We would like to use the RCI method (Jacobson and Traux 1991) to identify cognitive decline following coronary artery bypass grafting surgery. However we do not have a non-surgical control group and therefore plan to use a slightly modified version of the RCI where the test-retest reliability measure will be taken from published data from the test manufacturer, along with the SD of our study population at baseline (before surgery) to calculate the SE difference of the measurement. We will then use the individual participant scores to identify if they reliably declined post-surgery or if their change was due to measurement artefact.

To spell out the RCI method

• The RCI calculated based on the participants individual change score between two time-points divided by the SEdifference = (posttest score – baseline (pretest) score)/SE difference.
• Where the SE difference is calculated by = √2((SD baseline control√(1 – test-retest reliability of the measure))2

Can anybody see any large limitations with this modified method?