Is LSC appropriate for group analysis?
Dear Prof. Shepherd,
I am a student of XXX. I have a puzzle in the use of Lower Significant Change (LSC): LSC is used to monitor whether an individual’s BMD change is meaningful, but is it appropriate for a mean change of BMD of a subject group? (as the following table shows)
|
NECK BMD(mg/cm2) |
||||
|
? |
First Time |
|
Change |
Mean of Changes |
|
Subject1 |
0.668 |
0.71 |
0.042 |
0.019±0.017 |
|
Subject2 |
0.686 |
0.708 |
0.022 |
|
|
Subject3 |
0.712 |
0.724 |
0.012 |
|
|
Subject4 |
0.842 |
0.835 |
-0.007 |
|
|
Subject5 |
0.648 |
0.668 |
0.02 |
|
|
Subject6 |
0.768 |
0.777 |
0.009 |
|
|
Subject7 |
0.479 |
0.524 |
0.045 |
|
|
Subject8 |
0.744 |
0.748 |
0.004 |
|
|
Subject9 |
0.801 |
0.821 |
0.02 |
? |
LSC of NECK BMD of our technician with the DXA is 0.04mg/cm2. Does that mean the BMD change of the group during the 12 months is meaningless?
As you are a professor in densitometry, would you like to give me some directions?
Thank you a lot!
LSC is for individuals
Thanks for your question. "Least Significant Change" is a concept applicable only to individuals. This is the magnitude of change you must see to have 95% confidence that the change between the two measures is significantly greater than 0.
For populations, you have an increase in power becuase you are sampling the population multiple times. The simple way to do this is to take the difference between matched measures on each individual and perform a student's t-test to see if the mean difference is significant. GOOD LUCK!!!