Population‐Level Impact of Risk Factors on Stunting Among Children Under Five in Sub‐Saharan Africa (2015–2022)

ABSTRACT

Aim

To identify the correlates of stunting and investigate their population‐level impacts among children under 5 years in sub‐Saharan Africa.

Methods

Data from 179 572 children under five were included from nationally representative surveys conducted across 24 sub‐Saharan African countries (2015–2022). Multivariable logistic regression models and population‐level impacts of risk factors were estimated to highlight their contributions to stunting prevalence across the study regions.

Results

Stunting prevalence ranged from 18% to 54%, with rates exceeding 50% in Burundi. Lack of basic household amenities was consistently associated with stunting, accounting for 24%–57% of cases across regions. Limited access to mass media further contributed to stunting rates, highlighting the role of education and awareness in preventing malnutrition. Maternal characteristics, such as lack of education and health insurance, were also significant risk factors, with adjusted odds ratios (aORs) ranging from 1.44 to 2.24.

Conclusion

Socio‐economic disparities are key risk factors for stunting in sub‐Saharan Africa. These factors were statistically associated with a substantial proportion of stunting cases in high‐prevalence regions. Targeted interventions to reduce inequalities, improve maternal education and expand essential services access are vital to achieving global nutrition targets, including the Sustainable Development Goal of reducing child stunting by 40% by 2025.

Abbreviations

aOR

adjusted odds ratio

CI

confidence interval

DHS

Demographic and Health Surveys

LEAP

livelihood empowerment against poverty

mHealth

mobile health

OR

odds ratio

PAR%

population attributable risk percentage

PSU

primary sampling unit

SCTP

Social Cash Transfer Programme

SD

standard deviation

SDG 2

Sustainable Development Goal 2 (end hunger, achieve food security and improved nutrition and promote sustainable agriculture)

SES

socio‐economic status

SUN

scaling up nutrition

UNICEF

United Nations Children's Fund

VIF

variance inflation factor

WHO

World Health Organization

Summary.

• The burden of stunting among children under five is not well quantified across sub‐Saharan Africa.

• Data from 24 countries showed that stunting was strongly associated with low maternal education, poor household conditions and limited healthcare access, with notable regional variation.

• Addressing structural inequalities through maternal‐focused nutrition education, cash transfer programmes and improved service delivery could substantially reduce stunting and inform the design of future public health and nutrition strategies.

1. Introduction

Despite global improvements, children under age five in low‐ and middle‐income countries still have the highest burden of growth failures [1]. The latest WHO estimates for these regions highlight significant challenges in curbing child malnutrition and its enduring adverse health effects [2, 3, 4]. Low socio‐economic conditions, such as extreme poverty, are a primary indicator of inadequate healthcare and malnutrition [5, 6]. Stunting, defined by the WHO Child Growth Standards (2007) as a reduced height‐for‐age that falls below the norm for a child's age, represents a severe and often irreversible form of growth failure [6]. It not only impairs cognitive development but also increases the risk of cardiometabolic diseases [7, 8, 9].

In sub‐Saharan Africa, the rates of growth failures, such as stunting in children under five, are among the highest globally [10, 11, 12]. In response to these alarming rates, UNICEF has initiated the National Nutrition Policy [13, 14, 15]. This policy, which is part of a coordinated effort, aligns with the Global Scaling Up Nutrition Movement [16, 17]. The Strategic Action Plan developed under this policy was implemented from 2018 to 2022. Additionally, numerous non‐governmental organisations have collaborated with local communities and implemented both nutrition‐specific and nutrition‐sensitive interventions, aiming to comprehensively address the issue of child malnutrition [18, 19].

Understanding the prevalence and correlates of stunting is essential for assessing the prevention and care models. As comorbidities associated with growth failures continue to impact many regions within sub‐Saharan Africa, there is an urgent need to understand their most influential contributors. Our study investigated the interplay of factors associated with stunting using nationally representative surveys conducted across 24 sub‐Saharan countries between 2015 and 2022. In addition to the standard statistical methodologies, we also estimated the proportion of stunting in children attributable to several risk factors in a multi‐factorial setting, after accounting for their correlation structure [20]. Specifically, we aimed to estimate the extent of the relative contributions of established risk factors, such as a range of household characteristics that are indicators of low socio‐economic conditions, and maternal characteristics, on stunting rates.

Our study objectives align with the targets of the Sustainable Development Goal 2, which emphasises the need to monitor and address the prevalence of stunting, wasting and various forms of malnutrition in children under the age of five. This is the largest and most comprehensive study to investigate the individual‐ and population‐level impacts of the significant risk factors of stunting across sub‐Saharan countries.

2. Methods

2.1. Study Population

Our study analysed data from 179 572 women aged 15–49 years, who consented to enrol in the Demographic and Health Surveys conducted in four WHO regions (2015–2022): West Africa (Benin, Burkina Faso, Côte d'Ivoire, Gambia, Ghana, Liberia, Mali, Mauritania, Nigeria, Senegal and Sierra Leone) (93 304, 51%), East Africa (Burundi, Kenya, Madagascar, Malawi, Rwanda and Tanzania) (55 652, 31%), Central Africa (Angola, Cameroon and Gabon) (15 800, 9%) and Southern Africa (South Africa, Zambia and Zimbabwe) (14 816, 8%). All surveys used a stratified two‐stage cluster sampling design (https://dhsprogram.com). Clusters were randomly selected from enumeration areas delineated in the latest census, ensuring a spread across different geographical and demographic settings, including variations in province and urban or rural classifications. The survey data were weighted for the different selection probabilities, ensuring the representativeness of the national population in our analysis.

2.2. Measurements

Standardised questionnaires were used to collect a comprehensive range of characteristics at both household and individual levels (https://dhsprogram.com). In selected households, interviews were conducted with study participants aged 15–49 years. We considered a variety of maternal characteristics as potential correlates of stunting, including maternal age, marital status, education level (any schooling)/literacy, smoking status (current smoker vs. not) and health insurance status. We also considered residing in urban/rural areas, having a job/regular income and socio‐economic status index. Household characteristics and availability of basic amenities: access to water, toilets, electricity as well as availability of various mass media outlets, including radio/television, were also considered. Maternal care was measured using information on whether a woman received malaria prevention tablets or iron supplements, the number of antenatal visits and visits to healthcare facilities. Birthplace, the presence of a doctor, nurse and midwife were all considered as potential correlates of stunting.

2.2.1. Prevalence of Low Birth Weight, Stunting, Wasting and Underweight

Babies with a recorded birth weight < 2500 g are categorised as low birth weight [6]. In our study, we also assessed growth failures using three child anthropometric indicators: stunting, measured by height‐for‐age z‐scores; underweight, determined by weight‐for‐age z‐scores; and wasting, assessed through weight‐for‐height z‐scores. Children are classified as ‘normal’, ‘stunted’, ‘underweight’ or ‘wasted’ if their respective z‐scores fall below −2 standard deviations (SD) from the median. In our analysis, child death was defined as the death of any child reported in the past 5 years within the first month of life, whereas infant mortality was defined as the death of a baby before their first birthday.

2.3. Statistical Methods

All analyses incorporated sampling weights and accounted for the complex survey design, including clustering (primary sampling units) and stratification, using survey estimation procedures. The analysis was conducted using various visualisation techniques: (1) using the ‘heat maps’, we calculated country specific stunting rates, then classified countries according to the tertiles cut‐points: ‘lowest prevalence’ (< 20%) (first tertile), ‘moderate prevalence (20%–29%)’ (second tertile) and ‘highest prevalence (30%+)’ (top tertile) stunting rates countries. In a stratified analysis, we identified the correlates of stunting for the ‘lowest’, ‘moderate’ and ‘highest’ stunting group countries separately. We described the study population using percentages across the three categories of the stunting. After identifying the risk factors associated with the stunting rates, we calculated subjects specific total risk factors for stunting. We assigned a point ‘1’ if a woman has the risk factor, ‘0’ otherwise. We added these scores to create the total number subject‐specific score and categorised them using the cut‐points of women with < 4 points (11%), 4–5 points (14%), 6–7 points (19%), 8–9 points (27%) and 10+ points (28%).

Only complete cases were used in the analysis, and no imputation was applied. The primary outcome variable, stunting, was over 99% complete, and the minimal level of missingness across most variables allowed for robust estimates from complete‐case analysis. Although forward stepwise selection was employed in building the multivariable models, the initial selection of candidate variables was guided primarily by their established importance in the literature and relevance to child nutrition and stunting, rather than exclusively relying on statistical criteria. These models accounted for the complexities of the survey design. Variance inflation factors (VIFs) were calculated to assess multicollinearity, and all final models demonstrated acceptable levels, with VIFs below the commonly accepted threshold of 5 and mean VIFs ranging from 1.32 to 1.51 across regional models. This indicates low multicollinearity and supports the stability of the regression estimates. Additionally, the Hosmer–Lemeshow goodness‐of‐fit test was used to evaluate model calibration. Together, these diagnostics support the validity and robustness of the multivariable models used in the analysis.

Using the ‘heat maps’, we presented the risk‐by‐year specific prevalence of stunting for the five risk factors across the survey years from 2015 to 2022 (Figure 2). In these three‐dimensional visual presentations, we illustrated the prevalence of stunting among children under age five across the quintiles of the risk score categories for each survey year. In a stratified analysis by the WHO regions, logistic regression models were fitted across the increasing categories of the subject‐specific scores.

FIGURE 2.

Trends in prevalence of stunting by risk score categories.

2.3.1. Population‐Level Impacts of Risk Factors on Stunting

The population‐level impacts of the risk factors were estimated using the population attributable risk percentage (PAR%) in a multi‐factorial setting for the low socio‐economic indicators as well as maternal characteristics [20].

For a single risk factor (1/0):

$$\mathrm{PAR}\%=\frac{p\left(\text{OR}-1\right)}{p\left(\text{OR}-1\right)+1}=1-\frac{1}{{\sum }_{s=1}^2{p}_s{\text{OR}}_s}$$

s indexes the levels of a risk factor of interest, which was generalised to a multivariable model when there are several risk factors:

$$\mathrm{PAR}\%=\frac{{\sum }_{s=1}^S{p}_s\left({\text{OR}}_s-1\right)}{1+{\sum }_{s=1}^S{p}_s\left({\text{OR}}_s-1\right)+1}=1-\frac{1}{{\sum }_{s=1}^S{p}_s{\text{OR}}_s}$$

where OR _s and p _s , s = 1, …, S. The combinations of the prevalence's of several risk factors were considered, using the combination of exposures and ORs (Appendix S1).

3. Results

We analysed data from 179 572 children born over the past 5 years across four WHO regions in sub‐Saharan Africa: West Africa accounted for the majority with 93 304 children (52%), followed by East Africa with 55 652 children (31%), Central Africa with 15 800 children (9%) and Southern Africa with 14 816 children (8%) (Table 1). Significant regional differences were observed in the proportion of stunting, which was higher in East, Central and Southern Africa compared to West Africa (31%, 30% and 31% vs. 25%, respectively). However, wasting, underweight, low birth weight and child mortality were higher in West Africa than in the other regions. Over half of the populations in West and Southern Africa resided in areas with the lowest socio‐economic scores (50% and 52%, respectively), compared to 46% in Central Africa and 43% in East Africa. Additionally, West Africa recorded the highest proportion of women with no education (84%).

TABLE 1.

Characteristics of study populations by World Health Organization regions (N = 179 572).

	West Africa a	East Africa b	Central Africa c	Southern Africa d	p
	N = 93 304 (%)	N = 55 652 (%)	N = 15 800 (%)	N = 14 816 (%)
Stunting
No	75%	69%	70%	69%	< 0.001
Yes	25%	31%	30%	31%
Wasting
No	92%	95%	96%	96%	< 0.001
Yes	8%	5%	4%	4%
Underweight
No	83%	85%	86%	90%	< 0.001
Yes	17%	15%	14%	10%
Low birth weight
No	81%	85%	87%	87%	< 0.001
Yes	19%	15%	13%	13%
Child mortality (HH)
No	61%	68%	72%	67%	< 0.001
Yes	39%	32%	28%	33%
Socio‐economic conditions
Mother has no education	64%	28%	27%	9%	< 0.001
Lowest socio‐economic index	50%	43%	46%	52%	< 0.001
No medical aid/insurance	77%	68%	86%	78%	< 0.001
No access to healthcare (past 12 months)	41%	32%	50%	32%
Antenatal care
Blood pressure not measured	3%	20%	8%	4%	< 0.001
Urine specimens not assessed	10%	44%	12%	28%	< 0.001
Blood not taken	10%	14%	9%	3%	< 0.001
Iron tablet not taken	14%	21%	23%	8%	< 0.001
Malaria drug not taken	22%	41%	32%	5%	< 0.001
Birth
Birth home	30%	23%	39%	14%	< 0.001
No doctor present	91%	84%	81%	92%	< 0.001
No nurse present	82%	61%	76%	90%	< 0.001
No midwife present	34%	56%	49%	15%	< 0.001
Household amenities
No piped water	92%	96%	94%	86%	< 0.001
No flushed toilet	85%	95%	86%	78%	< 0.001
No electric	55%	72%	48%	60%	< 0.001
No fridge	82%	91%	73%	68%	< 0.001
No radio	43%	55%	56%	52%	< 0.001
No TV	60%	77%	46%	57%	< 0.001
No mobile phone	40%	51%	42%	37%	< 0.001
No internet access	84%	89%	82%	84%	< 0.001

^a Benin, Burkina Faso, Côte d'Ivoire, Gambia, Ghana, Liberia, Mali, Mauritania, Nigeria, Senegal, Sierra Leone.

^b Burundi, Kenya, Madagascar, Malawi, Mozambique, Rwanda, Tanzania.

^c Angola, Cameroon, Gabon.

^d South Africa, Zambia, Zimbabwe.

A significant majority of the study population lacked medical insurance, with rates varying from 68% to 86% (p < 0.001). Between 81% and 91% of births were reported to have occurred without the presence of a doctor or nurse. Despite these challenges, most women received antenatal care that included blood pressure monitoring, as well as urine and blood assessments. The majority also received iron supplements and malaria prophylaxis. Home births were significantly more common in Central Africa (39%) and West Africa (30%) compared to 23% in Southern Africa and 14% in East Africa. The majority of the study population lacked access to basic household amenities. Lack of piped water was low across the regions (86%–96%), as was no access to flush toilets (78%–95%), not having refrigerators and electricity (68%–91% and 55%–72%, respectively). Internet access was limited, with over 80% of participants across all regions not having any; lack of access to radio ownership varied from 43% in West Africa to 56% in East Africa; lack of television ownership ranged from 46% in Central Africa to 77% in East Africa; and 40% of West Africans to 51% of East Africans reported not owning a mobile phone (p < 0.001 for all comparisons).

3.1. Correlates of Stunting by Countries With Lowest, Moderate and Highest Rates

After we calculated the prevalence of stunting for each country separately, we categorised them as having the lowest (< 20%), moderate (20%–29%) and highest (30%+) stunting rates (Figure 1). Maternal characteristics, including the lack of education (aORs ranging from 1.44 to 2.24, p < 0.001) and absence of medical aid/insurance (aORs of 1.68–2.04 in regions with moderate and high stunting rates, p < 0.001) (Table 2), were significantly associated with increased odds of stunting across all regions. The population‐level impacts of these factors ranged from 14% to 39%. Additional low socio‐economic indicators, including the absence of basic household amenities and residing in areas with lower socio‐economic conditions, were all determined as significant correlates of stunting across all regions. At the population level, 24%–57% of stunting cases were statistically associated with children living in households without access to piped water.

FIGURE 1.

Prevalence of stunting by risk categories: lowest, moderate and highest risk.

TABLE 2.

Correlates of stunting stratified by countries according to their stunting prevalences: lowest, moderate and highest stunting rate regions.

	Lowest stunting regions a (< 20%)			Moderate stunting rate regions b (20%–29%)			Highest stunting rate regions c (30%+)
	aOR (95% CI)	p	PAR% (95% CI)	aOR (95% CI)	p	PAR% (95% CI)	aOR (95% CI)	p	PAR% (95% CI)
Education
None	1.44 (1.37, 1.50)	< 0.001	14 (12, 15)	1.63 (1.46, 1.81)	< 0.001	26 (23, 29)	2.24 (2.14, 2.36)	< 0.001	33 (29, 35)
Medical aid
No	1.01 (0.95, 1.07)	0.713	—	1.68 (1.46. 1.92)	< 0.001	38 (34, 42)	2.04 (1.64, 2.55)	< 0.001	39 (33, 43)
Child mortality (HH)
Yes	1.31 (1.24, 1.40)	< 0.001	6 (5, 7)	1.33 (1.21, 1.47)	< 0.001	8 (6, 10)	1.61 (1.50, 1.72)	< 0.001	14 (12, 16)
Socio‐economic status
Poorest	2.19 (2.09, 2.29)	< 0.001	32 (30, 33)	1.74 (1.68, 1.80)	< 0.001	24 (22, 26)	1.96 (1.88, 2.05)	< 0.001	22 (20, 24)
Piped water
No	1.50 (1.38, 1.62)	< 0.001	24 (22, 26)	1.73 (1.60, 1.89)	< 0.001	36 (34, 38)	3.04 (2.42, 3.83)	< 0.001	57 (52, 61)
Flushed toilet
No	1.83 (1.72, 1.94)	< 0.001	32 (30, 33)	2.44 (2.11, 2.83)	< 0.001	45 (42, 48)	2.95 (2.69, 3.24)	< 0.001	55 (52, 58)
Radio
No	1.24 (1.07, 1.45)	0.006	6 (4, 8)	1.27 (1.17, 1.37)	< 0.001	14 (12, 16)	1.58 (1.50. 1.66)	< 0.001	17 (14, 19)
TV
No	1.88 (1.80, 1.97)	< 0.001	24 (22, 26)	1.87 (1.79, 1.94)	< 0.001	31 (29, 33)	2.67 (2.48, 2.90)	< 0.001	40 (37, 43)
Electricity
No	1.81 (1.73, 1.90)	< 0.001	21 (19, 23)	1.81 (1.73, 1.88)	< 0.001	27 (25, 28)	2.08 (1.98, 2.18)	< 0.001	33 (20, 36)
Fridge no
No	1.85 (1.74, 1.96)	< 0.001	32 (30, 34)	2.06 (1.93, 2.19)	< 0.001	40 (38, 42)	2.75 (2.55, 2.95)	< 0.001	50 (47, 53)
Mobile phone
No	1.80 (1.71, 1.88)	< 0.001	14 (12, 16)	1.80 (1.73, 1.85)	< 0.001	21 (19, 23)	2.15 (2.05, 2.25)	< 0.001	32 (30, 34)
Internet access
No	1.97 (1.86, 2.07)	< 0.001	33 (30, 36)	2.79 (2.60, 3.01)	< 0.001	54 (52, 56)	3.23 (2.92, 3.56)	< 0.001	58 (56, 61)
Access to healthcare d
No	1.28 (1.21, 1.36)	< 0.001	5 (3, 7)	1.30 (1.23, 1.39)	< 0.001	5 (3, 7)	1.26 (1.18, 1.34)	< 0.001	8 (6, 10)

^a Kenya, Gambia, Ghana, Senegal.

^b Burkina Faso, Gabon, Côte d'Ivoire, Zimbabwe, Mauritania, South Africa, Mali, Cameroon.

^c Sierra Leone, Benin, Liberia, Tanzania, Madagascar, Malawi, Mozambique, Rwanda, Angola, Zambia, Burundi, Nigeria.

^d Over the past 12 months.

Furthermore, limited access to mass media tools, including radio, television, phones and the internet, was associated with increased stunting rates across all regions, with aORs ranging from 1.24 to 3.23.

The population‐level impacts of these characteristics on stunting rates were also high, especially in countries with moderate to high stunting rates. For instance, more than 50% of the stunted children in these regions were linked exclusively to limited internet access; lack of television and mobile phones also had significant impacts, associated with 21%–40% of stunting cases. Other factors, such as not receiving malaria tablets during pregnancy and home births, were significantly associated with increased odds of stunting, though their population‐level impacts were relatively minimal (PAR% ranging from 1% to 14%).

3.2. Temporal Trends in Stunting by Total Risk Score Categories

As outlined in Section 2, we developed a subject‐specific risk score for stunting by summing the risk factors listed in Table 2. The total score ranged from 0 to 15. Figure 2 illustrates the proportions of stunted children across the five risk score categories (< 4, 4–5, 6–7, 8–9 and 10+ points) across the survey years: 2015–2022. We observed a consistent linear increase in stunting rates with higher risk score categories across all survey years. For example, compared to women with less than four risk factors, those with 10 or more risk factors were more than five times more likely to have stunted children, with stunting rates increasing from 10% to 50%.

3.3. Region‐Specific Analysis of Stunting by Total Risk Score Categories

We further analysed the association of maternal risk scores, categorised into five groups (< 4, 4–5, 6–7, 8–9 and 10+ points), with stunting rates across West, East, Central and Southern African regions, as illustrated in Figure 3. A consistent linear trend was observed between increasing maternal risk score categories and higher odds ratios for stunting, with the lowest risk category (< 4 points) serving as the reference. Specifically, odds ratios for stunting increased from 1.80 to 3.90 in West Africa, 2.05–4.16 in East Africa, 1.49–2.38 in Central Africa and 2.16–3.53 in Southern Africa (ptrend < 0.001 for all regions).

FIGURE 3.

Association between stunting and risk score categories: odds ratios from logistic regression by sub‐Saharan African regions. ^§Benin, Burkina Faso, Côte d'Ivoire, Gambia, Ghana, Liberia, Mali, Mauritania, Nigeria, Senegal, Sierra Leone; ^§§Burundi, Kenya, Madagascar, Malawi, Mozambique, Rwanda, Tanzania; ^§§§Angola, Cameroon, Gabon; ^§§§§South Africa, Zambia, Zimbabwe.

4. Discussion

There was an overall decline in the prevalence of stunting across the region, from 41% in 2016 to 22% in 2022 and 20% in 2023, which is consistent with global trends. However, stunting prevalences varied significantly at the country level, with prevalences exceeding 30% in half of the sub‐Saharan countries. Specifically, Malawi, Madagascar and Burundi, three East African countries had some of the highest rates, at 35%, 39% and 55% respectively. Additionally, Nigeria and Angola had the highest rates in West and Central Africa with 36% and 39% respectively. Although the UNICEF (2019) reported a global stunting prevalence of 27% of children under five, our findings indicated much higher and more alarmingly rates which significantly exceed the Sustainable Development Goals' target for a 40% reduction in stunting by 2025, as outlined in the UNICEF's ‘Transforming our World 2015’ and ‘From‐the‐First‐Hour‐of‐Life‐2016’ reports. We identified several factors including maternal characteristics, as significant correlates of increased stunting prevalence across countries with varying stunting rates. Specifically, low socio‐economic indicators such as the lack of basic household amenities such as flush toilets, piped water, electricity and fridge were significantly associated with higher stunting rates. At a population‐level, theoretically, modifying these factors could prevent 24%–34% of the stunting cases in the countries with less than 20% stunting rates; this proportion increases to 36%–45% in countries with 20%–29% stunting rates. In countries, with the highest stunting rates (30%+), the low socio‐economic conditions were associated with approximately half of the stunted children.

Exposure to poverty, which restricts access to standard household amenities and healthcare services, is linked to child malnutrition and growth failures, including stunting in various populations. This limited access prevents the detection of malnutrition and its associated adverse health complications [21, 22]. Mothers' lack of education was identified as significant correlates of stunting, aligning with findings from other studies [23, 24]; however, its population‐level impact is unique to the current study, attributing 14%–33% of stunting cases to women with no education.

Our study found strong evidence linking limited access to social media via radio, television, and internet with child malnutrition. This may lead to poor awareness about proper nutrition and sanitation. These conditions can deprive children of essential nutrients, increasing vulnerability to disease, creating a cycle of malnutrition and illness [25].

In our study, women who reported not having medical insurance and who had no access to healthcare in the past 12 months also showed significantly higher odds of having stunted children. Approximately 80% of these women reported lacking health insurance, and over 40% had not accessed healthcare, potentially preventing them from receiving regular antenatal care and monitoring, which is crucial for a diet that supports a healthy pregnancy for themselves and their newborns. As previously reported, these findings highlight the interplay between socio‐economic status and health, demonstrating that lower income families might face significant barriers to accessing essential health services [21, 25]. Our findings also highlight the influence of gender dynamics on child nutrition. Limited maternal education, healthcare access and financial autonomy—often driven by broader gender inequalities—can restrict women's ability to make decisions that support their child's nutrition and health. Addressing these structural barriers through interventions that promote women's empowerment is essential for reducing stunting and improving child health outcomes across generations.

Our findings reveal a consistent temporal and geographical impact of risk factors on stunting rates across sub‐Saharan African countries. An increasing number of risk factors were consistently associated with higher odds of stunting, regardless of the region. For example, women with 10 or more risk factors were more than three times as likely to have stunted children compared to those with fewer than four risk factors, with adjusted odds ratios of 3.90 in West Africa, 4.16 in East Africa and 3.53 in Southern Africa. This pattern emphasises the pervasive influence of these risk factors throughout sub‐Saharan Africa, irrespective of geographic location. The prevalence of stunting remained consistently high among women with a higher number of risk factors across the survey years from 2015 to 2022, underscoring the significant and persisting impact of these risk factors on severe malnutrition regardless of the geographical region. For example, the prevalence of stunting was estimated to be 40%–50% among women who had 10+ risk factors. Consequently, this geographical and temporal analysis validates our findings, providing a robust approach to understanding the dynamics of stunting.

This is the first study to comprehensively present the population‐level impacts of these factors on child malnutrition specifically within sub‐Saharan Africa. Consequently, the results have significant relevance and importance at country level. Although these findings generally align with prior research, the specific impact of low socio‐economic indicators on stunting brings unique insights to sub‐Saharan countries. Despite global efforts to reduce poverty, our study found that rates of household hunger have remained high and stable, showing little improvement over time. According to the 2023 Food Security and Nutrition report, the cost of daily food is 110% higher than the daily minimum wage in some parts of West and Central Africa. The impact of the COVID‐19 pandemic has also accelerated food insecurity, exacerbating existing vulnerabilities and potentially increasing the prevalence of child malnutrition [26]. In sub‐Saharan Africa, household‐level food insecurity is reported as 20%, more than double the global average of 9% [14, 15]. These estimates highlight the significant poverty burden, emphasising that the region is substantially behind the Sustainable Development Goals aimed at ending hunger, achieving food security, improving nutrition and promoting sustainable agriculture by 2030.

To improve the relevance of our findings, we also highlight key implementation challenges that may explain why some nutrition interventions have not led to sustained reductions in stunting. These include poor programme coordination, limited collaboration between sectors and weak integration with local systems. To support more actionable insights, we provide examples of effective strategies. These include conditional cash transfer programmes in Malawi, community‐based nutrition education focusing on breastfeeding and complementary feeding, integrated programmes like Ethiopia's Seqota Declaration and Ghana's Livelihood Empowerment Against Poverty initiative and mobile health platforms used to deliver targeted nutrition and immunisation messages to caregivers [27, 28, 29]. These approaches underscore the importance of coordinated, context‐specific strategies in addressing the persistent burden of stunting across sub‐Saharan Africa.

4.1. Limitations

The data from the Demographic and Health Surveys used in our study have several limitations. Self‐reported data, as used in all the surveys, are potentially subject to recall or reporting biases due to social desirability, which may impact data accuracy. Additionally, although DHS surveys are designed to be nationally representative through stratified, multi‐stage cluster sampling, some groups may not be well represented in the sample due to various reasons, including conflict or household absence during survey visits. Although we used PAR% estimates to assess the population‐level impact of various risk factors, it is important to note that PAR% assumes a causal relationship between exposure and outcome. Given the cross‐sectional nature of the data, causality cannot be established, and PAR% values should be interpreted as indicating statistical associations rather than definitive causal effects. Therefore, caution is warranted when interpreting our findings, and conclusions should be understood as indicative of association rather than causation.

5. Conclusion

Our findings highlight the profound impact of low socio‐economic conditions on stunting, emphasising the need for targeted interventions that address these underlying issues. The public health implications of this study provide significant insights for evaluating and modifying the prevention strategies implemented in sub‐Saharan African countries. Effective policies and programmes must prioritise improving socio‐economic conditions to reduce stunting and achieve global health and nutrition goals.

Author Contributions

All authors contributed to the study conception and design. Data extraction, preparation and analysis were performed by H.W.; H.W. extracted the database; H.W., J.M. and S.N. prepared the first draft. V.G. and S.N. assisted with the second draft. All authors assisted in interpreting the results and guided statistical analysis. All authors interpreted the results and contributed to finalising the manuscript. All authors read and approved the final manuscript.

Disclosure

Computer Software Availability: Appendix S1 presented the codes generated in SAS Software to estimate PAR%.

Ethics Statement

This study utilised secondary data from the Demographic Health Survey (DHS) Program. Given that there was no direct or indirect interaction with study participants, ethical approval was not required. Information regarding the ethical considerations adhered to by the DHS Program is available at the DHS Program website on protecting the privacy of survey respondents (https://dhsprogram.com/methodology).

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Appendix S1

Data Availability Statement

Data are available in a public, open access repository. The dataset is available at https://dhsprogram.com/data/available‐datasets.