Dataset Information

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

ABSTRACT: It is becoming increasingly popular to produce high-resolution maps of vaccination coverage by fitting Bayesian geostatistical models to data from household surveys. Usually, the surveys adopt a stratified cluster sampling design. We discuss a number of crucial choices with respect to two key aspects of the map production process: the acknowledgement of the survey design in modeling, and the appropriate presentation of estimates and their uncertainties. Specifically, we consider the importance of accounting for urban/rural stratification and cluster-level non-spatial excess variation in survey outcomes, when fitting geostatistical models. We also discuss the trade-off between the geographical scale and precision of model-based estimates, and demonstrate visualization methods for mapping and ranking that emphasize the probabilistic interpretation of results. A novel approach to coverage map presentation is proposed to allow comparison and control of the overall map uncertainty. We use measles vaccination coverage in Nigeria as a motivating example and illustrate the different issues using data from the 2018 Nigeria Demographic and Health Survey.

SUBMITTER: Dong TQ

PROVIDER: S-EPMC9384691 | biostudies-literature | 2021 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

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

Dong Tracy Qi TQ Wakefield Jon J

Vaccine 20210403 18

It is becoming increasingly popular to produce high-resolution maps of vaccination coverage by fitting Bayesian geostatistical models to data from household surveys. Usually, the surveys adopt a stratified cluster sampling design. We discuss a number of crucial choices with respect to two key aspects of the map production process: the acknowledgement of the survey design in modeling, and the appropriate presentation of estimates and their uncertainties. Specifically, we consider the importance o ...[more]

PMID: 33824039

Similar Datasets

Project description:Coverage estimates of insecticide-treated nets (ITNs) are often calculated at the national level, but are intended to be a proxy for coverage among the population at risk of malaria. The analysis uses data for surveyed households, linking survey enumeration areas (clusters) with levels of malaria endemicity and adjusting coverage estimates based on the population at risk. This analysis proposes an approach that is not dependent on being able to identify malaria risk in a location during the survey design (since survey samples are typically selected on the basis of census sampling frames that do not include information on malaria zones), but rather being able to assign risk zones after a survey has already been completed.The analysis uses data from 20 recent nationally representative Demographic and Health Survey (DHS), Malaria Indicator Surveys (MIS), an AIDS Indicator Survey (AIS), and an Anemia and Malaria Prevalence Survey (AMP). The malaria endemicity classification was assigned from the Malaria Atlas Project (MAP) 2010 interpolated data layers, using the Geographic Positioning System (GPS) location of the survey clusters. National ITN coverage estimates were compared with coverage estimates in intermediate/high endemicity zones (i.e., the population at risk of malaria) to determine whether the difference between estimates was statistically different from zero (p-value <0.5).Endemicity varies substantially in eight of the 20 studied countries. In these countries with heterogeneous transmission of malaria, stratification of households by endemicity zones shows that ITN coverage in intermediate/high endemicity zones is significantly higher than ITN coverage at the national level (Burundi, Kenya, Namibia, Rwanda, Tanzania, Senegal, Zambia, and Zimbabwe.). For example in Zimbabwe, the national ownership of ITNs is 28%, but ownership in the intermediate/high endemicity zone is 46%.Incorporating this study's basic and easily reproducible approach into estimates of ITN coverage is applicable and even preferable in countries with areas at no/low risk of malaria and will help ensure that the highest-quality data are available to inform programmatic decisions in countries affected by malaria. The extension of this type of analysis to other malaria interventions can provide further valuable information to support evidence-based decision-making.

Project description:BackgroundAside from breastfeeding, there are little data on use of essential newborn care practices, such as thermal protection and hygienic cord care, in high mortality countries. These practices have not typically been measured in national household surveys, often the main source for coverage data in these settings. The Every Newborn Action Plan proposed early breastfeeding as a tracer for essential newborn care due to data availability and evidence for the benefits of breastfeeding. In the past decade, a few national surveys have added questions on other practices, presenting an opportunity to assess the performance of early breastfeeding initiation as a tracer indicator.MethodsWe identified twelve national surveys between 2005-2014 that included at least one indicator for immediate newborn care in addition to breastfeeding. Because question wording and reference populations varied, we standardized data to the extent possible to estimate coverage of newborn care practices, accounting for strata and multistage survey design. We assessed early breastfeeding as a tracer by: 1) examining associations with other indicators using Pearson correlations; and 2) stratifying by early breastfeeding to determine differences in coverage of other practices for initiators vs non-initiators in each survey, then pooling across surveys for a meta-analysis, using the inverse standard error as the weight for each observation.FindingsAssociations between pairs of coverage indicators are generally weak, including those with breastfeeding. The exception is drying and wrapping, which have the strongest association of any two interventions in all five surveys where measured; estimated correlations for this range from 0.47 in Bangladesh's 2007 DHS to 0.83 in Nepal's 2006 DHS. The contrast in coverage for other practices by early breastfeeding is generally small; the greatest absolute difference was 6.7%, between coverage of immediate drying for newborns breastfed early compared to those who were not.ConclusionsEarly initiation of breastfeeding is not a high performing tracer indicator for essential newborn care practices measured in previous national surveys. To have informative data on whether newborns are getting life-saving services, standardized questions about specific practices, in addition to breastfeeding initiation, need to be added to surveys.

Project description:BackgroundRecent outbreaks of measles and polio in low-income countries illustrate that conventional methods for estimating vaccination coverage do not adequately identify susceptible children. Immune markers of protection against vaccine-preventable diseases in oral fluid (OF) or blood may generate more accurate measures of effective vaccination history, but questions remain about whether antibody surveys are feasible and informative tools for monitoring immunization program performance compared to conventional vaccination coverage indicators. This study compares six indicators of measles vaccination status, including immune markers in oral fluid and blood, from children in rural Bangladesh and evaluates the implications of using each indicator to estimate measles vaccination coverage.MethodsA cross-sectional population-based study of children ages 12-16 months in Mirzapur, Bangladesh, ascertained measles vaccination (MCV1) history from conventional indicators: maternal report, vaccination card records, 'card+history' and EPI clinic records. Oral fluid from all participants (n=1226) and blood from a subset (n=342) were tested for measles IgG antibodies as indicators of MCV1 history and compared to conventional MCV1 coverage indicators.ResultsMaternal report yielded the highest MCV1 coverage estimates (90.8%), followed by EPI records (88.6%), and card+history (84.2%). Seroprotection against measles by OF (57.3%) was significantly lower than other indicators, even after adjusting for incomplete seroconversion and assay performance (71.5%). Among children with blood results, 88.6% were seroprotected, which was significantly higher than coverage by card+history and OF serostatus but consistent with coverage by maternal report and EPI records. Children with vaccination cards or EPI records were more likely to have a history of receiving MCV1 than those without cards or records. Despite similar MCV1 coverage estimates across most indicators, within-child agreement was poor for all indicators.ConclusionsMeasles IgG antibodies in OF was not a suitable immune marker for monitoring measles vaccination coverage in this setting. Because agreement between conventional MCV1 indicators was mediocre, immune marker surveillance with blood samples could be used to validate conventional MCV1 indicators and generate adjusted results that can be compared across indicators.

Project description:BackgroundEnumerating dog populations is essential to plan and evaluate rabies vaccination campaigns. To estimate vaccination coverage and dog population size in a Haitian commune, 15 sight-resight counts were conducted over two days following a government-sponsored vaccination campaign.MethodsDogs received temporary laminated collars and livestock wax marks on the head and sides at the time of rabies vaccination. After the vaccination campaign, pairs of surveyors walked pre-defined routes through targeted neighborhoods, photographing and recording characteristics and location of each dog seen on a standardized data sheet. On the second survey day, surveyors retraced the prior day's track, followed the same procedure, and indicated in addition whether they believed the dogs were resighted from the prior day. After completion of the field survey, two independent evaluators reviewed photographs and characteristics of each dog to assess which had vaccination marks and which were resighted. Surveyor and photo-reviewer sight-resight decisions were compared using Cohen's kappa, and population estimates were compared using Lincoln-Petersen 95% confidence intervals.ResultsField-surveyors identified dogs consistent with the photograph evaluations in 629 out of 800 instances (78.6%, Cohen's kappa of 0.12). Despite this inconsistency, the population estimates resulting from the field and final determinations were not significantly different at 1,789 (95% CI 1,677 to 1,901) and 1,978 (95% CI 1,839 to 2,118). Vaccination coverage was also the same at 55% and 56%; however, an observed vaccination mark loss of 13.8% suggests that the true coverage may have been closer to 64%.ConclusionUsing photos improved dog identification during the sight-resight study, leading to a higher population estimate. Despite using a 2-mark system to temporarily identify vaccinated dogs, a significant proportion had lost all identifying marks by the second day of field surveys. Efforts to estimate vaccination coverage using sight-resight surveys should consider improvement of marking techniques or better accounting for potential loss of marks in their free-roaming dog vaccination coverage assessments.

Dataset Information

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

Publications

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

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets