2018, Cilt 34, Sayı 4, Sayfa(lar) 265-271
Evaluation of concordance between measurement techniques using graphical methods and regression models with an application
Doğukan Özen
Ankara Üniversitesi, Veteriner Fakültesi, Biyoistatistik Anabilim Dalı, Ankara, Türkiye
Keywords: Bland Altman, Mountain Plot, Deming regression, Passing Bablok regression analysis

Aim: The aim of this study was to introduce the graphical (Bland Altman and Mountain Plot) and methodogical (Passing- Bablok and Deming Regression analysis) methods which are used to evaluate the rational and absolute agreement between two measurement techniques with the help of an application data set.

Materials and Methods: A total of 84 thymus tissue specimens provided by a researcher in the field were included in the study. TUNEL (Method 1) and Caspase 3 immunohistochemistry techniques (Method 2) were applied to the sections taken from the blocks of these tissue samples. The apoptotic cells in the tissue samples were counted using both methods. Bland Altman and Mountain Plot graphical methods and Passing-Bablok and Deming regression analyzes were used to examine the concordance of the two measurement methods.

Results: Results showed that there was a systematic constant error (bias) between the two measurement methods and the TUNEL method gave an average of 4.11 units higher than the Caspase 3 immunohistochemistry method. However, it was determined that there was no proportional bias between the two techniques. Also, there was a moderate positive correlation between the two methods (ρc=0.51, 95%CI: 0.41-0.60).

Conclusion: Erroneous approaches used in method comparison studies were discussed and alternative methods were presented with an application that can be used to obtain unbiased results.