found that high parity was significantly associated with higher uptake of ANC visits [
52]. This could be attributable to previous complications and/or adverse pregnancy outcomes. Furthermore, this could be due to influences of previous ANC visits, in case if they had.
The identified individual and regional level factors such as distance from health facility, socio-economic status, number of living children and women’s autonomy might be related to each other. They do not exist as separate factors in life. For example, in a systematic review, it was found that poor geographic access to health care was overlapped with poverty. In Uganda, those regions with the worst access to health care were the regions where the large segment of the population lived below poverty line [
58]. Distance was found to be a barrier in obtaining health care for 20% of the poorest as compared to only 9 % of the richest population in Uganda [
59]. In low-and middle-income countries, it was found that improving healthcare access could reduce socio-economic gaps in healthcare [
60].
This study linked population and health facility data to identify both the demand and supply side determinants of antenatal care use. This was not the case in most previous studies where they assessed the demand and supply side determinants separately. In addition to the standard multilevel analysis, this study identified geographical variations of ANC use as well as factors associated with these variations. Investigating ANC use geographically is very important for informed decision making and monitoring and evaluation purposes.
Even though this study had several methodological limitations, most of these were minimized. Problems related to sampled facilities, temporal differences between DHS and SPA surveys, and misclassifications errors were minimized [
32]. However, using a straight-line distance introduces bias and this would be reduced if a road network link was carried out. In case of road network analysis, the distance between DHS clusters and health facilities would not be affected by terrain characteristics. It reflects the road distance rather than the shortest distance between two points. Furthermore, analysis of sampled facilities and removing DHS clusters without geographic coordinate information might under or overestimate the study finding. For instance, the influence of distance on ANC use could be different if all health facilities were included. The estimated average straight-line distance to the nearest ANC facility would not be this much high.
Even though multilevel analysis should include weights at each level, this study did not considered sampling weights. The problem with this is that DHS does not provide separate weights for different levels, such as region, cluster or household-level weights. DHS only provides an average weight which is proportional to hv005 or v005. The GLIMMIX procedure, however, asks each level weight. It has OBSWEIGHT = option and WEIGHT = option. The GLIMMIX procedure does not provide any other solution when we have average weights.