Multilevel analysis of predictors of multiple indicators of childhood vaccination in Nigeria

listen audio

Study Justification:
This study aimed to identify and address factors that predict childhood vaccination coverage in Nigeria. The study used data from the 2018 Nigeria Demographic and Health Survey and geospatial data sets to analyze predictors of three vaccination outcomes: receipt of the first dose of Pentavalent vaccine (PENTA1), receipt of the third dose having received the first (PENTA3/1), and receipt of measles vaccine (MV) among children aged 12-35 months. The study aimed to provide insights into the factors influencing vaccination coverage and to inform strategies to increase vaccine uptake.
Highlights:
– The study found that factors associated with vaccination were similar for documented and recall evidence of vaccination.
– Health card/document ownership, receipt of vitamin A, and maternal educational level were significantly associated with each vaccination outcome.
– Socio-economic status, ethnic group, skilled birth attendance, lower travel time to the nearest health facility, and problems seeking health care were significantly associated with both PENTA1 and MV.
– Maternal religion and age were also related to specific vaccination outcomes.
– The study highlighted the importance of addressing socio-demographic and health care access factors to improve vaccination coverage in Nigeria.
Recommendations:
– Addressing inequities in coverage requires addressing factors such as health service quality and community attitudes, in addition to the factors measured in the study.
– Strategies should focus on improving health card/document ownership, promoting vitamin A supplementation, and increasing maternal educational attainment.
– Efforts should be made to improve socio-economic status, access to skilled birth attendance, and reduce travel time to health facilities.
– Community-level interventions should be implemented to address problems seeking health care and promote vaccination.
– Further research is needed to explore the impact of health service quality and community attitudes on vaccination coverage.
Key Role Players:
– National Population Commission (NPC)
– Inner City Fund (ICF) International
– The Measure DHS Program
– Nigeria National Health Research Ethics Committee
– ICF Institutional Review Board
– Ethics and Research Governance, University of Southampton
Cost Items for Planning Recommendations:
– Health card/document production and distribution
– Vitamin A supplementation programs
– Education programs to improve maternal educational attainment
– Infrastructure development to improve access to health facilities
– Training programs for skilled birth attendants
– Community engagement and awareness campaigns
– Research funding for further investigations into health service quality and community attitudes.

The strength of evidence for this abstract is 8 out of 10.
The evidence in the abstract is strong because it is based on a nationally representative cross-sectional survey and utilizes Bayesian multilevel regression models. The study includes a large sample size and considers multiple predictors of childhood vaccination. However, to improve the evidence, the abstract could provide more details on the specific findings and effect sizes of the predictors. Additionally, it would be helpful to include information on the limitations of the study and potential implications for policy and practice.

Background Substantial inequalities exist in childhood vaccination coverage levels. To increase vaccine uptake, factors that predict vaccination coverage in children should be identified and addressed. Methods Using data from the 2018 Nigeria Demographic and Health Survey and geospatial data sets, we fitted Bayesian multilevel binomial and multinomial logistic regression models to analyse independent predictors of three vaccination outcomes: receipt of the first dose of Pentavalent vaccine (containing diphtheria-tetanus-pertussis, Hemophilus influenzae type B and Hepatitis B vaccines) (PENTA1) (n = 6059) and receipt of the third dose having received the first (PENTA3/1) (n = 3937) in children aged 12-23 months, and receipt of measles vaccine (MV) (n = 11839) among children aged 12-35 months. Results Factors associated with vaccination were broadly similar for documented versus recall evidence of vaccination. Based on any evidence of vaccination, we found that health card/document ownership, receipt of vitamin A and maternal educational level were significantly associated with each outcome. Although the coverage of each vaccine dose was higher in urban than rural areas, urban residence was not significant in multivariable analyses that included travel time. Indicators relating to socio-economic status, as well as ethnic group, skilled birth attendance, lower travel time to the nearest health facility and problems seeking health care were significantly associated with both PENTA1 and MV. Maternal religion was related to PENTA1 and PENTA3/1 and maternal age related to MV and PENTA3/1; other significant variables were associated with one outcome each. Substantial residual community level variances in different strata were observed in the fitted models for each outcome. Conclusion Our analysis has highlighted socio-demographic and health care access factors that affect not only beginning but completing the vaccination series in Nigeria. Other factors not measured by the DHS such as health service quality and community attitudes should also be investigated and addressed to tackle inequities in coverage.

The study used data from the nationally representative cross-sectional 2018 Nigeria Demographic and Health Survey (NDHS) which was implemented by the National Population Commission (NPC) with technical assistance provided by Inner City Fund (ICF) International through The Measure DHS Program [23]. Data collection was conducted from 14 August to 29 December 2018 with pre-test conducted from 30 April to 20 May 2018. The survey utilised a stratified two-stage sampling approach. In the first stage, 1389 enumeration areas (EAs) or clusters were selected as the primary sampling units. At the second stage, 40427 households were selected. Stratification was achieved by separating each administrative level 1 area (i.e., the 36 states and the Federal Capital Territory) into urban and rural areas, and samples were selected independently within each stratum. Detailed information on the description of the methods employed in the study is available elsewhere [23]. The primary outcome variables/indicators in this study are receipt of PENTA1 vaccine (n = 6059) and receipt of PENTA3 having received PENTA1 (PENTA3/1—the converse of dropout, n = 3937) among children aged 12–23 months, and receipt of MV (n = 11839) among children aged 12–35 months. For each of the three indicators, we assessed the binomial outcome: any evidence of vaccination versus no evidence of vaccination. For each of PENTA1 and MV, we additionally assessed the multinomial outcome: no evidence of vaccination, card invalid/history evidence of vaccination and card valid vaccination, to assess potential variation in associations that could be caused by misclassifying vaccination status when a verbal history of vaccination is accepted [24] (see Table A in S1 File). Valid and invalid vaccine doses were defined according to WHO guidance [25]. The study considered covariates at the child-, household- and community-levels. The selection of these covariates was informed by literature on the predictors of vaccination coverage or of child health outcomes in general, expert opinion, and availability in the 2018 NDHS or other sources, as detailed in Table A in S1 File [6,11,17,21,26–35]. The study however excluded some pre-selected DHS covariates due to missingness or multicollinearity (see supplementary materials). These variables include preceding birth interval, antenatal and postnatal care, maternal receipt of tetanus toxoid vaccination (these variables had >15% missing cases for MV and were excluded from the analyses to make the results comparable across all the three indicators), maternal decision-making (whether mother decides health care, visits, and purchases)—and region of residence. Also, among similar covariates (e.g., mother’s occupation and employment status), one was selected for inclusion in our model based on the literature and expert opinion. The geospatial covariates retained in this study include travel time to health facility, enhanced vegetation Index (EVI) and livestock density index. Tertiles of the distribution of these covariate were used to allow similar number of observations for each tertile. We present further description about these covariates and their relevance in Fig A and Table B in S1 File [18–22,29,36,37]. Other geospatial covariates considered included distance to conflict locations, maximum number of conflicts, night light intensity, annual aridity index, maximum temperature, annual precipitation/rainfall, proximity to national borders, water, protected areas, and slope [18–20] were excluded from final models due to multicollinearity (as was EVI) or non-significance after adjusting for DHS variables. Cross-tabulations and single-level logistic regression analysis. We tabulated each outcome against each of the selected covariates separately to explore relationships and used Chi-squared tests to determine the significance of the associations. We then fitted frequentist single level simple logistic regression models to obtain the corresponding crude odds ratios (cORs) and associated 95% confidence intervals (CI). These results were later compared with results from the multiple multilevel analyses to determine changes in statistical significance and direction of effects. Multiple multilevel binomial regression analysis (any evidence of vaccination) and interaction effects. We fitted Bayesian multilevel [38,39] binomial regression models to estimate adjusted odds ratios (aORs) and corresponding 95% credible intervals, accounting for the hierarchical structure of the data (child/household, community, and stratum levels) and, intrinsically, the survey design (clustering and stratification) through the last two hierarchies (see Fig B in S1 File). A detailed description of the model is included in the supplementary information (S1 File). We investigated whether child/household covariate effects could be modified by the geospatial covariates by introducing interaction terms between both sets of covariates. To incorporate the interaction terms, we first fitted the main effects model using both DHS and geospatial covariates and then introduced the interactions between selected DHS and geospatial covariates sequentially, retaining only those that were significant in the final model. Multiple multilevel multinomial regression analysis (vaccination according to source of evidence). The study also employed a Bayesian multinomial multivariable multilevel modelling approach to estimate the adjusted relative risk (aRR) and associated 95% credible intervals for covariates significantly associated with PENTA1 and MV using the multinomial outcomes defined previously. No interaction terms were considered for the multinomial analyses due to model complexities and non-convergence challenges. We computed summary measures [30,40] of the amount of residual variation attributable to the hierarchies in the binomial models. These included the variance partitioning coefficient (VPC) which measures residual variation between clusters/communities in different strata; the median odds ratio (MOR) which quantifies residual community level variation in the likelihood of vaccination on the odds ratio scale, and the percent change in variance (PCV) which measures change in residual variation due to the inclusion of covariates in the models [41–45]. Also, although prediction was not the main goal, we evaluated the discriminatory or predictive power of the fitted models using the area under the receiver operating characteristic (AUROC) curve (see supplementary materials for details). All analyses were implemented in Stata version 16 [46], MLwiN version 3.05 [47], and the R programming language version 4.0.3 [48]. Additionally, we used the runmlwin [49] program to run the MLwiN multilevel modelling software from within Stata. We utilized MCMC algorithms with a burn-in length of 1000, a monitoring chain length of 60000, and thinning of 20. Convergence of the MCMC chains was assessed via visual inspection of the trace and autocorrelation plots of the parameters. Ethical approval was obtained from the Nigeria National Health Research Ethics Committee and the ICF Institutional Review Board for the main NDHS [23], and from the Ethics and Research Governance, University of Southampton, United Kingdom. Written informed consent was obtained from all study respondents. However, the data were analysed anonymously in the present study. All methods were performed in accordance with the relevant guidelines and regulations.

Based on the provided information, here are some potential innovations that could improve access to maternal health:

1. Mobile Health (mHealth) Solutions: Develop mobile applications or SMS-based systems to provide pregnant women with information about prenatal care, vaccination schedules, and nearby healthcare facilities. These platforms can also send reminders for appointments and provide access to telemedicine consultations.

2. Community Health Workers: Train and deploy community health workers to remote areas to provide maternal health education, antenatal care, and vaccination services. These workers can also help identify and refer high-risk pregnancies to appropriate healthcare facilities.

3. Telemedicine: Establish telemedicine networks to connect pregnant women in rural or underserved areas with healthcare providers who can offer remote consultations, monitor pregnancies, and provide guidance on vaccination schedules.

4. Transportation Solutions: Improve transportation infrastructure and services to ensure that pregnant women can easily access healthcare facilities for prenatal care, vaccinations, and emergency obstetric care. This could include providing affordable transportation options or implementing mobile clinics in remote areas.

5. Health Facility Upgrades: Invest in upgrading and equipping healthcare facilities, particularly in rural areas, to provide comprehensive maternal health services, including vaccination programs. This could involve improving infrastructure, training healthcare providers, and ensuring the availability of necessary medical supplies and equipment.

6. Public Awareness Campaigns: Launch targeted public awareness campaigns to educate communities about the importance of maternal health, including vaccination. These campaigns can address cultural beliefs, dispel myths, and promote the benefits of vaccination for both mothers and infants.

7. Data-driven Decision Making: Utilize data from national surveys, such as the Nigeria Demographic and Health Survey, to identify areas with low vaccination coverage and develop targeted interventions. This could involve mapping vaccination rates, identifying barriers to access, and implementing strategies to address these challenges.

8. Public-Private Partnerships: Foster collaborations between government agencies, healthcare providers, and private sector organizations to leverage resources and expertise in improving access to maternal health services. This could include initiatives to increase vaccine availability, improve healthcare infrastructure, and expand outreach programs.

It is important to note that these recommendations are based on the provided information and may need to be further tailored and evaluated in the specific context of Nigeria’s maternal health system.
AI Innovations Description
The study titled “Multilevel analysis of predictors of multiple indicators of childhood vaccination in Nigeria” aims to identify factors that predict vaccination coverage in children in order to increase vaccine uptake and address inequalities in childhood vaccination coverage levels. The study used data from the 2018 Nigeria Demographic and Health Survey and geospatial data sets.

The study employed Bayesian multilevel binomial and multinomial logistic regression models to analyze independent predictors of three vaccination outcomes: receipt of the first dose of Pentavalent vaccine (PENTA1), receipt of the third dose having received the first (PENTA3/1), and receipt of measles vaccine (MV) among children aged 12-35 months.

The study found that factors associated with vaccination were similar for documented evidence and recall evidence of vaccination. Health card/document ownership, receipt of vitamin A, and maternal educational level were significantly associated with each outcome. Other factors such as socio-economic status, ethnic group, skilled birth attendance, lower travel time to the nearest health facility, and problems seeking health care were also significantly associated with vaccination outcomes. Maternal religion and age were related to specific vaccination outcomes.

The study also highlighted the importance of addressing factors not measured by the survey, such as health service quality and community attitudes, to tackle inequities in vaccination coverage.

The study utilized a nationally representative cross-sectional survey conducted by the National Population Commission with technical assistance provided by ICF International. The survey employed a stratified two-stage sampling approach, selecting enumeration areas (EAs) or clusters as primary sampling units and households as the second stage. The survey collected data on vaccination status and various covariates at the child-, household-, and community-levels.

The analysis included both frequentist single-level logistic regression models and Bayesian multilevel regression models to estimate odds ratios and adjusted odds ratios, accounting for the hierarchical structure of the data and survey design.

The study provides insights into the predictors of childhood vaccination in Nigeria and suggests the need to address socio-demographic factors, health care access, and other factors to improve vaccination coverage. It also emphasizes the importance of considering geospatial factors in understanding vaccination outcomes.

Overall, the study’s findings can inform the development of innovative strategies to improve access to maternal health by addressing the identified predictors of childhood vaccination coverage in Nigeria. These strategies may include targeted interventions to improve health card/document ownership, increase awareness about the importance of vaccination and vitamin A supplementation, enhance maternal education, and address socio-economic disparities and barriers to seeking health care. Additionally, efforts to improve health service quality and address community attitudes towards vaccination can also contribute to improving access to maternal health.
AI Innovations Methodology
The study you provided focuses on analyzing predictors of childhood vaccination coverage in Nigeria. While the study does not directly address innovations to improve access to maternal health, I can provide you with some potential recommendations based on the findings and briefly describe a methodology to simulate the impact of these recommendations on improving access to maternal health.

Potential recommendations to improve access to maternal health based on the study:

1. Strengthen health card/document ownership: The study found that health card/document ownership was significantly associated with vaccination coverage. Promoting the use and ownership of health cards or documents can help improve access to maternal health services by ensuring that mothers have a record of their vaccinations and other health information.

2. Enhance maternal educational level: The study identified maternal educational level as a significant predictor of vaccination coverage. Investing in maternal education can empower women with knowledge about maternal health and vaccination, leading to improved access to maternal health services.

3. Improve health care access: Factors such as socio-economic status, skilled birth attendance, lower travel time to the nearest health facility, and problems seeking health care were significantly associated with vaccination coverage. Enhancing access to health care services, particularly in rural areas, can contribute to improved access to maternal health services.

Methodology to simulate the impact of recommendations on improving access to maternal health:

1. Define the target population: Determine the specific population for which you want to simulate the impact of the recommendations. This could be based on geographical location, socio-economic status, or any other relevant criteria.

2. Collect baseline data: Gather data on the current status of maternal health access in the target population. This can include information on vaccination coverage, health card/document ownership, maternal educational level, and other relevant factors.

3. Develop a simulation model: Create a simulation model that incorporates the potential recommendations identified earlier. This model should consider the interplay between factors such as health card/document ownership, maternal educational level, and health care access.

4. Input data and run simulations: Input the baseline data into the simulation model and run multiple simulations to assess the impact of the recommendations on improving access to maternal health. The simulations should consider different scenarios and variations in the implementation of the recommendations.

5. Analyze results: Analyze the results of the simulations to determine the potential impact of the recommendations on improving access to maternal health. This can include assessing changes in vaccination coverage, health care utilization, and other relevant indicators.

6. Validate and refine the model: Validate the simulation model by comparing the simulated results with real-world data, if available. Refine the model based on feedback and further insights gained from the analysis.

7. Communicate findings and implement recommendations: Present the findings of the simulation analysis to relevant stakeholders, such as policymakers, healthcare providers, and community organizations. Advocate for the implementation of the recommendations to improve access to maternal health based on the evidence generated from the simulations.

It’s important to note that the methodology described above is a general framework and may need to be adapted based on the specific context and data availability. Additionally, the simulation results should be interpreted with caution and considered alongside other evidence and expert opinions when making decisions about improving access to maternal health.

Partagez ceci :
Facebook
Twitter
LinkedIn
WhatsApp
Email