Measles spreads efficiently when susceptible people are concentrated in the same population, allowing outbreaks to expand rapidly once the virus is introduced. Vaccination interrupts that chain of transmission by reducing the number of people who can become infected and pass the virus onward. The first measles-containing vaccine dose is therefore a central marker of childhood protection, although a single dose does not provide complete protection for every child. The study’s estimated coverage of 58.3%, with a 95% confidence interval of 55.5% to 61.1%, indicates that many children remained outside the reach of routine or supplementary immunization services during the period represented by the survey. The researchers describe this coverage as below the level needed for herd protection. In practical terms, low overall coverage can conceal even more serious local vulnerabilities: a national average may appear moderate while particular districts or communities have far fewer vaccinated children and therefore greater potential for sustained transmission.

The investigators analyzed records for 3,436 children aged between 12 and 59 months. The survey data were collected during fieldwork in 2018 and 2019 and formed the basis of the 2020 Somalia Demographic and Health Survey. To estimate vaccination prevalence, the researchers applied the survey’s sampling weights, including the variable V005, so that the results reflected the design of the national survey rather than simply treating every sampled record as equally representative. The outcome was coded as vaccinated or unvaccinated according to whether the child had received MCV1. Because children were sampled within households and communities, the study used multilevel logistic regression rather than a single-level model. This approach is designed to separate associations linked to individual or household characteristics from variation associated with the broader community context. It cannot prove that a particular factor causes vaccination, but it can show how strongly measured characteristics are associated with the odds of receiving a dose while accounting for clustered data.

The geographic analysis revealed a strong spatial pattern. Global Moran’s I, a statistic used to test whether similar values are arranged near one another rather than distributed randomly, was 0.653, with a probability value below 0.001. That result indicates significant positive spatial autocorrelation: areas with relatively high vaccination levels tended to be near other areas with relatively high levels, while areas with lower coverage tended to cluster with other low-coverage areas. The researchers then used Getis-Ord Gi* hotspot analysis to identify concentrations of unusually high or low values and applied SaTScan Bernoulli models to search for statistically defined geographic clusters. Together, these techniques produced a map of concentrated vaccination advantage and disadvantage. Hotspots were identified in the northern regions of Togdheer, Sool, Sanaag and Bari. Coldspots appeared in the southern regions of Gedo, Lower Juba and Bakool, where low uptake was geographically concentrated rather than isolated.

The multilevel findings add detail to that map. Children aged 48 to 59 months had substantially higher adjusted odds of having received MCV1 than the comparison age group, with an adjusted odds ratio of 16.32. An odds ratio compares the odds of an outcome between groups after accounting for other variables in the model; it is not the same as a percentage increase in vaccination coverage. The strong age association may reflect the cumulative opportunity for a child to encounter vaccination services as time passes, although the survey design cannot establish the exact reason. Maternal education was also associated with uptake. Children whose mothers had secondary education had 1.83 times the adjusted odds of vaccination compared with the reference category. The authors interpret this pattern as consistent with the potential importance of health knowledge, communication with services and the ability to navigate vaccination schedules, while recognizing that education may also be connected with other social and economic conditions.

Household economic position and mobility were likewise associated with MCV1. Children from middle-income households had 1.88 times the adjusted odds of vaccination compared with the reference wealth group. The study also found higher adjusted odds among children in nomadic residence, with an adjusted odds ratio of 1.60. That result does not mean that all nomadic families experience easier access to immunization. Instead, it shows an association within this dataset and may reflect the reach of particular outreach activities, differences in service delivery or other characteristics measured or unmeasured in the survey. Regional differences were especially large: children in Togdheer had 6.3 times higher odds of vaccination than children in the reference region. Such contrasts emphasize that national vaccination strategies cannot rely solely on uniform service delivery. The same intervention may perform differently depending on transport routes, settlement patterns, local health infrastructure and the ability of families to return for scheduled care.

The researchers quantified the contribution of community-level conditions using the intra-cluster correlation coefficient, or ICC. The ICC indicated that 13.1% of the variance in vaccination status was attributable to differences between communities. In a hierarchical dataset, this measure helps show whether children living in the same community resemble one another in their likelihood of vaccination beyond what can be explained by their individual characteristics. A community contribution of this magnitude suggests that place-based factors matter: the presence and reliability of health facilities, outreach schedules, local information networks, security conditions, transport and the organization of mobile services may all influence whether families can obtain immunization. The analysis did not measure every one of these mechanisms directly, so the ICC should not be read as identifying a single cause. Rather, it supports the conclusion that interventions focused only on household behavior would leave part of the vaccination gap untouched.

The study’s conclusions direct attention to the southern coldspots and to groups that may be missed by facility-based programs. The authors recommend prioritizing geographically identified low-coverage areas, strengthening maternal health literacy and expanding mobile vaccination outreach for underserved rural and nomadic populations. A mobile approach can bring services closer to communities whose distance, movement or limited transport options makes routine clinic attendance difficult. Geospatial results can help planners decide where such activities are most urgently needed, while multilevel findings can help tailor communication and delivery strategies to local circumstances. The analysis is based on secondary, anonymized, publicly available survey data and therefore describes associations at the population level rather than tracking children over time. Even with that limitation, its combination of weighted survey estimates, spatial statistics and hierarchical regression provides a more targeted view than a single national coverage figure. For Somalia, the message is clear: closing the measles immunity gap will require not only more doses, but also a precise understanding of where access and protection remain weakest.

The study’s design is particularly useful for distinguishing national coverage from the distribution of coverage. A weighted prevalence describes the estimated share of eligible children who had received MCV1, whereas the spatial statistics examine whether vaccination levels are arranged in recognizable geographic patterns. These answer different questions: the first indicates the overall scale of the immunity gap, while the second helps determine whether low uptake is concentrated enough to justify place-specific planning. The significant Moran’s I result and the hotspot and coldspot analyses therefore complement, rather than duplicate, the regression findings.

The multilevel framework also changes how the reported associations should be interpreted. In a conventional regression, observations from the same community may be treated as independent even though they can share health facilities, outreach teams, information sources and environmental constraints. Accounting for clustering reduces the risk that such shared circumstances will be mistaken for purely individual effects. The community-level variance identified by the investigators consequently supports a delivery-oriented interpretation: improving vaccination may require changes in how services are organized and reached, not simply stronger demand among caregivers.

Age deserves careful consideration when translating the results into program decisions. Older children have had more time to encounter routine services or catch-up activities, so the age association may signal delayed rather than absent access. It also suggests that programs should actively identify children who have passed the usual vaccination contact point without receiving MCV1. This is different from assuming that younger children are inherently less likely to benefit; they may simply have had fewer opportunities by the time of observation. The cross-sectional survey cannot establish the sequence of events behind the association.

Similarly, the findings on education, wealth and residence should guide investigation rather than serve as fixed classifications of families. These characteristics can be connected to several pathways at once, including awareness, affordability of travel, trust, service availability and the timing of outreach. The study’s adjusted odds ratios show relationships after inclusion of measured variables, but they do not identify which pathway is responsible or demonstrate that changing one characteristic would produce a specific increase in uptake. Local follow-up could therefore help determine whether the most effective response is communication, transport support, improved session reliability, or a combination of approaches.

Because the analysis used existing anonymized survey data, it provides a broad population-level assessment without exposing participants to additional data collection. Its findings can establish priorities for vaccination planning, but implementation should still be accompanied by monitoring. Repeated coverage assessments and evaluation of outreach in the designated low-coverage areas would help determine whether geographic inequalities narrow over time and whether improvements reach children who remain unvaccinated, rather than merely shifting the national average.

Subject of Research: Geographic and social determinants of first-dose measles vaccination among children in Somalia

Article Title: Spatial patterns and determinants of childhood measles vaccination in Somalia a multilevel and geospatial analysis of the 2020 demographic and health survey

Article References: Abdillahi, A. M., Ali, I. H., Omer, H. A., Said, N. I., Bile, H. S., & Muse, A. H. (2026). Spatial patterns and determinants of childhood measles vaccination in Somalia a multilevel and geospatial analysis of the 2020 demographic and health survey. Discover Social Science and Health. https://doi.org/10.1007/s44155-026-00416-4

Image Credits: AI Generated

DOI: 10.1007/s44155-026-00416-4

Keywords: Measles vaccination, Somalia, MCV1, Child health, Geospatial analysis, Multilevel modeling, Vaccine coverage, Health inequities, Spatial, patterns, determinants, childhood