The Rectangle Target Plot: A New Approach to the Graphical Presentation of Accuracy of Systems for Self-Monitoring of Blood Glucose. (March 2016)

• Record Type: Journal Article
• Authors: Stephan, Peter; Schmid, Christina; Freckmann, Guido; Pleus, Stefan; Haug, Cornelia; Müller, Peter
• Is Part Of: Journal of diabetes science and technology. Volume 10:Number 2(2016:Mar.)
• Page Start: 343
• Page End: 349
• Publication Date: 2016-03
• Subjects: ISO 15197 -- rectangle target plot -- self-monitoring of blood glucose -- system accuracy; Diabetes -- Periodicals; Medical technology -- Periodicals; Diabetes Mellitus -- Periodicals; 616.462005
• DOI: 10.1177/1932296815612490
• Languages: English
• ISSNs: 1932-2968

Abstract

Background: The measurement accuracy of systems for self-monitoring of blood glucose (SMBG) is usually analyzed by a method comparison in which the analysis results are displayed using difference plots or similar graphs. However, such plots become difficult to comprehend as the number of data points displayed increases. This article introduces a new approach, the rectangle target plot (RTP), which aims to provide a simplified and comprehensible visualization of accuracy data. Methods: The RTP is based on ISO 15197 accuracy evaluations of SMBG systems. Two-sided tolerance intervals for normally distributed data are calculated for absolute and relative differences at glucose concentrations <100 mg/dL and ≥100 mg/dL. These tolerance intervals provide an estimator of where a 90% proportion of results is found with a confidence level of 95%. Results: Plotting these tolerance intervals generates a rectangle whose center indicates the systematic measurement difference of the investigated system relative to the comparison method. The size of the rectangle depends on the measurement variability. Conclusions: The RTP provides a means of displaying measurement accuracy data in a simple and comprehensible manner. The visualization is simplified by reducing the displayed information from typically 200 data points to just 1 rectangle. Furthermore, this allows data for several systems or several lots from 1 system to be displayed clearly and concisely in a single graph.