Multilevel analysis of women’s empowerment and maternal high-risk fertility behavior in Sub-Saharan Africa, 2011–2022 Authors: Jolene Tan, Shao-Tzu Yu Published: 2026 ABSTRACT BackgroundHigh-risk fertility behaviors remain a persistent public health challenge in sub-Saharan Africa, where maternal mortality and fertility rates are high. Women’s empowerment is an essential factor in understanding and addressing high-risk fertility behaviors. However, previous studies have often focused on singular aspects of empowerment, such as education or decision-making authority. This study adopts a multidimensional approach to examine the association between women’s empowerment and high-risk fertility behaviors across sub-Saharan Africa.MethodsWe used Demographic and Health Survey data from 29 sub-Saharan African countries collected between 2011 and 2022 (n = 179,430). Women’s empowerment was measured using principal component analysis to derive three dimensions of empowerment (decision-making, attitudes toward violence, and social independence). High-risk fertility behaviors were assessed based on childbirth before age 18, short birth intervals (< 24 months), and high birth parity (≥ 4 births). We used multilevel regressions to examine the association between empowerment dimensions and high-risk fertility behaviors while accounting for country-specific effects and controlling for potential determinants and risk factors.ResultsAcross the study population, substantial variation in women’s empowerment dimensions was observed across countries and population subgroups. Multilevel analyses showed significant associations between each empowerment dimension and reduced probabilities of engaging in high-risk fertility behaviors. For example, a one standard deviation increase in social independence was associated with lower odds of early childbirth (OR = 0.534; CI = 0.502–0.567), short birth intervals (OR = 0.796; CI = 0.761–0.832), and high birth parity (OR = 0.560; CI = 0.532–0.590). Similar negative associations were observed for the decision-making and attitudinal dimensions. These associations appeared strongest for high-risk behaviors occurring earlier in women’s reproductive lives.ConclusionsThe findings highlight the importance of women’s empowerment in reducing high-risk fertility behaviors across multiple dimensions and life stages, providing insights for potential policy interventions targeted at gender-based factors contributing to high-risk fertility behaviors to improve the region’s maternal and child health outcomes.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12884-026-09430-5. Introduction Although significant efforts have been made to reduce maternal mortality worldwide, a disproportionate number of maternal deaths continue to occur in low- and middle-income settings, accounting for approximately 95% of global maternal deaths due to complications from pregnancy or childbirth [1, 2]. Sub-Saharan Africa (SSA) bears a high burden, with this region alone accounting for approximately 70% of global maternal deaths [1, 3]. Understanding the key contributing factors to poor maternal health outcomes is imperative for developing targeted strategies to improve maternal health across SSA regions. In SSA, high-risk fertility behaviors continue to be closely linked to poor maternal health outcomes [4, 5]. High-risk fertility behaviors—characterized by narrow birth intervals, high birth order, and young maternal age at birth—have been reported to have detrimental effects on both mothers and children [6–8]. Specifically, infants born to extremely young mothers (below 18 years), within short birth intervals (less than 24 months after the previous birth), or to mothers with high parity (four or more children) are at a higher risk of mortality [9, 10]. Pregnancies before age 18 can lead to various complications, such as low birth weight [11] and premature birth [12]. In addition, teenage mothers are more likely to develop anemia [13]. Short birth intervals are associated with congenital abnormalities and adverse perinatal health outcomes [14, 15]. The prevalence and patterns of these behaviors, nonetheless, may vary considerably across countries in the region. Such variation may reflect differences in sociocultural norms, early marriage practices, educational opportunities, and access to reproductive health services across settings [6–15]. As such, addressing and mitigating high-risk fertility behaviors is regarded as a critical strategy to reduce maternal mortality and improve child outcomes across SSA [1–3]. Women’s empowerment may be considered a salient factor in understanding and managing high-risk fertility behaviors among women in SSA [16–19]. The concept of women’s empowerment originated from the feminist movement of the 1970s, and early literature defined it as an interactive process that allows individuals, particularly those considered powerless, to take control of their lives [20–22]. The focus is on enhancing human agency, which involves increasing individuals’ ability to gain control over or have the choice to access formal and informal resources [23]. Women’s empowerment has thus historically been regarded as a process through which improvements in women’s education, employment, and healthcare access can be achieved [24]. However, initial conceptualizations of empowerment often overlooked the broader social, institutional, and cultural factors fundamental to the empowerment process [25, 26]. In many low- and middle-income countries, women’s ability to exercise agency may continue to be constrained by context-specific practices and norms despite improvements in education and labor force participation [18, 27]. Older family members or husbands may still dominate decisions concerning accessing healthcare, visiting family and friends, and joining the workforce. These decisions relate to women’s capability to challenge institutionalized ideologies and practices, emphasizing how they can negotiate their roles and responsibilities within and beyond their local social contexts [28, 29]. As the understanding of empowerment has evolved, subsequent conceptualizations have recognized that women’s agency is often shaped by broader social and structural environments, including household power dynamics, spousal relationships, institutional arrangements, and cultural norms that may constrain or facilitate opportunities [25–29]. More recent perspectives therefore conceptualize women’s empowerment as a multidimensional process operating across individual, household, and broader social contexts. Empirically, recent work has only begun to examine how these dimensions may jointly interact and correlate with one another to provide a more complete understanding of women’s empowerment profiles [29–31]. However, despite well-established theoretical developments in envisioning women’s empowerment, empirical evidence supporting the link between the progress in empowerment and high-risk fertility behaviors remains limited and inconclusive [19]. Previous studies have suggested that women who are empowered—measured indirectly through women’s education, employment, and participation in household decisions—are more likely to engage in practices that reduce maternal high-risk fertility behaviors. These practices include using modern contraception [32], accessing antenatal care [33], ensuring skilled birth attendance [34], and providing adequate nutrition to their children [35]. Nonetheless, these studies often leverage a proximate approach to measure empowerment, focusing less on how the various dimensions of empowerment in the individual, household, and broader social environments may jointly interact. Well-established aggregate measures, such as the Global Gender Gap Index [36], Gender Development Index [37], and Gender Inequality Index [38], have created an opportunity for comparative work ranking countries based on gender equality achievements. However, these approaches often overlook the differences in how empowerment manifests in subpopulations across the SSA regions. Moreover, they fail to account for the complex social dynamics that may influence the link between empowerment processes and health risk behaviors and outcomes. This study examined the relationship between women’s empowerment and high-risk fertility behaviors. We aimed to advance the extant literature in two ways. First, the study considered the multidimensional aspect of women’s empowerment in personal, household, and societal domains and performed comparable analyses of maternal health risk behaviors across the SSA. Second, we employed a multilevel analysis to determine the association between varying empowerment dimensions and maternal health risk on a continuum (maternal age at childbirth, birth intervals, and birth parity) while accounting for country-specific effects and controlling for other potential determinants and risk factors identified in prior studies. Given the paucity of research in this area, our findings have important implications for understanding the role of women’s empowerment in reducing high-risk fertility behaviors at different life stages among women in SSA. Methods Data and sample We used data from the Demographic and Health Surveys (DHS) conducted across 29 SSA countries between 2011 and 2022. The DHS is administered in low- and middle-income countries at approximately five-year intervals. It uses a stratified two-stage sampling approach to obtain nationally representative samples that are comparable across countries. Forty out of 51 SSA countries have conducted at least one DHS survey. In this study, we excluded five countries with surveys only collected before 2010 (Botswana, Central African Republic, Eritrea, Eswatini, and São Tomé and Príncipe) and six countries with incomplete survey items (Congo, Ethiopia, Mauritania, Niger, South Africa, and Tanzania). For countries with multiple surveys, we used the latest available data. The complete list of countries and survey years can be found in Appendix Table A1. We pooled a sample of 222,281 partnered, reproductive-age women (15–49 years) who had given birth within the five years preceding the survey and were not sterilized. After excluding women with incomplete data on the variables of interest—specifically, those with missing information on empowerment measures—the final sample included 179,430 women. A comparison of the descriptive characteristics between the final analytic and excluded sample can be found in Appendix Table A2. Empowerment measures We adopted the Survey-based Women’s Empowerment Index (SWPER) developed by Ewerling et al. [39] to measure women’s empowerment. Conceptually, this approach perceives women’s empowerment as a multidimensional construct that may be more appropriately captured by evaluating women’s socioeconomic status and their perceptions of their agency in the relational, household, and health dimensions. Therefore, this approach extends the proximate approach to women’s empowerment by arguing that multiple dimensions may operate jointly to formulate diverse empowerment profiles. Empirically, the SWPER index leverages principal component analysis (PCA)—a data reduction technique—to estimate the diversity between and consistency within each identified empowerment dimension. This approach transforms the correlated empowerment indicators into a set of uncorrelated principal dimensions (or components); each component captures a specific variance in the data, allowing for the identification of underlying patterns and relationships within the multidimensional empowerment construct. As such, this flexible approach has broad applicability because it allows for comparisons among countries and population subgroups across the region. Here, we aligned with previous approaches to constructing a SWPER index by utilizing 14 items from the DHS. These items include women’s attitudes toward wife-beating (namely whether it is justified in specific scenarios); participation in household or personal decisions, such as purchasing household items, accessing health care, and visiting relatives; employment history in the past 12 months; educational attainment and differences in spousal education and age; and age at first cohabitation. Details regarding the specific items and their coding can be found in Appendix Table A3. High-risk fertility behavior measures We included three high-risk fertility behavior measures as our outcome variables. High-risk fertility behavior was assessed based on whether (1) the women’s age at childbirth is under 18 years (1 = yes, 0 = no), (2) their birth interval is less than 24 months after the previous birth (1 = yes, 0 = no), which was estimated using DHS birth history information and applied only to women with sufficient reproductive history for whom a previous birth was available, or (3) their birth parity is four or higher (1 = yes, 0 = no). In addition, we constructed another outcome measure that summarized the total number of high-risk fertility behaviors reported by the respondents, from 0 to 3. The criterion for respondents’ age exceeding 34 years at childbirth was excluded due to the majority of women not reaching 34 years at the time of the survey; moreover, only a small percentage (14%) of women had births beyond 34 years. Controls We controlled for a range of household and demographic characteristics that could be associated with women’s empowerment or maternal high-risk fertility behavior. These include sex of household head (female, male), household size (1–4, 5–10, > 10 members), place of residence (rural, urban), household wealth (quintiles), history of terminated pregnancy (no, yes), religion (no religion, Catholic, Christian, Muslim, traditional/spiritual/animist, other), five-year age groups, and survey period. These factors may influence maternal childbearing risk by affecting women’s sexual reproductive health decisions, such as using contraception to prevent pregnancy or delaying their sexual debut. Ethical approval We used secondary data with no identifiable information on survey respondents. Permission to use the data for this project was obtained from the DHS program [39]. The DHS survey protocols for each country received approval from the respective ethical review bodies. We carried out all data collection procedures in accordance with relevant guidelines and regulations. Analyses We conducted a principal component analysis on the 14 items related to women’s empowerment to determine the number of components to retain [30, 31, 40]. The descriptive statistics and tests of differences for the items used in the empowerment index and high-risk fertility behavior are provided in Appendix Table A4. Consistent with the SWPER index, we identified three dimensions of empowerment: decision-making, attitudes toward violence, and social independence. Decision-making relates to women’s involvement in decisions regarding the household, their personal health, and visits to relatives. The attitudinal dimension gauges women’s perspectives on the justification of wife-beating in different scenarios. Social independence includes factors contributing to women’s autonomy, such as education, media consumption, age at significant life events, and differences relative to their spouses. Scores for each dimension were normalized to a scale of 0 to 100 for ease of interpretation. We subsequently standardized these values in the regression analyses. Higher scores may translate into higher levels of personal agency, decision-making, and access to resources. We next performed three multilevel binary logistic regression models to predict high-risk fertility behaviors with respondents nested within their country of residence. These models can be formally summarized as1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}\:logit&\left({\pi\:}_{ij}\right)={\beta\:}_{0}+{\beta\:}_{1}\:{Dec}_{ij}+{\beta\:}_{2}\:{Att}_{ij}\\&+{\beta\:}_{3}\:{Indp}_{ij}{+\:\theta\:\beta\:}_{n}\:{Controls}_{nij}+{u}_{j}\end{aligned}$$\end{document} where πij represents the probability of each behavior for individual i in country j, including childbirth before age 18, a short birth interval, and a birth parity of four or higher. β0 is the intercept term, β1 Decij is the coefficient for the decision-making score for individual i in country j, β2 Attij is the coefficient for the attitudes toward violence score for individual i in country j, β3 Indpij is the coefficient for the social independence score for individual i in country j, uj represents the country-level random intercept that captures unobserved heterogeneity across countries, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\theta\:$$\end{document}βn Controlsnij indicates a vector of controls. We ran an additional multilevel multinomial logistic regression to examine the effect of women’s empowerment on the number of high-risk fertility behaviors. We applied sample weights to all analyses, which we performed using Stata 18. Results Table 1 presents the descriptive statistics for high-risk fertility behaviors. Across all three indicators, women who did not engage in high-risk fertility behavior had higher average empowerment scores than those who did. For example, women who gave birth at age 18 years or older had higher scores in decision-making (mean = 55.83; SE = 0.07), attitudes toward violence (mean = 41.27; SE = 0.05), and social independence (mean = 45.61; SE = 0.04) than women who gave birth before age 18 years (49.11, 37.33, and 40.61, respectively). Similar patterns were observed for birth interval and parity. Women with birth intervals of at least 24 months and those with fewer than four children consistently reported higher empowerment scores across all three dimensions than their counterparts with shorter birth intervals or higher parity. Table 1Descriptive statistics for high-risk fertility behavior across 29 SSA countries, 2011–22 (n = 179,430)VariablesChildbirth ≥ 18 yearsChildbirth < 18 yearsBirth interval ≥ 24 monthsBirth interval < 24 monthsBirth parity < 4Birth parity ≥ 4Mean/%SEMean/%SEMean/%SEMean/%SEMean/%SEMean/%SEEmpowerment dimension Decision-making55.830.0749.110.1055.690.0750.500.0956.000.0750.750.08 Attitudes toward violence41.270.0537.330.0740.590.0539.020.0640.800.0538.960.06 Social independence45.610.0440.610.0545.210.0442.080.0546.180.0541.300.04Age in years 15–192.220.0513.230.179.010.111.220.0510.810.120.030.01 20–2418.960.1424.710.2127.290.1711.350.1436.050.192.880.07 25–2927.080.1624.450.2127.980.1723.580.1931.940.1919.410.17 30–3422.830.1518.860.1918.830.1525.480.2014.260.1430.110.19 35–3917.620.1311.790.1611.290.1222.200.195.470.1027.780.19 40–448.390.105.540.114.440.0811.890.141.230.0514.790.15 45–492.910.061.430.061.160.044.290.090.240.025.000.09Sex of household head Female14.880.1312.940.1715.480.1412.400.1515.570.1412.660.14 Male85.120.1387.060.1784.520.1487.600.1584.430.1487.340.14Household size 1–432.500.1725.740.2241.090.1914.170.1649.290.207.690.11 5–1056.670.1860.510.2449.380.1970.660.2042.330.2076.460.17 > 1010.830.1113.750.179.520.1115.170.168.380.1115.850.15Place of residence Rural64.120.1874.030.2363.630.1973.010.2162.360.2073.390.20 Urban35.880.1825.970.2336.370.1926.990.2137.640.2026.610.20Household wealth status Lowest quintile19.370.1326.310.2118.870.1425.850.1918.200.1425.800.17 2nd lowest quintile19.430.1324.730.2119.920.1423.080.1819.350.1523.370.17 Middle quintile19.440.1421.130.2019.400.1520.890.1819.000.1521.180.17 2nd highest quintile20.420.1517.320.2020.700.1617.460.1820.860.1717.650.17 Highest quintile21.340.1610.510.1621.120.1712.720.1622.590.1812.000.15Ever had a terminated pregnancy No84.650.1386.460.1785.420.1485.000.1686.770.1483.460.16 Yes15.350.1313.540.1714.580.1415.000.1613.230.1416.540.16 Religion No religion1.590.042.060.071.670.051.870.061.550.051.980.06 Catholic18.380.1414.440.1817.310.1416.720.1717.610.1516.430.16 Christian57.500.1757.320.2458.420.1855.990.2259.020.1955.560.21 Muslim20.460.1424.250.2020.730.1523.180.1820.040.1523.700.17 Traditional/spiritual/animist1.170.031.120.050.970.031.430.050.830.031.540.05 Other0.890.030.820.050.910.040.820.040.950.040.780.04Estimates are weighted to the population using the sample weights. Mean values are reported for the three empowerment dimensions and percentages are reported for all other variables. The scores within each empowerment dimension have been normalized to a scale of 0 to 100, with higher scores indicating greater empowerment Table 1Descriptive statistics for high-risk fertility behavior across 29 SSA countries, 2011–22 (n = 179,430)VariablesChildbirth ≥ 18 yearsChildbirth < 18 yearsBirth interval ≥ 24 monthsBirth interval < 24 monthsBirth parity < 4Birth parity ≥ 4Mean/%SEMean/%SEMean/%SEMean/%SEMean/%SEMean/%SEEmpowerment dimension Decision-making55.830.0749.110.1055.690.0750.500.0956.000.0750.750.08 Attitudes toward violence41.270.0537.330.0740.590.0539.020.0640.800.0538.960.06 Social independence45.610.0440.610.0545.210.0442.080.0546.180.0541.300.04Age in years 15–192.220.0513.230.179.010.111.220.0510.810.120.030.01 20–2418.960.1424.710.2127.290.1711.350.1436.050.192.880.07 25–2927.080.1624.450.2127.980.1723.580.1931.940.1919.410.17 30–3422.830.1518.860.1918.830.1525.480.2014.260.1430.110.19 35–3917.620.1311.790.1611.290.1222.200.195.470.1027.780.19 40–448.390.105.540.114.440.0811.890.141.230.0514.790.15 45–492.910.061.430.061.160.044.290.090.240.025.000.09Sex of household head Female14.880.1312.940.1715.480.1412.400.1515.570.1412.660.14 Male85.120.1387.060.1784.520.1487.600.1584.430.1487.340.14Household size 1–432.500.1725.740.2241.090.1914.170.1649.290.207.690.11 5–1056.670.1860.510.2449.380.1970.660.2042.330.2076.460.17 > 1010.830.1113.750.179.520.1115.170.168.380.1115.850.15Place of residence Rural64.120.1874.030.2363.630.1973.010.2162.360.2073.390.20 Urban35.880.1825.970.2336.370.1926.990.2137.640.2026.610.20Household wealth status Lowest quintile19.370.1326.310.2118.870.1425.850.1918.200.1425.800.17 2nd lowest quintile19.430.1324.730.2119.920.1423.080.1819.350.1523.370.17 Middle quintile19.440.1421.130.2019.400.1520.890.1819.000.1521.180.17 2nd highest quintile20.420.1517.320.2020.700.1617.460.1820.860.1717.650.17 Highest quintile21.340.1610.510.1621.120.1712.720.1622.590.1812.000.15Ever had a terminated pregnancy No84.650.1386.460.1785.420.1485.000.1686.770.1483.460.16 Yes15.350.1313.540.1714.580.1415.000.1613.230.1416.540.16 Religion No religion1.590.042.060.071.670.051.870.061.550.051.980.06 Catholic18.380.1414.440.1817.310.1416.720.1717.610.1516.430.16 Christian57.500.1757.320.2458.420.1855.990.2259.020.1955.560.21 Muslim20.460.1424.250.2020.730.1523.180.1820.040.1523.700.17 Traditional/spiritual/animist1.170.031.120.050.970.031.430.050.830.031.540.05 Other0.890.030.820.050.910.040.820.040.950.040.780.04Estimates are weighted to the population using the sample weights. Mean values are reported for the three empowerment dimensions and percentages are reported for all other variables. The scores within each empowerment dimension have been normalized to a scale of 0 to 100, with higher scores indicating greater empowerment Descriptive statistics for high-risk fertility behavior across 29 SSA countries, 2011–22 (n = 179,430) Estimates are weighted to the population using the sample weights. Mean values are reported for the three empowerment dimensions and percentages are reported for all other variables. The scores within each empowerment dimension have been normalized to a scale of 0 to 100, with higher scores indicating greater empowerment High-risk fertility behaviors also varied across sociodemographic characteristics. Early childbirth was more common among younger women, with women aged 15–24 years accounting for a substantially larger share of births before age 18 than of births at age 18 years or older. Short birth intervals and high parity were concentrated among women aged 30 years and older, whereas women aged 20–29 years accounted for the largest proportion of women with lower parity and longer birth intervals. High-risk fertility behaviors were also more prevalent among women residing in rural areas, living in larger households, and belonging to lower-wealth households. For instance, women in the lowest wealth quintile accounted for 26.31% of births before age 18 years compared with 19.37% of births at age 18 years or older, while women in the highest wealth quintile accounted for only 10.51% of births before age 18 years but 21.34% of births at age 18 years or older. Similar socioeconomic gradients were observed for short birth intervals and high parity. Overall, the descriptive results suggest that lower levels of women’s empowerment and socioeconomic disadvantage are associated with a higher prevalence of high-risk fertility behaviors. Further, greater prevalence of high-risk fertility behaviors was generally observed in parts of East and Central Africa, and lower prevalence was more common in Southern Africa (Fig. 1). To provide additional context on the distribution of women’s empowerment, we present descriptive statistics for the three empowerment dimensions across sociodemographic characteristics in Appendix Table A5 and regional variations in empowerment dimensions across the 29 countries in Appendix Figure A1. Fig. 1Prevalence of high-risk fertility behaviors across 29 SSA countries, 2011–22 (n = 179,430). Note: Gray areas indicate no available data Prevalence of high-risk fertility behaviors across 29 SSA countries, 2011–22 (n = 179,430). Note: Gray areas indicate no available data To examine the link between empowerment dimensions and high-risk fertility behaviors, we included each dimension in three separate multilevel logistic regressions to predict fertility behavior outcomes. Figure 2 presents the average marginal effects of high-risk fertility behaviors by empowerment dimensions. Fig. 2Average marginal effects of empowerment dimensions on high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430). Note: Estimates are based on individual multilevel binary logistic regression models that control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. All estimates are adjusted for sample weights Average marginal effects of empowerment dimensions on high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430). Note: Estimates are based on individual multilevel binary logistic regression models that control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. All estimates are adjusted for sample weights Significant between-country heterogeneity was observed for all outcomes (Table 2). The intraclass correlation coefficients ranged from 4.7% to 8.6% in the fully adjusted models, indicating that a meaningful proportion of the variation in high-risk fertility behaviors was attributable to country-level differences. Likelihood-ratio tests comparing multilevel and single-level logistic models were statistically significant for all outcomes (p < 0.001), supporting the use of multilevel modeling. Country-level variance and ICC values decreased after adjustment for covariates across all outcomes, suggesting that part of the between-country variation in high-risk fertility behaviors was explained by individual and household characteristics. To further assess heterogeneity across countries, country-specific adjusted and unadjusted odds ratios are presented in Appendix Figures A2 and A3. The results were broadly consistent with the pooled estimates, with most countries showing significant associations between higher empowerment and lower odds of high-risk fertility behaviors. Table 2Multilevel model diagnostics for the association between women’s empowerment and high-risk fertility behaviors in 29 SSA countries, 2011–22 (n = 179,430)ModelCountry-level variance (SE)ICCLR test vs. single-level logisticLog likelihoodAICBICChildbirth < 18 years Null0.303 (0.080)0.084χ̄²=6664.06, p < 0.001-102,298.90204,607.80204,658.30 Full0.255 (0.068)0.072χ̄²=5086.66, p < 0.001-96,910.88193,871.80194,124.20Short birth interval < 24 months Null0.268 (0.071)0.075χ̄²=6,673.75, p < 0.001-115,109.64230,229.30230,279.80 Full0.163 (0.043)0.047χ̄²=5,589.20, p < 0.001-110,055.45220,160.90220,413.30Birth parity ≥ 4 Null0.311 (0.083)0.086χ̄²=4,459.11, p < 0.001-116,663.58233,337.20233,387.70 Full0.176 (0.021)0.051χ̄²=3,797.87, p < 0.001-111,801.41223,652.82223,905.20 Table 2Multilevel model diagnostics for the association between women’s empowerment and high-risk fertility behaviors in 29 SSA countries, 2011–22 (n = 179,430)ModelCountry-level variance (SE)ICCLR test vs. single-level logisticLog likelihoodAICBICChildbirth < 18 years Null0.303 (0.080)0.084χ̄²=6664.06, p < 0.001-102,298.90204,607.80204,658.30 Full0.255 (0.068)0.072χ̄²=5086.66, p < 0.001-96,910.88193,871.80194,124.20Short birth interval < 24 months Null0.268 (0.071)0.075χ̄²=6,673.75, p < 0.001-115,109.64230,229.30230,279.80 Full0.163 (0.043)0.047χ̄²=5,589.20, p < 0.001-110,055.45220,160.90220,413.30Birth parity ≥ 4 Null0.311 (0.083)0.086χ̄²=4,459.11, p < 0.001-116,663.58233,337.20233,387.70 Full0.176 (0.021)0.051χ̄²=3,797.87, p < 0.001-111,801.41223,652.82223,905.20 Multilevel model diagnostics for the association between women’s empowerment and high-risk fertility behaviors in 29 SSA countries, 2011–22 (n = 179,430) Overall, each dimension of empowerment was significantly associated with lower odds of engaging in high-risk fertility behaviors (Table 3). For every one standard deviation increase in the social independence score, for example, we found 46.6% lower odds of early childbirth (OR = 0.534; 95% CI = 0.502–0.567), 20.4% lower odds of a birth interval shorter than 24 months from the previous birth (OR = 0.796; 95% CI = 0.761–0.832), and 44.0% lower odds of having a birth parity of four or higher (OR = 0.560; 95% CI = 0.532–0.590). Table 3Multilevel logistic regression of empowerment dimensions on high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430)Childbirth < 18 yearsBirth interval < 24 monthsBirth parity ≥ 4OR95% CIOR95% CIOR95% CIDecision-making0.649***0.614–0.6860.806***0.785–0.8280.641***0.622–0.660Attitudes toward violence0.685***0.655–0.7170.841***0.822–0.8610.683***0.659–0.708Social independence0.534***0.502–0.5670.796***0.761–0.8320.560***0.532–0.590All models control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. All estimates are adjusted for sample weights. * p < 0.05; ** p < 0.01; *** p < 0.001. Table 3Multilevel logistic regression of empowerment dimensions on high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430)Childbirth < 18 yearsBirth interval < 24 monthsBirth parity ≥ 4OR95% CIOR95% CIOR95% CIDecision-making0.649***0.614–0.6860.806***0.785–0.8280.641***0.622–0.660Attitudes toward violence0.685***0.655–0.7170.841***0.822–0.8610.683***0.659–0.708Social independence0.534***0.502–0.5670.796***0.761–0.8320.560***0.532–0.590All models control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. All estimates are adjusted for sample weights. * p < 0.05; ** p < 0.01; *** p < 0.001. Multilevel logistic regression of empowerment dimensions on high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430) All models control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. All estimates are adjusted for sample weights. * p < 0.05; ** p < 0.01; *** p < 0.001. For the decision-making dimension, an increase of one standard deviation in decision-making scores was associated with 35.1% lower odds of early childbirth (OR = 0.649; 95% CI = 0.614–0.686), 19.4% lower odds of a short birth interval (OR = 0.806; 95% CI = 0.785–0.828), and 35.9% lower odds of having a birth parity of four or higher (OR = 0.641; 95% CI = 0.622–0.660). Similarly, for the attitudes dimension, a one standard deviation increase in attitudes toward violence scores was associated with 31.5% lower odds of early childbirth (OR = 0.685; 95% CI = 0.655–0.717), 15.9% lower odds of a short birth interval (OR = 0.841; 95% CI = 0.822–0.861), and 31.7% lower odds of having a birth parity of four or higher (OR = 0.683; 95% CI = 0.659–0.708). To further understand the importance of each empowerment dimension in mitigating high-risk fertility behaviors, we ran an additional multilevel multinomial logistic model to predict the cumulative numbers of behaviors reported by the respondents (Fig. 3). Average marginal effects showed that a one-standard-deviation increase in decision making, attitudes toward violence, and social independence was associated with a 5.8, 5.0, and 7.8%-point increase, respectively, in the probability of having no high-risk fertility behavior. All empowerment dimensions were also negatively associated with multiple high-risk fertility behaviors. Specifically, a one-standard-deviation increase in decision making, attitudes toward violence, and social independence was associated with a 1.5, 1.0, and 1.5%-point decrease in the probability of having two high-risk fertility behaviors, and a 4.7, 4.2, and 6.5%-point decrease in the probability of having three high-risk fertility behaviors. Fig. 3Average marginal effects of empowerment dimensions on the number of high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430). Note: Estimates are based on multilevel multinomial logistic regression models that control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. Estimates represent marginal effects indicating changes in probabilities associated with a one standard deviation increase in each empowerment dimension. All estimates are adjusted for sample weights Average marginal effects of empowerment dimensions on the number of high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430). Note: Estimates are based on multilevel multinomial logistic regression models that control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. Estimates represent marginal effects indicating changes in probabilities associated with a one standard deviation increase in each empowerment dimension. All estimates are adjusted for sample weights As these behaviors follow a continuum of reproductive stages (from childbirth before age 18 and short birth intervals to high birth parity), we subsequently isolated and examined the impact of these dimensions on each high-risk behavior—including childbirth under 18, short birth intervals, and high birth parity—in the absence of other high-risk factors (Table 4). We found significant inverse associations between all three empowerment dimensions and childbirth before age 18. Higher levels of decision-making (OR = 0.841; 95% CI = 0.799–0.885), attitudes toward violence (OR = 0.865; 95% CI = 0.832–0.899), and social independence (OR = 0.753; 95% CI = 0.710–0.798) were associated with lower odds of childbirth before age 18. Similar inverse associations were observed for short birth intervals, with higher levels of decision making (OR = 0.842; 95% CI = 0.767–0.913), attitudes toward violence (OR = 0.865; 95% CI = 0.813–0.914), and social independence (OR = 0.756; 95% CI = 0.706–0.805) associated with lower odds of a birth interval shorter than 24 months. We did not observe significant associations between empowerment dimensions and birth parity of four or higher when this behavior occurred in isolation. These findings suggest that empowerment may be more strongly associated with earlier reproductive outcomes, particularly childbirth before age 18 and short birth intervals. Table 4Multilevel logistic regression of empowerment dimensions on single high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430)Childbirth < 18 yearsBirth interval < 24 monthsBirth parity ≥ 4OR95% CIOR95% CIOR95% CIDecision making0.841***0.799–0.8850.842***0.767–0.9130.9530.801–1.105Attitudes toward violence0.865***0.832–0.8990.865***0.813–0.9140.9890.968–1.056Social independence0.753***0.71–0.7980.756***0.706–0.8050.9970.960–1.048All models control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. All estimates are adjusted for sample weights. * p < 0.05; ** p < 0.01; *** p < 0.001. Table 4Multilevel logistic regression of empowerment dimensions on single high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430)Childbirth < 18 yearsBirth interval < 24 monthsBirth parity ≥ 4OR95% CIOR95% CIOR95% CIDecision making0.841***0.799–0.8850.842***0.767–0.9130.9530.801–1.105Attitudes toward violence0.865***0.832–0.8990.865***0.813–0.9140.9890.968–1.056Social independence0.753***0.71–0.7980.756***0.706–0.8050.9970.960–1.048All models control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. All estimates are adjusted for sample weights. * p < 0.05; ** p < 0.01; *** p < 0.001. Multilevel logistic regression of empowerment dimensions on single high-risk fertility behaviors, 29 SSA countries, 2011–22 (n = 179,430) All models control for survey period, age, sex of household head, household size, place of residence, wealth index, history of terminated pregnancy, and religion. Models include a country-level random intercept. All estimates are adjusted for sample weights. * p < 0.05; ** p < 0.01; *** p < 0.001. Discussion High-risk fertility behavior poses a longstanding public health challenge in SSA [1, 3]. Despite progress in reducing maternal deaths, challenges remain in improving maternal and child health outcomes across the SSA regions. Progress in women’s empowerment is closely linked to better health outcomes, recognized not only as an end for improving women’s lives worldwide but also as a means to achieve a more gender-equitable society [16]. Yet, the relationship between women’s empowerment and high-risk fertility behavior remains relatively understudied. To address this gap, our study employed an existing measure of women’s empowerment (SWPER) that captures the holistic nature of empowerment in the attitudinal, social, and decision-making dimensions to study its complex relationships with high-risk fertility behavior outcomes. Our findings highlight the importance of women’s empowerment in reducing high-risk fertility behaviors across SSA. Aligned with previous research linking empowerment to positive family planning outcomes [18, 19], our models identified significant negative associations between empowerment progress and high-risk fertility behaviors. Specifically, women with higher scores in decision-making power and negative attitudes toward violence against women were less likely to experience early childbirth, shorter intervals between births, and a higher number of children overall. These findings corroborate previous research demonstrating that having more influence in household decision-making tends to lengthen the time between births, thereby reducing the risks associated with fertility-related health issues [18, 41]. Moreover, regional variations in empowerment levels suggest unique sociocultural, economic, and political factors that may shape empowerment within each context. With most of the existing literature focusing on regional-level analysis, our study highlights the need for more research into these sub-populations to provide tailored interventions for specific social groups, communities, regions, or countries. Our findings extend previous studies by showing that empowerment may be especially important in earlier reproductive stages. When considering only isolated high-risk behaviors—that is, when women reported just one behavior—the association between empowerment and high birth parity diminished. All empowerment dimensions proved critical in predicting women’s experience of multiple high-risk fertility outcomes. Together, these findings highlight that empowerment may be closely related to, or is reflected by, how young girls and women negotiate their fertility decisions with their spouses and families at earlier ages. The results also show how empowerment may be essential in preventing multiple potentially cumulative health risks. Such findings align with emerging policy frameworks emphasizing the importance of empowering young girls and adolescents regarding health and well-being in many low-resource settings [42, 43]. Crucially, we consistently observed that social independence was most strongly associated with reduced probabilities of early childbirth, shorter birth intervals, and higher birth parities. This dimension comprises multiple indicators, such as education level, access to information, and age at first cohabitation. These findings suggest that developing human capital, coupled with women’s early romantic relationships, may play a key role in mitigating fertility-related health risks in earlier life stages. Focusing on school enrollment and expanding access to reproductive health knowledge may provide a viable strategy for preventing and managing high-risk fertility behaviors among young girls and women. While education is widely recognized as a deterrent to high-risk fertility behavior, the consistent correlation of information access with these behaviors remains less evident [44]. This integrated approach to understanding women’s empowerment progress thus provides a more complete representation of social independence and its association with high-risk fertility behavior. From a policy perspective, these findings raise broader questions regarding how progress in women’s empowerment should be monitored. Existing policy frameworks often emphasize singular indicators of empowerment, particularly women’s education and economic participation [1–3]. However, our findings suggest that dimensions receiving less emphasis, such as social independence and decision-making, are also consistently associated with high-risk fertility behaviors. Incorporating multiple dimensions into policy discussions may therefore provide a more complete representation of women’s lived experiences and better track women’s empowerment progress. These findings also raise important questions for future work regarding whether progress across empowerment dimensions occurs concurrently or whether gains in some domains may lag behind others; more fundamentally, our findings echo recent discussions that, in consideration of such multidimensionality, revisit how best to conceptualize what constitutes an empowered woman (Tan J, Yu ST: Heterogeneity in women’s public and private agency across sub-Saharan Africa, unpublished). Our study has some limitations. First, its focus on currently married women excludes insights into fertility behavior among unmarried women. Although fertility primarily occurs within marriage in many settings, this restriction may limit the generalizability of our findings, and future studies including unmarried women could shed light on how empowerment and relationship status jointly impact fertility behaviors over the life course. Second, using cross-sectional DHS data limits our ability to establish causal relationships between empowerment and high-risk fertility behavior; our findings should therefore be interpreted as associations rather than causal effects. Subsequent longitudinal studies are needed to identify causal mechanisms and mediating factors. In addition, while age was included as part of our analytical strategy, high parity reflects cumulative reproductive exposure and may therefore vary across women at different life stages. Future work with longitudinal data may better account for differences in reproductive duration and fertility trajectories over time. Third, while our research considers dimensions of empowerment that partly reflect women’s relationships and interactions within partnered households, it may not capture broader social and community influences on women’s empowerment and fertility behaviors. Future research could incorporate different levels of analysis to explore how women’s social environment affects the association between empowerment and high-risk fertility behaviors. Conclusion In conclusion, our study contributes to the literature in three key ways. First, we address the underdevelopment of the study of empowerment and high-risk fertility behavior by highlighting the role of empowerment in reducing risk behaviors, which could help improve maternal and child health outcomes. Second, our use of a validated empowerment measure in SSA enables a holistic understanding that extends beyond single determinants and allows for comparability across studies. Finally, our findings highlight the importance of empowering women across various dimensions and potentially at different life stages. These findings provide insights that could inform potential policy interventions targeting gender-based factors that contribute to high-risk fertility behaviors to improve maternal and child health outcomes in SSA. Supplementary information Supplementary Material 1. Supplementary Material 1.