BlandAltmananalyse

Bland-Altman analysis, also known as the Bland-Altman method or Bland-Altman plot, is a statistical approach for assessing agreement between two quantitative measurement methods. It was introduced by J. Martin Bland and Douglas G. Altman in the 1980s, with a seminal publication in The Lancet in 1986. The method focuses on whether two methods can be used interchangeably, rather than on correlation or association alone.

The procedure involves paired measurements from the same subjects. For each pair, compute the difference between the two methods and the average of the two methods. The mean of the differences is the bias, indicating systematic deviation between methods, while the standard deviation of the differences reflects imprecision. A Bland-Altman plot graphs the differences (y-axis) against the averages (x-axis). The limits of agreement are the bias plus and minus 1.96 times the standard deviation of the differences, representing where about 95% of differences are expected if the differences are normally distributed.

Interpretation hinges on both the center and the spread of the differences, and on any patterns observed. If most differences lie within the limits of agreement and show no increasing or decreasing trend with magnitude, the methods may be considered in agreement within clinical tolerance. If proportional bias is present, transformations (such as logarithmic) or regression-based approaches may be employed, and confidence intervals can be computed for the limits of agreement. Extensions address repeated measurements, heteroscedasticity, or non-normal differences.

Applications are widespread in clinical chemistry, medical device validation, radiology, and other biomedical fields to evaluate new instruments, procedures, or diagnostic tests. Limitations include sensitivity to outliers, reliance on normality of differences, and the need to predefine clinically acceptable agreement rather than relying solely on statistical criteria.