To what extent does the adapted Women’s empowerment in nutrition index demonstrate psychometric validity and determine the nutritional empowerment among pregnant women in southern Ethiopia?

Abstract

Background

Women’s empowerment plays a crucial role in the wellbeing of families by affecting nutrition and health. Women’s Empowerment in Nutrition Index (WENI) was recently developed in rural South Asia, but requires validation in new contexts.

Objective

The study aimed to evaluate the psychometric validation of the WENI, the level of nutritional empowerment and factors associated with nutritional empowerment among pregnant women in Southern Ethiopia.

Design

A facility-based cross-sectional study design was employed.

Methods

The study was conducted among 392 pregnant women from March 10 to May 30, 2025. The structural validity of the WENI was assessed using Exploratory and Confirmatory Factor Analysis (EFA/CFA) to identify latent constructs of empowerment. Kaiser-Meyer-Olkin (≥ 0.60) and significant results in Bartlett’s Test (p < 0.05) were used for confirming sampling adequacy. Independent factors were identified using a hierarchical negative binomial multiple regression with adjusted incidence rate ratios (aIRRs) calculated, considering statistical significance at p < 0.05.

Results

Factor analysis revealed a six-factor structure with an internal consistency (Cronbach’s alpha = 0.86), and convergent and discriminant validities (AVE=0.50, CR=0.70). The model demonstrated strong fit indices (CFI= 0.94; RMSEA<0.06). A significant relationship was found between WENI scores and maternal and child nutrition outcomes (p<0.001), with 51.5% of women being nutritionally empowered. Empowerment was higher in women free from intimate partner violence (aIRR = 1.12), those receiving nutritional counseling (aIRR = 1.33). Conversely, empowerment rates were lower in women with no formal education (aIRR = 0.76) and those in the lowest wealth quintile (aIRR = 0.82).

Conclusion

This study confirmed that nutritional empowerment, measured via a 21-item tool, improves maternal health and food security. However, social norms and a “knowledge-agency gap” persist. Interventions must shift from basic education to actively promoting women’s control over resources, gender equality, and stronger community-based support systems to be truly effective.

Plain language summary

This study assessed how to measure nutritional empowerment, woman ability to make decisions that impact the health and nutrition of herself and her family’s among 392 pregnant women in Southern Ethiopia. By using a 21-item tool called the Women's Empowerment in Nutrition Index (WENI), researchers confirmed that the survey is an accurate and reliable way to measure these capabilities in the region. The findings showed that about 51.5% of the women surveyed were nutritionally empowered, a status that was directly linked to better overall health outcomes for both mothers and children. The study identified clear factors that influence whether a woman can exercise this control. Women were significantly more likely to be empowered if they received nutritional counselling or lived in households free from intimate partner violence. Conversely, empowerment was lower among women who had no formal education or lived in the lowest income households. Ultimately, the researchers concluded that while nutritional knowledge is important, it is not enough on its own. To bridge the gap between knowing about health and having the agency to act on it, future interventions must shift focus toward promoting gender equality, securing women control over household resources, and strengthening community-based support systems.

Introduction

Malnutrition affects over one billion people globally ¹ with 20-36% of pregnant women malnourished.^2,3 Undernourishment is more prevalent in rural (25%) than urban areas (15%). ⁴ This issue is linked to poor diet and nutrient deficiencies, endangering maternal ⁵ and fetal health. ⁶ Central to addressing these nutritional gaps is the concept of women’s empowerment, ⁷ as they influence household dietary choices. ⁸ However, persistent challenges include unequal access to resources and decision-making power, ⁹ many married women lacking control over their reproductive rights,^10,11 and limited economic power. ¹² Addressing these issues is vital for achieving sustainable development goal 2 (zero hunger), with gender-equitable households demonstrating better nutritional outcomes. ¹³

Women’s empowerment is a transformative process that enables individuals or groups to make life-defining choices and achieve desired outcomes.^14,15 Kabeer defines it as a journey from dependence to autonomy, facilitated by access to resources that allow women to exercise their agency. ¹⁶ The Food and Agricultural Organization of the United Nations (FAO) defines women’s empowerment as the process of enhancing women’s power and agency, enabling them to control their lives ultimately improving the well-being of their children and future generations. ⁹ True women empowerment is characterized by the exercise of agency across five power layers: internal power (self-worth and autonomy), resource control, life direction autonomy, collective action, and relationship-based accomplishments. ¹⁷ Therefore, empowerment is a process of awareness and capacity building leading to greater decision-making power and control and greater participation in transformative action. ¹⁸

To quantify the complex dynamics of women’s nutritional empowerment, this study utilized the Women’s Empowerment in Nutrition Index (WENI). WENI, developed through research in rural South Asia, defines nutritional empowerment as the ability to achieve health and well-being. ¹⁹ It follows Kabeer’s model, ¹⁶ focusing on resources, agency, knowledge, and achievements through 33 indicators reflecting women’s capabilities in food and health systems. By emphasizing nutrition-specific agency and resource control, WENI serves as a stronger predictor of maternal and family nutritional status compared to conventional empowerment measures. ¹⁹

While the WENI has been validated for women of reproductive age, there remains a critical gap in validated tools for pregnant women, who face unique physiological demands ⁶ and sociocultural factors that specifically affect their nutritional agency during gestation. ²⁰ Consequently, this study makes two primary contributions: first, it provides a foundational psychometric validation of the WENI within the Ethiopian context; and second, it examines the index’s application specifically among pregnant women—a population whose distinct household and nutritional status requires specific analytical consideration. To address these objectives, we evaluated the psychometric properties of the WENI, including construct, convergent, discriminant, and criterion validities, as well as its reliability. Furthermore, we measured nutritional empowerment levels during pregnancy and identified the key factors influencing this agency. By establishing this evaluative framework, our study provides essential evidence to guide policy interventions that integrate women’s empowerment into broader maternal health strategies.

Methods

Study design and setting

A hospital-based cross-sectional study was conducted from 10 March to 30 May 2025 at Bule Hora University Teaching Hospital (BHUTH) and Yabelo General Public Hospital, located in the West Guji and Borena Zone, respectively. BHUTH, situated in Bule Hora town - the capital of West Guji Zone, 467 km South of Addis Ababa-serves a population of 1,389,821. According to the 2023/24 Zonal Department Health Management Information System report, the Hospital has an annual record of over 3450 deliveries, and has 186 health professional employees. Yabelo General Public Hospital, located in Yabelo town, the capital city of Borena Zone, 570 km South of Addis Ababa. The hospital provides services to 926,690 people with 3306 deliveries reported in 2023/24.

Population and sample size determination

The source population comprised pregnant women attending antenatal care (ANC) follow-up at public Hospitals in West Guji and Borena Zones, Oromia region, Southern Ethiopia. The study included pregnant women who attended ANC follow up at public health facilities in Bule Hora and Yabello towns. We chose this population because pregnancy heightens nutritional vulnerability and intensifies household focus on maternal and fetal health. ²¹ By framing these physiological and social shifts as central to our analysis rather than confounding variables, we assess WENI as a reflection of both inherent empowerment and pregnancy-specific leverage.^22,23 Unmarried women were excluded from the study to control for the confounding effect of marital status on women’s empowerment, ensuring sample homogeneity as decision-making autonomy and socio-familial dependencies vary significantly by marital status, ²⁴ and allowing the study to effectively measure intra-household dynamics without distortion from marital status. To ensure sufficient statistical power and precision for the CFA of the cross-culturally adapted WENI tool, and considering the original 33 items as indicators, we calculated the required sample size following the 10:1 observations-to-indicator ratio guideline. ²⁵ This calculation initially provided a minimum of 330 sample size. Accounting 20% for non-response rate provided a final target sample size of 396, ensuring robust evaluation of the adapted tool’s construct validity. The final sample size was distributed proportionally between the two hospitals. Study participants, pregnant women attending ANC, were selected from each hospital using a simple random sampling technique based on the ANC registration book, ensuring comprehensive coverage.

Study variables

The primary outcomes of our study were:

• 1. The psychometric validation of the WENI among pregnant women, assessing construct validity, convergent, discriminant, and criterion validities, as well as its reliability

• 2. Quantification of nutritional empowerment among pregnant women: Measured by an adapted version of the WENI, incorporating 21 validated indicators scored dichotomously (1 for positive responses, 0 for negative).

• 3. Identification of factors affecting nutritional empowerment

Secondary outcomes

The secondary outcomes were aimed to assess the predictive validity of the adapted WENI scores by examining its association with key maternal nutritional outcomes.

• 1. Maternal nutritional status: Measured using Mid-Upper Arm Circumference (MUAC) taken on the non-dominant arm between the acromion and olecranon processes using a non-stretchable tape, recorded to the nearest millimetre and those pregnant women with MUAC < 23.0 cm were considered undernourished, while those with MUAC ≥ 23 cm were categorized as well-nourished. ⁵

• 2. Household food insecurity was evaluated through the Household Food Insecurity Scale (HFIS), a nine-item validated tool that assesses the past 30 days for both rural and urban households in Ethiopia. The scale provided three frequency response options, with total scores ranging from 0 to 27. Scores indicate the following: 0 means Food Security, 1–10 denotes Mild Food Insecurity, 11–16 reflects Moderate Food Insecurity, and 17–27 represents Severe Food Insecurity. ²⁶

• 3. Maternal Dietary Diversity (MDD): Evaluated through a 24-hour dietary recall, specifically measuring the consumption of 10 standardized food groups, and consumption from five or more food groups was considered to indicate adequate diet diversity. ²⁷

The independent variables were socio-demographic and economic factors (participant and spousal age, educational attainment, occupational status, household size, wealth index, and place of residence); and maternal obstetric and reproductive health factors (gravidity, parity, antenatal and postnatal care utilization, contraceptive use, and the prevalence of intimate partner violence).

Data collection tool, procedure and measurement and data quality

The sociodemographic and economic variables were gathered using a validated tool through face-to-face interviews, while the obstetric and reproductive characteristics and whether the mother received nutritional counseling during pregnancy were extracted from the ANC registration book by trained data collectors. Household wealth status was measured using a wealth index, a composite measure of long-term living standards, and categorized into three: lowest, middle, and highest wealth tertiles, based on the distribution of asset ownership and dwelling characteristics within the study population. ²⁸ Intimate Partner Violence (IPV) was measured across three distinct dimensions: physical violence (six items); psychological violence (four items); and sexual violence (three items). To ensure high-quality data, all questionnaires were thoroughly checked for completeness and logical errors. Data consistency checks were performed, and any inconsistent entries or responses were verified against the original questionnaires and corrected. The forward and backward translation of the original WEIN to Afaan Oromoo was conducted by two bilingual translators for further details on the translation process and pilot test (Supplementary file 1 Tool translation).

Statistical analysis

Data collected via the KoboToolbox digital platform were analyzed using STAT 14. IBM SPSS Statistics (Version 26), and Jamovi. Descriptive statistics characterized the participants’ socio-demographic, reproductive, and obstetric profiles. The psychometric properties of the WENI were evaluated starting with a Confirmatory Factor Analysis (CFA); however, due to poor initial fit indices, an Exploratory Factor Analysis (EFA) was subsequently conducted using Principal Axis Factoring (PAF) with an oblique promax rotation, ²⁹ which is specifically suited for categorical indicators. Since the indicator variables are binary variables, we employed a tetrachoric correlation to compute a matrix of correlations between the variables and then run a factor analysis on the matrix. Factor retention was determined via Parallel Analysis and scree plot inspection, maintaining a minimum factor loading of 0.30. ³⁰ Sampling adequacy was confirmed by a Kaiser–Meyer–Olkin (KMO) value > 0.60, ³¹ and a significant Bartlett’s Test of Sphericity (p < 0.01). ³² Model fit was assessed using established benchmarks: chi ² /df < 3.0, CFI and TLI > 0.90, RMSEA < 0.06, and SRMR< 0.08. ³³ Convergent validity was confirmed via three criteria: factor loadings > 0.50, ³⁴ an Average Variance Extracted (AVE) greater than 0.50, ³⁵ and a Composite Reliability (CR) of ≥0.70 ³⁶ ensuring the measures accurately represented their latent constructs. Discriminant validity was established via the Fornell-Larcker criterion (AVE>correlations), ³⁷ and a minimum 0.20 difference between primary and secondary loadings. ³⁴ Internal consistency reliability was assessed using Cronbach’s alpha (α), with coefficients of ≥0.70 deemed adequate. ³⁶ To assess criterion, specifically predictive validity, Receiver Operating Characteristic (ROC) curves and Area Under the Curve (AUC) were used across four outcomes: Household Food Security, MDD, MUAC, and Exclusive Breastfeeding (thresholds: 0.7–0.8 fair; 0.8–0.9 excellent). ³⁸ Additionally, Zero-Inflated Negative Binomial (ZINB) models addressed overdispersion and excess zeros when linking the WENI to MDD and food insecurity, while multivariable linear regression evaluated the association between empowerment and maternal MUAC.

Given the observed overdispersion (Mean = 12.72, Variance = 17.68), a Negative Binomial regression was utilized instead of a Poisson model to ensure accurate parameter estimates and avoid underestimating standard errors. ³⁹ The analysis followed a hierarchical, three-stage modeling strategy: Model 1 addressed distal structural factors, Model 2 focused on proximal individual determinants, and Model 3 integrated both into a final combined model. Variables were selected via bivariable screening (p < 0.25) and screened for multicollinearity (Variance Inflation Factors (VIF), < 10). The Full Model was identified as the superior fit based on AIC = 1341.914, BIC = 1433.053, and deviance value/df and Pearson goodness-of-fit tests value/df of 0.98. The Incidence Rate Ratios (IRR) and 95% Confidence Intervals (CIs) were calculated for all predictors to quantify the strength of the association, and a p-value < 0.05 to declare statistical significance. Data were screened for outliers and missing values. Sensitivity analyses were performed across prevalence thresholds (40%–80%) to optimize the empowerment classification.

Finally, a principal component analysis was used to create a wealth index, validated by Bartlett’s test of sphericity (p < 0.05) and a KMO measure of sample adequacy of 0.72. Q-Q (Quantile-Quantile) plot was used to assess the normality assumption in linear regression. The study adhered to the STROBE reporting guidelines ⁴⁰ (Supplementary file_2 STROBE Statement).

Ethical considerations

Ethical approval for this study was granted by the Bule Hora University Institutional Research Ethics Review Committee (IRERC) on January 22, 2025 (Ref. No.: BHU-IRERC/021/25). Written informed consent was obtained from all individual participants included in the study.

Results

Socio-demographic and economic characteristics

Out of the total calculated sample size of 396, there were 392 voluntary responses, achieving a 98.9% response rate. The mean (±SD) age of the participant was 24.9 (±4.9) years and the mean age of their partners was 30.0 years (±5.1). More than three-quarters of the participants resided in urban areas (76.3%). In terms of educational attainment, 37.4% of participants had completed primary education and 31.1% of their partners had completed secondary education. In terms of wealth status, one-quarter (25%) of participants were categorized as poor. Being a housewife (34.2%) and being employed (38.3%) were the most common occupations for the participants and their partners, respectively (Table 1).

Table 1.

Sociodemographic and economic characteristics of pregnant women in Southern Ethiopia, 2025.

Variables	Category	Frequency (392)	Percentage
Residence	Rural	93	23.7
	Urban	299	76.3
Participants age at study enrolment	18-19 years	103	26.3
	20-29	216	55.1
	30-34	52	13.3
	>=35	21	5.4
Maternal Education	No formal education	94	24
	Primary school	136	34.7
	Secondary school	135	34.4
	College & above	27	6.7
Partner Education	No formal education	71	18.1
	Primary school	104	26.5
	Secondary school	122	31.1
	College and above	95	24.2
Maternal occupation	Farmer/pastoralist	89	22.7
	Merchant	88	22.4
	Daily labourer	43	11
	Housewife	134	34.2
	Employed	38	9.7
Partner occupation	Employed	150	38.3
	Merchant	91	23.9
	Daily labour	48	12.2
	Farmer/pastoralist	103	26.3
Wealth Index	Lowest wealth quintile	94	24
	Middle wealth quintile	99	25.2
	Highest wealth quintile	199	50.8
Family size	1-4	272	69.4
	5-6	99	25.3
	7+	21	5.4

Obstetrics reproductive and intimate partner violence of the pregnant women

As indicated in Table 2, three-quarters of the participants, 302 (77%) were under the age of 20 years at the time of their first pregnancy. Nearly two-in-five (39%) of the participants had an interpregnancy interval of greater than 2 years. Similarly, five-in-seven (71.9%) of the participants had ever utilized family planning. Regarding IPV, 17.9%, and 23.5% of the participants reported experiencing at least one form of physical IPV and psychological IPV, respectively (Table 2).

Table 2.

Obstetrics and reproductive characteristics of the pregnant women in Southern Ethiopia, 2025.

Variables	Category	Frequency (n=392)	Percentage (%)
Recalled age at first pregnancy	15-19 years	302	77
	20-24	76	19.4
	>=25	14	3.6
Parity	Nullipara	117	29.8
	Primipara	104	26.5
	Multipara	171	43.6
Gravida	Primigravida	117	29.8
	Multigravida	273	70.2
Gestation in weeks	≤34	145	37
	>34	247	63
Ever utilized family Planning	Yes	282	71.9
	No	110	28.1
Interpregnancy Interval	24 months	107	27.3
	24–59 months	153	39
	Not applicable	132	33.7
Physical violence	No	322	82.1
	Yes	70	17.9
Sexual violence	No	369	94.1
	Yes	23	5.9
Psychological violence	No	300	76.5
	Yes	92	23.5
Over all intimate partner violence	No	281	71.5
	Yes	111	28.5
Age difference between wife and husband	Wife Older	3	0.8
	0-4 years gap	134	34.4
	5-9 years gap	200	51
	10+ years	55	14

Factor analysis (FA)

To evaluate the validity of the WENI, an initial CFA was performed to test the original theoretical structure. However, the model demonstrated a poor fit to the current data (RMSEA = 0.109, CFI = 0.745, TLI = 0.681 and SRMR = 0.112), suggesting that the existing framework was not applicable to pregnant women. Consequently, an EFA was conducted to identify an empirical latent structure tailored to the current sample. The EFA was conducted with the 34 indicators (Supplementary file 3 WEIN indicators) using Principal Axis Factoring with Promax rotation. The KMO measure (0.83) and Bartlett’s test (p < .001) supported the factorability of the data. Thus, based on parallel analysis, a six-factor solution was identified, explaining 55.6% of the total variance (Figure 1).

Figure 1.

The total variance explained for the WENI index tool among pregnant women attending Antenatal care in Public Hospital in Southern Ethiopia, 2025 (n=392).

Three items were removed for low primary factor loadings (<0.30), and one item was eliminated due to a low KMO measure (0.42). Additionally, seven items were dropped as their communality values were below 0.30, indicating inadequate variance explanation (Supplementary Table 1). Following a revised structure, a CFA revealed the initial model had poor fit due to highly correlated error terms. A modification index facilitated improvements, allowing error terms to covariance within the same factor as justified theoretically. Post-modification, model fit indices were acceptable: RMSEA = 0.05, CFI = 0.944, TLI = 0.931, SRMR = 0.05, and a Chi-square/df Ratio of 2.5, adhering to recommended limits. All items showed significant correlation (P < 0.001). A six-factor model emerged, merging “Health Knowledge” and “Institution” into Factor 3, with strong item loadings on the new factor (0.67, 0.57, and 0.69 for items 1, 3, and 5, respectively) (Supplementary Table 2).

Internal reliability analysis

The overall Cronbach’s alpha for the 21 items of the WENI index is 0.86, indicating strong internal reliability. Additionally, the Corrected Item-Total Correlation values range from 0.312 to 0.650, demonstrating that the scale items are reliable and well-constructed as an effective measurement tool (Supplementary Table 3).

Convergent and discriminant validity

Convergent validity was established with CR values between 0.72 and 0.88, and AVE values for all factors above 0.50, ranging from 0.57 to 0.82. Additionally, all standardized factor loadings were significant (p<.001) and values ranged from 0.50 to 0.96. Discriminant validity was validated via the Fornell-Larcker criterion, showing that the square root of the AVE for each latent construct (diagonal elements) was higher than its highest correlation with any other construct in the model (Table 3).

Table 3.

Item total correlations, composite reliability, the square root of the average variance extracted.

Factor	No. of items	Composite reliability	Average variance extracted	Factor correlation matrix
				1	2	3	4	5	6
1	5	.88	0.59	0.76	​	​	​	​	​
2	5	.78	0.57	.307	0.75	​	​	​	​
3	5	.76	0.65	.669	0.478	0.80	​	​	​
4	2	.81	0.82	.258	0.154	0.453	0.90	​	​
5	2	.79	0.81	.393	0.081	0.460	0.209	0.90	​
6	2	.72	0.75	.352	0.052	0.452	0.225	0.489	0.86

Criterion validity

The ROC analysis revealed that the WENI is a statistically strong predictor of maternal health and nutrition outcomes, though its discriminative power varied across four maternal health and nutrition domains. The highest performance was observed for Household Food Security (AUC = 0.796, 95% CI: 0.74-0.84) and MDD (AUC = 0.777, 95% CI: 0.73-0.82) both approaching the “excellent” discrimination threshold (Figure 2(a)–(d)).

Figure 2.

(a). ROC curve illustrating the performance of the WENI in predicting Maternal MUAC. (b). ROC curve illustrating the performance of the WENI in predicting MDD adequacy. (c). ROC curve illustrating the performance of the WENI in predicting Household Food Security. (d). ROC curve illustrating the performance of the WENI in predicting prior history of EBS.

Similarly, the ZINB analysis revealed that higher empowerment levels significantly reduce food insecurity while improving dietary diversity. Specifically, each one-unit increase in the WENI score was associated with a 6% decrease in food insecurity counts (IRR=0.94; 95%CI=0.91–0.96) and a 5% increase in maternal food group consumption (IRR = 1.05, 95% CI: 1.04–1.06). Notably, integrated agency and resource control (Factor 2) showed the strongest impact, correlating with a 26% reduction in food insecurity severity (IRR = 0.74; 95%CI=0.63–0.88) and an 89% increase in dietary diversity (IRR = 1.89, 95% CI: 1.72–2.07) (Supplementary file 5, Figures 1 and 2). Linear regression results further identified the WENI score as a positive predictor of maternal health, where a one-unit increase corresponded to a 0.17 unit rise in MUAC (B =0.17 95% CI: 0.13, 0.20; p<0.05). While tangible resource-based dimensions (Factors 1 and 2) were strongly associated with improved MUAC, food and health knowledge (Factors 3 and 4) did not reach statistical significance (Supplementary file 5, Figure 3).

Figure 3.

A sensitivity analysis determining the threshold cut-off points for women’s empowerment score.

Sensitivity analysis & threshold selection for prevalence of Women’s empowerment

We conducted a dual-stage analysis. First, we performed a sensitivity analysis by calculating the prevalence of empowerment across demographic strata at 10% increments (40%–80%). As the “empowerment” threshold rises from 40% to 80%, the rate of empowered women declines sharply, from 81% at 40% to 24% at 80%. The most significant drop occurs between the 50% and 60% thresholds (Figure 3). We identified the 60% threshold as the optimal cutoff, as it demonstrated the greatest discriminatory power in capturing the expected variance across residency and educational status, while avoiding the ceiling and floor effects observed at other levels; urban women exhibit greater resilience, with 59% remaining empowered at 60%, compared to just 27% in rural areas (Supplementary Table 4). Moreover, to validate this threshold, we performed a ROC analysis to assess the classification accuracy of our index. The index demonstrated good predictive validity for household food security with an AUC of 0.796 (Figure 2 (c)), suggesting that the 60% threshold is a reliable indicator for capturing socioeconomic and food-related agency within the household. Similarly, the index showed a good predictive performance for MDD (AUC = 0.68) (Figure 2(a)), which is likely attributable to the multifactorial nature of maternal nutritional status where biological and environmental stressors often operate independently of the agency-based indicators measured by the WENI. By utilizing the Youden Index to optimize our cutoff, we confirmed the 60% threshold as the most statistically robust point for classifying empowerment across these varying outcomes, ensuring an appropriate balance between sensitivity and specificity in our sample population.

Accordingly, the estimated overall proportion of pregnant women who were nutritionally empowered was 51.5% (95%CI: 46.5% to 56.5%). Empowerment levels demonstrated significant heterogeneity across dimensions: The vast majority of pregnant women showed empowerment in food agency (92.6%) and health agency (94.4%). In contrast, empowerment was significantly lower concerning resource and institutional access (50.8%) and health knowledge & institutional linkage (60.5%) (Figure 4).

Figure 4.

Dimension-specific empowerment levels among pregnant women in Southern Ethiopia.

Factors associated with nutritional empowerment in pregnant women

Initially, bivariable negative binomial regression analyzed 16 independent variables, retaining those with a p-value < 0.25 for multivariable hierarchical models. Thus, variables with a p-value > 0.25, such as mothers’ age, age at first pregnancy, partner’s age, gestational age at first ANC and district, were excluded from the subsequent model (Table 4).

Table 4.

Bivariable negative binomial regression analysis showing the determinants of women’s empowerment among pregnant women in Southern Ethiopia 2025.

Variables	Category	cIRR (95% Cl)
Mothers age in years	​	1.00 (0.99 - 1.01)
Age at first pregnancy in years	​	1.02 (0.96 - 1.07)
Age of partner	​	1.001 (0.92 - 1.08)
Family size	​	0.98 (0.96 - 1.007)
District	Yabello	1.00 (0.93 - 1.07)
	Bule Hora	1
Residence	Rural	0.83 (0.76 - 0.90)
	Urban	1
Maternal Education	No formal education	0.71 (0.61 - 0.83)
	Primary school	0.82 (0.71 - 0.96)
	Secondary school	0.90 (0.77 - 1.04)
	College & above	1
Partner Education	No formal education	0.81 (0.73 - 0.91)
	Primary school	0.90 (0.81 - 1.00)
	Secondary school	1.00 (0.91 - 1.10)
	College and above	1
Husband occupation	Employed	1.29 (1.18 - 1.42)
	Merchant	1.26 (1.14 - 1.40)
	Daily labourer	1.23 (1.10 - 1.39)
	Farmer/pastoralist	1
Wealth Index	Poorest	1
	Middle	1.40 (1.29 - 1.51)
	Richest	1.34 (1.23 - 1.45)
Parity	Nullipara	0.73 (0.49 - 1.10)
	Primipara	1.14 (1.01 - 1.28)
	2-3	1.15 (1.03 - 1.28)
	4 and above	1
Gravidia	Primigravida (1)	0.74 (0.49 - 1.11)
	Multigravida (2-4)	1.16 (1.05 - 1.28)
	Grand multigravida (>=5)	1​
Gestational age in week	<34	0.91 (0.85 - 0.98)
	>=34	1
Nutritional counselling during pregnancy	Yes	1.37 (1.24 - 1.52)
	No	1
Intimate partner violence	No	1.20 (1.11 - 1.31)
	Yes	1
GA at first ANC	Within 12 Weeks	1.01 (0.91 - 1.13)
	Within 16 Weeks	1.00 (0.90 - 1.12)
	Within 20 Weeks	0.97 (0.88 - 1.06)
	Above 20 Weeks	1

In the distal multivariable model (Model 1), socio-economic status was a primary determinant of women nutritional empowerment. Specifically, participants in the lowest wealth quintile exhibited a significantly lower incidence rate of women nutritional empowerment compared to the highest wealth quintile (aIRR = 0.74, 95% CI: 0.64–0.86). Other socio-economic factors, including urban residence, partner’s education, and husband’s occupation, did not show statistically significant associations with the women nutritional empowerment at the distal level.

In the proximal model (Model 2), both education and occupational status showed significant associations. Women with no formal education had a 20% lower incidence rate (aIRR: 0.80; 95% CI: 0.67–0.97) compared to those with college-level education or above. Moreover, receiving nutritional counseling during pregnancy was associated with a 30% increase in the rate of the women nutritional empowerment (aIRR: 1.30; 95% CI: 1.16–1.45) compared to those who did not receive counseling. The absence of IPV was associated with a significantly higher rate of the women nutritional empowerment. Women reporting no IPV had a 14% higher rate (aIRR: 1.14; 95% CI: 1.04–1.24) compared to those who reported experiencing IPV.

In the final combined Negative Binomial model, both distal socio-economic factors and proximal clinical factors remained significant predictors of the outcome. The poorest wealth quintile (aIRR: 0.82; 95% CI: 0.70–0.95) and “no formal education (aIRR: 0.76; 95% CI: 0.62–0.94) were associated with significantly lower incidence rates. Conversely, nutritional counseling during pregnancy was the strongest positive predictor, associated with a 33% increase in the incidence rate (aIRR = 1.33; 95% CI: 1.19–1.50). The absence of IPV was associated with an increased rate of the women nutritional empowerment. Women reporting “No” IPV had a 12% higher rate compared to those experiencing IPV (aIRR: 1.12; 95% CI: 1.03–1.22 in Model 3). Model fit statistics (Log-Likelihood: -644.957) confirm that a combined approach of distal and proximal factors provides the most comprehensive explanation of the data (Table 5).

Table 5.

Results of the block-based hierarchical negative binomial regression analysis identifying the determinants of women’s empowerment among pregnant women in Southern Ethiopia 2025.

Independent variables	Model 1 (Distal)	Model 2 (Proximal)	Model 3 (combined)
	aIRR (95% CI)	aIRR (95% CI)	aIRR (95% CI)
Socio-economic/distal
Wealth Index (Ref:)	​	​	​
Poorest	0.74 (0.64–0.86)***	—	0.82 (0.70–0.95)*
Middle	0.92 (0.84–1.01)	—	0.96 (0.88–1.05)
Residence (Ref: Rural)	​	​	​
Urban	1.13 (0.98–1.31)	—	1.06 (0.92–1.23)
Partner’s Education (Ref: College and above)	​	​	​
No formal education	0.93 (0.81–1.07)	—	1.06 (0.88–1.28)
Primary	0.96 (0.85–1.08)	—	0.97 (0.85–1.11)
Secondary/Higher	1.00 (0.90–1.10)	—	0.96 (0.86–1.07)
Husband occupation (Ref:Farmer/pastoralist)	​	​	​
Employed	1.12 (0.96–1.31)	—	0.99 (0.78–1.25)
Merchant	1.10 (0.95–1.28)	—	1.00 (0.79–1.26)
Daily labourer	1.12 (0.96–1.32)	—	1.00 (0.78–1.27)
Individual/Proximal
Woman’s Education (Ref: College & above)	​	​	​
No formal education	​	0.80 (0.67–0.97)*	0.76 (0.62–0.94)*
Primary	—	0.89 (0.76–1.05)	0.93 (0.77–1.11)
Secondary/Higher	—	0.93 (0.79–1.08)	0.95 (0.80–1.12)
Nutritional counselling during pregnancy (Ref: No)	​	​	​
Yes	—	1.30 (1.16–1.45)***	1.33 (1.19–1.50)***
Intimate partner violence (Ref: Yes)	​	​	​
No	​	1.14 (1.04–1.24)**	1.12 (1.03–1.22)**
Parity (Ref: 4 and above)	​	​	​
Nullipara	​	0.75 (0.50–1.13)	0.73 (0.48–1.10)
Primipara	​	0.91 (0.73–1.14)	1.02 (0.79–1.31)
Multipara	​	0.94 (0.76–1.17)	1.01 (0.81–1.27)
Gestational age in week (Ref: >=34 week)	​	​	​
<34	​	0.96 (0.89–1.03)	0.98 (0.91–1.06)
Model Fit Statistics	​	​	​
Log-Likelihood	-685.589	-650.159	-644.957 (Best Fit)
Deviance	1.228	0.97	0.98
AIC	1395.178	1332.319	1341.914
BIC	1437.242	1388.404	1433.053
Likelihood Ratio Test (chi^2)	50.775***	121.663***	132.039***

Notes: Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001. aIRR (Adjusted Incidence Rate Ratio).

Discussion

This study confirms that the WENI is a highly capable tool for identifying the nuances of nutritional agency among pregnant women. By verifying a six-factor structure with adequate fit indices, we provide evidence that nutritional empowerment is a complex, multifaceted phenomenon that can be reliably measured (alpha=0.86). Moreover, the ROC curve analyses confirm that empowerment is not merely a social construct but a measurable determinant of health, particularly concerning HFS (AUC = 0.796) and past EBF (AUC = 0.792). Our results reveal that while roughly half of the participants (51.5%) exhibited empowerment, significant disparities persist. The data suggest that empowerment is heavily contingent upon social determinants; specifically, educational and economic marginalization act as primary barriers, while the provision of nutritional counselling and the presence of a safe, violence-free domestic environment are critical predictors of high empowerment counts. The WENI appears to be a useful tool for identifying vulnerable groups and measuring the potential contribution of gender-transformative interventions toward improving maternal health outcomes.

The final structure of the WENI tool among pregnant women consists of 21 items across six factors. While this deviates from the original theoretical structure, ¹⁹ it appears to demonstrate improved psychometric performance within the present context-specific. This aligns with previous validation efforts particularly the Abridged WENI that a condensed set of 20 indicators can maintain high predictive utility. ⁴¹ This shift suggests that the latent constructs of empowerment are multidimensional and sensitive to context specific populations. ⁴² These factor differences can be attributed to demographic shifts; specifically, the seven-factor structure identified in the original Alkire-Foster study ⁴³ was condensed into a six-factor model through factor analysis in the present study of pregnant women. ⁴⁴ While the emergent model shows improved validity, EFA and CFA were conducted on the same sample. Therefore, this is a preliminary calibration requiring future validation with independent samples.

Our findings reveal a positive relationship between women’s empowerment and improvements in dietary diversity, food security, and maternal nutritional status, corroborated by A-WENI evidence linking empowerment to enhanced BMI and MUAC.^7,42 Empowered women can procure nutritious food, thereby alleviating food insecurity^45–47; additionally, decision-making authority over resources predicts better dietary diversity.^48,49 Data shows that control over household income greatly enhances dietary diversity and food security, ⁵⁰ while land ownership is also positively correlated with food security. ⁵¹ Time allocation is crucial, as pregnant women with adequate leisure time demonstrate higher rates of dietary diversity, MUAC, and food security compared to those with limited time. ⁵² Furthermore, social norms and integration play vital roles; women in supportive environments achieve better dietary metrics than those in restrictive ones. ⁵³ Social capital is influential, as women without group affiliations experience higher food insecurity. ⁵⁴ Food taboos affect a significant portion of women, limiting their dietary options. Overall, the results indicate that gender-based social constraints and power dynamics are key factors impacting maternal malnutrition, with women’s empowerment being a measurable determinant of health and household food security. ⁵⁵

Our findings suggest a positive association between empowerment and child health outcomes, particularly in birth spacing and infant nutrition. Women with control over their resources and agency have a 74% rate of maintaining a birth interval of 24 months or more, compared to 40% for those without such control. Additionally, 78% of empowered women use contraceptives, indicating that increased autonomy enhances contraceptive negotiation with partners. ⁵⁶ Empowerment also correlates with higher prior history of EBF rates, with empowered women achieving 88% in health knowledge and 80% in food knowledge, versus 60% and 17% for non-empowered women. The findings emphasize the need for health interventions to tackle socio-economic and relational barriers to agency, ensuring lasting improvements in maternal and child health practices. ⁵⁷

Our study reveals that 51.5% of pregnant women demonstrate nutritional empowerment, aligning with global trends, ¹² and slightly exceeding India's rate of 51.2%, ⁴¹ as well as surpassing 33.8% found in a multi-country study of low- and middle-income nations. ⁵⁸ Differences in empowerment levels may arise from the study’s focus on pregnant women rather than all reproductive-age women, indicating potential shifts in agency during pregnancy. ⁵⁹ Participants show high personal agency and food knowledge (86.2%) but face systemic barriers, evidenced by a significant gap between high food (92.6%) and health agency (94.4%) versus lower resource access (50.8%). This reflects the Social-Ecological Model, emphasizing the influence of community and policy factors on individual behaviors. ⁶⁰ Our study result reveals a discrepancy between high food knowledge (86.2%) and lower integrated agency & resource control (54.8%), indicating that knowledge alone does not empower individuals. While participants are knowledgeable about nutrition, their ability to manage necessary resources is significantly less, reflecting the need for changes in household or community power dynamics. The 51.5% empowerment estimate likely reflects selection bias, as it is derived from a population already accessing facility-based ANC. Given that non-attending women face greater socioeconomic and structural barriers, their actual empowerment levels are likely significantly lower. Therefore, maternal health interventions must pivot toward targeted outreach strategies for these marginalized, non-facility-attending populations to effectively address the most vulnerable groups.

Educational attainment is the most significant predictor of empowerment among pregnant women, with those lacking formal education scoring 24% lower on empowerment scales compared to their educated counterparts, reinforcing prior studies on the topic. ⁶¹ At the 60% threshold, 70% of college-educated women reported empowerment compared to 34% of those without formal education, indicating a significant benefit from higher education. This leads to greater autonomy in health decisions, financial independence, and civic engagement ⁶² and bargaining power within the household. ⁶³ However, a lack of collective agency was noted, with 51% of uneducated women not participating in community organizations and 56% having not engaged in civic activities. Therefore, there is a need to recruit uneducated pregnant women into Women’s Self-Help Groups (SHGs) to enhance their collective agency and health communication. ⁶⁴

Our findings reveal a significant socioeconomic gradient in women’s empowerment, with an 18% decline in nutritional empowerment scores for women in the lowest wealth quintile compared to the highest. Additionally, 12.8% of women in the lowest quintile lack control over household purchases, in contrast to 4.5% of those in the highest quintile. This supports the resource theory of power, indicating that greater economic status enhances women’s influence in domestic matters. ⁶⁵ Women in the highest quintile also have higher engagement in social visits (82%) and leisure time (77%) compared to lower quintile women (57% and 24%, respectively), suggesting that temporal scarcity limits social capital accumulation, ⁶⁶ and hampers collective agency to tackle systemic inequities, as per Kabeer’s empowerment framework. ¹⁶

Nutritional counselling during pregnancy was the strongest positive predictor of empowerment in this study. Specifically, women who received nutritional counselling had a 33% higher empowerment count compared to those who did not, aligned with existing evidence. ⁶⁷ Pregnant women receiving such counselling exhibited a 98% increase in health knowledge, compared to a 72% disempowerment rate in those without counselling. This underscores the need for targeted nutritional education by healthcare providers. ⁶⁸ Additionally, while mobile phones can enhance service delivery ⁶⁹ , yet 71% of women in our study lack access to digital information about government nutritional and financial schemes, reflecting a feminist theory perspective on digital exclusion linked to patriarchal control. ⁷⁰ Furthermore, 61.5% of women required partner permission to visit health centres, highlighting institutional barriers that restrict access to essential nutrition education and prenatal supplementation. This result indicates that engaging with healthcare services and nutritional counselling can enhance a woman’s empowerment score through greater knowledge, social support from providers, and increased confidence in health-related decision-making.

Our study indicates that pregnant women experiencing an absence of IPV show higher rates of nutritional empowerment, aligning with prior research that highlights the negative impact of violence on women’s agency.^71,72 Our findings also show that disempowerment regarding food and health, as well as restricted mobility, doubles the risk of violence, reinforcing the theory that IPV operates as a mechanism of patriarchal control. ⁷³ Specifically, 48% of women with restricted mobility report IPV, compared to 21% of those with greater mobility, highlighting the importance of freedom of movement as a protective factor. ⁷⁴ Furthermore, women facing IPV exhibit higher rates of undernutrition (27%) and severe food insecurity (24%) compared to non-IPV counterparts (15% and 7%, respectively), aligning with recent global literature. ⁷⁵ These findings underscore the need for IPV screening in nutrition programs and gender-sensitization training to improve health outcomes.

Limitations of the study

Despite providing significant insights into empowerment indices among pregnant women in Southern Ethiopia, given that our study relied on hospital-based sampling, our participants may possess higher levels of health-seeking behavior, mobility, and family support compared to those not attending facilities, potentially overrepresenting women with greater health agency and support. This approach likely overlooks the most marginalized women who face barriers to healthcare, thus framing empowerment narrowly within the context of marriage and temporary support during pregnancy. As a result, the findings may not reflect the experiences of all women. Additionally, the reliance on self-reported past history for certain variables introduces the potential for recall bias. Furthermore, the study did not assess test-retest reliability, and its conclusions could be influenced by specific cultural contexts. Finally, focusing exclusively on married women limits the generalizability of our findings. While this approach enabled a targeted analysis of intra-household dynamics, it omits unmarried pregnant women, who may face distinct vulnerabilities. Future research should consider community-based sampling to better understand empowerment pathways for marginalized populations.

Conclusion

This study validated the WENI for pregnant women in Southern Ethiopia, establishing a six-factor, 21-item structure with reliable psychometric properties (alpha = 0.86; CFI = 0.944). It found that nutritional empowerment significantly predicts maternal MUAC, dietary diversity, and food security, as well as reproductive and child health outcomes such as longer birth spacing and increased breastfeeding rates. While roughly half of the participants (51.5%) exhibited empowerment, significant disparities and issues such as a “knowledge-agency gap,” limited resource control, and restrictive social norms persist. Formal education and household wealth were identified as key determinants of autonomy. The study emphasizes the adverse effects of IPV and the need for effective interventions beyond basic nutrition education, recommending increased women’s resource control, gender-sensitization in prenatal care, and community engagement through SHGs. While our study is limited by hospital-based sampling, it provides a strong framework for addressing the complexities of empowerment to enhance maternal and child well-being in Ethiopia.

ORCID iD

Anteneh Fikrie https://orcid.org/0000-0002-9809-3326

Ethical Considerations

This study was one component of a project on ‘Empowering women for healthier futures: a multi-methodological study on women’s empowerment, decision-making and adverse pregnancy outcomes in southern Ethiopia”. It was conducted in strict adherence to the ethical principles outlined in the Declaration of Helsinki. Ethical approval for this study was granted by the Bule Hora University Institutional Research Ethics Review Committee (IRERC) on January 22, 2025 (Ref. No.: BHU-IRERC/021/25). Name of ethics committee: Bule Hora University Institutional Research Ethics Review Committee (BHU-IRERC). Approval Reference Number: BHU-IRERC/021/25. Approval Date: January 22, 2025.

Consent to Participate

Written informed consent was obtained from all individual participants included in the study. All participants took part on a strictly voluntary basis. They were fully informed about the study’s purpose, procedures, potential risks, and benefits through a detailed information sheet, and written informed consent was obtained from every individual. Participants were explicitly assured of their right to withdraw at any time without penalty. To ensure privacy and confidentiality, all collected data were immediately de-identified and coded, and personal identifiers were stored separately and securely.

Data Availability Statement

All relevant data supporting the findings of this study are available within the manuscript or its associated supplementary files. Additional data requests may be addressed to the corresponding author.*