Background: Ethiopia houses the second largest population of female adolescents in Africa. Adolescent childbearing can have detrimental effect to the health and wellbeing of women and their offspring. This study examined trends, sub-national variations and determinants of early childbearing (i.e. childbearing before age 20) in Ethiopia. Methods: Data from the 2000-2011 Ethiopia Demographic and Health Surveys and from the 2014-2016 Performance Monitoring and Accountability surveys were pooled for this analysis. Based on the year the women reached puberty, five different cohorts were reconstructed that date back to the early 1970s. Kaplan-Meier methodology was used to estimate the cumulative probability of early childbearing and a Cox proportional hazard regression model to examine the associated factors. Results: The cumulative probability of early childbearing declined by approximately two-fifth in the past four decades, from 57.6 to 35.3%. The occurrence of early childbearing varies substantially by region. In the most recent period, it ranged from 9.6% in Addis Ababa to 59% in Benishangul-Gumuz. Early childbearing risk was reduced by 95% for women who did not marry before the age of 20years compared to those who married before the age of 18years. For adolescents who married at the age of 18 and 19years, early childbearing risk decreased by 60 and 78%, respectively. During the same period, there was a parallel decline in the cumulative probability of early marriage (i.e., before the legal age of 18years) from 55.3 to 28.7%. Compared with adolescents with no education, those with elementary and secondary or higher education had a 50 and 82% lower risk of early childbearing, respectively. Conclusions: Early childbearing declined in Ethiopia, largely driven by a parallel reduction in early marriage. However, a large portion of adolescents are still facing early childbearing, and the situation is more dismal in some regions than others. A further reduction in early childbearing is warranted by enforcing the law on the minimum marriage age and expanding secondary and higher education for females. These efforts should give greater emphasis to regions where early childbearing is markedly high.
This study is based on an analysis of pooled national data from six rounds of surveys from two different sources. The data were from the 2000, 2005 and 2011 Ethiopia Demographic and Health Surveys (EDHS) and the 2014, 2015 and 2016 Ethiopia Performance Monitoring and Accountability (EPMA) surveys. These six surveys were selected for this pooled analysis following inventory of all available national and sub-national household surveys that were available in the country at the time this study was initiated and through consultation of the Ethiopia Central Statistical Agency (CSA). The inclusion criteria of the surveys in this pooled analysis were: national and sub-national representativeness of the surveys; similarities in sampling design; the use of the same national sampling frame for clusters selection; samples drawn for the survey by the Ethiopia Central Statistical Agency (CSA); comparability of survey questionnaires for focus variables of this analysis; and availability of raw data for the pooled analysis. Inclusion in the pooled analysis was not restricted by survey date as far as the surveys met the inclusion criteria. Conversely, we excluded any other surveys that did not meet the inclusion criteria. The Demographic and Health Survey (DHS) is one of the largest programs producing nationally representative household surveys that provide data for a wide range of monitoring and impact evaluation indicators in the areas of population, health, and nutrition throughout the developing world [20]. The Ethiopia DHS is part of the worldwide MEASURE DHS project, which is funded by the United States Agency for International Development (USAID) and other development partners. The country began to implement the DHS in 2000. The survey has been implemented by the CSA, and ICF International provided technical assistance through the MEASURE DHS project. The DHS collected and reported data in five-year intervals. The EPMA surveys are conducted annually to fill gaps in the timely availability of current and reliable information on population dynamics, reproductive health, family planning, and maternal health, among a few other topics. The lengthy gap between successive DHSs, which restricts the ability of planners and programmers to make timely adjustments to policies and programs based on these data, is the rationale for conducting the PMA surveys [21]. The first EPMA survey was conducted in 2014, which was followed by the 2015 and 2016 rounds. The EPMA is led by the Addis Ababa University’s School of Public Health at the College of Health Sciences, in collaboration with regional universities, the Federal Ministry of Health and the Ethiopia Central Statistics Agency. Overall direction and support is provided by the Bill & Melinda Gates Institute for Population and Reproductive Health at Johns Hopkins Bloomberg School of Public Health and funded by the Bill & Melinda Gates Foundation. The EDHS and EPMA surveys are based on nationally representative multi-stage cluster sampling, in which the country’s census enumeration areas served as clusters or primary sampling units in both surveys. Both surveys are designed to provide estimates for several indicators for the entire country and, separately, for individual regions. The CSA was responsible for the selection of sample clusters for both surveys using the standard DHS sampling methodology and maintains the comparability of the two surveys. Both surveys raw data with details of data dictionaries and data structures are available in the public domain upon request. Pooling the data from these surveys provides a large data set to allow robust statistical analyses with high precision. The DHS data used for this study are openly available and can be downloaded at http://www.measuredhs.com/data/available-datasets.cfm?inputSearch=ETHIOPIA. Upon permission from PMA2020, we downloaded the three rounds of the PMA data from http://www.pma2020.org/dataset-download. Pertinent to this study are the data on the timing of first birth, first marriage, number of children ever born, contraceptive behavior, and basic demographic and socio-economic characteristics of the women, which were gathered from both the EDHSs and EPMA surveys in a similar manner. The main outcome variable of interest was early childbearing, which was measured by the cumulative probability of having a first birth before the age of 20 years. We also estimated the cumulative probability of having a first birth by a given adolescent age, such as by age 15, 16, 17, 18 or 19 years, based on the responses of women aged 20–49 years who participated in the different surveys. The data from the six surveys were pooled to create cohorts of women since the early 1970s. For women whose ages ranged from 20 to 49 years in each of the surveys, the year at which they reached puberty (age 10 years) was computed by subtracting the number of years elapsed since they reached an age of 10 years from the date of the surveys. For instance, a woman who was 40 years old in the 2016 survey and another woman who was 35 years old in the 2011 survey both reached puberty in 1986. These women in turn would experience the period of adolescence (age 10–19 years) during the period from 1986 to 1995. Although these women came from two different surveys, they represented a cohort of women who reached puberty around the same year and were thus assigned to the same cohort. Accordingly, five successive cohorts were reconstructed, each representing different periods of entry into the period of adolescence – i.e., 1971–1981; 1982–1987; 1988–1993; 1994–1999; and 2000–2005. The 2000–2005 cohort represents the most recent one and comprised women who were 20–26 years old in 2016. We excluded those women who reached puberty prior to 1971 from the analyses due to the small sample size. Women who belonged to the same cohort were destined to pass through similar social, economic and related transformations. In this respect, we can mention a few important landmarks that are relevant to the health, population and developmental activities of the country over the past decades and to which the various cohorts were exposed. First, the Ethiopian government issued a number of policies in 1993, including those targeting health, population and women. Second, successive health sector development programs were launched and implemented since the mid-1990s. Third, beginning in the year 2000, the government of Ethiopia has implemented several community-based programs to expand access to primary health care services throughout the country, the most important of which was the launch of the health extension program in 2003. We examined the relevance of selected background and proximate factors for early childbearing. These factors include the region (categorized into 11 regional states), urban-rural residence, maternal education, and household wealth. We used three categories for women’s educational status: no education, elementary education (1–6 years of schooling), and secondary or higher education (7 plus years of schooling). The EDHS and EPMA survey raw data were provided with wealth index variables that were constructed to rank households using principal component analysis. The wealth quintiles encompassed five categories – poorest, poor, medium, rich, and richest. Proximate factors included in our analyses were marital status and contraceptive use during the adolescent period. We created four marital status categories: (1) married before 18 years old, (2) married at 18 years old, (3) married at 19 years old and (4) not married before 20 years old. This categorization of marital status in part references the Ethiopian minimum marriage age of 18 years. Contraceptive use is a dichotomous variable that measures whether the women used a contraceptive during adolescence (i.e., before 20 years old). The data for the women were transformed into person-years, in which individuals contributed the number of person-years they lived before having a first birth during adolescence. The outcome variable (early childbearing) was dichotomized as 1 or 0. If a woman had her first birth before age 20 years, she was coded as 1; otherwise, she was coded as 0. We used a life-table analysis using Kaplan-Meier (KM) methodology to compute the cumulative probability of having a first birth before 20 years of age. We also estimated the cumulative probabilities of having a first birth at ages of 10, 11, 12…19 years. The median age at first birth was also estimated using the same approach. In addition, the KM method was also used to estimate the cumulative probability of early marriage (i.e., before the legal age of 18 years). We present temporal trends in the cumulative probability of early childbearing across women’s cohorts at the national level and separately for each region. The trend analysis for each region excluded those women who did not spend their adolescent period (age 10–19 years) in their current place of residence (i.e., current region). The EDHS and EPMA survey collected data for the place of birth, mobility and number of years that the respondent continuously lived in the current place of residence. Using this information, we were able to identify women who spent their adolescent period in the region they resided at the time of the survey and those who did not. Due to the high influx of people to Addis Ababa, the capital city, we did not present a separate trend analysis for the city. The data on mobility revealed that approximately 60% of the women who are currently residing in Addis Ababa were born somewhere else and of whom half did not spend their adolescent years in the city. This result is in stark contrast to the other regions, where approximately 85% of the survey respondents were still living in their birthplaces, and 5% reported that they moved out of their birthplaces but were still living in the same region of other districts. In addition, 6% of the women lived in their birthplaces until 19 years of age before they moved out to other regions of the country. These together comprise 96% of the sample of women who were suitable for the trend analysis. The regional trend analysis excluded women who did not spend their adolescent years in the regions they were residing at the time of the survey. Temporal trends in the cumulative probability of early childbearing across the cohorts were tested for statistical significance using the log-rank test. To examine determinants of early childbearing, multivariate analyses were performed using the Cox proportional hazards regression model. We present the adjusted hazard ratio (HR) and p-value for each covariate. We performed three separate models of background and proximate factors. The first model included only the background factors, i.e., region, education, residence (urban/rural), and wealth. As the main proximate determinant of fertility, we added marital status to the background factors in a second model. The third model additionally included contraceptive use, but it was restricted to only those women who married during adolescence. All three models were applied to the most recent cohort of women. We used Stata version 12 (Stata Corporation, College Station, TX, USA) for data management and analyses. The Survey command in STATA was used to delineate the strata and primary sampling unit. All proportions, rates and hazard ratios were weighted for the sampling probabilities.
N/A
DIMA AI Care
