Determinants of Minimum Acceptable Diet Among Children Aged 6–23 Months in Somaliland: A Cross‐Sectional Study Using Multivariable Logistic Regression

ABSTRACT

Background and Aims

Appropriate infant and young child feeding (IYCF) practices are critical for reducing undernutrition. In Somaliland, child stunting (21%) and wasting (11%) remain significant public health challenges (Somaliland Health and Demographic Survey [SHDS] 2020). This study evaluates the determinants of the minimum acceptable diet (MAD) among children aged 6–23 months to provide evidence for targeted nutritional interventions.

Methods

Data from the SHDS 2020 were analyzed. The analytical sample comprised 3336 children aged 6–23 months with complete dietary recall data. MAD was calculated based on the updated WHO/UNICEF 2021 guidelines, requiring both minimum dietary diversity (MDD: ≥ 5 of 8 food groups) and minimum meal frequency (MMF). Covariates including child age, maternal age, education, and household wealth were selected based on theoretical relevance and prior literature. Multivariable logistic regression models, adjusted for the complex survey design using Stata 17, identified significant predictors. The analytic sample size used in the final regression was 3336 mother–child pairs.

Results

IYCF practices remain suboptimal. Among children aged 6–23 months, only 6.3% (95% CI: 4.8–7.8) met the MAD criteria. The low prevalence of MAD appears primarily related to low dietary diversity (15.0%) rather than meal frequency (54.1%). In the multivariable model, child age, maternal education, and household wealth were significant predictors. Children in the richest quintile had 6.03 times higher odds of meeting MAD (AOR = 6.03, 95% CI: 2.82–12.90, p < 0.001) compared with those in the poorest quintile.

Conclusion

MAD is strongly associated with socioeconomic status and maternal literacy. Public health strategies should guide policy toward the future evaluation of nutrition‐sensitive agriculture and social protection programs to increase access to micronutrient‐dense food groups.

1. Introduction

The first 1000 days of life, from conception to 2 years, represent a critical developmental window. Optimal infant and young child feeding (IYCF) practices—specifically the introduction of a minimum acceptable diet (MAD) from 6 to 23 months—are globally recognized as the most effective strategies to combat childhood stunting, wasting, and mortality [1, 2]. According to the Somaliland Health and Demographic Survey (SHDS) 2020, child stunting (21%) and wasting (11%) remain high in the region [3]. Despite global progress, the Somaliland region continues to experience profound nutritional insecurity, often exacerbated by climate‐related disasters and socioeconomic fragility [4, 5].

Current data suggest that while caregivers in the region often prioritize minimum meal frequency (MMF), there is a common difficulty in achieving systemic failure to achieve minimum dietary diversity (MDD), leading to widespread micronutrient deficiencies [6, 7]. Existing literature indicates that socioeconomic status, maternal education, and urbanicity are the primary architects of these disparities [8, 9]. This study evaluates the determinants of MAD among children aged 6–23 months using a nationally representative dataset to provide evidence‐based insights for pivoting policy from simple caloric intake toward nutrient‐dense dietary diversity [10, 11].

While previous studies have noted high rates of malnutrition in the Horn of Africa, the specific drivers of complementary feeding in the post‐2020 Somaliland context remain poorly defined. Specifically, there is a lack of evidence regarding how maternal “cognitive agency” and market access influence dietary diversity versus meal frequency.

This study addresses the knowledge gap by evaluating the determinants of MAD among children aged 6–23 months using a nationally representative dataset. We aim to provide evidence‐based insights to shift policy from simple caloric intake toward nutrient‐dense dietary diversity [9].

2. Methodology

2.1. Study Design and Data Source

This cross‐sectional study utilized the SHDS 2020. The survey employed a multi‐stage stratified cluster sampling design to capture data across 352 clusters.

2.2. Sample Size and Inclusion

The sample was restricted to children aged 6–23 months (n = 3336). To minimize recall bias and maintain independence of observations, only the most recent live‐born child was selected for each mother. This was achieved by selecting records where the birth index variable (“bidx”) was equal to 1. Child age was calculated in months based on the difference between the date of the interview and the child's recorded date of birth. Children with missing or incomplete dietary recall data were excluded from the regression analysis.

2.3. Indicator Construction

Following WHO/UNICEF 2020 guidelines [11, 12]:

• MDD: Achieved if the child consumed ≥ 5 of 8 food groups: (1) breastmilk; (2) grains, roots, and tubers; (3) legumes and nuts; (4) dairy products; (5) flesh foods (meat/fish/poultry); (6) eggs; (7) vitamin A‐rich fruits/vegetables; and (8) other fruits/vegetables.

• MMF: For breastfed infants 6–8 months, ≥ 2 feedings; for breastfed children 9–23 months, ≥ 3 feedings; and for non‐breastfed children 6–23 months, ≥ 4 feedings.

• MAD: For breastfed children, MAD is met if both MDD and MMF are achieved. For non‐breastfed children, MAD requires MDD, MMF, and at least two milk feedings.

2.4. Data Analysis

Analyses were performed in Stata 17. We applied survey weights using the “svyset” command. Multivariable logistic regression was used to calculate adjusted odds ratios (AORs). Covariates (child age, maternal age, maternal education, residence, wealth, sex of child, and antenatal care [ANC] visits) were selected based on prior literature and theoretical relevance. The design‐adjusted Wald test was used to assess the overall association between the included covariates and MAD. p‐Values are reported as p < 0.001, p < 0.01, or exact values for p $\ge$ 0.001.

2.5. Ethics Approval and Consent to Participate

This study is a secondary analysis of publicly available, anonymized data from the 2020 SHDS. The original SHDS survey protocol was approved by the Ministry of Planning and National Development of the Somaliland Government and the ICF Institutional Review Board in the United States. Informed consent was obtained from all participants (or their legal guardians for children) before the original data collection. Accordingly, no further ethical approval was required for this secondary analysis.

3. Results

The study population for the MAD analysis comprised 3336 mother–child pairs. The cohort was predominantly urban‐based (62.0%), and 81.0% of mothers reported having no formal education. Only 30.3% of mothers attended the recommended four or more ANC visits (Table 1).

Table 1.

Descriptive characteristics of the analytical sample (aged 6–23 months).

Characteristic	Category	Frequency, n (%)
Child age group	6–11 months	1104 (33.1)
	12–17 months	1118 (33.5)
	18–23 months	1114 (33.4)
Household wealth quintile	Poorest	973 (23.3)
	Poorer	860 (20.6)
	Middle	748 (17.9)
	Richer	882 (21.1)
	Richest	715 (17.1)
Maternal education level	No formal education	3386 (81.0)
	Primary education	596 (14.3)
	Secondary education or higher	196 (4.7)
Maternal age group (years)	15–24	1212 (29.0)
	25–34	1880 (45.0)
	35–49	1086 (26.0)
Antenatal care (ANC) visits	< 4 visits	2911 (69.7)
	≥ 4 visits	1267 (30.3)
Place of residence	Rural	1088 (26.0)
	Urban	2591 (62.0)
	Nomadic	499 (12.0)
Sex of child	Male	2242 (53.7)
	Female	1936 (46.3)
Current breastfeeding status	Currently breastfeeding	2845 (68.1)
	Not currently breastfeeding	1333 (31.9)

The multivariable logistic regression analysis for children aged 6–23 months showed that both household wealth and maternal education were significant predictors of achieving a MAD. The model, adjusted for all variables listed in Table 2, showed a dose–response relationship for wealth; children in the richest quintile had 6.03 times higher odds (AOR: 6.03; 95% CI: 2.82–12.90; p < 0.001) of meeting MAD criteria compared to those in the poorest quintile. Maternal literacy also showed a strong association, with mothers having secondary or higher education being nearly three times more likely (AOR: 2.94; 95% CI: 1.14–7.54; p = 0.025) to provide an adequate diet than those with no formal education. Table 2 presents the full model results for every adjusted covariate.

Table 2.

Determinants of minimum acceptable diet (MAD) among children 6–23 months. The multivariable model showed a significant overall association (adjusted Wald F(11, 205) = 3.94, p < 0.001).

Determinant	Category	Adjusted odds ratio (AOR)	95% confidence interval	p‐value
Child age	6–11 months (ref)	1.00	—	—
	12–17 months	1.42	0.85–2.38	0.181
	18–23 months	2.15	1.12–4.11	0.021*
Household wealth	Poorest (ref)	1.00	—	—
	Poorer	0.078	0.011–0.573	0.012*
	Middle	0.045	0.005–0.402	0.006**
	Rich	0.253	0.049–1.318	0.102
	Richest	6.03	2.82–12.90	< 0.001***
Maternal education	No education (ref)	1.00	—	—
	Primary	0.310	0.054–1.777	0.188
	Secondary	2.430	0.737–8.019	0.144
	Higher	3.127	0.539–18.161	0.203
Maternal age	35–49 years (ref)	1.00	—	—
	15–24 years	0.821	0.239–2.827	0.754
Antenatal care	< 4 visits (ref)	1.00	—	—
	≥ 4 visits	5.787	1.649–20.306	0.006**
Residence	Rural (ref)	1.00	—	—
	Urban	0.903	0.180–4.539	0.901
	Nomadic	0.205	0.021–2.035	0.175
Sex of child	Male (ref)	1.00	—	—
	Female	1.965	0.609–6.344	0.257

^* p < 0.05

^** p < 0.01

^*** p < 0.001.

Table 3 summarizes the performance of core IYCF indicators. While the majority of children (54.1%) met the MMF, the estimated prevalence of MDD was critically low at 15.0%. Consequently, only 6.3% of children aged 6–23 months achieved the MAD. An assessment of the eight WHO/UNICEF food groups reveals that feeding patterns are heavily skewed toward grains, roots, and tubers (72.4%), while micronutrient‐dense protective groups—specifically vitamin A‐rich fruits/vegetables (11.4%) and eggs (6.2%)—are infrequently consumed.

Table 3.

Dietary quality indicators for children 6–23 months.

Indicator/Food group	Prevalence (%)	95% confidence interval
Infant and young child feeding (IYCF) indicators
Minimum meal frequency (MMF)	54.1	51.7–56.4
Minimum dietary diversity (MDD)	15.0	12.8–17.2
Minimum acceptable diet (MAD)	6.3	4.8–7.8
WHO eight food groups consumed
Breastmilk	78.2	75.1–81.0
Grains, roots, and tubers	72.4	69.1–75.7
Legumes, nuts, and seeds	18.2	15.5–20.9
Dairy products	41.5	38.2–44.8
Flesh foods (meat, poultry, fish, and organ meats)	22.8	20.1–25.5
Eggs	6.2	4.9–7.5
Vitamin A‐rich fruits and vegetables	11.4	9.2–13.6
Other fruits and vegetables	14.5	12.1–17.2

4. Discussion

The observed difference between MMF (54.1%) and MDD (15.0%) suggests that, while caregivers meet the quantitative feeding frequency, they may face barriers to meeting the qualitative requirements for micronutrient‐dense foods. The wealth‐gradient results suggest that dietary diversity could be related to household purchasing power. Furthermore, the significant association with ANC ($\ge$ 4 visits [AOR = 5.79, p = 0.006]) reinforces the potential importance of healthcare contact in improving feeding practices. However, these associations require further investigation, as the cross‐sectional design precludes causal inference. Our findings suggest that current feeding practices may be influenced by patterns that prioritize calorie‐dense staples over nutritional optimization.

The observed discrepancy between MMF and MDD suggests that caregivers are generally meeting the feeding frequency necessary to satisfy hunger but may lack the means to meet the qualitative requirements for micronutrient‐dense foods. The data reveal a stark lopsidedness in dietary composition: 72.4% of children consume staples, yet the consumption of nutrient‐dense “protective” foods, such as eggs (6.2%) and vitamin A‐rich fruits (11.4%), remains low. This suggests that current feeding practices may be constrained by the availability of various food groups, underscoring a need for targeted Behavioral Change Communication (BCC) that shifts the focus toward dietary quality [13].

The clear wealth gradient observed in the results suggests that dietary diversity in Somaliland is strongly associated with household purchasing power. In this context, nutritional quality may be limited by economic capacity: while wealthier households may have the resources to purchase a diverse basket of goods, lower income families may be more constrained. For these households, caloric density is often prioritized, which potentially leads to “hidden hunger”—a state where energy needs are met, but micronutrient deficiencies persist [14].

Furthermore, the significant correlation between maternal education and MAD reinforces the importance of the education‐nutrition relationship. Beyond basic economic capacity, maternal education may improve knowledge of nutrition and feeding decisions. Educated caregivers appear better equipped to navigate resource constraints and make more informed trade‐offs to prioritize nutrient‐dense food groups. The fact that MMF significantly outperforms MDD indicates that while caregivers have internalized the importance of frequent feeding, this knowledge has not been successfully paired with the economic security or specific culinary knowledge required to diversify the child's diet [15].

4.1. Limitations

First, the cross‐sectional design means that causality between determinants and feeding practices cannot be inferred. Second, indicators were based on 24‐hour recall, which may not reflect habitual intake. Finally, factors such as market distance and food pricing were not directly measured in the SHDS dataset.

5. Policy Implications

Given the observed association between household wealth, maternal education, and MAD, future studies could assess if the following strategies improve dietary diversity among young children in Somaliland:

• 1.Nutrition‐sensitive agriculture: Assessing the potential of home‐gardening and small‐scale poultry production to increase the availability of eggs and produce.
• 2.Literacy‐linked communication: Evaluating context‐specific BCC that simplifies nutritional concepts for caregivers with low levels of formal education.
• 3.Social protection programs: Investigating whether targeted cash transfers or food vouchers alleviate economic barriers to protein‐rich food.

6. Conclusion

This study concludes that achieving a MAD in Somaliland is strongly associated with household wealth and maternal education. The low prevalence of MAD (6.3%) appears primarily related to a deficiency in dietary diversity rather than meal frequency. Therefore, public health strategies should consider multisectoral interventions to address the identified socioeconomic barriers. Finally, because this study used a cross‐sectional design, causality between these determinants and nutritional outcomes cannot be inferred.

Author Contributions

Mohamed Said Hassan: conceptualization, investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, project administration, data curation, supervision, resources. Asma Abdirahman: writing – original draft, writing – review and editing, supervision, methodology, data curation.

Funding

The authors have nothing to report.

Disclosure

All authors have read and approved the final version of the manuscript. Mohamed Said Hassan had full access to all of the data in this study and takes complete responsibility for the integrity of the data and the accuracy of the data analysis.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Transparency Statement

Mohamed Said Hassan affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Data Availability Statement

The SHDS 2020 dataset is owned by the Somaliland Ministry of Planning and National Development. Data are available from the corresponding author upon reasonable request, subject to permission from the relevant authority.