Spatial distribution and multilevel determinants of unvaccinated children in Somaliland: analysis of the 2020 Somaliland Health and Demographic Survey (SLHDS)

Abstract

Background

Children who have never received any vaccination represent one of the clearest markers of health system inequity. In this study, unvaccinated status was defined as caregiver report that a child under five years had never received any routine vaccine, reflecting complete non-vaccination rather than dose-specific (e.g., DTP1-based) definitions. This study examined the prevalence and determinants of unvaccinated children in Somaliland using nationally representative data.

Methods

We analyzed data from 1,104 children under five years from the 2020 Somaliland Health and Demographic Survey (SLHDS). Multilevel logistic regression models were fitted to assess individual-, household-, and community-level factors associated with being unvaccinated.

Results

The prevalence of unvaccinated children was 79.9%. In the fully adjusted model, children of mothers with secondary education had significantly lower odds of being unvaccinated compared with those whose mothers had no education (AOR = 0.32; 95% CI: 0.14–0.74). Antenatal care attendance was strongly associated with reduced odds of non-vaccination (1–3 visits: AOR = 0.37; 95% CI: 0.28–0.52; ≥4 visits: AOR = 0.40; 95% CI: 0.26–0.64). Older children were less likely to be unvaccinated compared to infants aged 0–8 months (9–23 months: AOR = 0.37; 95% CI: 0.19–0.72; 24–59 months: AOR = 0.32; 95% CI: 0.19–0.58). Children not currently breastfeeding had lower odds of being unvaccinated (AOR = 0.68; 95% CI: 0.52–0.90). At the community level, residence in areas with high ANC utilization was associated with substantially lower odds of non-vaccination (AOR = 0.37; 95% CI: 0.25–0.57). Descriptive mapping revealed marked regional variation, with particularly high prevalence observed in Awdal and Sool.

Conclusion

Complete non-vaccination in Somaliland is strongly associated with maternal education, engagement with antenatal care, child age, breastfeeding status, and community-level health service utilization. Integrated maternal–child health strategies and targeted regional interventions are urgently needed to reduce the exceptionally high burden of unvaccinated children.

Introduction

Childhood immunization is a cornerstone of global public health, widely recognized as one of the most successful and cost-effective interventions for preventing mortality and morbidity from a host of infectious diseases [1–3]. Zero-dose children, as defined by UNICEF (2025), are those who have missed out on all routine immunizations, signifying profound inequities in access to essential health services [4]. Globally, approximately 6.3 million children under five die each year, with Sub-Saharan Africa disproportionately affected, accounting for about half of the 1.5 million global vaccine-preventable deaths annually [5, 6]. While global immunization efforts have impressively saved an estimated 154 million lives over the past 50 years and significantly reduced infant deaths by over 50% in the African Region, monumental challenges remain. During the pandemic years, an estimated 67 million children missed one or more essential vaccines, and coverage for critical vaccines like measles is still below outbreak prevention thresholds, highlighting ongoing vulnerabilities, especially in resource-limited settings [7].

Globally, the challenge of reaching every child with essential immunizations persists, with an estimated 14.5 million children under one year of age not receiving any basic routine vaccine doses in 2023, categorizing them as “zero-dose”. This represents a concerning increase from 13.9 million in 2022 and a significant rise of nearly 2.7 million compared to the 12.9 million zero-dose children recorded pre-pandemic in 2019 WHO [8]. Critically, over half of these children reside in countries affected by conflict, fragility, or instability, where access to fundamental health services is severely compromised, and the vast majority are concentrated in low- and middle-income countries, particularly within Africa and South-East Asia WHO [8, 9].

Africa faces a significant crisis in childhood immunization, leading the world in the number of children missing out on routine vaccinations. In 2021 alone, 8.7 million African children were classified as ‘zero-dose,’ having never received a single vaccine, out of a total 12.7 million under-vaccinated children on the continent [10]. This problem is geographically concentrated, with half of the 20 countries worldwide most affected by zero-dose children situated in Africa. Within this context, Nigeria and Ethiopia, Somaliland’s neighbor, stand out, with these two nations alone home to over 3.3 million zero-dose children—representing 40% of the continent’s total [10, 11]. Studies in Ethiopia, for instance, show that while full immunization coverage was low (e.g., 39% in Debie et al., 2020), significant proportions of children were also likely zero-dose or incompletely vaccinated, with spatial clustering of these vulnerable groups [1, 12, 13].

Inspired by the demonstrated success of large-scale global efforts to control specific infectious diseases, the World Health Organization (WHO) launched the Expanded Program on Immunization (EPI) at the World Health Assembly in 1974 [11, 14, 15]. The initial goal of EPI was to systematically apply and expand this successful model to immunize children against other key vaccine-preventable diseases (VPDs), thereby establishing a crucial global framework to reduce child mortality and morbidity from these illnesses [15]. The Expanded Program on Immunization (EPI) in Somaliland commenced in 1995, established in collaboration with WHO and UNICEF [16]. However, findings from the Somaliland Demographic Health Survey indicate persistent challenges, with merely 13% of children under the age of two having completed their immunization schedule [17]. Such low coverage contributes to an elevated under-five mortality rate of 117 per 1,000, a significant portion of which is attributable to vaccine-preventable diseases [18].

Identifying the geographical distribution and determinants of zero-dose status is crucial for effective public health interventions. National-level statistics often obscure significant subnational variations and inequities [10]. Research in similar contexts has consistently shown that zero-dose prevalence is not uniform. For example, a study in Kikwit, DRC, found a 16.3% zero-dose prevalence with significant spatial clustering and identified hotspots [11]. Similarly, analyses across 43 LMICs revealed substantial geographical disparities, with some regions in Africa and Asia exhibiting extremely low immunization rates [11]. In Kenya, distinct hotspots of zero-dose children were identified in northern arid and semi-arid lands [19], while studies in Ethiopia also highlighted geographical clustering of incomplete immunization [12, 13].

Previous studies across various low- and middle-income countries, including contexts like Ethiopia, Kenya, and the DRC, indicate that the factors contributing to zero-dose status include socio-demographic aspects such as maternal education and household wealth; child-specific factors such as age; health service access indicators like distance to facilities and ANC attendance; caregiver knowledge of VPDs; and media exposure, including mobile phone usage [1, 2, 11–14, 19]. While considerable research exists on general immunization coverage in the Horn of Africa, there is a specific need to focus on zero-dose children in Somaliland and to explore the spatial patterns and determinants associated with this status. Traditional regression analyses, while informative, may not fully capture the geographically varying nature of these determinants. Geospatial mapping, on the other hand, provides powerful tools to visualize the distribution of zero-dose children, identify high-risk areas (hotspots), and potentially uncover how determinants exert their influence differently across space.

Therefore, this study aims to utilize geospatial mapping techniques to assess the distribution and determinants of unvaccination status among children under five years in Somaliland, based on data from the 2020 Somaliland Demographic and Health Survey (SLHDS). By identifying geographical hotspots and examining associated factors, this research seeks to provide actionable evidence for policymakers and public health practitioners to develop targeted strategies to reach these highly vulnerable children, thereby strengthening immunization programs and advancing health equity in Somaliland.

Methods

Study design and setting

This study employed a cross-sectional design, utilizing quantitative secondary data analysis and geospatial mapping techniques to examine the distribution and identify the determinants unvaccination status among children in Somaliland. Somaliland is located in the Horn of Africa, bordered by Djibouti, Ethiopia, and Somalia, with a coastline along the Gulf of Aden. Administratively, the country is divided into six major regions (Awdal, Marodi-Jeex, Sahil, Sool, Sanaag, and Togdheer), which are further subdivided into districts.

Data source

The data for this study were derived from the Somaliland Health and Demographic Survey (SLHDS) 2020. This survey, the first of its kind in Somaliland, was conducted by the Central Statistics Department of the Ministry of Planning and National Development (MoP&ND) and the Ministry of Health Development (MoHD), with technical and financial assistance from the United Nations Population Fund (UNFPA) and other partners. Fieldwork for the main survey was largely undertaken between August 2018 and December 2019.

The SLHDS 2020 was designed to be nationally representative, covering all six administrative regions and stratifying by urban, rural, and nomadic populations using a multi-stage cluster sampling approach. For urban and rural areas, Enumeration Areas (EAs) were created from digitized dwelling structures via satellite imagery, followed by household listing and systematic selection. For nomadic populations, Temporary Nomadic Settlements (TNS) identified by local informants served as primary sampling units, with subsequent household listing and selection. Data were collected using standardized Household and Ever-Married Woman’s Questionnaires via Computer-Assisted Personal Interviewing (CAPI). For the present study, individual child records, their vaccination status (ascertained from vaccination cards or maternal recall), relevant socio-demographic and economic characteristics of the child, mother, and household, along with geospatial cluster coordinates (EAs/TNSs), were extracted from the SLHDS 2020 datasets.

Variables

Outcome variable

The primary outcome variable was the child’s vaccination status, derived from the SLHDS 2020 question “ever had vaccinated.” Children reported as having received at least one routine childhood vaccine were classified as vaccinated, while those reported as having never received any vaccine were classified as unvaccinated. Children with missing, “don’t know,” or non-response values for this variable were excluded from the analysis. As missingness may be related to vaccination status, this approach may introduce bias and is acknowledged as a limitation. Vaccination card and dose-specific information were incomplete and could not be consistently used; therefore, classification relied on caregiver report. The outcome reflects complete non-vaccination rather than vaccine-specific coverage.

Predictor variables

Predictor variables were selected a priori based on documented associations with childhood vaccination in the literature and their availability in the SLHDS 2020 dataset. Variables were grouped into individual/household-level and community-level factors to align with the multilevel modeling framework. In this analysis, communities were defined as survey clusters (enumeration areas) used in the SLHDS sampling design, and random intercepts were specified at this level.

At the individual and household level, predictors included child characteristics (sex of the child) and maternal factors, including the mother’s highest educational attainment (no education, primary, secondary, or higher), current pregnancy status (yes/no), current breastfeeding status (yes/no), number of antenatal care (ANC) visits during the most recent pregnancy (none, fewer than four visits, or four or more visits), and exposure to mass media. Household-level characteristics included place of delivery (home or health facility) and household wealth index, categorized as poor, middle, or rich based on SLHDS asset scores.

Community-level predictors captured broader contextual influences and included place of residence (urban or rural), administrative region of residence (Awdal, Marodi-Jeex, Sahil, Togdheer, Sool, or Sanaag), and community-level ANC utilization, defined as the proportion of women within each survey cluster who attended at least four ANC visits and categorized as high or low.

Data management and analysis

Data source and preparation

This study utilized secondary data from the Somaliland Health and Demographic Survey (SLHDS) 2020, a nationally representative survey employing a multi-stage cluster sampling design. Rigorous data quality assurance procedures were integral to the primary data collection. The primary outcome was a binary variable: ‘Un-vaccinated’ (children who had received no routine vaccines) versus ‘vaccinated’ (children who had received at least one). Cases with missing data on the outcome or key predictor variables were excluded from the final analysis.

Statistical analysis

Because the outcome variable was binary (unvaccinated vs. vaccinated), multilevel logistic regression models were used to examine associations between child non-vaccination and individual-, household-, and community-level factors, while accounting for the hierarchical structure of the SLHDS data. Children were nested within communities (clusters), and random intercepts were specified at the community level to capture unobserved contextual variation.

Covariates were selected a priori based on existing literature and conceptual relevance to childhood vaccination, including child age, maternal characteristics, household socioeconomic factors, and community-level indicators. These variables were entered sequentially into four models: Model 0 (null model with no predictors), Model I (individual and household-level factors), Model II (community-level factors), and Model III (full model including all predictors).

Model fit and comparison were assessed using Akaike’s Information Criterion (AIC) and Bayesian Information Criterion (BIC), with lower values indicating better fit. Fixed effects were reported as Adjusted Odds Ratios (AORs) with 95% confidence intervals. Community-level variation was assessed using the Intra-class Correlation Coefficient (ICC) derived from the random-effects variance.

Results

This study utilized data from a total unweighted sample of 1,104 children under five years of age. The descriptive analysis of children included in the study reveals significant variations in vaccination status across different socio-demographic and regional characteristics in Somaliland (Table 1). The distribution of unvaccinated and vaccinated children under five years of age by socio-demographic characteristics in Somaliland. The largest proportions of children were from Togdheer (29.98%), Sanaag (27.47%), and Sool (21.98%), with Togdheer accounting for the highest share of unvaccinated children (40.15%). Children were almost equally distributed between urban (49.89%) and rural (50.11%) residences, with a slightly higher proportion of unvaccinated children residing in urban areas. Most children were born to mothers with no formal education (78.60%), and this group accounted for the majority of unvaccinated children (70.39%). Male children constituted a slightly higher proportion of the sample than females, with no marked difference in vaccination status by sex. The majority of mothers were not currently pregnant (81.27%), although a higher proportion of unvaccinated children had mothers who were currently pregnant. More than half of the children were currently breastfeeding; however, a lower proportion of unvaccinated children were breastfeeding compared to vaccinated children. Most children were aged 24–59 months (80.83%), and this age group also represented the highest proportion of unvaccinated children. Home delivery was common (71.30%), and more than half of unvaccinated children were delivered at home. Nearly all households reported exposure to mass media. Regarding maternal health service utilization, most mothers did not attend any antenatal care (ANC) visits, and unvaccinated children were less likely to have mothers who attended one or more ANC visits compared to vaccinated children.

Table 1.

Distribution of Unvaccinated Status among Children Under Five by Socio-Demographic Characteristics, SLHDS 2020

			Vaccination Status
			Unvaccinated	Vaccinated
Variable	Categories	Total N (Weighted %)	Yes(%)	No(%)
Regions	Awdal	48 (4.36%)	10(2.04%)	38 (6.08%)
	Marodijex	107 (9.68%)	62 (13.11%)	45. (7.13%)
	Sahil	72 (6.53%)	33 (6.94%)	39 (6.23%)
	Togdheer	331 (29.98%)	189(40.15%)	142 (22.44%)
	Sool	243 (21.98%)	70(14.93%)	173(27.21%)
	Sanaag	303 (27.47%)	107(22.82%)	196 (30.92%)
Place of residence	Urban	551 (49.89%)	243 (51.71%)	308(48.55%)
	Rural	553(50.11%)	227 (48.29%)	326 (51.45%)
Highest Educational level	No education	868 (78.60%)	331(70.39%)	5537 (84.68%)
	Primary	188 (17.05%)	101(21.46%)	87 (13.78%)
	Secondary	40(3.62%)	32(6.73%)	8(1.32%)
	Higher	8 (0.73%)	7(1.42%)	1(0.23%)
Sex of child	Male	608(55.05%)	252 (54.20%)	353 (55.69%)
	Female	496(44.95%)	215 (45.80%)	281(44.31%
Currently pregnant	Yes	207 (18.73%)	105 (22.34%)	102(16.05%)
	No	897(81.27%)	364(77.66%)	533(83.95%)
Currently breastfeeding	Yes	575 (52.08%)	209 (44.59%)	365 (57.63%)
	No	529 (47.92%)	260(55.41%)	268 (42.37%)
Child Age	0–8 months	80(7.24)	18(3.82%)	62(9.77%)
	9–23 months	131(11.93%)	58(12.31%)	73(11.65%)
	24–59 months	893(80.83%)	395(83.86%)	498(78.58%)
Place of delivery	Home	787 (71.30%)	258(54.83%)	529(83.51%)
	Health facility	317(28.70%)	212(45.17%)	105(16.49%)
Wealth index	Poor	528 (47.82%)	137(29.15%)	391(61.66%)
	Middle	135(12.19%)	64(13.59%)	71(11.15%)
	Rich	441(39.99%)	269(57.26%)	172(27.19%)
Media exposure	No	12 (1.08%)	8(1.72%)	4(0.61%)
	Yes	1,092 (98.92%)	462(98.28%)	630(99.39%)
ANC visit	No ANC visit	662 (59.97%)	180(38.28%)	482(76.03%)
	1–3 visits	291 (26.31%)	178(37.82%)	113(17.78%)
	4 or more visits	151 (13.72%)	112(23.89%)	39(6.18%)

A multilevel logistic regression analysis was conducted to examine both individual and community-level factors associated with zero-dose vaccination among children in Somaliland, as detailed .Table 2 presents the multilevel logistic regression analysis examining factors associated with the unvaccinated status of children under five years of age in Somaliland. Results are reported as Adjusted Odds Ratios (AORs) with corresponding 95% Confidence Intervals (CIs). Interpretation focuses on Model 3, which includes individual-, household-, and community-level variables and therefore provides the most robust estimates.

Table 2.

Multilevel logistics regression analysis of individual and community level factors associated Unvaccinated Status among Children Under Five Somaliland

Variables	Model 0	Model 1	Model 2	Model 3
		AOR (95%CI)	AOR (95%CI)	AOR (95%CI)
Region
Awdal			1	1
Marodijex			0.39 (0.183 − 0.841) *	0.54 (0.249 − 1.18)
Sahil			0.45 (0.228 − 0.909) *	0.77 (0.377 − 1.60)
Togdheer			0.54 (0.277- 0.1.05)	0.69 (0.351 − 1.35)
Sool			0.85 (0.448 − 1.63)	0.98 (0.510-0.1.91)
Sanaag			0.67 (0.354 − 1.28)	0.75 (0.386 − 1.46)
Place of resident
Rural			1	1
Urban			0.99 (0.771 − 1.27)	0.96 (0.743 − 1.24)
Highest educational level
No education		1		1
Primary		0.98 (0.686 − 1.39)		0.91(0.689 − 1.39)
Secondary		0.30 (0.130- 0.709) **		0.32 (0.139-0.745) **
Higher		0.54 (0.113 − 2.62)		0.56 (0.118 − 2.71)
Sex of child
Male		1		1
Female		0.88 (0.686 − 1.12)		0.88 (0.690 − 1.13)
Child age
0–8 months		1
9–23 months		0.38 (0.199- 0.747) **		0.37 (0.193-0.720) ***
24–59 months		0.31 (0.180- 0.564) ***		0.32 (0.185-0.579) ***
Currently pregnant
Yes		1		1
No		1.07(0.755 − 1.53)		1.09(0.7652-0.1.555)
Currently breastfeeding
Yes		1		1
No		0.67 (0.509- 0.890) **		0.68 (0.517- 0.904)**
Place of delivery
Home		1		1
Health facility		0.69 (0.497- 0.970) *		0.72 (0.513-0.1.01)
Wealth index
Poor		1		1
Middle		0.81 (0.527 − 1.24)		1.16(0.734 − 1.83)
Rich		0.54 (0.382- 0.765) **		0.85 (0.574 − 1.28)
Media exposure
No		1		1
Yes		2.03(0.531 − 7.83)		1.96(0.518 − 7.42)
ANC visit
No ANC visit		1		1
1–3 Visits		0.32 (0.241- 0.443) ***		0.37 (0.278-0.515) ***
4 or more visits		0.32 (0.210- 0.509) ***		0.40 (0.262-0.637) ***
Community ANC utilization
Low Community ANC Utilization			1
High Community ANC Utilization			0.20 (0.151-. 288)***	0.37 (0.248-0.570) ***

Maternal education and prenatal health-seeking behavior were significant determinants of childhood vaccination status. Children born to mothers with secondary education had significantly lower odds of being unvaccinated compared with those whose mothers had no formal education (AOR = 0.32, 95% CI: 0.14–0.74, p < 0.01). Antenatal care (ANC) attendance emerged as the strongest individual-level predictor. Mothers who attended 1–3 ANC visits (AOR = 0.38, 95% CI: 0.28–0.51, p < 0.001) or four or more visits (AOR = 0.41, 95% CI: 0.26–0.64, p < 0.001) were significantly less likely to have unvaccinated children compared to mothers who did not attend any ANC visits. These findings underscore the critical role of maternal contact with health services in improving childhood immunization coverage.

Child age was strongly associated with vaccination status. Compared to infants aged 0–8 months, children aged 9–23 months (AOR = 0.37, 95% CI: 0.19–0.72) and those aged 24–59 months (AOR = 0.33, 95% CI: 0.18–0.58) had significantly lower odds of being unvaccinated (p < 0.001), indicating increased vaccination uptake as children grow older. In addition, children who were not currently breastfeeding had 32% lower odds of being unvaccinated (AOR = 0.68, 95% CI: 0.52–0.90, p < 0.01) compared with those who were breastfeeding at the time of the survey. At the community level, overall utilization of maternal health services was a significant predictor of vaccination status. Children residing in communities with high ANC utilization had substantially lower odds of being unvaccinated (AOR = 0.38, 95% CI: 0.25–0.57, p < 0.001) compared to those in low-utilization communities. Although regional location (Marodijex and Sahil) and facility-based delivery were significant in Models 1 and 2, their associations were attenuated and no longer statistically significant in the fully adjusted model.

Table 3 presents a comparison of four nested models (Model 0, Model I, Model II, and Model III). Model fit was evaluated using the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and log-likelihood values. Among the models, Model III had the lowest AIC (1592.795) and the lowest BIC (1713.395), indicating improved fit compared to the preceding models. The log-likelihood values also increased progressively across models, with Model III showing the highest value (− 773.397), further supporting its improved fit. Based on these model comparison criteria, Model III was selected as the final model for interpretation.

Table 3.

Model comparison and random effect analysis result

Parameters	Model 0	Model I	Model II	Model III
AIC	1747.19	1605.224	1674.048	1592.795
BIC	1757.677	1689.12	1721.239	1713.395
Log likelihood	-871.5952	-786.61178	-828.02394	-773.39727

Figure 1 presents the spatial distribution of the prevalence of unvaccinated children under five across regions in Somaliland. Substantial regional variation is observed. The highest prevalence of non-vaccination is found in Awdal (80.1%), followed by Sool (71.1%), indicating particularly high vulnerability in the westernmost and southeastern regions. Sanaag also exhibits a high prevalence (64.6%), while Sahil shows a moderate level of non-vaccination (54.8%). In contrast, Maroodi Jeex (42.3%) and Togdheer (43.0%) record comparatively lower prevalence, although levels remain unacceptably high across all regions. Overall, the map highlights pronounced geographic inequalities in childhood vaccination coverage, underscoring the need for region-specific strategies to improve immunization access and reduce zero-dose prevalence in Somaliland.

Fig. 1.

Spatial Distribution of the Prevalence of Unvaccinated Children Under Five in Somaliland

Discussion

This study provides a critical multilevel analysis of unvaccinated children in Somaliland, revealing an alarmingly high prevalence of 79.88% among children under five. Overall, the findings indicate that low maternal education, limited engagement with antenatal care services, and residence in communities with poor health-service utilization are the primary barriers to childhood vaccination in Somaliland. These results highlight the importance of integrated interventions that strengthen maternal education and expand ANC coverage at both individual and community levels as effective strategies to reduce the proportion of unvaccinated children. The prevalence observed in Somaliland is substantially higher than reported in many Sub-Saharan African countries and pooled LMIC estimates, dwarfing the 16.5% zero-dose prevalence found across 33 Sub-Saharan African countries and the pooled 7.7% across 92 LMICs [11, 20]. The situation appears far more severe than in neighboring countries like Ethiopia, where studies have documented significant challenges with incomplete and untimely vaccination but have not reported zero-dose rates of this magnitude [6, 28, 29]. While countries like Togo have shown slow but steady progress in reducing their zero-dose burden over decades [21], the sheer scale of the problem in Somaliland underscores the critical importance of the government’s commitment to “scale up priority interventions such as the vaccination program” [22].

Our multilevel analysis identified powerful determinants that align with established evidence. The protective effect of maternal education is a consistent theme across the region, validated by research in Ethiopia, Kenya, and Nigeria [5, 19, 27]. and by Somaliland’s National Health Policy, which identifies “High illiteracy” as a key social determinant of health [23]. Similarly, contact with the health system via Antenatal Care (ANC) and facility-based delivery emerged as a highly significant protective factor. This finding is particularly poignant in light of global evidence that nearly half of children defined as “penta-zero dose” have had at least one other vaccine contact, framing ANC and delivery as critical but often missed opportunities for immunization [23, 26].

Socioeconomic status was another critical determinant, reflecting deep-rooted inequities central to Somaliland’s UHC goals [24]. This wealth-based disparity is a pervasive driver of under-vaccination across Africa [6, 20]. Importantly, this inequity has consequences that extend far beyond immediate health. Research from India demonstrates that being a zero-dose child is a marker of systemic disadvantage that predicts poorer long-term learning attainment in preadolescence, with effects most pronounced in the poorest communities [25]. This reframes Somaliland’s zero-dose crisis as a fundamental threat to its future human capital.

Descriptive mapping in this study revealed regional variation in the prevalence of unvaccinated children, with higher proportions observed in Awdal and Sool. Similar regional disparities in immunization coverage have been documented in other African settings, including Kenya [19], Zambia [26], Nigeria [27, 31], and Ethiopia [27], where geographic inequalities reflect differences in access to health services and socioeconomic conditions. Although the present analysis was descriptive rather than inferential, the observed regional variation underscores the importance of geographically sensitive programmatic planning within the broader health system strengthening efforts outlined in Somaliland’s UHC Roadmap [24].

Cluster-level variation was observed in the empty model and decreased substantially after adjustment for individual and community-level covariates, indicating that much of the between-cluster variability was explained by measured factors. This finding is powerfully illustrated by evidence that the negative developmental impacts of being zero-dose are moderated by community context, demonstrating that a child’s environment shapes their life-course outcomes [25]. The significant unexplained community-level variance in our model confirms that unobserved characteristics—such as the systemic weaknesses outlined in Somaliland’s policy documents—play a crucial role, reinforcing the need for community-based approaches, a conclusion also reached in multilevel analyses in Nigeria and across SSA [13, 20, 27].

This study’s strengths lie in its use of a recent, nationally representative dataset and robust analytical methods. Limitations must be acknowledged, including the cross-sectional design and reliance on maternal recall, a practice necessitated by the “low card retention rate” officially noted in Somaliland’s EPI policy [22]. Furthermore, this study defined unvaccinated status based on caregiver report that the child had never received any vaccine. This differs from the standard WHO definition of zero-dose based on non-receipt of DTP1 and should be considered when comparing estimates [28].

In conclusion, this study provides an evidence base to operationalize the very policies designed to address Somaliland’s zero-dose crisis. Our findings strongly support a dual strategy that is not only data-driven but also directly aligned with national health priorities. First, broad initiatives to enhance female education and promote ANC utilization directly support the goals outlined in the National Health Policy III and UHC Roadmap. Second, targeted programmatic attention to regions such as Awdal and Sool, where higher prevalence was observed, may help prioritize immunization strengthening efforts [19, 26]. By leveraging these multilevel insights, Somaliland can turn the ambitious goals of its national policies into life-saving—and life-altering—realities for its most vulnerable children, a critical step toward strengthening its health system and aligning with the global Immunization Agenda 2030 [25] .

Policy implications

The exceptionally high prevalence of unvaccinated children in Somaliland highlights an urgent need for strengthened and context-specific immunization strategies. Descriptive prevalence estimates and regional maps indicate substantial geographic variation, with some regions consistently showing higher proportions of unvaccinated children. Public health programs should prioritize these high-burden regions through targeted service delivery approaches, including outreach vaccination activities, mobile clinics, and community-based engagement, to address access barriers in underserved areas.

The strong protective associations observed for maternal education, antenatal care (ANC) attendance, and facility-based delivery underscore the importance of integrating immunization services within maternal and child health platforms. Policies that promote girls’ education and improve women’s access to ANC and skilled delivery services may have substantial downstream benefits for childhood vaccination uptake. Strengthening routine immunization counseling during ANC visits and ensuring timely vaccination at or shortly after birth can help reduce the number of children who remain entirely unvaccinated.

Socioeconomic disparities remain a critical determinant of immunization exclusion. Children from poorer households were significantly more likely to be unvaccinated, indicating that zero-dose status reflects broader structural disadvantage. Policy responses should therefore extend beyond the health sector to include social protection measures, reduction of indirect costs associated with accessing care, and targeted support for vulnerable households. Community-level variation in unvaccinated prevalence further highlights the importance of supporting community health workers and local leadership to promote vaccination, address misinformation, and facilitate linkage to services.

Finally, routine use of prevalence-based mapping can support immunization planning by identifying areas with persistently low coverage and guiding resource allocation. Integrating such descriptive geographic monitoring into national immunization planning can strengthen Somaliland’s progress toward equitable service delivery and align with national health and universal health coverage priorities.

Conclusion

Somaliland faces an exceptionally high burden of childhood non-vaccination, with nearly four in five children under five years of age having received no routine immunizations. This prevalence far exceeds regional and global averages, underscoring a critical public health challenge. Descriptive analyses reveal notable regional variation in unvaccinated prevalence, while multilevel regression results highlight the importance of maternal education, household wealth, antenatal care utilization, and facility-based delivery as key protective factors against zero-dose status.

Community-level influences also contribute meaningfully to vaccination outcomes, emphasizing the need for interventions that address both individual and contextual determinants. Together, these findings provide actionable evidence for policymakers to strengthen immunization programs through integrated maternal–child health services, socioeconomic support for vulnerable populations, and targeted strategies in high-prevalence regions. Addressing these gaps is essential for advancing immunization equity and improving child health outcomes in Somaliland.

Abbreviations

ANC

Antenatal Care

AOR

Adjusted Odds Ratio

AIC

Akaike Information Criterion

BIC

Bayesian Information Criterion

CI

Confidence Interval

ICC

Intraclass Correlation Coefficient

LISA

Local Indicators of Spatial Association

SLHDS

Somaliland Health and Demographic Survey

OR

Odds Ratio

Authors’ contributions

H.M.D. conceptualized the study, conducted the data analysis, and drafted the initial manuscript. F.I.M. contributed to advanced statistical analysis, spatial modeling, and interpretation of findings. A.H.K. contributed to data management, interpretation of results, and manuscript writing. S.N. provided statistical guidance, contributed to methodological development, and critically revised the manuscript. All authors read and approved the final manuscript.

Funding

The authors did not receive any specific funding for this work.

Data availability

The data used in this study were obtained from the Central Statistics Department, Ministry of Planning, Somaliland. The dataset is not publicly available due to institutional restrictions but can be accessed upon reasonable request by contacting the Central Statistics Department, Ministry of Planning, Somaliland. The corresponding author can provide further details upon request.

Declarations

Ethics approval and consent to participate

Ethical approval for the 2020 Somaliland Health and Demographic Survey (SLHDS) was granted by the Ministry of Planning and National Development, Somaliland. Permission to use the anonymized dataset for secondary analysis was obtained from the Central Statistics Department, Ministry of Planning. As this study involved secondary analysis of de-identified data, no additional ethical approval was required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.