Background

Every year, an estimated 21 million girls aged 15 to 19 years in developing regions become pregnant and approximately 12 million of them give birth [1]. Globally, adolescent birth rate has decreased from 64.5 births per 1000 adolescent girls (15–19 years) in 2000 to 41.3 births per 1000 adolescent girls in 2023. However, rates of change have been uneven in different regions of the world with the sharpest decline in Southern Asia and slower declines in the Latin American and Caribbean (LAC) and Sub-Saharan Africa (SSA) regions. Although declines have occurred in all regions, SSA and LAC continue to have the highest rates globally at 99.4 and 52.1 births per 1000 women, respectively in 2022. In the WHO African Region, the estimated adolescent birth rate was 97 per 1000 adolescent girls in the European Region [2].

Africa is home to the world’s youngest and rapidly growing population. In 2019, the continent had an estimated 230 million young people aged between 15 and 24, constituting 19% of the global youth population. Projections indicate that by 2030, the number of youths residing in Africa will have surged by 42 percent in 2023 [3]. Alarming trends suggest that the total number of teenage pregnancies is expected to increase by 2030, with sub-Saharan Africa projected to witness a higher prevalence. Notably, the African nations with the highest prevalence of teenage pregnancies include Niger, Mali, Angola, Mozambique, Guinea, Chad, and Cote d’Ivoire [4]. This is particularly concerning given that the region already leads in both teenage pregnancies and child marriages [5].

For over four decades, Zambia has grappled with persistently high fertility rates [6, 7]. However, there has been a noteworthy decline in the total fertility rate, decreasing from 6.5 children in 1992 to 4.7 children in 2018 [7, 8]. Approximately 29% of adolescents aged 15 to 19 in Zambia have already given birth, with 6% of these births occurring among 15-year-olds and a substantial 58% among those aged 19 [6]. Significant variations exist in the percentage of adolescent girls aged 15 to 19 engaging in childbearing, ranging from 14.9% in Lusaka to 42.5% in the Southern Province in 2018 [7].

Various studies have shown that the key determinants of fertility among the older adolescents and younger adults is individual’s current age, type of residential area, educational attainment, contraceptive utilization, and socioeconomic status. Research indicates that adolescent fertility often stems from the denial of sexual and reproductive health rights for young girls. Zambia grapples with one of the world’s highest child marriage rates, with a slight decrease from 31.7% in 2024 to 29% in 2018 [9]. Notably, 16.5% of girls aged 15 to 19 are married, and 31.4% of those aged 20 to 24 married before turning 18 [10]. The 2007 Zambia Demographic Health Survey revealed that among the general population aged 15–19 years, 12.3% and 16.2% of women and men respectively had sexual debut before the age 15, with sexual activities comes with pregnancies and eventually children [11]. In response to critical issues like communicable and non-communicable diseases, early marriages, and pregnancies, Zambia implemented the Adolescents Health Strategic Plan (2011–2015), endorsed by the Adolescent Friendly Health Services (ADFHS), to tackle challenges, including adolescent fertility [12].

Adolescent mothers aged between 10 and 19 face higher risks of eclampsia, puerperal endometritis and systemic infections than women aged 20 to 24 years, and babies of adolescent mothers face higher risks of low birth weight, preterm birth and severe neonatal condition. Moreover, research has underscored the correlation between age and maternal mortality, with a prevalence rate of 13% observed among individuals aged between 10 and 19 years [13]. Preventing pregnancy among adolescents and pregnancy related mortality and morbidity are foundational to achieving positive health outcomes across the life course and imperative for achieving the sustainable development goals (SDGs) related to maternal and newborn health [14]. In light of these concerning trends, this study aims to investigate the socio-economic disparities and associated factors of fertility among adolescents aged 15 to 19 years in Zambia.

Methods

This study constitutes a secondary analysis of microdata utilizing national-level data sourced from the Zambia Demographic and Health Survey (ZDHS) program. The ZDHS is a comprehensive, nationally representative household survey conducted by the Zambia Statistics Agency in collaboration with global partners, including ICF International and the United States Agency for International Development (USAID). The survey employs a two-stage sampling process, initially selecting enumeration areas (EAs) and subsequently households. The nature of the DHS data enables the comparison of variables over time, facilitating the monitoring of changes in indicators across various geographical regions [15].

The study included all individual files of participants aged 20 and above from selected households who had consented to take part in the research. The study adopted a complete case analysis were missing values were removed automatically at every point of analysist. Detailed methods employed in the DHS are comprehensively documented elsewhere [7]. For this specific study, we extracted all pertinent variables from the adolescent girls aged 15 to 19 years data files (individual recode) 2018 ZDHS datasets. The data under examination pertains to the population of adolescent’s aged 15–19 years. Data collection took place from 18 July 2018 to 24 January 2019 [14]. Data was accessed 1^st of October and 20^th of November 2023. The authors in this study did not have access to information that could identify individual participants during or after data collection (see Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Description of sample derivation criteria.

For this specific study, we extracted all pertinent variables from the women’s data files (individual recode) 2018 ZDHS dataset. The data under examination pertains to the population of adolescent women in reproductive age group (15–19). Women 20 years and above were dropped in the study.

https://doi.org/10.1371/journal.pgph.0002739.g001

Data analysis

For descriptive purposes, frequencies and percentages were computed for categorical variables using sample weighting for accurate representation. To determine association between the outcome variable (“ever given birth”) and the categorical variables, the Uncorrelated Design Based Chi-square test (Rao–Scott Chi-square test) was used. This was selected to take into account clustering. The study furthermore utilized the survey multilevel mixed effect ordinal logistic regression (proportional odds model) to determine the factors associated with fertility among adolescents. This approach accounted for the hierarchical structure of the data, with adolescents aged 15–19 nested within households and households nested within clusters. The outcome variable (ever given birth; no child, 1 child and 2+ children) is ordinal and using multinomial regression would give less efficient estimates. The study used an investigator led approach, all variables were selected from wide range literature. The probability F-test showed that the adopted model explained the outcome better than the null model (model without the explanatory variables), P<0.0001. In addition, the Brant test was utilized to examine whether the model had fulfilled the parallel lines assumption. A significant test statistic would suggest a violation of the parallel regression assumption. Consequently, the proportional odds model (p = 0.6083) was adopted with confidence, as it met the assumption. The log likelihood ratio test, Akaike Information Criteria (AIC) and the Bayesian Information Criteria (BIC) were sufficiently explored to select the best fit model. The variance for the predictors was less than 5 imposing no worry on multicolinearity in the model (see Table 1). Stata version 14.2 was used for the analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Multicolinearity test.

https://doi.org/10.1371/journal.pgph.0002739.t001

The proportional odds model is expressed in the logit form as:

Where, π_j(x) = π(Y ≤ j|x₁, x_2, …,x_p), which is the probability of being at or below category j given a set of predictors, j = 1, 2, …, j– 1. α are the cut of points and β₁, β₂, …, β_p are logit coefficients. To estimate the ln(odds) of being at or below the jth category, the Proportional Odds model can be rewritten as:

This model is based on the assumption of consistent effects represented by β for each logit [21, 22]. To find the magnitude and presence of relationships, odds ratios and their corresponding 95% confidence intervals were computed. The proportionality of odds for the dependent variable was evaluated using the Brant Test.

Results

The study’s findings reveal several notable patterns among the participants. Firstly, it is evident that a higher proportion of participants (95.61%) with no prior history of children were 15 years old. In contrast, a majority (37.26%) of participants who had two or more children were 19 years old. Furthermore, the data demonstrates that the majority of adolescents who gave birth between the ages of 15 to 17 comprised those who had only one or more children. Additionally, a significant proportion of these adolescents with one or at least children originated from rural areas (25.62% and 4.34% respectively). In terms of marital status, 60.2% of adolescents with a history of one childbirth were married, as opposed to those who were not. Similarly, a majority of adolescents with a history of multiple childbirths were married, accounting for 14.86%%. Moreover, the study’s findings suggest a significant prevalence of adolescents from low-income households (poorest) with a high proportion of having 2 or more children (6.12%), in contrast to those adolescents from wealthier (richest) backgrounds (0.28%). This association was found to be statistically significant, with a p-value of less than 0.0001 as shown in Table 2 below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Variables distribution and association of adolescent fertility in Zambia 2018.

https://doi.org/10.1371/journal.pgph.0002739.t002

Adolescent girls with no history of a childbirth

In urban settings, the prevalence of adolescents without a history of childbirth was slightly higher than in rural settings, with the exception of the southern province. In the southern province, a notable proportion of adolescent girls in rural areas reported no history of childbirth compared to their urban counterparts. Additionally, a significant difference in the population of adolescents without a history of childbirth was observed in the eastern province, as depicted in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Adolescent girls with no history of a childbirth.

Generated by excel.

https://doi.org/10.1371/journal.pgph.0002739.g002

Adolescent girls with a history of one childbirth

Additionally, the study’s findings indicate a higher count of adolescents with a history of one childbirth hailing from rural settings, with the exception of the southern province, where there is a greater proportion of adolescents in urban settings who have experienced childbirth. Furthermore, Luapula province documented the lowest percentage of adolescents with a history of one childbirth, and in urban areas, the lowest proportion was observed in Lusaka province as shown in Fig 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Adolescent girls with a history of one childbirth.

Generated by excel.

https://doi.org/10.1371/journal.pgph.0002739.g003

Adolescent girls with history of at least two childbirths

The study’s results in Fig 4 reveal a higher percentage of adolescents in rural settings with a history of at least two childbirths across all provinces, except for the Northern Province. Noteworthy variations were observed in the central, eastern, Luapula, Lusaka, and northwestern provinces. Specifically, the Northern Province exhibited a higher proportion of adolescents in rural settings with a record of at least two childbirths, followed by the central province. In urban settings, the Western Province took the lead, followed by Muchinga.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Adolescent girls with history of at least two childbirths.

Generated by excel.

https://doi.org/10.1371/journal.pgph.0002739.g004

Multilevel mixed effect ordinal logistic regression

The study investigated the factors associated with adolescent fertility using the multilevel mixed effect ordinal logistic regression. The findings suggest that a year increase in the age of adolescents increased the odds of being beyond a particular category of fertility (Multiple child births), given the effects of all other predictors are held constant. In other words, older adolescents were associated with increased odds of having multiple children, (AOR, 1.50; 95% CI, 1.31–1.72; P<0.0001) and this was statistically significant. In the same vein, adolescents with a higher level of education (primary, secondary and tertiary education) had reduced odds of being beyond a particular fertility category compared to those with no level of education (AOR, 0.47; 95% CI, 0.23–0.97; AOR, 0.2; 95% CI, 0.10–0.47; AOR, 0.03; 95% CI, 0.00–0.54 respectively).

Furthermore adolescents who were married, used contraceptives, coming from the poorer backgrounds or had an exposure to family planning (FP) messages had increased odds of having multiple children (being beyond a particular fertility category) (AOR, 2.56; 95% CI, 1.78–3.67; AOR, 3.09; 95% CI, 2.20–4.32; AOR, 1.68; 95% CI, 1.11–2.53 and AOR, 1.18; 95% CI, 1.13–1.22 respectively). Similarly adolescent told about FP at a health facility had increased odds of 2.77 of being beyond a particular category of fertility (higher number of births) compared to those were not told (95% CI, 2.04–3.76). Holding all things constant, a year increase in age at first sex increased the odds of having multiple children by a factor of 1.18 times (AOR, 1.18; 95% CI, 1.13–1.22). Among the contextual factors, that is residential area, community poverty, community level of education, community age at first birth, community access to FP message, community employment and community idea number of children), only community age at first birth was predicting multiple births among adolescents. In other words, adolescents from a high community age at first birth had increased odds of being beyond a particular category of fertility compared to those from a low community age at first birth (AOR, 1.59; 95% CI, 1.01–2.52), see Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Multilevel mixed effect ordinal logistic regression.

https://doi.org/10.1371/journal.pgph.0002739.t003

Discussion

This study analyzed the predictors of fertility among older adolescents aged between 15 and 19 in Zambia. Utilizing a multilevel ordinal logistic regression on data from the 2018 Zambia Demographic and Health Surveys, the study aimed to gain deeper insights into the factors contributing to elevated adolescent fertility in this age group. The study identified demographic and socio-economic disparities in adolescent fertility, revealing that a majority of adolescents with one or more children were aged 19. About 60% of married adolescents had one child, and approximately 14% in rural areas had more than one child. Rural areas showed significant increases in adolescents with multiple children. Furthermore, those with no formal education, married, residing in rural areas, low-income, or unemployed were more likely to have multiple children. This aligns with studies in Kenya, Nigeria, Malawi, and Tanzania [23–26]. However, the study noted that the percentage of adolescents with one child was higher than in Malawi, Uganda, Tanzania, Ethiopia, Rwanda, Eritrea, and Ghana [27–30].

The study revealed that certain individual factors such as education, marital status, wealth index, contraceptive use, exposure to family planning messages, and receiving family planning education at health facilities were significantly associated with higher adolescent fertility in Zambia. Notably, only community age at first birth emerged as a significant predictor of higher adolescent fertility. Among the models considered, model three, which incorporated both individual and community-level variables, was accepted. This decision was based on its superior fit, as evidenced by smaller values of AIC, BIC, and Log likelihood in comparison to the null model (without individual and community level variables) and model two (including only individual-level variables).

Furthermore, the study findings indicate that older adolescents face an increased risk of having more than one child. This aligns with observations from studies conducted in Kenya, Nigeria and Ghana, where it was noted that older adolescents are more susceptible to higher fertility compared to their younger counterparts [23, 27, 31]. A similar trend was identified in a study conducted in Indonesia, revealing a positive correlation between the age of adolescents and birthrate [32]. Similarly, our study revealed that adolescents who were older at the time of their first sexual encounter and those from communities with a higher community age at birth were more likely to have more than one child. This finding contradicts a study conducted in Uganda, which found that individuals who engaged in early sexual activity and delayed childbirth were less likely to experience higher fertility compared to those who underwent these events early [23, 33].

Our findings indicate that adolescent girls who completed primary, secondary, and tertiary education were less likely to have more children compared to their counterparts with no formal education. This observation is consistent with studies conducted in Nigeria, where it was established that adolescents aged 15–19 with primary education or higher exhibited lower risks of having children in comparison to adolescents with no formal education [25, 34]. This suggests that education plays a crucial role in enhancing autonomy, decision-making abilities, and economic independence. Consequently, it contributes to delaying marriage and sexual debut while increasing knowledge about contraception, as noted in studies such as those by Mohr et al. [35].

Marital status in the study similarly predicted adolescent fertility. The findings suggests that married adolescent girls had a higher chance of having multiple children compared to those who were not. This aligns with research conducted in various African countries, including Burundi, Ethiopia, Nigeria, Kenya, Congo, and Central Africa, where early marriages have been identified as a significant factor contributing to high fertility rates [23, 36–38]. This implies that early child marriages play a central role in driving increased adolescent fertility. The pressure exerted on married adolescents by their families and partners to start their own families contributes to this phenomenon. Teenage pregnancies resulting from early marriages pose heightened risks, including increased chances of maternal mortality during childbirth and elevated rates of neonatal and infant deaths among young mothers. Girls aged between 15 and 19 face double the likelihood of dying during childbirth compared to women aged 20 and above. This risk is further compounded by factors such as HIV, making complications during pregnancy and childbirth the primary cause of death for young women in this age group. Additionally, pregnant adolescents are more susceptible to serious health issues such as eclampsia, puerperal endometritis, and systemic infections compared to adults [39].

Furthermore adolescents who used contraceptives, were exposed to family planning (FP) or received FP education at health facilities were more likely to have more than one child compared to those who did not use contraceptives. This aligns with research conducted in Kenya and Zambia, which indicated that adolescents using contraceptives were at a higher risk of experiencing higher fertility [23, 40]. This phenomenon could be attributed to the observation that adolescents often initiate contraceptive use after having their first child. Furthermore, a contributing factor may be the cultural norms that discourage having children out of wedlock, particularly prevalent in rural areas, leading many girls who become pregnant to enter early marriages [41].

In contrasts to the use of contraceptives, exposure to family planning (FP) or receiving FP the findings in Ethiopia, Namibia, and Burundi, where research indicated that teenagers who receive sexual education through media channels, such as television and radio, are likely to abstain from sexual intercourse due to knowledge about the dangers of early engagement, such as sexually transmitted infections and risky pregnancies leading to potential harm [36, 42, 43]. This inconsistency might be attributed to the perceptions held by young people that family planning messages are primarily intended for older married individuals and do not apply to them, leading them to disregard or avoid such messages [37].

Residence did not emerge as a significant predictor of adolescent fertility in the study, a result that aligns unexpectedly with studies conducted in Ethiopia, Uganda, and Mexico. These studies similarly found that the place of residence was not a significant predictor of higher adolescent fertility in Zambia [44, 45]. In contrasts to the findings in the descriptive results that shows a higher proportion of adolescents with at least one childbirth residing in rural areas across majority of the provinces. The differences are much wider among adolescents with at least two multiple childbirths.

Furthermore, as a way combating fertility among adolescents the government of Zambia has introduced Comprehensive Sexuality Education (CSE) in schools. Additionally, the Government of Sweden, through the Centre for People (PeaCe) health program, has supported the Ministry of Health (MoH) [46]. A key strategy of the program aims to reduce adolescent pregnancy by increasing access to quality Adolescent Health (ADH) information and promoting behavior change through outreach and facility-based engagement models. With support from the Clinton Health Access Initiative (CHAI), MoH has trained healthcare workers and peer educators (PEs) in selected provinces to provide people-centered adolescent health services, including sexual and reproductive health (SRH) and information services, both at the community and facility levels [47].

Acknowledgments

We wish to express our sincere gratitude to the Zambia Statistics Agency and the DHS Program for granting permission to utilize the 2018 ZDHS dataset.