Worldwide about 13.5 million children have never received a vaccine at all. These are known as zero-dose children. Reaching them is central to the Immunization Agenda 2030, a global drive to ensure that everyone has access to the vaccines they need. Gavi, the vaccine alliance, has built its strategy around finding and reaching them.
Reaching zero-dose children starts with knowing where they are: putting them on a map.
Geographers offer a starting point – the first law of geography is that objects closer to another are more alike than those further apart. This principle underpins geospatial health, the study of how where people live shapes their health. If we know something about one place, we can often infer the same thing about its neighbours. This includes how many unvaccinated children live in areas surveys haven’t reached.
We are part of the REACH-OUT team, a project which aims to develop cost-effective strategies to deliver vaccines to zero-dose children. Our expertise includes spatial epidemiology, health economics and implementation science.
In a recent study, we assessed how geospatial data and methods are being used to find zero-dose children, and where the gaps are. We reviewed 102 studies across 68 low- and lower-middle-income countries where, according to WHO and UNICEF estimates of immunisation coverage, more than 80% of the world’s zero-dose children are.
Our findings show that geospatial approaches are being applied in child vaccination coverage modelling. But few focus specifically on zero-dose children. Reaching them needs more complete and connected data, and tools in the hands of those closest to the children.
Hidden in the averages
A country can post high overall immunisation coverage and still mask small pockets of zero-dose children.
For instance, according to Kenya’s Demographic and Health Surveys (2022), nationally, only 2.1% of children aged 12-23 months were zero-dose. But at county level it’s different. Two counties, Mandera and Garissa, recorded zero-dose prevalence above 36%.
High-resolution mapping is one way to find zero-dose children who remain exposed to preventable diseases. But it requires first pinning down who counts as zero dose.
There are two definitions.
First, Gavi defines a zero-dose child as one who has not received even a single vaccine shot. This definition is very strict and often hard to implement due to lack of appropriate data. Surveys are conducted every three to five years, while routine health facility data does not link records to individual children, making it hard to track the children in need.
Second, zero-dose children are measured as those missing the first dose of the diphtheria, tetanus and pertussis vaccine (DTP1) by their first birthday. DTP1 is recommended at six weeks, marking the first point of a child’s contact with the routine immunisation system. It is also widely administered and routinely reported through national immunisation systems, allowing comparison of zero-dose burden across countries.
The difference in the populations captured by the two definitions is large: globally, the first measure identifies nearly twice as many children as the second. So a country’s zero-dose burden (and its strategy) will differ depending on which measure is used.
Zero-dose children are often the hardest to reach. They cluster in marginalised settings – remote rural areas, conflict zones, informal urban settlements, and mobile and migrant families – mostly in low- and lower-middle-income countries. Finding them takes more than knowing which districts have low coverage; it means locating the settlements and households where they live.
A growing field, but an uneven picture
Geospatial modelling for childhood vaccination coverage is progressing fast. Most studies that we reviewed were published after 2022. But the evidence base is still thin. Only one in five studies looked specifically at zero-dose children; most examined routine immunisation coverage more broadly. The evidence is also geographically skewed as most studies were from just three countries: Nigeria, Ethiopia and India. Some countries, including those with a heavy zero-dose burden such as Sudan and Yemen, were barely represented. This reflects broader disparities in global health research capacity and priority-setting.
The model is only as good as the data
One of our clearest findings was the huge reliance on household survey data. Routine immunisation data from health facilities, collected continuously through systems like DHIS2 – a widely used digital platform for collecting, managing and analysing routine health facility data – were rarely used.
Surveys are valuable but limited. They run every three to five years, a snapshot in time. They are designed for national or regional estimates rather than for pinpointing pockets of unvaccinated children.
Routine data collected continuously at the facility level offer greater local detail, but can be undermined by incomplete reporting and unreliable estimates of how many children are eligible for vaccination.
Read more:
HIV in Malawi: digital filing system saved lives and boosted care – research
The studies we reviewed often failed to represent the very children most likely to miss out. A few covered conflict-affected communities, nomadic groups, refugees, cross-border migrants or urban-slum dwellers. These populations are more likely to live far from health facilities, move frequently, or depend on easily disrupted services. A model lacking remoteness or conflict data cannot estimate the children affected by them. Thus, the children we most need to find become the ones data are least able to see.
Methods matter as much as data. More than half the studies leaned on simpler methods like clustering and hotspot analysis. These methods show where zero-dose children have been counted, not how many live where no survey reached. Only a handful used geostatistical methods which integrate types of information to predict where vaccination gaps are likely.
What next?
Three steps would make the biggest difference.
First, combine the strengths of survey and routine data. Each captures only part of the picture. Together they can show where children are missed. But both need strengthening. Surveys are financially fragile, currently sustained by an emergency grant after funding collapsed as part of the wider upheaval in global health financing. Routine data must be redesigned to follow individual children rather than monthly totals, and paired with better estimates of how many children each facility should be reaching.
Second, feed the models what they’re missing. Reaching children in conflict zones, informal settlements or on the move means giving models information on conflict, displacement and remoteness. Some of this information exists, in conflict event databases, displacement tracking, and satellite maps of settlements.
Third, build capacity to generate and use these maps where the burden is high. The maps that matter most will come from people in the hardest-hit countries: researchers, health workers and the community leaders.
Ann Muthoni Njogu receives funding from the European Union under The Global Health EDCTP3 Joint Undertaking (GH EDCTP3 JU)—Grant Agreement number 101159477. Views and opinions expressed are however, those of the authors only and do not necessarily reflect those of the European Union or GH EDCTP3. Neither the European Union nor the granting authority can be held responsible for them.
The work was funded by the European Union under the Global Health EDCTP3 Joint Undertaking (GH EDCTP3 JU)—Grant Agreement number 101159477. Views and opinions expressed are, however, those of the authors only and do not necessarily reflect those of the European Union or GH EDCTP3. Neither the European Union nor the granting authority can be held responsible for them. Peter Macharia receives funding from Fonds voor Wetenschappelijk Onderzoek (FWO), Belgium, under a Senior Postdoctoral Fellowship (Grant number: 1201925N).
Carlo Federici does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.
Prof. Screw
Oscar Pistorius. Not the first South African with a race problem.
Chasm Face
Droewors – dried sausage [droo-ah-vors] A term used to describe a dried sausage, very similar to biltong or the German bratwurst or mettwurst.