Using Shiny to support data analysis in health services research

By Dr Nicole White

Nicole White Shiny Apps

Shiny is a coding package available in “R”, a popular free software for statistical data analysis. The Shiny package helps users share their R analysis code as an interactive, user-friendly web page. A major benefit of using Shiny is that it provides an accessible way for researchers to apply newly developed tools in their work without needing knowledge of R. Depending on the application, Shiny can provide tools for exploring and visualising existing datasets or allow users to input their own data to create customised summaries for use in publications and other outputs.

The AusHSI statistics team has developed two new Shiny applications for health services researchers to support the uptake of best-practice research methods. Available applications aim to help researchers synthesise evidence for literature reviews and health economic modelling.

Visual Checklist is a visualisation tool for summarising study adherence to reporting checklists as a useful addition to systematic and scoping reviews. The application takes data inputted by the user and displays included studies’ adherence against selected reporting checklist criteria. Templates for common reporting checklists are available for download with the application. Interactive data summaries can examine results for a subset of included studies and/or reporting checklist items. R code is generated from the inputs provided for advanced users who wish to customise visualisations further.

The application was recently used as part of AusHSI’s work with the Digital Health CRC, to summarise reporting quality across published health economic evaluations of computerised decision support systems.

ShinyPrior supports the estimation of probability distributions for use in health economic modelling. Estimating probability distributions is a key step in health economic modelling. However, estimating required distributions can be difficult when researchers only have access to summary data on model inputs, such as costs. ShinyPrior applies different statistical methods to convert summary data entered by the user into a corresponding probability distribution. Results are provided as visualisations and summary tables, which can be customised and downloaded for future use, for example, in publications.

This application is now being used in CHD LIFE+, a national project that is using simulation modelling to understand different models of care for neurodevelopmental support in congenital heart disease.