A Comparison of Markov and Mechanistic Models for Soil-Transmitted Helminth Prevalence Projections in the Context of Survey Design.

Globally, there are over 1 billion people infected with soil-transmitted helminths (STHs), mostly living in marginalized settings with inadequate sanitation in sub-Saharan Africa and Southeast Asia. The World Health Organization recommends an integrated approach to STH morbidity control through improved access to sanitation and hygiene education and the delivery of preventive chemotherapy (PC) to school-age children delivered through schools. Progress of STH control programs is currently estimated using a baseline (pre-PC) school-based prevalence survey and then monitored using periodical school-based prevalence surveys, known as Impact Assessment Surveys (IAS). We investigated whether integrating geostatistical methods with a Markov model or a mechanistic transmission model for projecting prevalence forward in time from baseline can improve IAS design strategies. To do this, we applied these 2 methods to prevalence data collected in Kenya, before evaluating and comparing their performance in accurately informing optimal survey design for a range of IAS sampling designs. We found that, although both approaches performed well, the mechanistic method more accurately projected prevalence over time and provided more accurate information for guiding survey design. Both methods performed less well in areas with persistent STH hotspots where prevalence did not decrease despite multiple rounds of PC. Our findings show that these methods can be useful tools for more efficient and accurate targeting of PC. The general framework built in this paper can also be used for projecting prevalence and informing survey design for other neglected tropical diseases.

Geostatistical modelling enables efficient safety assessment for mass drug administration with ivermectin in Loa loa endemic areas through a combined antibody and LoaScope testing strategy for elimination of onchocerciasis.

The elimination of onchocerciasis through community-based Mass Drug Administration (MDA) of ivermectin (Mectizan) is hampered by co-endemicity of Loa loa, as individuals who are highly co-infected with Loa loa parasites can suffer serious and occasionally fatal neurological reactions from the drug. The test-and-not-treat strategy of testing all individuals participating in MDA has some operational constraints including the cost and limited availability of LoaScope diagnostic tools. As a result, a Loa loa Antibody (Ab) Rapid Test was developed to offer a complementary way of determining the prevalence of loiasis. We develop a joint geostatistical modelling framework for the analysis of Ab and Loascope data to delineate whether an area is safe for MDA. Our results support the use of a two-stage strategy, in which Ab testing is used to identify areas that, with acceptably high probability, are safe or unsafe for MDA, followed by Loascope testing in areas whose safety status is uncertain. This work therefore contributes to the global effort towards the elimination of onchocerciasis as a public health problem by potentially reducing the time and cost required to establish whether an area is safe for MDA.

MBGapp: A Shiny application for teaching model-based geostatistics to population health scientists.

User-friendly interfaces have been increasingly used to facilitate the learning of advanced statistical methodology, especially for students with only minimal statistical training. In this paper, we illustrate the use of MBGapp for teaching geostatistical analysis to population health scientists. Using a case-study on Loa loa infections, we show how MBGapp can be used to teach the different stages of a geostatistical analysis in a more interactive fashion. For wider accessibility and usability, MBGapp is available as an R package and as a Shiny web-application that can be freely accessed on any web browser. In addition to MBGapp, we also present an auxiliary Shiny app, called VariagramApp, that can be used to aid the teaching of Gaussian processes in one and two dimensions using simulations.

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.

Global, regional and national epidemiology and prevalence of child stunting, wasting and underweight in low- and middle-income countries, 2006-2018.

In 2016, undernutrition, as manifested in childhood stunting, wasting, and underweight were estimated to cause over 1.0 million deaths, 3.9% of years of life lost, and 3.8% of disability-adjusted life years globally. The objective of this study is to estimate the prevalence of undernutrition in low- and middle-income countries (LMICs) using the 2006-2018 cross-sectional nationally representative demographic and health surveys (DHS) data and to explore the sources of regional variations. Anthropometric measurements of children 0-59 months of age from DHS in 62 LMICs worldwide were used. Complete information was available for height-for-age (n = 624,734), weight-for-height (n = 625,230) and weight-for-age (n = 626,130). Random-effects models were fit to estimate the pooled prevalence of stunting, wasting, and underweight. Sources of heterogeneity in the prevalence estimates were explored through subgroup meta-analyses and meta-regression using generalized linear mixed-effects models. Human development index (a country-specific composite index based on life expectancy, literacy, access to education and per capita gross domestic product) and the United Nations region were explored as potential sources of variation in undernutrition. The overall prevalence was 29.1% (95% CI 26.7%, 31.6%) for stunting, 6.3% (95% CI 4.6%, 8.2%) for wasting, and 13.7% (95% CI 10.9%, 16.9%) for underweight. Subgroup analyses suggested that Western Africa, Southern Asia, and Southeastern Asia had a substantially higher estimated prevalence of undernutrition than global average estimates. In multivariable meta-regression, a combination of human development index and United Nations region (a proxy for geographical variation) explained 54%, 56%, and 66% of the variation in stunting, wasting, and underweight prevalence, respectively. Our findings demonstrate that regional, subregional, and country disparities in undernutrition remain, and the residual gaps to close towards achieving the second sustainable development goal-ending undernutrition by 2030.

Evaluating survey designs for targeting preventive chemotherapy against Schistosoma haematobium and Schistosoma mansoni across sub-Saharan Africa: a geostatistical analysis and modelling study.

BACKGROUND: Schistosomiasis control programmes primarily use school-based surveys to identify areas for mass drug administration of preventive chemotherapy. However, as the spatial distribution of schistosomiasis can be highly focal, transmission may not be detected by surveys implemented at districts or larger spatial units. Improved mapping strategies are required to accurately and cost-effectively target preventive chemotherapy to remaining foci across all possible spatial distributions of schistosomiasis. METHODS: Here, we use geostatistical models to quantify the spatial heterogeneity of Schistosoma haematobium and S. mansoni across sub-Saharan Africa using the most comprehensive dataset available on school-based surveys. Applying this information to parameterise simulations, we assess the accuracy and cost of targeting alternative implementation unit sizes across the range of plausible schistosomiasis distributions. We evaluate the consequences of decisions based on survey designs implemented at district and subdistrict levels sampling different numbers of schools. Cost data were obtained from field surveys conducted across multiple countries and years, with cost effectiveness evaluated as the cost per correctly identified school. RESULTS: Models identified marked differences in prevalence and spatial distributions between countries and species; however, results suggest implementing surveys at subdistrict level increase the accuracy of treatment classifications across most scenarios. While sampling intensively at the subdistrict level resulted in the highest classification accuracy, this sampling strategy resulted in the highest costs. Alternatively, sampling the same numbers of schools currently recommended at the district level but stratifying by subdistrict increased cost effectiveness. CONCLUSIONS: This study provides a new tool to evaluate schistosomiasis survey designs across a range of transmission settings. Results highlight the importance of considering spatial structure when designing sampling strategies, illustrating that a substantial proportion of children may be undertreated even when an implementation unit is correctly classified. Control programmes need to weigh the increased accuracy of more detailed mapping strategies against the survey costs and treatment priorities.