Modeling and presentation of vaccination coverage estimates using data from household surveys

Improving the spatial and temporal resolution of burden of disease measures with Bayesian models

This paper contributes to the field by addressing the critical issue of enhancing the spatial and temporal resolution of health data. Although Bayesian methods are frequently employed to address this challenge in various disciplines, the application of Bayesian spatio-temporal models to burden of disease (BOD) studies remains limited. Our novelty lies in the exploration of two existing Bayesian models that we show to be applicable to a wide range of BOD data, including mortality and prevalence, thereby providing evidence to support the adoption of Bayesian modeling in full BOD studies in the future. We illustrate the benefits of Bayesian modeling with an Australian case study involving asthma and coronary heart disease. Our results showcase the effectiveness of Bayesian approaches in increasing the number of small areas for which results are available and improving the reliability and stability of the results compared to using data directly from surveys or administrative sources.

Smoothed model-assisted small area estimation of proportions

In countries where population census data are limited, generating accurate subnational estimates of health and demographic indicators is challenging. Existing model-based geostatistical methods leverage covariate information and spatial smoothing to reduce the variability of estimates but often ignore the survey design, while traditional small area estimation approaches may not incorporate both unit-level covariate information and spatial smoothing in a design consistent way. We propose a smoothed model-assisted estimator that accounts for survey design and leverages both unit-level covariates and spatial smoothing. Under certain regularity assumptions, this estimator is both design consistent and model consistent. We compare it with existing design-based and model-based estimators using real and simulated data.

Mapping the prevalence of cancer risk factors at the small area level in Australia

BACKGROUND

Cancer is a significant health issue globally and it is well known that cancer risk varies geographically. However in many countries there are no small area-level data on cancer risk factors with high resolution and complete reach, which hinders the development of targeted prevention strategies.

METHODS

Using Australia as a case study, the 2017-2018 National Health Survey was used to generate prevalence estimates for 2221 small areas across Australia for eight cancer risk factor measures covering smoking, alcohol, physical activity, diet and weight. Utilising a recently developed Bayesian two-stage small area estimation methodology, the model incorporated survey-only covariates, spatial smoothing and hierarchical modelling techniques, along with a vast array of small area-level auxiliary data, including census, remoteness, and socioeconomic data. The models borrowed strength from previously published cancer risk estimates provided by the Social Health Atlases of Australia. Estimates were internally and externally validated.

RESULTS

We illustrated that in 2017-2018 health behaviours across Australia exhibited more spatial disparities than previously realised by improving the reach and resolution of formerly published cancer risk factors. The derived estimates revealed higher prevalence of unhealthy behaviours in more remote areas, and areas of lower socioeconomic status; a trend that aligned well with previous work.

CONCLUSIONS

Our study addresses the gaps in small area level cancer risk factor estimates in Australia. The new estimates provide improved spatial resolution and reach and will enable more targeted cancer prevention strategies at the small area level. Furthermore, by including the results in the next release of the Australian Cancer Atlas, which currently provides small area level estimates of cancer incidence and relative survival, this work will help to provide a more comprehensive picture of cancer in Australia by supporting policy makers, researchers, and the general public in understanding the spatial distribution of cancer risk factors. The methodology applied in this work is generalisable to other small area estimation applications and has been shown to perform well when the survey data are sparse.

Mapping the timeliness of routine childhood vaccination in The Gambia: A spatial modelling study

INTRODUCTION

Timeliness of routine vaccination shapes childhood infection risk and thus is an important public health metric. Estimates of indicators of the timeliness of vaccination are usually produced at the national or regional level, which may conceal epidemiologically relevant local heterogeneities and makeitdifficultto identify pockets of vulnerabilities that could benefit from targeted interventions. Here, we demonstrate the utility of geospatial modelling techniques in generating high-resolution maps of the prevalence of delayed childhood vaccination in The Gambia. To guide local immunisation policy and prioritize key interventions, we also identified the districts with a combination of high estimated prevalence and a significant population of affected infants.

METHODS

We used the birth dose of the hepatitis-B vaccine (HepB0), third-dose of the pentavalent vaccine (PENTA3), and the first dose of measles-containing vaccine (MCV1) as examples to map delayed vaccination nationally at a resolution of 1 × 1-km2 pixel. We utilized cluster-level childhood vaccination data from The Gambia 2019-20 Demographic and Health Survey. We adopted a fully Bayesian geostatistical model incorporating publicly available geospatial covariates to aid predictive accuracy. The model was implemented using the integrated nested Laplace approximation-stochastic partial differential equation (INLA-SPDE) approach.

RESULTS

We found significant subnational heterogeneity in delayed HepB0, PENTA3 and MCV1 vaccinations. Specificdistricts in the central and eastern regions of The Gambia consistentlyexhibited the highest prevalence of delayed vaccination, while the coastal districts showed alower prevalence forallthree vaccines. We also found that districts in the eastern, central, as well as in coastal parts of The Gambia had a combination of high estimated prevalence of delayed HepB0, PENTA3 and MCV1 and a significant population of affected infants.

CONCLUSIONS

Our approach provides decision-makers with a valuable tool to better understand local patterns of untimely childhood vaccination and identify districts where strengthening vaccine delivery systems could have the greatest impact.

Timeliness of routine childhood vaccination among 12-35 months old children in The Gambia: Analysis of national immunisation survey data, 2019-2020

The Gambia's routine childhood vaccination programme is highly successful, however, many vaccinations are delayed, with potential implications for disease outbreaks. We adopted a multi-dimensional approach to determine the timeliness of vaccination (i.e., timely, early, delayed, and untimely interval vaccination). We utilised data for 3,248 children from The Gambia 2019-2020 Demographic and Health Survey. Nine tracer vaccines administered at birth and at two, three, four, and nine months of life were included. Timeliness was defined according to the recommended national vaccination windows and reported as both categorical and continuous variables. Routine coverage was high (above 90%), but also a high rate of untimely vaccination. First-dose pentavalent vaccine (PENTA1) and oral polio vaccine (OPV1) had the highest timely coverage that ranged from 71.8% (95% CI = 68.7-74.8%) to 74.4% (95% CI = 71.7-77.1%). Delayed vaccination was the commonest dimension of untimely vaccination and ranged from 17.5% (95% CI = 14.5-20.4%) to 91.1% (95% CI = 88.9-93.4%), with median delays ranging from 11 days (IQR = 5, 19.5 days) to 28 days (IQR = 11, 57 days) across all vaccines. The birth-dose of Hepatitis B vaccine had the highest delay and this was more common in the 24-35 months age group (91.1% [95% CI = 88.9-93.4%], median delays = 17 days [IQR = 10, 28 days]) compared to the 12-23 months age-group (84.9% [95% CI = 81.9-87.9%], median delays = 16 days [IQR = 9, 26 days]). Early vaccination was the least common and ranged from 4.9% (95% CI = 3.2-6.7%) to 10.7% (95% CI = 8.3-13.1%) for all vaccines. The Gambia's childhood immunization system requires urgent implementation of effective strategies to reduce untimely vaccination in order to optimize its quality, even though it already has impressive coverage rates.

Mapping the distribution of zero-dose children to assess the performance of vaccine delivery strategies and their relationships with measles incidence in Nigeria

Geographically precise identification and targeting of populations at risk of vaccine-preventable diseases has gained renewed attention within the global health community over the last few years. District level estimates of vaccination coverage and corresponding zero-dose prevalence constitute a potentially useful evidence base to evaluate the performance of vaccination strategies. These estimates are also valuable for identifying missed communities, hence enabling targeted interventions and better resource allocation. Here, we fit Bayesian geostatistical models to map the routine coverage of the first doses of diphtheria-tetanus-pertussis vaccine (DTP1) and measles-containing vaccine (MCV1) and corresponding zero-dose estimates in Nigeria at 1x1 km resolution and the district level using geospatial data sets. We also map MCV1 coverage before and after the 2019 measles vaccination campaign in the northern states to further explore variations in routine vaccine coverage and to evaluate the effectiveness of both routine immunization (RI) and campaigns in reaching zero-dose children. Additionally, we map the spatial distributions of reported measles cases during 2018 to 2020 and explore their relationships with MCV zero-dose prevalence to highlight the public health implications of varying performance of vaccination strategies across the country. Our analysis revealed strong similarities between the spatial distributions of DTP and MCV zero dose prevalence, with districts with the highest prevalence concentrated mostly in the northwest and the northeast, but also in other areas such as Lagos state and the Federal Capital Territory. Although the 2019 campaign reduced MCV zero-dose prevalence substantially in the north, pockets of vulnerabilities remained in areas that had among the highest prevalence prior to the campaign. Importantly, we found strong correlations between measles case counts and MCV RI zero-dose estimates, which provides a strong indication that measles incidence in the country is mostly affected by RI coverage. Our analyses reveal an urgent and highly significant need to strengthen the country's RI program as a longer-term measure for disease control, whilst ensuring effective campaigns in the short term.

Conditional probability and ratio-based approaches for mapping the coverage of multi-dose vaccines

Many vaccines are often administered in multiple doses to boost their effectiveness. In the case of childhood vaccines, the coverage maps of the doses and the differences between these often constitute an evidence base to guide investments in improving access to vaccination services and health system performance in low and middle-income countries. A major problem often encountered when mapping the coverage of multi-dose vaccines is the need to ensure that the coverage maps decrease monotonically with successive doses. That is, for doses i $$ i $$ and j $$ j $$ , i < j ⇒ p i ( s ) ≥ p j ( s ) $$ i<j\Rightarrow {p}_i\left(\boldsymbol{s}\right)\ge {p}_j\left(\boldsymbol{s}\right) $$ , where p i ( s ) $$ {p}_i\left(\boldsymbol{s}\right) $$ is the coverage of dose i $$ i $$ at spatial location s $$ \boldsymbol{s} $$ . Here, we explore conditional probability (CP) and ratio-based (RB) approaches for mapping p i ( s ) $$ {p}_i\left(\boldsymbol{s}\right) $$ , embedded within a binomial geostatistical modeling framework, to address this problem. The fully Bayesian model is implemented using the INLA and SPDE approaches. Using a simulation study, we find that both approaches perform comparably for out-of-sample estimation under varying point-level sample size distributions. We apply the methodology to map the coverage of the three doses of diphtheria-tetanus-pertussis vaccine using data from the 2018 Nigeria Demographic and Health Survey. The coverage maps produced using both approaches are almost indistinguishable, although the CP approach yielded more precise estimates on average in this application. We also provide estimates of zero-dose children and the dropout rates between the doses. The methodology is straightforward to implement and can be applied to other vaccines and geographical contexts.

Model building and assessment of the impact of covariates for disease prevalence mapping in low-resource settings: to explain and to predict

This paper provides statistical guidance on the development and application of model-based geostatistical methods for disease prevalence mapping. We illustrate the different stages of the analysis, from exploratory analysis to spatial prediction of prevalence, through a case study on malaria mapping in Tanzania. Throughout the paper, we distinguish between predictive modelling, whose main focus is on maximizing the predictive accuracy of the model, and explanatory modelling, where greater emphasis is placed on understanding the relationships between the health outcome and risk factors. We demonstrate that these two paradigms can result in different modelling choices. We also propose a simple approach for detecting over-fitting based on inspection of the correlation matrix of the estimators of the regression coefficients. To enhance the interpretability of geostatistical models, we introduce the concept of domain effects in order to assist variable selection and model validation. The statistical ideas and principles illustrated here in the specific context of disease prevalence mapping are more widely applicable to any regression model for the analysis of epidemiological outcomes but are particularly relevant to geostatistical models, for which the separation between fixed and random effects can be ambiguous.

District-level estimation of vaccination coverage: Discrete vs continuous spatial models

Health and development indicators (HDIs) such as vaccination coverage are regularly measured in many low- and middle-income countries using household surveys, often due to the unreliability or incompleteness of routine data collection systems. Recently, the development of model-based approaches for producing subnational estimates of HDIs using survey data, particularly cluster-level data, has been an active area of research. This is mostly driven by the increasing demand for estimates at certain administrative levels, for example, districts, at which many development goals are set and evaluated. In this study, we explore spatial modeling approaches for producing district-level estimates of vaccination coverage. Specifically, we compare discrete spatial smoothing models which directly model district-level data with continuous Gaussian process (GP) models that utilize geolocated cluster-level data. We adopt a fully Bayesian framework, implemented using the INLA and SPDE approaches. We compare the predictive performance of the models by analyzing vaccination coverage using data from two Demographic and Health Surveys (DHS), namely the 2014 Kenya DHS and the 2015-16 Malawi DHS. We find that the continuous GP models performed well, offering a credible alternative to traditional discrete spatial smoothing models. Our analysis also revealed that accounting for between-cluster variation in the continuous GP models did not have any real effect on the district-level estimates. Our results provide guidance to practitioners on the reliability of these model-based approaches for producing estimates of vaccination coverage and other HDIs.

Space-time smoothing of mortality estimates in children aged 5-14 in Sub-Saharan Africa

To meet the SDG requirement of spatial disaggregation of indicators, several methods have been developed to generate estimates of under-five mortality at the sub-national level. The reliability of sub-national mortality estimates in children aged 5-14 with the available survey data has not been evaluated so far. We generate Admin-1 sub-national estimates of the risk of dying in children aged less than five (5q0) and those aged 5 to 14 years old (10q5). We use 96 Demographic and Health Surveys (DHS) in 20 Sub-Saharan countries having at least 3 surveys designed to be representative at a sub-national level. The estimates account for the complex sample design of DHS and HIV-related biases in young children. A Bayesian space-time model previously developed for under-five mortality is used to smooth estimates across space and time in both age groups to reduce problems associated with data sparsity. The posterior distributions of the probability 10q5 are used to compute coefficients of variation and assess precision. Sufficiently precise estimates are retained to study the sub-national relationship between age-specific mortality rates (5q0 and 10q5), accounting for uncertainty in sub-national levels. Out of 1,132 space-time estimates, 62.3% are considered sufficiently precise with high heterogeneity across countries. Across all periods, sub-national estimates of mortality in children aged 0-4 are highly correlated with those in older children and young adolescents but this correlation is largely driven by the mortality decline. Within specific periods of time, it is often impossible to assess the relationship between mortality rates in the two age groups at the sub-national level, except in Nigeria, Ethiopia, Cameroon, Senegal and Zambia. As increased attention is devoted to survival after age 5, more research is needed to ensure that sub-national areas with specific interventions required for older children can be correctly identified.