https://orcid.org/0000-0002-2911-2095
Figures
Abstract
Background
Teenage pregnancy remains one of the major reproductive health problems, especially in sub-Saharan African countries. It can lead to maternal and neonatal complications and social consequences. The proportion of teenage pregnancy differs across regions of Ethiopia. Thus, this study aimed to determine the spatial variation in determinants of teenage pregnancy among adolescents aged 15–19 years in Ethiopia using the 2019 Demographic and Health Survey (DHS).
Methods
This study included a total weighted sample of 2165 teenage girls aged 15 to 19 years. A mixed-effect binary logistic regression model was employed to consider the hierarchical nature of the DHS data using STATA version 17. Adjusted odds ratios with 95% confidence intervals are reported, and a p-value less than 0.05 was used to identify significant predictors. The spatial analysis was conducted with ArcGIS version 10.7 and Python 3. To identify factors associated with the hotspots of teenage pregnancy, a multiscale geographically weighted regression (MGWR) was performed. Spatial regression models were compared using adjusted R2, the corrected Akaike information criterion (AICc), and the residual sum of squares (RSS).
Results
The prevalence of teenage pregnancy among adolescents aged 15 to 19 years was 12.98% (95% CI: 11.6%, 14.5%). It was spatially clustered throughout the country with a significant Moran’s I value. Significant hotspot areas were detected in central and southern Afar; northern, central, and western Gambela; northeastern and southern central Oromia; and the eastern Somali region. The MGWR analysis revealed that the significant predictors of spatial variations in teenage pregnancy were being illiterate and being married. Based on the multivariable multilevel analysis, age 17 (AOR = 3.54; 95% CI: 1.60, 7.81), 18 (AOR = 8.21; 95% CI: 3.96, 17.0), 19 (AOR = 15.0; 95% CI: 6.84, 32.9), being literate (AOR = 0.57; 95% CI: 0.35, 0.92), being married (AOR = 22.8; 95% CI: 14.1, 37.0), age of household head (AOR = 0.98; 95% CI: 0.98, 0.99) and residing in the Gambela region (AOR = 3.27; 95% CI: 1.21, 8.86) were significantly associated with teenage pregnancy among adolescents aged 15 to 19.
Conclusion
Teenage pregnancy is a public health problem in Ethiopia. Policymakers should prioritize addressing early marriage and improving teenage literacy rates, with a focus on the Gambela region and other hotspot areas. It is crucial to implement policies aimed at transforming the traditional practice of early marriage and to take measures to enhance literacy levels and promote awareness about sexual and reproductive health at the family and school levels. This will help ensure that young people have the opportunity to pursue education and make informed decisions about their reproductive health.
Citation: Tebeje TM, Abebe M, Aragaw FM, Seifu BL, Mare KU, Shewarega ES, et al. (2024) A multiscale geographically weighted regression analysis of teenage pregnancy and associated factors among adolescents aged 15 to 19 in Ethiopia using the 2019 mini-demographic and health survey. PLoS ONE 19(9): e0310025. https://doi.org/10.1371/journal.pone.0310025
Editor: Alfredo Luis Fort, Independent Consultant, UNITED STATES OF AMERICA
Received: August 23, 2023; Accepted: August 22, 2024; Published: September 6, 2024
Copyright: © 2024 Tebeje et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data used in this research are publicly available from the measure DHS program via an online request at https://dhsprogram.com/data/dataset_admin/login_main.cfm?CFID=10818526&CFTOKEN=c131014a480fe56-4E0C6B7F-F551-E6B2-50.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: AICc, corrected Akaike information criterion; AOR, adjusted odds ratio; CI, Confidence interval; CSA, central statistical agency; DHS, Demographic and Health Survey; EAs, Enumeration areas; EDHS, Ethiopian Demographic and Health Survey; GWR, Geographically weighted regression; HHH, Household Head; ICC, Intraclass correlation coefficient; LMICs, low- and middle-income countries; MOR, median odds ratio; MGWR, Multiscale Geographically Weighted Regression; PCV, Proportional change in variance; RSS, Residual sum of squares; OLS, Ordinary least squares; SSA, Sub-Saharan Africa; SNNPR, Southern Nations Nationalities and Peoples Region
Introduction
The highest number of teenage childbirths occurs in developing countries, where many girls marry early in life due to poverty, limited educational opportunities, cultural and social norms, and weak enforcement against child marriage [1–3]. Approximately 20 countries in sub-Saharan Africa (SSA) and Asia are responsible for 82% of adolescent maternal mortality [4]. This high rate could have resulted from the lower utilization of maternal health services by adolescents, coupled with their physical immaturity and discrimination [5,6]. Girls aged 15 to 19 years have a 28% greater risk of maternal mortality compared to women aged 20 to 24 years, influenced by physiological and bio demographic factors [7,8]. Pregnancy-related conditions are a significant cause of death among adolescent girls aged 15–19 years worldwide [9]. Girls aged 15 to 19 years make up 11% of all births worldwide [10]. Although there has been a decline in the rate of adolescent births, the total number of births has not decreased as much due to the growing adolescent population [11]. Each year, adolescents from low- and middle-income countries experience an estimated 21 million pregnancies, resulting in 12 million births [12]. Half of these pregnancies are unintended, and 55% of the unintended pregnancies often end in unsafe abortions [13]. Annually, 3.2 million unsafe abortions take place among adolescents aged 15 to 19, accounting for 15% of the total global incidence [14].
Adolescents have a significant unmet need for sexual and reproductive health care. A total of 43% of girls aged 15 to 19 years have an unmet need for modern contraception, which is higher than that of women of reproductive age, who have a 24% unmet need [12]. Adolescent pregnancies lead to health consequences such as increased risks of eclampsia, puerperal endometritis, and systemic infections. There are also social consequences, including stigma, rejection, violence by partners or parents, and dropping out of school, which are commonly observed [15]. Additionally, adolescent pregnancy is related to lower child survival, which is attributable to biological and social risks [16].
Teenage pregnancy is a significant public health issue in Ethiopia, where there is limited knowledge and access to reproductive health services, restricted educational opportunities, and unwanted marriage at a young age [17]. The country ranks third in terms of adolescent maternal deaths [4]. Data from the 2000, 2005, 2011, and 2016 Ethiopian Demographic and Health Survey (EDHS) show that teenage pregnancy rates in Ethiopia were 16%, 17%, 12%, and 13%, respectively [18–21]. This overall decline demonstrates how public health interventions have effectively altered societal perceptions of early marriage and provided adolescents with access to reproductive health services [22]. However, the inconsistent trends indicate the need for continuous efforts. When broken down by region, the highest percentage was in the Afar region (23%), and the lowest percentage was in the capital, Addis Ababa (3%) [21].
According to previous research, factors related to adolescent pregnancy in LMICs include lower education and socioeconomic status, child marriage, not utilizing contraception, child sexual abuse and early sexual activity, age, religious beliefs, limited knowledge of reproductive and sexual health, gender inequality, physical or sexual violence, lower family income, being in the 18–19 age group, maternal history of teenage pregnancy, and inadequate communication with parents about reproductive health issues [23–28].
Fertility and childbearing among young people are among the most neglected issues in Ethiopia [29]. Ethiopian efforts for adolescents and youth are dispersed across ministries, lack coordination and meaningful youth involvement, have insufficient funding, focus on specific projects, and ineffectively implement policies [30]. Some similar studies have been conducted in the country [31–35], but they were based on the 2016 EDHS. Four of the aforementioned studies [31–34] investigated the trend and determinants of teenage pregnancy and did not take into account the geographical variation in teenage pregnancy in Ethiopia. However, one of them [35] examined regional disparities and studied the factors that influence regional disparities in teenage pregnancy using geographically weighted regression (GWR). Our study used the most recent data from the mini-EDHS 2019 to provide updated information. Additionally, this study was conducted to address the limitation of the previous study, which used GWR to assess the relationship between teenage pregnancy and its determinants. GWR assumes that the relationship between the outcome and its predictors operates at the same spatial scale. This disregards the fact that different relationships may occur at various scales due to heterogeneity or nonstationarity in geospatial relationships. To address this methodological gap, we used Multiscale Geographically Weighted Regression (MGWR), which considers multiple scales and relaxes the assumption of GWR by determining parameter-specific bandwidths (the spatial scale at which the underlying spatial processes operate). This allows the relationships between independent and dependent variables to vary at different spatial scales [36–38]. The purpose of this study was to better understand the spatial context of the determinants of teenage pregnancy using an explicit multiscale approach. Therefore, our study aimed to explore the spatial determinants of teenage pregnancy among adolescents aged 15 to 19 years using the most recent EDHS.
Materials and methods
The 2019 mini-EDHS was conducted in Ethiopia through a community-based cross-sectional study. After Nigeria, Ethiopia is the second most populous country in Africa, with a total population of almost 126.5 million people in 2023 and an area of 1.1 million square kilometers. Children under the age of 15, individuals aged 15 to 64, and those aged 65 or above represent 44%, 52%, and 4% of the total population, respectively [39,40]. The country is subdivided into nine regions and two city administrations.
The 2019 mini-EDHS sample was stratified and selected in two stages. Each region was stratified into urban and rural areas, yielding 21 sampling strata. Samples of enumeration areas (EAs) were selected independently in each stratum in the two stages. The first stage involved a selection of 305 EAs, with 93 in urban areas and 212 in rural areas. In the second stage, a fixed number of 30 households per cluster was selected with an equal probability of systematic selection from the newly created household listing [39]. We utilized the individual data (IR) set from the mini-EDHS 2019 for the analysis, and we obtained longitude and latitude coordinates at the cluster or EA level. The Ethiopian administrative boundaries shapefile was obtained from the Central Statistical Agency (CSA).
The outcome variable is teenage pregnancy, defined as women aged 15–19 years who had a live birth, who were pregnant with their first child, or who had begun childbearing at the time of the survey [41]. It is dichotomized as 0 = no (those who did not experience teenage pregnancy) and 1 = yes (those who experienced teenage pregnancy). The percentage of women who had begun childbearing was calculated by dividing the number of women who either had a birth or who were pregnant at the time of the interview by all women aged 15–19 years [41]. The independent variables included were age, marital status, religion, sex of the household head, education level, wealth index, media exposure, and contraceptive utilization. The community-level variables were residence and region; we also created aggregate community-level variables, which were generated by aggregating individual-level factors at the cluster level: community education, community poverty, community contraceptive utilization, community media exposure, and community percentage of marriage.
Data processing and analysis
Factors associated with teenage pregnancy.
Due to the hierarchical nature of the EDHS data, teenagers within the same cluster exhibit similar characteristics compared to those in another cluster. Therefore, a multilevel binary logistic regression model that accounts for the heterogeneity between clusters was constructed. To assess the clustering effect, the intraclass correlation coefficient (ICC) was estimated. The ICC showed a significant clustering effect (ICC = 20.9%). The median odds ratio (MOR), which is the median value of the odds ratio between the highest risk area and lowest risk area when randomly picking out two areas, is the increased risk that would exist when moving to a different area with a higher level of risk [42]. The proportional change in variance (PCV) is the variation in the dependent variable explained by all individual- and community-level factors [43].
Four models were constructed for the multilevel logistic regression analysis. The first model was a null model without explanatory variables to determine the degree of cluster variation in teenage pregnancy. The second model was fitted with individual-level variables, the third model was fitted with community-level variables, and the fourth model was fitted with both individual- and community-level variables at the same time. Variables with p values < 0.25 in the bi-variable analysis were added to the multivariable model. The associations between the dependent and independent variables were assessed and are presented as adjusted odds ratios (AORs) with 95% confidence intervals (CIs) at p values < 0.05.
Spatial analysis.
A spatial analysis was conducted by using ArcGIS version 10.7 and Python 3. To assess whether the spatial distribution of teenage pregnancy in Ethiopia is dispersed, clustered, or randomly distributed, global spatial autocorrelation (global Moran’s I) was utilized. A Moran’s I value < 0.05 indicated that the distribution of teenage pregnancy in Ethiopia was nonrandom [44]. Hotspot analysis was conducted to identify statistically significant clustering areas using Getis-ordGi* statistics, which resulted in z scores and p-values indicating where high and low values of teenage pregnancy clustered spatially [45].
To comprehend the association between the density of a certain event and other various sociodemographic and environmental aspects of the population, spatial regression models are important [46]. After the identification of the high-risk areas (hotspot areas) of teenage pregnancy, factors that contributed to the spatial clustering of teenage pregnancy in these areas were determined at the EA level by a global and local spatial regression model: ordinary least squares (OLS) and GWR and MGWR, respectively. The spatial nonstationarity of the proportion of teenage pregnancy was ascertained by using global spatial autocorrelation before fitting the OLS, GWR, and MGWR models.
First, OLS was fitted to choose variables that are appropriate for the spatial variation of teenage pregnancy. Before proceeding to the GWR model, we first checked the six assumptions of the OLS model using exploratory regression. Exploratory regression is a data mining tool that helps us determine which combinations of explanatory variables meet all necessary OLS diagnostics [47]. We checked for the expected relationship of the explanatory variables, the statistical significance of each explanatory variable, the randomness of residuals (which should be normally distributed with no spatial patterns), the statistical nonsignificance of the Jarque-Bera statistics, freedom from multicollinearity in the explanatory variables (VIF < 7.5), and the model performance (strength of the adjusted R-squared) [48]. The Koenker Breusch–Pagan (BP) test was significant, indicating nonstationarity or heterogeneity in the relationship between the dependent and independent variables. As a result, we utilized local spatial regression models (GWR and MGWR), which account for variations in the relationships between variables across different spatial locations, for this study [49].
A geographically weighted regression model was used to determine the predictor variables for teenage pregnancy among adolescents aged 15 to 19 years. The proportion of the outcome variable and all predictors was calculated for each cluster. Classical GWR assumes that all of the processes being modeled operate at the same spatial scale. However, the MGWR reduces this assumption by allowing different processes to operate at different spatial scales by deriving an optimal bandwidth vector in which each element indicates the spatial scale at which a particular process takes place. Hence, MGWR is preferable because covariate-specific bandwidths are obtained rather than a single average bandwidth and provides valuable information on the scale at which different processes operate. Model calibration and bandwidth vector selection in MGWR were conducted using a back-fitting algorithm. Model fitness was compared among the global and local models using the Akaike information criterion (AIC), residual sum of squares (RSS), and adjusted R-squared values [50,51].
Ethical approval and consent to participate
The data were obtained from the Demographic and Health Surveys (DHS) Program and can be freely accessed from the program website (www.dhsprogram.com). As the study was a secondary data analysis of publicly available data from the MEASURE DHS program, ethical approval and participant consent were not necessary for this particular study. We requested the use of the DHS Program, and permission was granted to download and use the data for this study. Approval for the use of the data was obtained from the Measure DHS program, and the dataset was downloaded from https://dhsprogram.com/data/available-datasets.cfm. The IRB-approved procedures for DHS public-use datasets do not allow the identification of respondents, households, or sample communities. Geographic identifiers only go down to regional levels, and each EA has a primary sampling unit (PSU) number without labels. Surveys collect GIS coordinates for the entire enumeration area, preventing specific enumeration areas from being identified.
Results
Descriptive results
A total of 2165 teenage girls aged 15 to 19 years were included in the study. The median age of the respondents was 17 years. One-fifth (20.4%) of the participants were married, and only 9.5% of the girls utilized contraceptives. One-third of the girls were urban residents, while a quarter of them (25.1%) were illiterate. Regarding regional states, the majority of the participants were from the Oromia region (851, 39.3%), and the minority of the participants were from the Harari region (6, 0.29%) (Table 1).
Prevalence of teenage pregnancy in Ethiopia.
The prevalence of teenage pregnancy among adolescents aged 15 to 19 years was 12.98% (95% CI: 11.63%, 14.46%). The highest percentage of teenage pregnancy was observed in the Afar region (28.3%), followed by the Gambella region (27.4%). The lowest prevalence was observed in Addis Ababa (3.8%) (Table 1).
Spatial analysis
Global spatial autocorrelation analysis revealed that the spatial distribution of teenage pregnancy in Ethiopia significantly varied across the country, with a global Moran’s index value of 0.04 and a p-value < 0.001. The Z score of 4.71 indicated that there was less than a 1% likelihood that the clustered pattern could be a result of random chance. (Fig 1).
Significant hotspot areas (areas with a high proportion of teenage pregnancy) of teenage pregnancy were observed in central and southern Afar; northern, central, and western Gambela; northeastern and southern central Oromia; and the eastern Somali region. Significant cold spot areas (areas with a low proportion of teenage pregnancy) were identified in central and southern Amhara, central Oromia, Addis Ababa, and northern SNNPR. (Fig 2).
Shapefile source: (CSA, 2013; https://africaopendata.org/dataset/ethiopia-shapefiles); Map output: Own analysis using ArcGIS Software.
Spatial regression analysis.
To explore the assumptions of spatial regression and to estimate the coefficients of the selected variables on teenage pregnancy, OLS regression was conducted. The OLS model identified predictors of each hotspot of teenage pregnancy. The joint F-statistics and Wald statistics were significant, which shows that the model was statistically significant. The Koenker statistics was statistically significant, which is indicative of nonstationarity or heterogeneity of the relationship between the dependent and the independent variables across the study areas. Therefore, GWR was applied because it assumes that the relationship between the independent and dependent variables has spatial heterogeneity (as confirmed by Koenker statistics). No multicollinearity was observed between the selected explanatory variables (Table 2).
The local models, GWR and MGWR, were implemented to capture spatial heterogeneity. Both of the local models had better model fitness than the global model. The adjusted R2 improved from 0.547 (OLS) to 0.634 (GWR) and 0.691 (MGWR). The adjusted R2 indicates that the GWR and MGWR have 8.7% and 14.4% increased explanatory power, respectively, compared to the OLS. The residual sum of squares (RSS), which indicates unexplained variations, was very high in the OLS model (134.6) and was 93.7 (GWR) and 78.4 (MGWR). The local regression showed an AICc of 630.4, while the GWR reduced the AICc to 618.6, and the MGWR further decreased the AICc to 570.4. Relying on this information, it was confirmed that the local models improved the model performance (Table 3).
The GWR model has a single bandwidth of 103; thus, 103 nearest neighbors are considered to inform the construction of parameter estimates at each local regression point. This implies that the GWR analysis focused on the 103 nearest neighboring data points to calculate parameter estimates. Using the 103 bandwidths for each variable, 63.4% of the observations were correctly classified. In the case of MGWR, the optimum bandwidth for each variable is computed. By allowing multiple bandwidths in MGWR, the model theoretically accounts for an optimal number of neighbors for each parameter estimate, thus allowing better predictions for the response variables. In MGWR, the bandwidths for each parameter are 111 for poor wealth index, 43 for being married, 301 for being Muslim, 301 for being illiterate, 292 for community poverty and 161 for community illiteracy. These specific bandwidths enable the model to adjust to the unique spatial characteristics of each variable (Table 4).
The figures below show the mapped GWR and MGWR coefficient estimates for the intercept and the covariates. In Fig 3, the local intercept in the MGWR is interpreted as the value of the outcome that would be expected if every location had exactly the same average value of each covariate. Based on the mapping of the intercept coefficients, the local intercept in the MGWR model can be interpreted as showing the level of teenage pregnancy that would be expected in each region if each of the regions had an average population profile or held all covariates constant. There is a high rate of teenage pregnancy in Gambella and some parts of western and northeastern Oromia. In contrast, there was a lower rate of teenage pregnancy in central and western Amhara. The intercept in the GWR was not significantly different from zero, but there was an outlier with a positive relationship in southern Afar at the border with Oromia.
GWR (left) and MGWR (right) parameter estimates for the intercept showing local patterns of spatial heterogeneity. The gray dots are not significantly different from zero. Shapefile source: (CSA, 2013; https://africaopendata.org/dataset/ethiopia-shapefiles); Map output: Own analysis using Python Software.
The parameter estimates for the proportion of being married were positively associated with teenage pregnancy in Tigray, Amhara, Afar, Somali, eastern SNNPR, Benishangul, Gambela, Harari, Dire Dawa and eastern Oromia. The GWR model showed a positive relationship with the outcome variable across the study area with no spatial heterogeneity (Fig 4).
GWR (left) and MGWR (right) parameter estimates for the proportion of married women showing local patterns of spatial heterogeneity. The gray dots are not significantly different from zero. Shapefile source: (CSA, 2013; https://africaopendata.org/dataset/ethiopia-shapefiles); Map output: Own analysis using Python Software.
The proportion of being illiterate does not show any spatial heterogeneity, as it has a positive association with teenage pregnancy across the country. This implies that being illiterate is associated with teenage pregnancy throughout the country, with no spatial differences. In the GWR model, it was not significantly different from zero because it has no statistically nonzero parameter estimate (Fig 5).
GWR (left) and MGWR (right) parameter estimates for the proportion of illiterate women showing local patterns of spatial heterogeneity. The gray dots are not significantly different from zero. Shapefile source: (CSA, 2013; https://africaopendata.org/dataset/ethiopia-shapefiles); Map output: Own analysis using Python Software.
Multilevel analysis of factors associated with teenage pregnancy
Multilevel mixed-effect logistic regression analysis was conducted to identify individual- and community-level variables that were significantly associated with teenage pregnancy. The ICC in the null model was 0.209, which shows that 20.9% of the variation in teenage pregnancy is explained by community-level variables (between-cluster variability). According to the MOR in the null model (2.43), there was a variation in teenage pregnancy between the clusters. When a teenage girl is switched from a low-risk cluster to a high-risk cluster, she is 2.43 times more likely to have a teenage pregnancy. According to the final model (model 4), the PCV showed that 72% of the total variability in teenage pregnancy was explained by both individual- and community-level factors. (Model 4) was the best-fitting model because it attained the highest log likelihood (in negative numbers) and the lowest deviance (Table 5).
Based on the findings of Model 4, the factors that were found to have a significant association with teenage pregnancy were the age of the girls, literacy status, marital status, age of the household head and residence in the Gambela region (Table 6).
The odds of having teenage pregnancy among girls aged 17, 18 and 19 years were 3.5 times greater (AOR = 3.54; 95% CI: 1.60, 7.81), 8.2 times greater (AOR = 8.21; 95% CI: 3.96, 17.0) and 15 times greater (AOR = 15.0; 95% CI: 6.84, 32.9), respectively, than teenagers aged 15 years. Literate teenagers had a 43% (AOR = 0.57; 95% CI: 0.35, 0.92) lower risk of teenage pregnancy than did those who were illiterate. Compared with single teenagers, married teenagers had 22.8 (AOR = 22.8; 95% CI: 14.1, 37.0) times greater odds of having a teenage pregnancy. With a one-year increase in the age of the household head, the odds of experiencing teenage pregnancy decreased by 2% (AOR = 0.98; 95% CI: 0.98, 0.99). Teenagers residing in the Gambela region had 3.27 (AOR = 3.27; 95% CI: 1.21, 8.86) times greater odds of having a teenage pregnancy than did those residing in the Tigray region. However, religion, education level, contraceptive use, wealth index, media exposure, and community variables, except region, were not statistically significant (Table 6).
Discussion
The spatial statistics showed that the spatial distribution of teenage pregnancy among adolescents aged 15 to 19 years had significant variation in Ethiopia. The hotspot areas of teenage pregnancy were identified in central and southern Afar, northern, central and western Gambela, northeastern and southern central Oromia and the eastern Somali region. A possible explanation might be that most of these areas are pastoralists and semi pastoralists, characterized by seasonal mobility and lower utilization of contraceptives [52]. They also live in traditional settings and strongly adhere to traditional cultural values and beliefs, which leads to poor sexual and reproductive health outcomes and early marriage [53].
According to the spatial regression analysis of the global model, the significant predictors of teenage pregnancy hotspots among adolescents aged 15 to 19 years were marital status and illiteracy. Local models were applied to the same predictors used in OLS to explore the local spatial variation in the relationship with the proportion of teenage pregnancy. As the MGWR model had the smallest AICc and RSS values and the highest adjusted R2 value, it is a statistically preferable local model, which is in line with previous studies [54].
The MGWR analysis revealed a significant positive relationship between the proportion of being a married women and teenage pregnancy in central and southern Afar, most parts of Amhara, Tigray and Benishangul, eastern SNNPR, Gambela, Harari, Dire Dawa, and eastern Oromia. This is because adolescents who are married are exposed to early sexual debut, which results in teenage pregnancy [31]. In most parts of Ethiopia, specifically rural areas, it is common for married girls to become pregnant soon after marriage to prove their fertility. Additionally, family planning utilizations are usually disapproved by the partner, in-laws and the community as a whole as a result of social norms, and little awareness and negative perception also play a role [29,55].
The proportion of being illiterate was significantly positively associated with teenage pregnancy throughout the country according to the MGWR model. This is because literacy increases the understanding of reproductive and sexual health through reading printed materials and contraceptive utilization [26]. Teenagers who are illiterate will have lower self-esteem and want to be accepted and validated by society through early childbearing [56].
Based on the multivariable multilevel logistic regression, literacy, marital status, age of the girls, age of the household head and residence in the Gambela region were significantly associated with teenage pregnancy among adolescents aged 15 to 19 years. Literate teenagers have a lower risk of pregnancy during adolescence than do illiterate teenagers. This finding is in line with studies conducted in Ethiopia [26], Zambia [57], Philadelphia, the USA [58] and Africa [59]. Giving an emphasis on female literacy will reduce teenage pregnancy. Teenagers who are married are more likely to experience teenage pregnancy than those who are single. This is supported by previously conducted similar studies in Ethiopia [26,31,33], Zambia [57], Uganda [60] and Nigeria [61]. As the frequency of sexual activity is greater among married teenagers, there is a greater likelihood of pregnancy in the absence of contraception than among those who are not married [62].
With respect to age, teenagers aged 17, 18 and 19 years had a greater risk of experiencing teenage pregnancy than did those aged 15 years. This finding is similar to those of studies conducted in Ethiopia [26] and Kenya [63]. This is because as age increases, the probability of engaging in sexual intercourse and marriage, which are proximal determinants of fertility, increases. Consequently, the chances of pregnancy and childbearing will also increase [64]. As the age of the household head increases, the chance of a teenage pregnancy decreases. This finding is consistent with a study from Rwanda [65]. This is because when household heads are older and more mature, they support teens under their household by giving them good parental support, including advising their teen girls on risky sex and fulfilling financial needs [66,67]. In terms of region, the odds of experiencing teenage pregnancy among adolescents aged 15 to 19 years who lived in the Gambela region were greater than those among adolescents in the Tigray region. This is attributable to the region’s high unmet need for family planning, greater use of polygamous practices, and early sexual initiation among adolescents, which exacerbates this condition [68].
The key strength of this study was the use of the most recent weighted nationally representative data with a relatively large sample size. To generalize findings at the national level by adjusting the data’s hierarchical nature, a multilevel model analysis was performed. The use of spatial and multiscale geographically weighted regression analysis assists in identifying factors that contribute to spatial variation in teenage pregnancy. This research, however, has certain limitations that must be considered when interpreting the findings. Important variables such as age at first sex and marriage and demand for family planning were not included since the mini-EDHS of 2019. In addition, because of the cross-sectional nature of the EDHS data, the cause–effect relationship between the dependent and independent variables cannot be shown.
Conclusion
There is significant spatial variation in teenage pregnancy among adolescents aged 15 to 19 years. The proportion of being illiterate and the proportion of being married were significantly positively associated with teenage pregnancy throughout the country according to the MGWR model. In the multivariable multilevel logistic regression, literacy, marital status, age of the girls, age of the household head and residence in the Gambela region were significantly associated with the outcome variable. To address this issue, stronger actions must be taken to implement the law, raising the minimum age for marriage to 18 years. Teenage marriage, which is the main cause of teenage pregnancy in Ethiopia, should be changed by addressing cultural norms that prioritize maintaining family status in the community.
Priority should also be given to enhancing levels of literacy and sexual and reproductive health education. In addition to increasing girls’ enrollment in formal schools, the government of Ethiopia must increase adult education programs for young women outside of school by incorporating reproductive health issues into the program. The population must also become aware of the consequences and adverse outcomes of teenage pregnancy. Generally, policies and implementations that focus on preventing early marriage, educating teenage girls, and creating awareness of reproductive health issues need to be strengthened.
Acknowledgments
We would like to acknowledge the measure DHS program for providing the dataset to use the mini-EDHS 2019.
References
- 1.
Kothari MT, Wang S, Head SK, Abderrahim N. Trends in adolescent reproductive and sexual behaviors [Internet]. Calverton, Maryland: ICF International; 2012. https://www.dhsprogram.com/pubs/pdf/CR29/CR29.pdf.
- 2. Abera M, Nega A, Tefera Y, Gelagay AA. Early marriage and women’s empowerment: the case of child-brides in Amhara National Regional State, Ethiopia. BMC Int Health Hum Rights [Internet]. 2020;20(1):30. Available from: pmid:33317540
- 3. Asmamaw DB, Tafere TZ, Negash WD. Prevalence of teenage pregnancy and its associated factors in high fertility sub-Saharan Africa countries: a multilevel analysis. BMC Womens Health [Internet]. 2023;23(1):23. Available from: pmid:36650514
- 4. Nove A, Matthews Z, Neal S, Camacho AV. Maternal mortality in adolescents compared with women of other ages: evidence from 144 countries. Lancet Glob Heal [Internet]. 2014 Mar 1;2(3):e155–64. Available from: pmid:25102848
- 5. Mekonnen T, Dune T, Perz J. Maternal health service utilisation of adolescent women in sub-Saharan Africa: a systematic scoping review. BMC Pregnancy Childbirth [Internet]. 2019;19(1):366. Available from: pmid:31638927
- 6.
United Nations Population Fund. Adolescent pregnancy [Internet]. https://esaro.unfpa.org/en/topics/adolescent-pregnancy.
- 7. Blanc AK, Winfrey W, Ross J. New Findings for Maternal Mortality Age Patterns: Aggregated Results for 38 Countries. PLoS One [Internet]. 2013 Apr 16;8(4):e59864. Available from: pmid:23613716
- 8. Conde-Agudelo A, Belizán JM, Lammers C. Maternal-perinatal morbidity and mortality associated with adolescent pregnancy in Latin America: Cross-sectional study. Am J Obstet Gynecol [Internet]. 2005 Feb 1;192(2):342–9. Available from: pmid:15695970
- 9.
World Health Organization. Trends in maternal mortality 2000 to 2020: estimates by WHO, UNICEF, UNFPA, World Bank Group and UNDESA/Population Division. Geneva; 2023.
- 10.
Council NR, Population C on. Growing up global: The changing transitions to adulthood in developing countries. National Academies Press; 2005.
- 11.
UNICEF. Progress for children: A report card on adolescents. New York UNICEF. 2012;(10).
- 12.
Darroch JE, Woog V, Bankole A, Ashford LS. Adding it up: costs and benefits of meeting the contraceptive needs of adolescents. 2016.
- 13.
Sully EA, Biddlecom A, Darroch JE, Riley T, Ashford LS, Lince-Deroche N, et al. Adding it up: investing in sexual and reproductive health 2019. 2020.
- 14. Shah IH, Åhman E. Unsafe abortion differentials in 2008 by age and developing country region: high burden among young women. Reprod Health Matters. 2012;20(39):169–73. pmid:22789095
- 15.
World Health Organization. SDG Target 3.7 Sexual and reproductive health [Internet]. https://www.who.int/data/gho/data/themes/topics/sdg-target-3_7-sexual-and-reproductive-health.
- 16. Noori N, Proctor JL, Efevbera Y, Oron AP. The Effect of Adolescent Pregnancy on Child Mortality in 46 Low- and Middle-Income Countries. BMJ Glob Heal [Internet]. 2022 May 1;7(5):e007681. Available from: http://gh.bmj.com/content/7/5/e007681.abstract.
- 17.
Erulkar A, Ferede A, Ambelu W, Girma W, Amdemikael H, GebreMedhin B, et al. Ethiopia young adult survey: A study in seven regions. 2010.
- 18.
Central Statistical Authority/Ethiopia, ORC Macro. Ethiopia Demographic and Health Survey 2000 [Internet]. Addis Ababa, Ethiopia: Central Statistical Authority/Ethiopia and ORC Macro; 2001. http://dhsprogram.com/pubs/pdf/FR118/FR118.pdf.
- 19.
Central Statistical Agency/Ethiopia, ORC Marco. Ethiopia Demographic and Health Survey 2005 [Internet]. Addis Ababa, Ethiopia: Central Statistical Agency/Ethiopia and ORC Macro; 2006. http://dhsprogram.com/pubs/pdf/FR179/FR179.pdf.
- 20.
Central Statistical Agency/Ethiopia, ICF International. Ethiopia Demographic and Health Survey 2011 [Internet]. Addis Ababa, Ethiopia: Central Statistical Agency/Ethiopia and ICF International; 2012. http://dhsprogram.com/pubs/pdf/FR255/FR255.pdf.
- 21.
Central Statistical Agency—CSA/Ethiopia, ICF. Ethiopia Demographic and Health Survey 2016 [Internet]. Addis Ababa, Ethiopia: CSA and ICF; 2017. http://dhsprogram.com/pubs/pdf/FR328/FR328.pdf.
- 22. Alamneh Gebeyehu A, Teshome AA, Teshager W, Teshome F, Tiruneh M, Dessie AM, et al. Trends change in teen pregnancy among adolescent women in Ethiopia based on Ethiopian demographic and health surveys: Multivariate decomposition analysis. PLoS One. 2023;18(6):e0287460. pmid:37352189
- 23. Kaphagawani NC, Kalipeni E. Sociocultural factors contributing to teenage pregnancy in Zomba district, Malawi. Glob Public Health [Internet]. 2017 Jun 3;12(6):694–710. Available from: pmid:27687242
- 24.
World Health Organization. Adolescent pregnancy [Internet]. 2023. https://www.who.int/news-room/fact-sheets/detail/adolescent-pregnancy.
- 25.
Nuru Nabwire O. The association between the level of education and teenage pregnancy in Kenya A secondary analysis of Kenya Demographic and Health Survey 2014. 2023.
- 26. Bitew DA, Akalu Y, Belsti Y, Diress M, Gela YY, Belay DG, et al. Predictors of underage pregnancy among women aged 15–19 in highly prevalent regions of Ethiopia: a multilevel analysis based on EDHS, 2016. Sci Rep [Internet]. 2023;13(1):857. Available from: pmid:36646737
- 27. Imamura M, Tucker J, Hannaford P, Da Silva MO, Astin M, Wyness L, et al. Factors associated with teenage pregnancy in the European Union countries: a systematic review. Eur J Public Health. 2007;17(6):630–6. pmid:17387106
- 28. Ayele BG, Gebregzabher TG, Hailu TT, Assefa BA. Determinants of teenage pregnancy in Degua Tembien District, Tigray, Northern Ethiopia: A community-based case-control study. PLoS One [Internet]. 2018 Jul 25;13(7):e0200898. Available from: pmid:30044850
- 29.
Chuta N. Neglected experiences: Fertility and childbearing among young people in Ethiopia. 2021; https://www.younglives.org.uk/publications/neglected-experiences-fertility-and-childbearing-among-young-people-ethiopia.
- 30. Admassu TW, Wolde YT, Kaba M. Ethiopia has a long way to go meeting adolescent and youth sexual reproductive health needs. Reprod Health. 2022 Jun;19(Suppl 1):130. pmid:35698082
- 31. Kassa GM, Arowojolu AO, Odukogbe ATA, Yalew AW. Trends and determinants of teenage childbearing in Ethiopia: evidence from the 2000 to 2016 demographic and health surveys. Ital J Pediatr [Internet]. 2019;45(1):153. Available from: pmid:31783884
- 32. Birhanu BE, Kebede DL, Kahsay AB, Belachew AB. Predictors of teenage pregnancy in Ethiopia: a multilevel analysis. BMC Public Health [Internet]. 2019;19(1):601. Available from: pmid:31101101
- 33. Kefale B, Yalew M, Damtie Y, Adane B. A Multilevel Analysis of Factors Associated with Teenage Pregnancy in Ethiopia. Int J Womens Health. 2020;12:785–93. pmid:33116928
- 34. Fikadu Y. Determinants of teenage pregnancy in Ethiopia: A Case–control study, 2019. Issues (St Louis Mo). 2019 Dec 13;17:112.
- 35. Tigabu S, Liyew AM, Geremew BM. Modeling spatial determinates of teenage pregnancy in Ethiopia; geographically weighted regression. BMC Womens Health [Internet]. 2021;21(1):254. Available from: pmid:34167542
- 36. Chang Chien YM, Carver S, Comber A. Using geographically weighted models to explore how crowdsourced landscape perceptions relate to landscape physical characteristics. Landsc Urban Plan [Internet]. 2020;203:103904. Available from: https://www.sciencedirect.com/science/article/pii/S0169204620304667.
- 37. Gao Y, Zhao J, Han L. Exploring the spatial heterogeneity of urban heat island effect and its relationship to block morphology with the geographically weighted regression model. Sustain Cities Soc [Internet]. 2022;76:103431. Available from: https://www.sciencedirect.com/science/article/pii/S2210670721007046.
- 38. Fotheringham AS, Yu H, Wolf LJ, Oshan TM, Li Z. On the notion of ‘bandwidth’ in geographically weighted regression models of spatially varying processes. Int J Geogr Inf Sci [Internet]. 2022 Aug 3;36(8):1485–502. Available from:
- 39.
Ethiopian Public Health Institute—EPHI, Federal Ministry of Health—FMoH, ICF. Ethiopia Mini Demographic and Health Survey 2019 [Internet]. Addis Ababa, Ethiopia: EPHI/FMoH/ICF; 2021. https://www.dhsprogram.com/pubs/pdf/FR363/FR363.pdf.
- 40.
World Population Review. Countries in Africa [Internet]. 2023. https://worldpopulationreview.com/country-rankings/countries-in-africa.
- 41.
Croft TN, Marshall AMJ, Allen CK, Arnold F, Assaf S, Balian S. Guide to DHS statistics. Rockv ICF. 2018;645.
- 42. Merlo J, Chaix B, Ohlsson H, Beckman A, Johnell K, Hjerpe P, et al. A brief conceptual tutorial of multilevel analysis in social epidemiology: using measures of clustering in multilevel logistic regression to investigate contextual phenomena. J Epidemiol Community Health. 2006 Apr;60(4):290–7. pmid:16537344
- 43. Merlo J, Yang M, Chaix B, Lynch J, Råstam L. A brief conceptual tutorial on multilevel analysis in social epidemiology: investigating contextual phenomena in different groups of people. J Epidemiol Community Health [Internet]. 2005 Sep 1;59(9):729–736. Available from: http://jech.bmj.com/content/59/9/729.abstract. pmid:16100308
- 44. Chen Y. New Approaches for Calculating Moran’s Index of Spatial Autocorrelation. PLoS One [Internet]. 2013 Jul 12;8(7):e68336. Available from: pmid:23874592
- 45. Zulu LC, Kalipeni E, Johannes E. Analyzing spatial clustering and the spatiotemporal nature and trends of HIV/AIDS prevalence using GIS: the case of Malawi, 1994–2010. BMC Infect Dis [Internet]. 2014;14(1):285. Available from: pmid:24886573
- 46.
Srinivasan S. Spatial Regression Models BT—Encyclopedia of GIS. In: Shekhar S, Xiong H, editors. Boston, MA: Springer US; 2008. p. 1102–5.
- 47.
Esri. How Exploratory Regression works [Internet]. https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-statistics/how-exploratory-regression-works.htm.
- 48. Tareke AA, Gashu KD, Endehabtu BF. Geographical weighted regression analysis of delayed antenatal care initiation and its factors among all reproductive-aged women in Ethiopia, 2016. Contracept Reprod Med [Internet]. 2022;7(1):23. Available from: pmid:36316744
- 49.
Charlton M, Fotheringham AS, Brunsdon C. Geographically weighted regression white paper. Kildare Natl Univ Irel Maynooth. 2009;1–14.
- 50. Fotheringham AS, Yang W, Kang W. Multiscale Geographically Weighted Regression (MGWR). Ann Am Assoc Geogr [Internet]. 2017 Nov 2;107(6):1247–65. Available from:
- 51. Li Z, Fotheringham AS, Oshan TM, Wolf LJ. Measuring Bandwidth Uncertainty in Multiscale Geographically Weighted Regression Using Akaike Weights. Ann Am Assoc Geogr [Internet]. 2020 Sep 2;110(5):1500–20. Available from:
- 52. Lakew Y, Reda AA, Tamene H, Benedict S, Deribe K. Geographical variation and factors influencing modern contraceptive use among married women in Ethiopia: evidence from a national population based survey. Reprod Health [Internet]. 2013;10(1):52. Available from: pmid:24067083
- 53.
van der Kwaak A, Baltissen G, Plummer D, Ferris K, Nduba J. Understanding nomadic realities: case studies on sexual and reproductive health and rights in Eastern Africa. KIT Publishers; 2012.
- 54. Amegbor PM, Zhang Z, Dalgaard R, Sabel CE. Multilevel and spatial analyses of childhood malnutrition in Uganda: examining individual and contextual factors. Sci Rep [Internet]. 2020;10(1):20019. Available from: pmid:33208763
- 55. Ketema H, Erulkar A. Married Adolescents and Family Planning in Rural Ethiopia: Understanding Barriers and Opportunities. Afr J Reprod Health. 2018 Dec;22(4):26–34. pmid:30632719
- 56.
Reading Partners. Early Reading Deficiencies Create Conditions that Enhance Teen Pregnancy Risk [Internet]. 2013. https://readingpartners.org/blog/early-reading-deficiencies-creating-conditions-that-enhance-teen-pregnancy/.
- 57. Phiri M, Kasonde ME, Moyo N, Sikaluzwe M, Simona S. A multilevel analysis of trends and predictors associated with teenage pregnancy in Zambia (2001–2018). Reprod Health [Internet]. 2023;20(1):16. Available from: pmid:36653839
- 58. Bennett IM, Frasso R, Bellamy SL, Wortham S, Gross KS. Pre-teen literacy and subsequent teenage childbearing in a US population. Contraception [Internet]. 2013 Apr 1;87(4):459–64. Available from: pmid:23245352
- 59. Odejimi O, Bellingham-Young D. Teenage pregnancy in Africa: trends and determinants in the 21st century. J Heal Soc Care Improv 2016 Vol 1 issue 1 [Internet]. 2016;1(1):12–20. Available from: https://wlv.openrepository.com/bitstream/2436/620300/1/TeenagePregnancyinAfrica-TrendandDeterminantsinthe21stCentury.pdf.
- 60. Ochen AM, Chi PC, Lawoko S. Predictors of teenage pregnancy among girls aged 13–19 years in Uganda: a community based case-control study. BMC Pregnancy Childbirth [Internet]. 2019;19(1):211. Available from: pmid:31234816
- 61. Ayoade MA. Spatiotemporal Patterns of Teenage Pregnancy in Nigeria: Evidence from the 2008, 2013 and 2018 NDHS. Pap Appl Geogr [Internet]. 2021 Apr 3;7(2):161–82. Available from:
- 62.
World Health Organization. Early marriages, adolescent and young pregnancies. Geneva WHO. 2012;1–4.
- 63. Were M. Determinants of teenage pregnancies: the case of Busia District in Kenya. Econ Hum Biol. 2007 Jul;5(2):322–39. pmid:17446148
- 64. Alemayehu T, Haider J, Habte D. Determinants of adolescent fertility in Ethiopia. Ethiop J Heal Dev. 2010;24(1).
- 65. Uwizeye D, Muhayiteto R, Kantarama E, Wiehler S, Murangwa Y. Prevalence of teenage pregnancy and the associated contextual correlates in Rwanda. Heliyon. 2020 Oct;6(10):e05037. pmid:33083588
- 66.
Nwogwugwu NC. Socio-demographic determinants of adolescent fertility in Zambia. University of the Witwatersrand, Faculty of Humanities; 2013.
- 67. Domenico DM, Jones KH. Adolescent pregnancy in America: Causes and responses. J Vocat Spec Needs Educ. 2007;30(1):4–12.
- 68. Bol KN, Negera E, Gedefa AG. Pregnancy among adolescent girls in humanitarian settings: a case in refugee camp of Gambella regional state, community-based cross-sectional study, Southwest Ethiopia, 2021. BMJ Open [Internet]. 2022 Nov 1;12(11):e064732. Available from: http://bmjopen.bmj.com/content/12/11/e064732.abstract. pmid:36396321