Immunization as a mediator of health disparities among children with disabilities in Nigeria: real-world evidence from MICS 2021 using causal mediation analysis

¹Human Services Research Institute, Cambridge, MA, USA, ²Federal Medical Center Nguru, Yobe State, Nigeria

^&Corresponding author
Hussaini Zandam, Human Services Research Institute, Cambridge, MA, USA

Introduction    

Children with disabilities face disproportionate health challenges worldwide, experiencing higher rates of infectious diseases and poorer health outcomes compared to their peers without disabilities [1]. In sub-Saharan Africa, where infectious disease burden remains high and healthcare access limited, these disparities are particularly pronounced [2]. Nigeria, with an estimated 1.2 million children with disabilities, represents a critical context for understanding and addressing these health inequities [3]. Traditional approaches to improving health outcomes for children with disabilities have focused primarily on enhancing healthcare access and treatment services [4]. However, emerging evidence suggests that prevention-oriented interventions may offer greater potential for reducing health disparities while providing superior cost-effectiveness [5]. Immunization programs, which reach approximately 78% of children globally and represent one of the most successful public health interventions, may serve as particularly important prevention platforms for vulnerable populations [6].

The theoretical foundation for immunization programs as prevention pathways extends beyond direct vaccine-preventable disease protection. Immunization contacts provide opportunities for health system engagement, preventive care delivery, health education, and early identification of health risks [7]. For children with disabilities, who often face barriers to routine healthcare access, immunization programs may represent crucial entry points into the healthcare system and sources of comprehensive preventive care [8]. Despite the potential importance of immunization programs as prevention pathways, empirical evidence quantifying their role in mediating health disparities among children with disabilities remains limited. Most existing research has focused on immunization coverage disparities rather than examining how immunization programs influence health outcomes [9]. Understanding the extent to which immunization programs mediate healthcare needs could inform targeted interventions and policy priorities for reducing health inequities.

Nigeria provides an ideal setting for investigating these relationships. The country has substantial immunization infrastructure through its National Primary Health Care Development Agency, yet significant coverage gaps persist, particularly among vulnerable populations [10]. The 2021 Multiple Indicator Cluster Survey (MICS6) provides comprehensive data on both disability status and health outcomes, enabling robust causal inference about immunization-mediated prevention pathways [11]. This study addresses a critical gap in the literature by quantifying the extent to which immunization programs mediate healthcare needs among children with disabilities in Nigeria. Using advanced causal mediation analysis, we examine whether immunization programs serve as prevention pathways that reduce illness burden beyond their direct vaccine-preventable disease effects. Our findings have important implications for prevention-focused strategies and immunization program design in low- and middle-income countries.

Methods     

Study design and data source: we conducted a population-based mediation analysis using data from Nigeria's Multiple Indicator Cluster Survey (MICS6) conducted in 2021 by the National Bureau of Statistics in collaboration with UNICEF [11]. MICS6 employed a two-stage stratified cluster sampling design to ensure national representativeness, with proportional to size sampling of enumeration areas followed by systematic sampling of households within selected areas. The survey achieved a household response rate of 99.1% and included comprehensive modules on child health, immunization, and functioning. Our analysis focused on children aged 2-4 years, as this age group has completed most basic immunization schedules while remaining vulnerable to infectious diseases of interest. The final analytical sample included 11,494 children (173 with disabilities, 11,312 without disabilities) with complete data on key variables.

Disability assessment: disability status was assessed using the UNICEF/Washington Group Child Functioning Module, which has been validated for population-based surveys in low- and middle-income countries [12]. The module assesses functional difficulties across four domains: seeing, hearing, walking, and communication. Children were classified as having a disability if they had “a lot of difficulty” or “cannot do at all” in any domain, consistent with international standards for disability identification in population surveys [13].

Immunization assessment: immunization status was assessed through maternal recall and vaccination card review, where available. We constructed a comprehensive immunization coverage score based on the Nigerian national immunization schedule, including BCG, three doses of DPT, three doses of oral polio vaccine, and measles vaccine. Complete basic immunization was defined as receipt of all scheduled vaccines by age-appropriate timeframes. We also created a continuous immunization score (0-7) to examine dose-response relationships.

Health outcome assessment: our primary outcomes were the two-week incidence of three common childhood illnesses: diarrheal episodes, fever, and acute respiratory infections (ARI). These conditions were selected because they represent major causes of morbidity among children in Nigeria and are potentially preventable through immunization-related mechanisms [14]. Outcomes were assessed through maternal reports using standardized questionnaires with recall periods shown to provide reliable estimates in similar populations [15]. Diarrheal episodes were defined as three or more loose or watery stools per day. Fever was defined as subjective or measured temperature elevation. Acute respiratory infections were defined using a composite measure including cough with fast breathing, difficulty breathing, or chest indrawing, consistent with Integrated Management of Childhood Illness guidelines [16].

Statistical analysis: we compare demographics and socioeconomic characteristics of children with and without disabilities using the chi-square test for categorical and the t-test for continuous variables. We calculated the prevalence for each child illness by disability status and estimated unadjusted and adjusted risk ratios using modified Poisson regressions. We used inverse probability weighting (IPW) to estimate causal mediation effects, accounting for confounding in both the exposure-mediator and mediator-outcome relationships [17]. The IPW approach requires careful estimation of propensity scores to ensure valid causal inference.

Propensity score estimation

Propensity scores were estimated using multivariable logistic regression models predicting immunization coverage based on observed confounders. We developed separate propensity score models for the exposure-mediator relationship (disability status ? immunization coverage) and implemented additional models for sensitivity analyses.

Primary propensity score model: the logistic regression model for immunization coverage included the following covariates: logit(P(Immunization = 1)) = β0 + β1(Disability Status) + β2(Child Age) + β3(Child Sex) + β5(Maternal Age) + β6(Maternal Education) + β7(ECD Attendance) + β8(Household Wealth Quintile) + β9(Urban/Rural Residence) + β10(Geographic Region) β12(Household Size) + β13(Number of Children) + β14(Maternal Employment) + β15(Access to Media) + e

Variable selection process: covariate selection followed a systematic approach combining clinical knowledge, statistical criteria, and causal inference principles. We included all variables that were: (1) potential confounders of the disability-immunization relationship based on subject matter knowledge, (2) predictors of immunization coverage with p<0.20 in univariate analysis, and (3) not potential mediators or colliders based on causal pathway analysis. Model selection used backward elimination with p<0.10 retention criteria, while ensuring all core confounders remained in the final model regardless of statistical significance.

Model diagnostics and validation: propensity score model performance was evaluated using multiple diagnostic approaches. Model discrimination was assessed using the C-statistics (area under the ROC curve), with values >0.70 considered adequate for propensity score applications. Calibration was evaluated using the Hosmer-Lemeshow goodness-of-fit test and calibration plots comparing predicted versus observed immunization rates across deciles of propensity scores. We examined propensity score distributions across disability status groups to ensure adequate overlap and common support. Extreme propensity scores (<0.05 or >0.95) were identified and sensitivity analyses conducted with and without these observations. Balance diagnostics compared standardized mean differences for all covariates before and after propensity score weighting, with differences <0.10 considered acceptable balance.

Inverse probability weighting implementation: IPW weights were calculated as the inverse of the propensity scores: Weight = 1/P(Immunization|Covariates) for immunized children and Weight = 1/(1-P(Immunization|Covariates)) for non-immunized children. To prevent extreme weights from dominating the analysis, we implemented weight stabilization using marginal probabilities and weight truncation at the 1st and 99th percentiles.

Stabilized weights were calculated as: SW = [P(Immunization)] / [P(Immunization|Covariates)] for immunized children and SW = [1-P(Immunization)] / [1-P(Immunization|Covariates)] for non-immunized children, where P(Immunization) represents the marginal probability of immunization in the sample.

Mediation analysis framework

The mediation analysis proceeded in three steps: (1) estimation of the exposure-mediator relationship (disability ? immunization), (2) estimation of the mediator-outcome relationship (immunization ? illness episodes), and (3) calculation of indirect effects and proportion mediated using the product of coefficients method with bootstrap confidence intervals [18].

Exposure-mediator model: we estimated the effect of disability status on immunization coverage using weighted logistic regression with IPW weights derived from the propensity score model described above. This model included disability status as the primary exposure and the same set of confounders used in propensity score estimation.

Mediator-outcome model: the relationship between immunization coverage and illness episodes was estimated using weighted logistic regression models for each health outcome (diarrhea, fever, ARI). These models included immunization coverage as the primary predictor, disability status, and all confounders. IPW weights ensured that the mediator-outcome relationship was estimated free from confounding.

Mediation effect calculation: indirect effects were calculated as the product of coefficients from steps 1 and 2: Indirect Effect = a x β, where a represents the exposure-mediator effect and β represents the mediator-outcome effect. The proportion mediated was calculated as: Proportion Mediated = (Indirect Effect / Total Effect) × 100%.

Bootstrap confidence intervals (1,000 replications) were used to account for uncertainty in both coefficient estimates and their product. The bootstrap procedure maintained the complex survey design by resampling primary sampling units with replacement while preserving survey weights and clustering.

Results     

Sample characteristics

Table 1 and Table 1.1 present children and mother's, and household characteristics by disability status. Children with disabilities were slightly older and had similar sex distribution compared with children without disabilities. Mothers of children with disabilities had lower education levels and were less likely to be employed. Households of children with disabilities were larger, poorer, and children were less likely to attend early childhood programs.

Vaccination coverage

Immunization coverage was significantly lower among children with disabilities across all vaccines (Table 2). Complete basic immunization coverage was 30.9% among children with disabilities compared to 38.4% among children without disabilities (difference: -7.5, 95% CI: -9.4 to -4.2, p=0.003). The largest gaps were observed for DPT3 (-3.4 percentage points) and Polio3 (-3.8 percentage points).

Illness episode

Children with disabilities had higher rates of illness compared with children without disabilities (Table 3). Diarrhea affected 17% of children with disabilities versus 11% of those without, with an adjusted risk ratio of 1.43 (95% CI: 1.17-1.75, p<0.001). Fever occurred in 42% versus 33%, adjusted RR 1.48 (1.22-1.81, p<0.001), and acute respiratory infections in 45% versus 33%, adjusted RR 1.64 (1.30-2.12, p<0.001). All differences remained significant after adjusting for age, sex, maternal education, wealth, residence, and region.

Primary mediation analysis

Table 4 presents the core mediation analysis results. Children with disabilities experienced higher rates of illness compared with children without disabilities. The total excess risk ranged from 52% for diarrhea episodes to 77% for acute respiratory infections. Direct effects accounted for much of this difference, while immunization mediated a substantial portion, with 35% of the excess diarrhea, 41% of fever, and 42% of acute respiratory infections explained by vaccination coverage. The combined analysis of all illness episodes indicated a 113% higher burden among children with disabilities, with 42% mediated through immunization. Both direct and indirect effects were statistically significant across all outcomes.

Discussion     

This study provides robust evidence that children with disabilities in Nigeria experience both lower immunization coverage and higher rates of infectious illnesses compared with children without disabilities, highlighting immunization as a critical mediator of health disparities. Across all vaccines examined, including BCG, DPT1, DPT3, Polio1, Polio3, and measles, children with disabilities consistently had lower coverage, ranging from 29% to 36%, compared with 34% to 41% among children without disabilities. Complete basic vaccination coverage was similarly lower (33% vs. 37%, p = 0.003). These disparities align with global evidence showing that children with disabilities are systematically under-immunized due to multiple barriers, including limited accessibility of healthcare services, caregiver perceptions of vaccine appropriateness, and social stigma [19,20].

The consequences of these immunization gaps are reflected in elevated rates of illness. Adjusted analyses revealed that children with disabilities had a 43% higher risk of diarrhea, 48% higher risk of fever, and 64% higher risk of acute respiratory infections, even after controlling for demographic and socio-economic covariates. These results are consistent with findings from studies in low- and middle-income countries, where children with disabilities face increased susceptibility to infectious diseases due to both structural vulnerabilities and reduced access to preventive healthcare [21,22]. Mediation analysis demonstrated that immunization accounted for 35-42% of the excess burden of illness, indicating that vaccination represents a substantial, though partial, protective pathway. The highest mediated proportion was observed for acute respiratory infections (42%), consistent with evidence that routine vaccines such as measles and DPT confer significant protection against respiratory pathogens [23,24]. These findings emphasize that while immunization alone cannot fully eliminate disparities, it is a quantifiable and modifiable intervention capable of reducing excess disease burden among children with disabilities. The persistence of significant direct effects suggests that other social determinants, such as malnutrition, environmental exposures, and healthcare accessibility, also contribute to the elevated risk, as reported in similar studies [25].

Our results reinforce the importance of disability-inclusive immunization strategies. The partial mediation by vaccination suggests that interventions targeting vaccine uptake through outreach programs, caregiver education, and health system strengthening could substantially reduce illness disparities. These findings are consistent with WHO's recommendations for integrating disability considerations into national immunization programs and with evidence from targeted interventions in Bangladesh and Kenya, which demonstrated improvements in vaccine coverage and health outcomes among children with disabilities when barriers to access were explicitly addressed [19,20].

This study has several limitations. The cross-sectional design of the MICS survey limits causal inference; while mediation analysis quantifies indirect effects, unmeasured confounding cannot be ruled out. Disability status was caregiver-reported using the Child Functioning Module, which may be subject to misclassification. Illness episodes were self-reported, introducing potential recall bias. Moreover, contextual factors such as local healthcare infrastructure, caregiver knowledge, and social support were not captured but are likely important determinants of both vaccination and illness risk. Despite these limitations, the study provides robust population-level evidence quantifying the mediating role of immunization in health disparities for children with disabilities.

Conclusion     

The findings demonstrate that children with disabilities in Nigeria face a dual disadvantage: lower immunization coverage and higher illness burden. Immunization serves as a critical protective pathway, mediating a substantial portion of excess risk, and underscores the potential impact of targeted vaccination interventions. Future research should examine longitudinal and interventional designs to clarify causal pathways and evaluate the effectiveness of disability-inclusive immunization programs. Integrating immunization efforts with complementary interventions, including improved nutrition, hygiene, and primary healthcare access, is essential to reduce disparities further.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Hussaini Zandam: conceptualized the study, designed the methodology, conducted the analysis, interpreted results, and drafted the manuscript. Abubakar Umar Abdurrahman: Contributed to study design, data interpretation, and critical revision of the manuscript. All authors read and approved of the final manuscript.

Tables     

Table 1: description of children and maternal characteristics by disability status of children

Table 1.1: description of household characteristics of children by disability status

Table 2: immunization coverage patterns by disability status and vaccine type

Table 3: rates, unadjusted and adjusted risk ratios of illness episodes by disability status

Table 4: mediation analysis by type of illness

References