Design and Analysis of Elimination Surveys for Neglected Tropical Diseases

Abstract

As neglected tropical diseases approach elimination status, there is a need to develop efficient sampling strategies for confirmation (or not) that elimination criteria have been met. This is an inherently difficult task because the relative precision of a prevalence estimate deteriorates as prevalence decreases, and classic survey sampling strategies based on random sampling therefore require increasingly large sample sizes. More efficient strategies for survey design and analysis can be obtained by exploiting any spatial correlation in prevalence within a model-based geostatistics framework. This framework can be used for constructing predictive probability maps that can inform in-country decision makers of the likelihood that their elimination target has been met, and where to invest in additional sampling. We evaluated our methodology using a case study of lymphatic filariasis in Ghana, demonstrating that a geostatistical approach outperforms approaches currently used to determine an evaluation unit’s elimination status.

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

“…However, if previous spatial data are available, alternative sampling strategies may be more cost effective and appropriate. For example, geostatistical sampling designs utilise information on the spatial structure of disease prevalence to design more efficient surveys [ 23 ]. Exploiting spatial correlation between locations, this approach uses spatially regulated sampling within a model-based geostatistical framework to estimate prevalence surfaces.…”

Supporting elimination of lymphatic filariasis in Samoa by predicting locations of residual infection using machine learning and geostatistics

“…Our study is a first step towards generating evidence about the potential for using spatial models to inform sampling strategy. Further investigations are required to explore whether the gains from a model-driven targeted sampling approach are maintained when compared to other approaches such as adaptive cluster sampling [18][19][20] , and under what conditions. The lower odds ratio results for predictions made only on the randomly sampled PSUs indicate that the model may perform differently in lower prevalence settings.…”

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

“…Further, as this modelling approach was developed using open-source software, this analysis can be easily modi ed to re ect actual administrative boundaries. Although we chose to analyse the most commonly used two stage cluster-based survey designs, analysis of actual geolocated country data would enable assessment of alternative sampling approaches, such as geostatistical survey designs [24]. Additionally, although this study analysed the largest available database of school-based surveys for schistosomiasis, this did not allow the evaluation of how spatial patterns changed over time or in response to control measures.…”