cp-R, an interface the R programming language for clinical laboratory method comparisons

Background and objective: Clinical scientists frequently need to compare two different bioanalytical methods as part of assay validation/monitoring. As a matter necessity, regression methods for quantitative comparison in clinical chemistry, hematology and other clinical laboratory disciplines must allow for error in both the x and y variables. Traditionally the methods popularized by 1) Deming and 2) Passing and Bablok have been recommended. While commercial tools exist, no simple open source tool is available. The purpose of this work was to develop and entirely open-source GUI-driven program for bioanalytical method comparisons capable of performing these regression methods and able to produce highly customized graphical output. Methods: The GUI is written in python and PyQt4 with R scripts performing regression and graphical functions. The program can be run from source code or as a pre-compiled binary executable. The software performs three forms of regression and offers weighting where applicable. Confidence bands of the regression are calculated using bootstrapping for Deming and Passing Bablok methods. Users can customize regression plots according to the tools available in R and can produced output in any of: jpg, png, tiff, bmp at any desired resolution or ps and pdf vector formats. Bland Altman plots and some regression diagnostic plots are also generated. Correctness of regression parameter estimates was confirmed against existing R packages. Results: The program allows for rapid and highly customizable graphical output capable of conforming to the publication requirements of any clinical chemistry journal. Quick method comparisons can also be performed and cut and paste into spreadsheet or word processing applications. Conclusions: We present a simple and intuitive open source tool for quantitative method comparison in a clinical laboratory environment. (C) 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.