HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Electronic Letters to:

Letters to the Editor: Katy Dewitte, Colette Fierens, Dietmar Stöckl, and Linda M. Thienpont Application of the Bland–Altman Plot for Interpretation of Method-Comparison Studies: A Critical Investigation of Its Practice Clin Chem 2002; 48: 799-801 [Full text] [PDF]

eLetters: Submit a response to this article

Electronic letters published:

Some additional references to comparison of methods.

Jacek Dmochowski   (26 June 2002)

Another weakness of the Bland-Altman method

bruce e siskowski   (14 June 2005)

Some additional references to comparison of methods. 26 June 2002
Jacek Dmochowski, statistician University at Buffalo SUNY Send letter to journal: Re: Some additional references to comparison of methods. sttjacek{at}buffalo.edu Jacek Dmochowski	Comparison of measuring instruments (methods) has a long history in statistical literature. In the basic measurement model we can compare relative bias (difference between means) and possibly precision/variability. In statistical literature these considerations started with Grubbs measurement model(JASA, 1948). Another very useful review can be found in Technometrics, 1973 paper by Grubbs titled "Errors of Measurement, Precision, Accuracy and the Statistical Comparison of Measuring Instruments". Later concept of R&R (Repeatability and Reproducibility) studies was developed, and usually this topic is part of Quality Control textbooks, for example by Vardeman and Jobe. One of reasonably recent articles on related subject is an article by Christensen and Blackwood "Tests for Precision and Accuracy of Multiple Measuring Devices", Technometrics 1993. Chemical or biochemical methods are of course more complicated because there are many different sources of variation, but with properly designed experiments effects of these different sources can be estimated (statistical buzzwords: variance components, random effect models or mixed models). Another probably important aspect are methods which interfere with measuring subject. Statistical references on these topics are many, and I will be more than happy to provide interested readers with additional references. Sincerely, Jacek Dmochowski
Another weakness of the Bland-Altman method 14 June 2005
bruce e siskowski, engineering director Reichert inc. Send letter to journal: Re: Another weakness of the Bland-Altman method bsiskowski{at}reichert.com bruce e siskowski	Another major weakness of the Bland-Altman method is that it only looks at bias and totally ignores within treatment repeatability variation. Here is an example of comparing two medical devices: Assume Device #1 reads an individual's pulse rate as 30, 60, 90 pulses/min and Device #2 reads 59, 60, 61 pulses/min. If this trend continued across all people tested over a range of nominal values, the Bland-Altman method would say these devices agree (based on average bias trends only). The method total ignores the fact that Device #1 has 30 times the repeatability variation. If bias is all you care about, Maloney and Rastogi proposed a better method in 1970 that can replace the subjective Bland-Altman method that came 13 years later. For those that perform device/method comparison studies that allow for replication (repeat runs) as done for blood pressure monitoring or intra-ocular pressure monitoring, the Maximum Likelihood Estimator (MLE) method can be used to determine true agreement for bias, variation and interaction quantities.