Abstract

Introduction

Pneumococcal disease is a serious global public health concern. The primary causative agent of severe illnesses such as pneumonia, meningitis, acute otitis media, and bacteremia is the pneumococcus bacterium. The pneumococcal conjugate vaccine is a key strategy to reduce the burden of pneumococcal disease. Understanding the spatial distribution of complete childhood pneumococcal conjugate vaccine utilization and its associated factors is crucial for designing strategies to improve vaccination implementation. Therefore, this study aimed to determine the spatial distribution of complete childhood pneumococcal conjugate vaccination coverage and identify its determinants in Ethiopia.

Method

A spatial and multilevel analysis was conducted using data from the 2019 Ethiopian Mini Demographic and Health Survey. The analysis included a total of 2,055 weighted children. The association between the outcome variable and the explanatory variables was determined by calculating adjusted odds ratios at a 95% confidence interval. Explanatory variables were considered significantly associated with the outcome if the p-value was less than 0.05.

Result

The prevalence of complete childhood pneumococcal conjugate vaccination in Ethiopia was 53.94% (95% CI: 51.77, 56.08). Higher complete childhood pneumococcal vaccination coverage was observed in the Addis Ababa, Tigray, Amhara, Benishangul-Gumuz, and Oromia regions, while lower coverage was seen in the Afar, Somali, and SNNPR regions of Ethiopia. Factors significantly associated with complete childhood pneumococcal conjugate vaccination included maternal age, antenatal care visits, place of delivery, region, community women鈥檚 literacy level, community poverty level, and community antenatal care utilization.

Conclusion

The distribution of complete childhood pneumococcal conjugate vaccination exhibited spatial variability across Ethiopia. Approximately half of children aged twelve to thirty-five months received the full dose of the childhood pneumococcal conjugate vaccine in the country. Several factors were identified as statistically significant determinants of complete childhood pneumococcal conjugate vaccination, including maternal age, antenatal care visits, place of delivery, region, community women鈥檚 literacy level, community poverty level, and community ANC utilization. Therefore, policies and strategies aimed at combating pneumococcal disease should consider these determinants and address areas with low vaccination coverage.

Introduction

Pneumococcal disease is a serious global public health concern. The primary microorganism causing serious illnesses such as pneumonia, meningitis, acute otitis media, and bacteremia is pneumococcus [1].

Worldwide, lower respiratory infections (LRI) are estimated to be responsible for 100,000 incidents and one hundred deaths annually per million children under the age of five. Pneumococcal pneumonia accounts for more than half (52.3%) of deaths due to LRI [2]. The pneumococcal conjugate vaccine is a strategy of choice to reduce the disease burden [3]. The World Health Organization (WHO) recommends that children receive the pneumococcal conjugate vaccine (PCV) in three doses [4].

The World Health Organization (WHO) recommends a three-dose schedule for administering pneumococcal conjugate vaccines (PCV) to children since infanthood onward [5]. This schedule can be either 2p鈥�+鈥�1 (two primary doses followed by a booster) or 3p鈥�+鈥�0 (three primary doses without a booster). The vaccination process can begin as early as 6 weeks of age. When choosing between the two schedules, countries should take into account programmatic factors, including vaccination timeliness and expected coverage [5].

Despite the World Health Organization鈥檚 recommendation for universal PCV vaccination, an estimated 51% of infants globally had not completed the full pneumococcal conjugate vaccine series by 2020 [6]. While coverage of the PCV booster dose has seen some improvement, nearly three-quarters of infants worldwide still did not receive the recommended booster shot as of the latest data [7]. The complete dose coverage of PCV in Africa varies from 68% in 2022 to 70% of 2017 despite expected coverage of 90% [8]. In Ethiopia, 74% of children aged 12鈥�23 months received the first dose of the pneumococcal conjugate vaccine (PCV1), while 60% received the third dose of PCV series by 23 months of age as per the Ethiopian mini demographic and health survey (2019) [9]. Geographically significant variation of complete basic childhood vaccination is observed in Ethiopia [10]. Among the factors that affect its coverage, maternal age, maternal education, religion, place of delivery, ANC visit, region, and residence were identified as significantly associated with complete basic childhood vaccination [10, 11].

Limited research has been conducted on the geographical variations and factors that influence vaccine coverage for specific vaccines included in the national expanded program on immunization (EPI) [12, 13]. However, to the best of the researchers鈥� knowledge, there is no up-to-date data available on the distribution and determinants of PCV vaccination across the administrative regions of Ethiopia. Understanding the patterns of complete childhood PCV utilization would be crucial for designing strategies to help improve implementation of PCV in Ethiopia. Thus, this study aimed to determine the spatial distribution of complete childhood pneumococcal conjugate vaccination and its associated factors in Ethiopia.

Method

Data

The secondary analysis of Ethiopian mini demographic and health survey (EMDHS) 2019 data was done to determine spatial distribution, prevalence and determinants of complete childhood pneumococcal conjugate vaccine in Ethiopia. The dataset was accessed from DHS official website () through formal registration and request. Registration is the sole requirement for access to the dataset. This study used children鈥檚 recode data which was extracted from the EMDHS 2019 dataset. Following data extraction, variables were cleaned, recoded and the weighting was done using the weighting variable. Whilst STATA version 14 and Microsoft excel 2013 were used for data management, ArcGIS version 10.8 was used for mapping complete childhood pneumococcal conjugate vaccination in Ethiopia. The data were weighted using the following variables to aid in drawing accurate conclusions: sample weight (v005), primary sampling unit (v021), and stratum (v023). The study was conducted in Ethiopia; an East African country. There are eleven provinces in Ethiopia, namely, Tigray, Afar, Amhara, Oromia, Somali, Benishangul, SNNPR, Gambela, Harari, Addis Ababa and Dire Dawa.

The source population for this study were all children aged 12 to 35 months in Ethiopia. The study population consisted of all children aged 12 to 35 months living in the selected enumeration areas (EAs) or clusters. The 2019 EDHS used a two-stage cluster sampling technique. In the first stage, 305 clusters/EAs were randomly selected using probability sampling, stratified by urban and rural areas. In the second stage, 30 households per cluster were selected using probability sampling. Mothers of all children under five years old in each selected household were asked about the vaccination status of their children.

Variables

The outcome variable for this study was complete childhood PCV vaccination coverage, dichotomized as 鈥測es鈥� for children who received all three recommended doses of the PCV since 6 weeks after birth, and 鈥渘o鈥� for those who received less than three doses. Since the DHS data has a hierarchical structure, the study included independent variables at two levels. The individual (household) level variables were maternal age, maternal education, household wealth status (wealth index), antenatal care (ANC) visits during pregnancy, place of delivery, and birth order. The community-level variables included residence, region, community women鈥檚 literacy level, community poverty level, and community ANC utilization.

The researchers took a multi-step approach to creating community-level measures of illiteracy, ANC and poverty from the individual-level data. First, the individual-level variables of maternal educational status and household wealth status (wealth index) were recategorized. Next, cross-tabulations were performed between these recategorized individual-level variables and the cluster variable using Stata statistical software. This allowed the researchers to compute the proportion of illiteracy and poverty within each cluster or community. These community-level proportions of illiteracy and poverty were then calculated in Microsoft Excel. Finally, the computed community-level proportions were imported back into Stata and combined with the original dataset containing the individual-level variables. The community-level proportions of illiteracy and poverty were subsequently categorized into levels. The variables in this study were categorized and coded (recoded) based on analytical framework used by EMDHS 2019.

Spatial analysis

We employed spatial autocorrelation and spatial clustering techniques to assess the spatial dependency and clustering of complete childhood pneumococcal vaccination in Ethiopia. These methods allowed us to identify the extent to which the vaccination coverage was spatially correlated and whether there were any clustering patterns present in the data.

To further investigate the geographic distribution of complete childhood pneumococcal conjugate vaccine utilization in Ethiopia, we used an optimized hot spot analysis technique. This approach helped identify both the hot spot (high-coverage) and cold spot (low-coverage) areas, providing insights into the spatial heterogeneity of the vaccination program.

We also utilized the spatial SaTScan statistics, which employs a circular scanning window that moves across the study area. This technique was used to determine any statistically significant clusters of the outcome variable, complete childhood pneumococcal conjugate vaccination. By fitting the case, control, and geographic coordinate data to the Bernoulli model, we were able to calculate the Log Likelihood Ratio (LLR), Relative Risk (RR), and p-values for each potential cluster to identify areas with significantly higher or lower than expected vaccination coverage.

Finally, we employed a spatial interpolation technique called Kriging to predict the complete childhood pneumococcal conjugate vaccination coverage in unsampled areas of Ethiopia, based on the observed data from the sampled regions. This allowed us to estimate the vaccination coverage for the entire country, even in areas where data was not directly collected.

Multilevel mixed effect analysis

The nature of variables in demographic and health survey data is hierarchical into two levels. The first hierarchy contains variables within the cluster and the second hierarchy contains variables between clusters. This character of DHS data allowed us to use multilevel or mixed effect logistic regression instead of ordinary logistic regression. There are four levels of models in multilevel logistic regression; null model, model I, model II and model III. Null model-a model without exposure variables, was used to determine variability of complete childhood pneumococcal conjugate vaccine utilization among clusters. In model I, the association of individual level factors with outcome was assessed and in model II, the association of community level factors with outcome variable was assessed. The final model (model III) was fitted to assess the association of mixed (individual and community) factors with complete childhood pneumococcal conjugate vaccination. The regression equation for multilevel (mixed effect) logistic model is given as follows [14];

$$\eqalign{\log ({{\prod ij} \over {1 - \prod ij}}) & = \beta o + \beta 1x1ij + \cdots \cr & \quad + \beta nxnij + uoj + eij \cr}$$

Where; 蟺ij is the probability of having complete childhood pneumococcal conjugate vaccine uptake, (1-蟺ij) is the probability of not having complete childhood pneumococcal conjugate vaccine uptake, 尾o is log odds of the intercept, 尾1, 鈥� 尾n are effect sizes of individual and community-level factors, x1ij鈥� xnij are independent variables of individuals and communities. The quantities uoj and eij are random errors at cluster levels and individual levels, respectively.

The measure of variation (random effect) was estimated by determining the median odds ratio (MOR), intra-cluster correlation coefficient (ICC), and proportional change in variance (PCV*). The MOR quantifies unexplained cluster heterogeneity. When two enumeration areas (clusters) are chosen at random, it is the median value of the odds ratio between a cluster with high odds of complete childhood pneumococcal conjugate vaccine immunization and a cluster with low odds of complete childhood pneumococcal conjugate vaccine immunization. The ICC measures the variation in complete childhood pneumococcal conjugate vaccine immunization across clusters. The PCV* is a measure of the total variation in complete childhood pneumococcal conjugate vaccine immunization as a result of individual and/or community-level variables.

The parameters-MOR, ICC and PCV* are equated as follows [15]; $\:MOR={\:e}^{0.95\sqrt{VC}}$, $\:ICC=\frac{VC}{VC+3.29}\times\:100\%$ and PCV*=(Vnull-VC)/Vnull脳100%), where; VC=variance of the cluster for respective model and Vnull= variance of the null model.

Model comparison was conducted by examining the log likelihood (LL) and deviance, the latter being equal to -2 (log likelihood) due to the nested nature of the model. The measures of association (fixed effects) were used to estimate the strength of the relationships, which were presented as adjusted odds ratios (AORs) with 95% confidence intervals and a statistical significance threshold of p鈥�&濒迟;鈥�0.05.

Result

Socio demographic characteristics of study participants

In this study, a total of 2,055 children (1,009 males and 1,047 females) aged 12 to 35 months were included in the analysis. Approximately half (50.6%) of the children were born to mothers with no formal education, while the majority (81.2%) were born to mothers aged 20鈥�35 years old. About three-fourths (74.6%) of the children were born to mothers who had received ANC visits. Additionally, half (50.2%) of the study participants were born at a health facility. Nearly three-fourths (72.7%) of the participants resided in rural areas. Furthermore, more than half (54.6%) of the subjects belonged to a community with a high level of women鈥檚 literacy (as shown in Table听1).

Table 1 Socio-demographic and economic characteristics of study participants (n鈥�=鈥�2,055)

Prevalence of complete childhood pneumococcal conjugate vaccination in Ethiopia

The prevalence of complete childhood pneumococcal conjugate vaccination in Ethiopia was 53.94% (95% confidence interval: 51.77鈥�56.08%). Notably, the highest coverage (90.27%) was observed in Addis Ababa, while the lowest coverage (20.21%) was found in the Somali region of the country (see Fig.听1).

Fig. 1

Regional coverage of complete childhood pneumococcal conjugate vaccination in Ethiopia; SNNPR: Southern Nations Nationalities and Peoples Region

Spatial autocorrelation and clustering

The output of the spatial autocorrelation analysis shows that the distribution of complete childhood pneumococcal conjugate vaccination was significantly varying across Ethiopia (Moran鈥檚 Index: 0.437599, Z-score: 9.561061, p-value鈥�<鈥�0.0001) (Fig.听2). This indicates that there was a spatial variation in complete childhood pneumococcal conjugate vaccine utilization in Ethiopia. The high clustering of complete pneumococcal conjugate vaccination was observed in the Getis-Ord General G statistic (Z-score鈥�=鈥�6.533907, p-value鈥�<鈥�0.0001) (Fig.听3).

Fig. 2

Spatial autocorrelation of complete childhood pneumococcal conjugate vaccination in Ethiopia

Fig. 3

Spatial clustering of complete childhood pneumococcal conjugate vaccination in Ethiopia

Hot spot analysis of complete childhood pneumococcal conjugate vaccination in Ethiopia

The distribution of hot and cold spots for complete childhood pneumococcal conjugate vaccination in Ethiopia was determined using optimized hotspot analysis. The study found that the average distance between a cluster and at least one neighboring cluster was 30听km, while the maximum calculated band distance from a neighboring cluster for clustering complete childhood pneumococcal conjugate vaccination was 60听km. After conducting the optimized hotspot analysis, researchers identified 100 statistically significant spots鈥�58 hotspots and 42 cold spots. Notably, Addis Ababa, Tigiray, Amhara, Benishangul-Gumuz, and Oromia provinces were identified as hotspots for complete childhood pneumococcal vaccination, whereas Afar, Somali, and SNNPR were cold spot areas (as shown in Fig.听4).

Fig. 4

Hot spot analysis of complete childhood pneumococcal conjugate vaccination in Ethiopia, PCV: Pneumococcal Vaccine

SaTScan analysis of complete childhood pneumococcal conjugate vaccination in Ethiopia

In the SaTScan analysis, a total of 281 significant clusters were identified. Among these, 119 primary (most likely) clusters were located at coordinates 11.166027 North and 36.342701 East, with a radius of 397.04听km. Children living within the primary cluster were 76% more likely to have complete pneumococcal conjugate vaccination compared to children outside the cluster (relative risk鈥�=鈥�1.76 and log-likelihood ratio鈥�=鈥�101.99, p-value鈥�<鈥�0.001) (see Table听2). The most likely clusters for complete childhood pneumococcal conjugate vaccination were predominantly found in Tigray, Amhara, Benishangul Gumuz, Addis Ababa, Harari, Dire Dawa, and some parts of Oromia (as shown in Fig.听5).

Table 2 SaTScan analysis of complete childhood pneumococcal conjugate vaccination in Ethiopia

Fig. 5

SaTScan analysis of complete childhood pneumococcal conjugate vaccination in Ethiopia, PCV: Pneumococcal Vaccine

Prediction of complete childhood pneumococcal conjugate vaccination in Ethiopia by using kriging interpolation

Figure听6 below displays the output of kriging interpolation analysis for complete childhood pneumococcal conjugate vaccination in Ethiopia. The map uses color coding to indicate predictions: green represents areas predicted to have high vaccination rates, while red and orange indicate areas with lower predicted vaccination rates. Specifically, the provinces of Addis Ababa, Amhara, the western part of Benishangul Gumuz, Dire Dawa, Harari, the southeast of Gambela, and Oromia (west and north parts) as well as Tigray were predicted to have higher rates of complete childhood pneumococcal vaccination. In contrast, Afar, Somali, SNNPR, the northern part of Benishangul Gumuz, Oromia (east and south parts), and some portions of Gambela were regions with lower predicted rates of complete childhood pneumococcal vaccination in Ethiopia (as shown in Fig.听6).

Fig. 6

Kriging interpolation of complete childhood pneumococcal conjugate vaccination in Ethiopia, PCV: Pneumococcal Vaccine

Determinants of complete childhood pneumococcal conjugate vaccination in Ethiopia

The fixed effect (Measures of Association)

According to the results of the best fitted model (Model III) of the mixed effect logistic regression analysis, a total of seven factors - including individual or household level (maternal age, ANC visit, place of delivery) and community level (region, community women literacy level, community poverty level and community ANC utilization) - were found to be significantly associated with complete childhood pneumococcal conjugate vaccination in Ethiopia.

In this respect, the odds of complete childhood pneumococcal conjugate vaccination were significantly higher among certain demographic groups. For instance, children born to mothers in the 20鈥�35 years (AOR鈥�=鈥�2.40, 95% CI: 1.61, 3.62) and 36鈥�49 years (AOR鈥�=鈥�2.42, 95% CI: 1.45, 4.03) age groups had 2.40 and 2.42 times higher odds of vaccination, respectively, compared to those born to mothers aged 15鈥�19 years.

In addition, children born to mothers who had antenatal care (ANC) visits during pregnancy had 2.35 (AOR鈥�=鈥�2.35, 95% CI: 1.81, 3.08) times higher odds of complete vaccination compared to those whose mothers did not have any ANC visits. Additionally, babies born at health facilities had 1.51 (AOR鈥�=鈥�1.51, 95% CI: 1.20, 1.92) times higher odds of vaccination than those born at home.

Community-level factors were also important predictors. The odds of complete pneumococcal conjugate vaccination were significantly elevated among children living in Tigray (AOR鈥�=鈥�2.96, 95% CI: 1.41, 6.05), Amhara (AOR鈥�=鈥�3.83, 95% CI: 1.98, 7.43), Benishangul (AOR鈥�=鈥�2.70, 95% CI: 1.36, 5.29), Addis Ababa (AOR鈥�=鈥�5.65, 95% CI: 2.43, 13.14), and Dire Dawa (AOR鈥�=鈥�2.11, 95% CI: 1.15, 4.20) regions compared to those in the Somali region. Children living in communities with high women literacy levels had 1.23 (AOR鈥�=鈥�1.21, 95% CI: 0.91, 1.69) times higher odds of complete vaccination compared to low literacy communities. Additionally, the odds were 1.54 (AOR鈥�=鈥�1.54, 95% CI: 1.10, 2.17) times higher for children in low poverty communities versus high poverty communities. Finally, children in communities with high ANC utilization had 1.48 (AOR鈥�=鈥�1.48, 95% CI: 1.07, 2.08) times higher odds of complete vaccination compared to the reference group (see Table听3).

The random effect (measures of variation) and model fit statistics

The results of the random effect analysis in the null model indicate that the variations between clusters account for 35% of the total variation in complete childhood pneumococcal conjugate vaccination. Specifically, the intracluster correlation coefficient (ICC) is estimated to be 0.35 (with a 95% confidence interval of 0.29 to 0.43). Additionally, the unexplained heterogeneity, as measured by the median odds ratio (MOR), is 3.63 in the null model. This means that if a child is randomly selected from two separate clusters, a child from a cluster with a high likelihood of complete childhood pneumococcal vaccination is 3.63 times more likely to be fully vaccinated compared to a child from a cluster with a low likelihood of complete vaccination.

In model I, 21% of the overall variation in complete childhood pneumococcal vaccination can be attributed to variation between clusters. The corresponding ICC value is 0.21 (with a 95% confidence interval of 0.16 to 0.28). In addition, more than half of the variability in complete childhood pneumococcal vaccination can be attributed to individual factors (PCV鈥�=鈥�51.28%) in model I. Furthermore, approximately 67% of the variability in complete childhood pneumococcal vaccination is associated with community-level factors (67.35%). The results from the random effect analysis of the final model revealed that both individual and community-level factors jointly account for approximately 74% of the variability in complete childhood pneumococcal conjugate vaccination (PCV鈥�=鈥�73.60%). Additionally, the unexplained heterogeneity is significantly reduced to 1.94.

To determine the best-fitted model, we consider the large value of the likelihood ratio test statistic (LLR) and the lowest value of deviance. Based on these criteria, the final model (model III) emerges as the most suitable choice. It exhibits the highest LLR (LLR = -1571.75) and the lowest deviance (deviance鈥�=鈥�3,143.5), as summarized in Table听4.

Table 3 Multilevel mixed effect regression analysis of factors for complete childhood pneumococcal conjugate vaccination in Ethiopia

Table 4 Measures of variation and model fit statistics of complete childhood pneumococcal vaccination in Ethiopia

Discussion

This study disclosed coverage, spatial distribution, and determinants of complete childhood pneumococcal conjugate vaccination in Ethiopia using EMDHS 2019 dataset.

The prevalence of complete childhood pneumococcal conjugate vaccine (PCV) immunization in Ethiopia was 53.94% (95% confidence interval: 51.77鈥�56.08). This prevalence exceeded that reported in a previous study conducted in Ethiopia (49.1) [16], as well as the prevalence in the Western Pacific (19%) and South-East Asia (29%) WHO regions [8]. The higher coverage of complete childhood PCV immunization observed in our study, compared to the previous study in Ethiopia, may be attributed to the utilization of more recent data from the Ethiopian Mini Demographic and Health Survey (EMDHS) conducted in 2019. Generally, immunization coverage tends to be higher when based on more recent DHS data.

Furthermore, the higher PCV vaccination coverage observed in our study compared to the Western Pacific and South-East Asian WHO regions could be influenced by the study setting. Our research focused on a single country, whereas these WHO regions aggregated data from several low and middle-income nations. Despite challenges posed by political instability, such as civil unrest, government changes, and protests, our findings suggest an improvement in PCV vaccination rates.

However, it鈥檚 important to note that our vaccination coverage was still lower than that reported in a study from Singapore (70.70%) [17] and the overall coverage in other WHO regions: Africa (68%), Americas (74%), Eastern Mediterranean (54%), and Europe (82%) [8]. This inconsistency may be related to variations in health infrastructure, socio-economic factors, and cultural differences among countries. Consequently, children aged 12 to 35 months in Ethiopia had relatively lower coverage of complete PCV vaccination.

The spatial analysis of this study revealed that complete childhood pneumococcal conjugate vaccine immunization varied across Ethiopia. Similar finding were evidenced by previous studies [10, 18, 19]. Accordingly, the higher complete childhood pneumococcal conjugate vaccination was observed in Addis Ababa, Tigiray, Amahara, Beninshangul Gumuz, and Oromia provinces, while the lower vaccination was observed in the Afar, Somali, and SNNPR regions of Ethiopia. The possible reason for low PCV immunization in these areas is low health care access and transportation infrastructure. Poor mothers鈥� healthcare-seeking behaviours, inadequate knowledge about childhood immunization, limited access to vaccines, and misconceptions about immunization among nomadic and pastoralist communities could all contribute to low immunization coverage in these locations [20, 21]. Given that nomadic and pastoralist populations travel from place to place periodically, access to health care services, including immunization, is challenging [22, 23].

In the multi-level (mixed effect) logistic regression analysis of this study; maternal age, ANC visit, place of delivery, region, community women literacy level, community poverty level and community ANC utilization were significantly associated with complete childhood pneumococcal conjugate vaccination in Ethiopia.

In congruent with previous studies [10], maternal age was significantly associated with complete childhood pneumococcal conjugate vaccine utilization. Accordingly, the odds of complete childhood pneumococcal conjugate vaccination were higher among children born to mothers in age groups of 20鈥�35 years and 36鈥�49 years compared to babies born to mothers in age group of 15鈥�19 years. As maternal age increases, there may be a proportional increase in the use of maternal health care services such as ANC visits, deliveries in medical facilities, and PNC visits as an initial course in immunization for children [24].

Antenatal care utilization was significantly associated with a complete schedule of childhood pneumococcal conjugate vaccination, both at the household and community level. When compared to children born to mothers who had not had ANC visits during pregnancy, the odds of complete pneumococcal conjugate vaccination were higher among children born to mothers who had ANC visits. In addition, the odds of complete childhood pneumococcal conjugate vaccination were higher among children living in a community with a high level of ANC utilization compared to children living in a community with a low level of ANC utilization. Previous studies have also reported the same [10, 16]. This could be due to the fact that ANC visits might have improved the mothers鈥� knowledge and attitude towards the importance of vaccinating their children. Moreover, prior experience and health education from professionals during an ANC visit might build trust towards vaccination services, which is important for reducing the vaccine confidence gap in the community [16].

Regarding the place of delivery, babies born at health facility had higher odds of complete childhood pneumococcal conjugate vaccination compared to babies born at home. This was in agreement with previous study [10]. One probable explanation could be that babies delivered in medical institutions receive the BCG and Polio 0 vaccines at birth, and mothers can receive additional information about complete PCV immunization schedules [25].

Administrative region was significantly associated with complete childhood pneumococcal conjugate vaccination. Previous related study also supports this finding [16]. The odds of complete pneumococcal conjugate vaccination were higher among children living in Tigray, Amhara, Benishangul, Addis Ababa, and Dire Dawa regions compared to children living in Somali region. In Somali region, weak health infrastructure and low density of population were reported as the challenges for immunization service delivery [26]. In addition, Somali region is particularly a region of dominant pastoralist nomadic community [26].

The odds of complete childhood pneumococcal conjugate vaccination were higher among children living in a community of high women literacy level compared to children living in a community of low women literacy level. Previous study also reported similar finding [18]. It is not surprising that children living in a community of educated women have higher odds of PCV vaccination, since educated mothers are wise enough about the benefits of vaccinating their children and ways of disease prevention. This demonstrates the need for schooling in communities with low literacy. Education-based programs have been successful in raising vaccination awareness and increasing demand [27].

Moreover, taking children living in a community of high poverty level as reference, the odds of complete childhood pneumococcal conjugate vaccination were higher for children living in a community of low poverty level. Other studies have also found similar finding [15, 28]. Even though expanded program on immunization offers free immunizations and ongoing public outreach to disadvantaged mothers and infants, there may be a correlation between wealthier communities鈥� increased access to healthcare, improved health-seeking behaviors, and increased child care practices [29].

Although this study employed nationally representative data and appropriate models, it is essential to consider the following limitations. Firstly, the study relied on secondary data, which precluded the inclusion of variables that could be associated with outcomes. Secondly, the study findings may have been influenced by spatial analysis biases, such as map area bias and projection bias.

Conclusion

According to this study, the distribution of complete childhood pneumococcal conjugate vaccination coverage was found to be spatially variable across different regions of Ethiopia. Notably, the study revealed that only approximately half of children aged 12 to 35 months had received the full recommended doses of the childhood pneumococcal conjugate vaccine nationwide. The study identified several statistically significant determinants of complete childhood pneumococcal conjugate vaccination, including maternal age, antenatal care (ANC) visit frequency, place of delivery, region of residence, community-level women鈥檚 literacy, community poverty level, and community ANC utilization. These findings underscore the need for policymakers and public health strategists to take these identified determinants and the observed geographic disparities in vaccination coverage into account when developing and implementing policies and interventions aimed at addressing pneumococcal disease in the country.