Statistics Glossary → Limits of agreement (Bland-Altman)
Limits of agreement (Bland-Altman)concept
The range within which about 95% of the differences between two measurement methods applied to the same units are expected to fall.
In the Bland-Altman approach, the difference (D) and the mean (A) of the two methods are computed for each unit. The mean difference shows the systematic bias, and the limits of agreement show how widely the individual differences spread: LoA = mean difference ± 1.96 × standard deviation of the differences. The differences are plotted against the means, so one can see whether the bias and spread change with the level of measurement.
A high correlation between two methods does not mean they agree; correlation only measures how they vary together. Whether the limits are acceptable is a clinical or technical decision, not a statistical one, and should be fixed before seeing the data. The method assumes the differences are roughly normal and that bias and spread are constant across the measurement range; otherwise a log transformation or regression-based limits are used. The limits are themselves sample estimates and should be reported with confidence intervals.
An everyday example
Blood glucose measured by a new fingertip device is compared with the laboratory value in 60 patients. The mean difference is 3 mg/dL and the standard deviation of the differences is 6 mg/dL, so the limits of agreement run from about -9 to +15 mg/dL. If clinicians had decided beforehand that a ±10 mg/dL difference is acceptable, the device does not agree well enough by that standard.
In MerQur
The core output of the Bland-Altman Agreement Analysis; 'Additional output: Regression-based LoA' is used when the mean or spread of the differences changes with the level of measurement, and the 'Clinical acceptance limit Δ' reports whether the limits fall within a predefined ±Δ.
Analyses where this concept appears
Bland-Altman Agreement Analysis
Related concepts
Proportional biasLin's concordance correlation coefficient (CCC)