Using repeated measurements

When repeated measurements (replicates) are made for each subject, it is inefficient to estimate average bias and limits of agreement using only the first measurement, rather than all measurements.

If replicates are available, it is sensible to use the mean of the replicates to estimate average bias. However, when estimating the limits of agreement, the reduction in the standard deviation due to averaging of the measurements must be considered and adjusted for if necessary.

  1. On the analysis task pane, select the Mean of replicates option.
  2. On the Fit Differences task pane, in the Difference between drop-down list, select single measurements.
    The limits of agreement will be estimated for the difference between single measurements by each method. This is standard practice when reporting patient results for PEFR. The mean measurements option uses the mean of the replicates to compute the limits of agreement. However, this will lead to narrower limits of agreement (due to the reduction in standard deviation mentioned above) and should only be used when it is standard practice to use the mean of multiple measurements as the patient result.
  3. Click Recalculate.
    The results are recalculated and the analysis report updates.

The mean difference estimate is 6.0, which indicates that the Mini Wright meter reads on average 6.0 l/min higher than the Wright meter. The confidence interval is narrower because the measurement error was reduced by using the mean of the replicates.

(click to enlarge)

The limits of agreement estimate an interval of -67.8 to 79.8, which indicates that the Mini Wright meter may measure as much as 67.8 l/min below and 79.8 l/min above the large meter. This would be unacceptable for clinical purposes. Again, the confidence intervals are narrower due to the use of more measurements.