Optimising immunisation equity: an analytical framework for geographical categorisation and prioritisation

¹Global Health Practice Centre of Excellence for Child Survival and Development, UNICEF, Nairobi, Kenya, ²Immunization Systems and Delivery Strengthening, UNICEF, Nairobi, Kenya, ³Health Emergency, Preparedness and Response, Health Section, Programme Group, UNICEF, Nairobi, Kenya, ⁴Global Health Expertise, UNICEF, Lagos, Nigeria, ⁵Food Security and Agriculture Sector, WFP country office, Beirut, Lebanon

^&Corresponding author
Mohammed Diaaeldin Omer, Global Health Practice Centre of Excellence for Child Survival and Development, UNICEF, Nairobi, Kenya

Introduction    

Globally, under-five mortality declined substantially from 77 to 37 deaths per 1,000 live births between 2000 and 2023, largely due to expanded access to low-cost, life-saving public health interventions, including immunisation. However, progress slowed markedly between 2015 and 2023 [1]. Reductions in child mortality have not been uniform, with persistent disparities across geographic, ethnic, socioeconomic, educational, and fragility contexts, underscoring the need to ensure that all children, regardless of who they are or where they live, have an equal chance of survival [2]. Since its establishment, the Expanded Programme on Immunisation has made significant contributions to child mortality reduction and is widely regarded as one of the most impactful public health programmes [3]. Despite this, global coverage of the first dose of diphtheria-tetanus-pertussis vaccine (DTP1 ) stagnated between 2011 and 2015 and began to decline in 2019. The COVID-19 pandemic further reversed progress by exposing and exacerbating health system gaps, including through the unintended effects of public health and social measures [4]. Recovery from this setback has required renewed attention to equity, consistent with Immunisation Agenda 2030 (IA2030), which prioritises equitable access and protection of vulnerable populations [5-7].

Equity in immunisation refers to ensuring fair opportunities for vaccination irrespective of socioeconomic status, race, ethnicity, geographic location, or other social determinants of health [5,7,8]. Achieving equity requires addressing structural barriers and tailoring services to population needs rather than applying blanket approaches [9-11]. For operational purposes, immunisation programmes tend to prioritise intervention geographies based on the absolute number of unvaccinated children, guiding resource allocation toward high-population areas. While this can improve aggregate coverage, it may obscure relative needs, reinforce inequities, and neglect smaller, often more vulnerable populations. Growing consensus highlights the need for more nuanced prioritisation criteria to ensure that immunisation programmes are both effective and equitable [5,7,8,12,13]. Population size often dominates prioritisation decisions regardless of coverage levels. For example, a country with DTP1 coverage of 90% and a birth cohort of one million may be prioritised over one with 65% coverage and a birth cohort of 250,000, despite much higher relative risk among children in the latter.

This article proposes a novel analytical framework for geographic categorisation and prioritisation to advance equity-informed immunisation strategies. Incorporating both zero-dose prevalence and absolute numbers of unvaccinated children, the article outlines the conceptual basis, methodology, and applications of the analytical framework. While developed for immunisation, the framework is adaptable to other health programmes requiring equity-driven prioritisation. Its operational feasibility is demonstrated using 2024 WHO/UNICEF Estimates of National Immunisation Coverage data for 194 countries [14].

Methods     

Conceptual foundations and the framework's hypothesis: this analytical framework is rooted in equity principles, particularly distributive justice, which prioritises fair resource allocation over purely utilitarian approaches aimed at maximising aggregate outcomes [13,15]. Distributive justice in health resource allocation requires prioritising populations facing relative disadvantages to avoid reinforcing inequities [16]. Whereas absolute need reflects the overall burden based on the total number of affected individuals, relative need captures the proportion affected, signalling systemic under-service. An equity-oriented approach combines both measures [scale and intensity] to ensure interventions address not only the scale of problems but also their distribution, reducing the risk of missing marginalised groups [9-11]. Building on the equity principle, our analytical framework integrates two measures to enhance equity in immunisation programs: prevalence of unvaccinated individuals (intensity) and the absolute number of unvaccinated individuals (scale). To illustrate the framework, this paper used two immunisation outcome measures: (i) the prevalence of zero-dose children and (ii) the absolute number of zero-dose children. The framework posits that integrating these variables enables the construction of analytical categories that systematically classify geographic locations or units according to their joint distribution rather than their individual characteristics. The bi-dimensional categorisation allows for a more nuanced interpretation of variation by capturing the combined influence of the two variables. Furthermore, it enhances geographic prioritisation by identifying and ranking locations based on the magnitude of the joint effect of both absolute burden and relative disadvantage. Taken together, the categorisation and ranking processes provide a logical and operational structure to inform program design, strategic decision-making, and equitable resource allocation.

Variables: to introduce and illustrate the analytical framework, two independent variables are employed: 1) The absolute number of unvaccinated children for the first dose of Diphtheria, Tetanus toxoid and Pertussis-containing vaccine, DTP1 (zero-dose number), and 2) The prevalence of unvaccinated children for DTP1 (zero-dose prevalence). The zero-dose number is calculated by subtracting the annual number of children who received the first dose of the DTP vaccine in a given geographical area from the total number of surviving children in the same year. Similarly, the zero-dose prevalence is derived by subtracting the observed proportion of DTP1 coverage from 100 per cent. These variables represent the scale and intensity constructs, respectively, that operationally inform and influence immunisation programming at different levels [5,6].

Analytical geographical categorisation: to establish analytical geographical categorisation, a bivariate analysis with dichotomisation was employed. Zero-dose prevalence and zero-dose number were used as the X and Y axes, respectively. Each variable was dichotomised using a predefined cut-off point, resulting in a 2×2 cross-tabulation that illustrates the distribution of geographical units under analysis, based on their observed data values. Four options are proposed to determine the cut-off points across both axes:

Median-based option: this option uses the median values of both variables to define the cut-off points. It typically produces four analytical quadrants with a relatively balanced distribution of units.

Mean-based option: this option relies on the mean values of both variables. While straightforward, it is sensitive to outliers and heterogeneity, which may distort the distribution across quadrants.

Purposive threshold option: this option suggests applying purposive cut-off points for both variables. It offers flexibility to account for factors such as local contexts, policy objectives, and alignment with specific standards or targets.

Combined option: this option suggests using a combination of the above options (e.g. applying the mean for the zero-dose number and the median or a purposive threshold for zero-dose prevalence). This hybrid approach offers enhanced flexibility when developing categorical analyses that must satisfy multiple analytical or policy criteria.

To ensure that all geographical units are assigned unambiguously to a specific category, and to avoid instances where observed values lie exactly on the cut-off lines, it is advisable to subtract a minimal fractional value from the selected cut-off point when an observed value is equal to that threshold. This adjustment enhances clarity and consistency in interpretation. In this paper, we used the median-based and the purposive threshold options as illustrative examples for establishing geographical categorisation.

Geographical ranking: the analytical framework incorporates a mathematical formula designed to generate a geographical ranking of units based on two measures representing the scale and intensity of the variable of interest - absolute number and prevalence. These measures are combined into a single composite score for each geographical unit, from which a descending ranking is generated, such that higher composite scores correspond to higher ranks. The higher ranks indicate greater relative burden and higher programmatic priority. To demonstrate the method, zero-dose number and zero-dose prevalence are employed as measures associated with DTP1 vaccination.

A geographical ranking of 194 countries whose immunisation data were available to WHO and UNICEF for the year 2024 was generated based on their computed composite scores. The analysis utilised data extracted from the 2024 WUENIC [14]. Country-level estimates of surviving infants for 2024 were obtained from the World Population Prospects 2024 [17]. A standard DTP1 coverage threshold was applied against observed coverage values to enable a weighted linear combination of normalised variables. A detailed description of the formula is provided under the results section. In addition, we generated a separate geographical ranking of the same 194 countries based exclusively on the absolute number of zero-dose children in the 2024 birth cohort. Countries were ranked in descending order, such that those with the highest absolute numbers of zero-dose children ranked highest. We compared the two rankings and quantified the differences in country positions between the two methods.

Results     

The analytical framework introduces a new methodology for conducting pro-equity geographical prioritisation for immunisation programs by providing a comparative analytical approach for geographical categorisation and geographical ranking.

Analytical geographical categorisation: Figure 1 illustrates the analytical geographical categories and the distribution of 61 geographical units using two approaches for determining cut-off points across the two independent variables: zero-dose prevalence and the absolute number of zero-dose children. The median-based and purposive-threshold approaches are presented to demonstrate how geographical units are distributed across the analytical categories under each method for determining the cut-off point. Using these cut-off points, four distinct analytical geographical categories were established:

Category 1: high zero-dose prevalence and high zero-dose number (high intensity and large scale).

Category 2: high zero-dose prevalence and low zero-dose number (high intensity and low scale).

Category 3: low zero-dose prevalence and high zero-dose number (low intensity and large scale).

Category 4: low zero-dose prevalence and low zero-dose number (low intensity and low scale).

The analytical categories are described according to their defining characteristics to facilitate systematic comparison. General technical considerations are provided to guide and inform evidence-based programmatic decision-making. Table 1 provides a descriptive overview of the analytical categories and a summary of general technical considerations.

Geographical ranking: to support a structured geographical ranking, a mathematical formula was developed based on the analytical framework's concept and equity principles:

Where:

Composite Score_x = The computed composite score for geographic unit (x), representing its relative priority based on the integrated burden (absolute number and reversed prevalence of the variable of interest).

N_x = Observed value of absolute number for the variable of interest in geographical unit (x) during a specified period.

P_x = Observed value of reversed prevalence for the variable of interest in geographic unit (x) during the specified period (Reversed prevalence typically refers to 1 − observed prevalence, or an equivalent transformation that reflects the intensity of the gap.).

C_x = Observed coverage for the variable of interest in geographic unit (x) during a specified period.

SC = A standardized coverage threshold of the variable of interest.

In this formulation, the ratio (SC/C_x) adjusts each unit's score according to its deviation from the standardized performance target. Squaring this ratio amplifies disparities in coverage, ensuring that units with substantial performance gaps are assigned proportionally higher composite scores, consistent with pro-equity prioritisation principles.

For the geographical ranking demonstrated in this paper, we applied the analytical framework using annual DTP1 coverage and the absolute number of DTP1 zero-dose children as the variables of interest. DTP1 zero-dose prevalence was derived as the inverse of DTP1 coverage and defined as the proportion of children who did not receive DTP1. The framework was operationalised using the 2024 WUENIC dataset comprising 194 countries, demonstrating its practical application for immunisation programme prioritisation. The specific formula used in this analysis is presented below:

Where:

Composite Score_x = Composite Score of geographic unit (x).

N_x = Observed value of annual absolute number of zero-dose children for DTP1 in geographic unit (x).

P_x = Observed value of annual reversed DTP1 coverage (zero-dose prevalence for DTP1) in geographic unit (x).

C_x = Observed value of annual DTP1 coverage in geographic unit (x).

90 = IA2030 minimum recommended annual national coverage of DTP1.

Table in Annex 1 compares the priority rankings of 194 countries derived from the analytical framework with rankings based solely on the absolute number of zero-dose children. The analysis shows that 89 countries moved upward in the framework-based ranking relative to the ranking based solely on the absolute number of zero-dose children, while 97 countries moved downward and 8 countries exhibited no change in position (Upward ranking corresponds to higher ranks, which indicate greater relative burden and higher programmatic priority). Several noteworthy shifts emerge from the comparison. Papua New Guinea and the Central African Republic appear among the top ten countries under the analytical framework ranking, despite not being included within the top 20 when ranked solely by the absolute number of zero-dose children. In contrast, India and Pakistan rank 24^th and 31^st, respectively, under the framework-based ranking, compared with 2^nd and 10^th under the absolute zero-dose ranking. Similarly, Venezuela (Bolivarian Republic of), Bolivia (Plurinational State of), and Somalia rank 13^th, 17^th, and 18^th in the framework-based ranking but occupy positions 22^nd, 34^th, and 21^st, respectively, in the absolute zero-dose ranking. Conversely, China, Brazil, and Cameroon, ranked 16^th, 17^th, and 20^th in the absolute zero-dose approach, fall outside the top 20 under the framework ranking, occupying positions 46^th, 34^th, and 22^nd, respectively.

Figure 2 visualises the magnitude and direction of changes in country ranking positions under the analytical framework relative to rankings based solely on the absolute number of zero-dose children. The observed positional changes range from a maximum upward shift of +43 positions for Suriname to a maximum downward shift of -37 positions for Japan. The comparison of the magnitude and direction of changes in country rankings between the two approaches revealed substantial differences. Under the framework-based approach, 44 countries - predominantly those with smaller populations - moved upward by more than 10 ranking positions, including Suriname (+43), Estonia (+40), Sao Tome and Príncipe (+39), Gabon (+28), and Guinea-Bissau (+27). In contrast, 38 countries declined by more than 10 positions, including several large-population and high-income countries such as India(-22), China (-30), the United States of America (-32), and Japan (-37). Additionally, 37 countries experienced minimal or no change in ranking, with shifts ranging from 0 to ±2 positions. A further 32 countries moved upward by 3-10 positions, whereas 43 countries shifted downward by 3-10 positions.

Discussion     

We introduce a pro-equity analytical framework for geographical categorisation and prioritisation that responds to the persistent need to translate equity principles into operational immunisation decision-making. The framework integrates problem scale, represented by the absolute number of zero-dose children, and problem intensity, represented by zero-dose prevalence. Considering these dimensions jointly helps avoid two common limitations of single-indicator prioritisation: overlooking smaller populations with disproportionately high relative need when decisions are based solely on absolute burden, and underestimating large numbers of missed children when decisions are based only on prevalence. The framework supports geographical categorisation and prioritisation, differentiated intervention design, resource allocation, and comparative performance monitoring.

The framework's dual-lens structure provides a practical response to a longstanding question in global immunisation: should priority be given to countries with the largest numbers of unvaccinated children or those with the highest prevalence of unvaccinated children? By integrating both dimensions, the framework offers a more equity-oriented basis for decision-making.

The categorisation component provides a complementary programmatic lens and suggests analytical descriptions. Geographical units with both high zero-dose prevalence and high zero-dose numbers require comprehensive responses that address widespread access barriers and systemic programme weaknesses. Units with high prevalence but low absolute numbers may be marginalised - for example, remote or conflict-affected areas, or areas dominated by mobile, vulnerable, and otherwise underserved populations. The needs in such areas may be obscured by their small population size. Units with low prevalence but high absolute numbers require sustained high coverage alongside targeted interventions tailored to the needs of missed children. Units with both low prevalence and low absolute numbers remain important for maintaining gains, detecting emerging gaps, and documenting practices that may be transferable to other settings. These category profiles, presented in Table 1, are indicative and should be interpreted considering local context, health-system capacity, population characteristics, and other relevant considerations.

Determining appropriate cut-off points for generating geographic categories using the analytical framework requires careful consideration of contextual factors, the intended purpose of the analysis, data quality, and the degree of heterogeneity across geographic units. While annual estimates of the independent variables may be used to define cut-off points, multi-year averages offer a more stable measure of persistent inequities and are better suited to guiding medium-term programmatic decisions. Nevertheless, averaging across years can mask recent changes in performance. Future applications of the framework should therefore select the analytical time window according to the intended decision-making objective and, where feasible, conduct sensitivity analyses to evaluate the robustness of priority classifications and priority rankings.

Application of the framework to 2024 WUENIC data illustrates its practical implications for priority ranking. Country rankings produced by the analytical framework differed markedly from those based solely on the absolute number of zero-dose children, resulting in substantial changes in country positions (Figure 2). These differences underscore the importance of examining the contextual, programmatic, and methodological factors that may contribute to the observed changes in country rankings. The findings also have significant implications for priority-setting and resource allocation, highlighting the potential value of this analytical approach in informing immunisation strategies and broader health programme planning.

The notable shifts in country rankings are primarily driven by the inclusion of zero-dose prevalence as an important determinant of geographical prioritisation. An upward shift in ranking indicates comparatively poorer immunisation programme performance, whereas a downward shift suggests relatively stronger performance. Poor performance may result from a range of adverse contextual factors, including humanitarian crises, fragile health systems, economic shocks, and chronic underinvestment, among others. For example, several countries experiencing such conditions, including Papua New Guinea, the Central African Republic, and Timor-Leste, demonstrated substantial upward movement in the framework-based ranking, reflecting disproportionately high levels of programmatic vulnerability despite their smaller population sizes. In contrast, several economically stronger and highly populated countries, including China, India, the United States, France, Germany, and Japan, moved downward in the rankings. This downward movement reflects relatively strong immunisation programme performance, with higher DTP1 coverage mitigating the influence of large absolute numbers of zero-dose children.

Although immunisation indicators are used to demonstrate the framework, the underlying approach is adaptable to other public health programmes in which both the number and proportion of affected individuals are relevant to equitable priority setting. The framework can be applied at global, regional, national, and subnational levels, provided that the selected indicators, thresholds, and weighting approaches are aligned with the purpose of the analysis and the quality of available data. Applying the framework at the subnational level requires careful consideration, as administrative coverage data are often affected by inaccuracies in population estimates and reporting quality. These limitations may bias estimates of zero-dose prevalence and burden, potentially influencing categorisation and prioritisation. Data quality should therefore be assessed before applying the framework to subnational analyses. The framework is intended to complement, rather than replace, existing immunisation planning and analytical approaches. As a structured decision-support tool, its outputs should be interpreted within context and complemented by evidence on the social, political, economic, humanitarian, and epidemiological factors that influence programme performance and priority setting.

Conventional burden-based rankings identify geographical areas with the largest numbers of missed children, while coverage-based assessments highlight areas with relatively low immunisation performance. Other approaches, including coverage and equity assessments, programme reviews, zero-dose mapping, and bottleneck analyses, provide important insights into underserved populations and barriers to immunisation access. The added value of the proposed framework is its integration of both absolute and relative need into categorical and ranked outputs, providing an additional tool for prioritisation. For example, whereas immunisation coverage and equity assessments reveal rural-urban disparities at the subnational level, the framework offers a complementary metric for geographical classification beyond the rural-urban classification, applicable at subnational, national, regional and global levels. Its output should therefore be interpreted alongside, rather than in place of, other analytical approaches. To support implementation, an operational electronic toolkit has been developed to facilitate analysis, document analytical assumptions, and enable periodic updates as new data become available.

Limitations: the framework is sensitive to data quality and may generate inaccurate classifications or rankings when applied to incomplete or unreliable data, particularly at the subnational level. Data quality assessment and validation are therefore important prerequisites for its application.

This study used DTP1 coverage and the estimated number of zero-dose children as demonstration variables. Although DTP1 is a widely accepted proxy for access to immunisation services, it does not reflect vaccination completion, timeliness, or broader life-course immunisation needs. However, the framework can be adapted to other indicators as appropriate.

The priority ranking was based on a single year (2024) of WUENIC data and may therefore be influenced by annual fluctuations in programme performance, population estimates, or reporting practices. Multi-year averages may provide more stable estimates for medium-term planning and resource allocation. Furthermore, because WUENIC estimates are released with an approximate eight-month lag, framework outputs should be complemented with recent contextual information when informing short-term decisions.

Finally, the framework does not incorporate all factors relevant to priority setting, such as feasibility, costs, health system capacity, or contextual considerations. Its outputs should therefore be interpreted alongside other evidence and analytical approaches rather than used as a standalone basis for decision-making.

Conclusion     

This study presents a novel analytical framework that supports equity-informed immunisation planning by integrating both the absolute number and prevalence of zero-dose children. Compared with conventional burden-based approaches, the framework offers a more comprehensive assessment of immunisation inequities and a practical method for identifying priority settings and populations. The findings demonstrate the framework's potential to inform prioritisation, resource allocation, differentiated technical support, and performance monitoring. Its flexibility enables application across geographic and administrative levels, making it relevant not only to immunisation programmes but also to broader public health initiatives. By incorporating both absolute and relative measures of need, the framework can strengthen evidence-based, equity-oriented decision-making and support more effective targeting of resources to underserved populations.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Mohammed Diaaeldin Omer, Antoinette Ba Nguz and Ida Marie Ameda: conceptualisation. Mohammed Diaaeldin Omer, Antoinette Ba Nguz, Ida Marie Ameda, Rashad Sheikh, Nasir Yusuf, Ammar Elhagali and Dereje Haile: methodology and validation. Mohammed Diaaeldin Omer, Rashad Sheikh, Nasir Yusuf, Ammar Elhagali and Dereje Haile: visualisation. Mohammed Diaaeldin Omer: original draft writing. Mohammed Diaaeldin Omer, Rashad Sheikh, Nasir Yusuf, Ammar Elhagali, Dereje Haile: data curation. Mohammed Diaaeldin Omer, Rashad Sheikh, Nasir Yusuf, Ammar Elhagali, Dereje Haile, Antoinette Ba Nguz and Ida Marie Ameda: writing, review and editing. Mohammed Diaaeldin Omer: formal analysis. All authors have read and agreed to the final manuscript.

Acknowledgments     

We acknowledge WHO and UNICEF for the WUENIC data, Microsoft Copilot for language editing and generating Figure 1, and Elias Mekuria for improving Annex 1 and graph visualisation.

Table and figures     

Table 1: analytical framework: geographical categories, characteristics, and associated technical considerations
Figure 1: illustration of geographical categorization and distribution of geographical units (D1-D61) under two cut-off point determination approaches: median-based (top) and purposive threshold (bottom)
Figure 2: changes in country ranking positions under the analytical framework compared with absolute number of zero-dose ranking

Supplementary materials     

Annex 1: comparative priority country rankings based on 2024 DTP1 vaccination metrics: analytical framework versus absolute zero-dose number ranking (N=194) (PDF-950KB)

References