Improving Detection of Rapid Cystic Fibrosis Disease Progression-Early Translation of a Predictive Algorithm Into a Point-of-Care Tool

Abstract

The clinical course of cystic fibrosis (CF) lung disease is marked by acute drops of lung function, defined clinically as rapid decline. As such, lung function is monitored routinely through pulmonary function testing, producing hundreds of measurements over the lifespan of an individual patient. Point-of-care technologies aimed at improving detection of rapid decline have been limited. Our aim in this early translational study is to develop and translate a predictive algorithm into a prototype prognostic tool for improved detection of rapid decline. The predictive algorithm was developed, validated and checked for 6-month, 1-year, and 2-year forecast accuracies using data on demographic and clinical characteristics from 30 879 patients aged 6 years and older who were followed in the U.S. Cystic Fibrosis Foundation Patient Registry from 2003 to 2015. Predictions of rapid decline based on the algorithm were compared to a detection algorithm currently being used at a CF center with 212 patients who received care between 2012-2017. The algorithm was translated into a prototype web application using RShiny, which resulted from an iterative development and refinement based on clinician feedback. The study showed that the algorithm had excellent predictive accuracy and earlier detection of rapid decline, compared to the current approach, and yielded a prototype platform with the potential to serve as a viable point-of-care tool. Future work includes implementation of this clinical prototype, which will be evaluated prospectively under real-world settings, with the aim of improving the pre-visit planning process for CF point of care. Likely extensions to other point-of-care settings are discussed.