Variance and correlation patterns
We estimated that the SI measured a clinically significant mean
difference change of 0.06 as clinically significant. The variance of the
difference between time 0 and time 1 was estimated to be 0.0097 based on
a previous study conducted by Borovac and
colleagues.25 We used this variance of difference as
an estimate for the variances of the SI measures within subjects. Based
on the previous data, with a specified minimal change in SI, a mean
difference of 0.06 between the SI measures was considered clinically
significant. A sample size of 252 patients per group, or a total of 504
patients, would give a power of at least 0.8 for testing the
hypothesis.30.
The sample size estimation for the longitudinal analysis was performed
using the General Linear Mixed Model Power and Sample Sizesoftware GLIMMPSE (URL: http://glimmpse.samplesizeshop.org/). All
other analyses were performed using the R statistical software package
(version 4.0.3). This report was written following the Enhancing the
QUality and Transparency of Health Research (EQUATOR) STROBE guidelines.