Development of an open source tool for mapping disease clusters

Abstract

BackgroundMapping is an important step in investigations of infectious disease outbreaks. Despite advances in molecular characterisation of pathogens that allow delineation of outbreaks with greater precision, geospatial methods that complement these analyses are rarely used in real time. Barriers include lack of flexibility, expense of specialised software, and the need for trained personnel. This study aimed to develop a tool with free, open source software that could be used by public health professionals to display clusters of disease on a map, and to demonstrate its utility through the exemplar of molecular clusters of tuberculosis in London. MethodsWe developed a bespoke interactive mapping tool using Shiny, a web application framework for the statistical software package, R. Mapping was enabled using Leaflet, a JavaScript library for interactive mapping, through the R package leaflet. Data were extracted from the Enhanced Tuberculosis Surveillance System for cases of tuberculosis with a residential postcode in London, who were part of a molecular cluster, and were notified between Jan 1, 2010, and Dec 31, 2013. FindingsData subsets can be interactively displayed on the basis of cluster name, notification date, demographics, and reported risk factors. In addition to mapping, epidemic curves and summary tables are automatically produced. The tool was used to explore 3194 cases of tuberculosis in 767 clusters in London, and allowed rapid overview of the geographical, temporal, and epidemiological features to support cluster investigation. InterpretationGeographical displays of molecular data can enhance understanding of disease transmission. Advantages of the tool developed in this study include: flexibility, allowing user-defined subsets of data in any geographical location to be displayed; potential adaptability to other geocoded health or other contextual data; and a web interface, which is user-friendly, needs minimal training to operate, and does not require upload of identifiable information to the internet. A potential limitation is maintenance of code, although the use of open source software also benefits from support and improvement from the wider developer community. Future work will involve testing of the tool with potential users to assess its utility in practice. FundingCMS is funded by the Farr Institute of Health Informatics Research.