Linear regression analysis for comparing two measurers or methods of measurement: But which regression?

Abstract

1. There are two reasons for wanting to compare measurers or methods of measurement. One is to calibrate one method or measurer against another; the other is to detect bias. Fixed bias is present when one method gives higher (or lower) values across the whole range of measurement. Proportional bias is present when one method gives values that diverge progressively from those of the other. 2. Linear regression analysis is a popular method for comparing methods of measurement, but the familiar ordinary least squares (OLS) method is rarely acceptable. The OLS method requires that the x values are fixed by the design of the study, whereas it is usual that both y and x values are free to vary and are subject to error. In this case, special regression techniques must be used. 3. Clinical chemists favour techniques such as major axis regression ('Deming's method'), the Passing-Bablok method or the bivariate least median squares method. Other disciplines, such as allometry, astronomy, biology, econometrics, fisheries research, genetics, geology, physics and sports science, have their own preferences. 4. Many Monte Carlo simulations have been performed to try to decide which technique is best, but the results are almost uninterpretable. 5. I suggest that pharmacologists and physiologists should use ordinary least products regression analysis (geometric mean regression, reduced major axis regression): it is versatile, can be used for calibration or to detect bias and can be executed by hand-held calculator or by using the loss function in popular, general-purpose, statistical software.

Telemetry-based oxygen sensor for continuous monitoring of kidney oxygenation in conscious rats

“…Ordinary least-squares regression analysis and analysis of covariance were used to determine whether the linear relationships between PO 2 and current changed during the period of implantation (28). Between animal means Ϯ SE of PO2 measured over each 24-h period were calculated.…”

“…Absolute changes in kidney PO2 from both systems were compared using the ordinary least products regression method, using a loss function to obtain 95% confidence intervals. This method for comparing methods of measurement has the benefit of distinguishing between fixed and proportional bias (28). ANOVA was also used to test for systematic bias, between the two methods of measurement, across the five levels of inspired oxygen (10,18,21,40, and 80%).…”

“…After the solution had been bubbled for at least 5 min, bubbling was stopped, the Clark electrode was placed back into the solution, and the measurement was taken from both systems over a period of at least 5 min. Ordinary least squares regression analysis was performed to generate linear relationships between PO 2 and current for each electrode (28).…”

The discordant method: a novel approach for differential correlation

“…Linear models have been shown to be effective, however, there are deficiencies when there is large differences in variability between groups, which may be relevant when examining -omics data from different types of platforms and/ or data from humans or non-experimental model systems. It has been shown that large variability results in incorrect slope estimates (Cornbleet and Gochman, 1979;Ludbrook, 2010). Furthermore, slope estimates can be different depending on what feature is considered the dependent or independent variable in the linear model.…”

An Index to Objectively Score Supraglottic Abnormalities in Refractory Asthma

“…7,8 We conducted these analyses using R . 9 Using an SGI cutoff of 10 units, 6 we estimated k between each new reader and the control reader (SAS/STAT software package, version 9.2 of the SAS System for Windows XP; SAS Institute Inc).…”