Background: Spatial analysis methods, including model-based geostatistics (MBG), small-area estimation (SAE), and cluster detection, are increasingly used to map subnational immunisation coverage and identify geographic inequalities in low- and middle-income countries. However, the extent to which these methods capture the multidimensional determinants of immunisation uptake, and whether their outputs inform programme decisions in practice, remains unclear. Methods: We conducted a critical scoping review following the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews guidelines, systematically searching PubMed and Google Scholar for studies applying spatial statistical methods to childhood immunisation coverage or equity. Findings were synthesised using a combination of descriptive summary and thematic and interpretive synthesis. Results: We included 50 studies from the 421 papers identified. Spatial methods have successfully revealed subnational coverage inequalities that national averages obscure, and studies developed in collaboration with national programme teams, integrating routine health system data alongside household surveys, produced the most operationally relevant outputs. However, most studies relied exclusively on survey data with a limited incorporation of supply-side determinants, and few discussed how uncertainty in estimates should constrain downstream use. Although a growing number of studies articulated clear implementation pathways, confirmed programmatic uptake of spatial outputs remained largely undocumented. The emergence of machine learning approaches (8 of 50 studies) offers predictive gains but introduces additional challenges around transparency and quality assurance for governance use. Conclusions: Spatial methods are becoming more frequently used for immunisation but are more likely to contribute to immunisation equity goals when co-produced with programme teams, matched to decision-relevant geographies, and accompanied by transparent documentation of model assumptions and limitations. Future research should prioritise quality frameworks for algorithm-assisted health estimates and systematic evaluation of whether spatial outputs improve decision-making relative to existing data sources.
Allorant et al. (Mon,) studied this question.