Dear Colleagues
I am assisting a clinical colleague with the analysis of data collected from
around 80 patients. The same data was collected (electronically) on 3
consecutive days. The same instrument was used on all 3 days and the
measurement of interest (an index usually ranging between 0 and 100) was
made by a human scorer interpreting the data and according to standard
criteria (for each day).
What we would like to do is assess whether we really need three days data.
Could we use just the first? I have completed a simple repeated measures
ANOVA with day as a within subject factor (there was no difference).
I have since thought that a “measures of agreement” approach may be
appropriate here. I realise that the “method” is the same here but it could
be argued that what we want to know is whether the index for day 2 “agrees”
with day 1 and more importantly the limits of the agreement. I felt that the
limits obtained this way would be clinically more relevant than say a
confidence interval around the mean difference between day 2 and day 1.
I would welcome any thoughts as to whether this approach is flawed or not.
Susan Charman
MRC Statistician
Papworth Hospital NHS Trust
Papworth Everard
Cambridge
CB3 8RE
_________________________________________________________________
Get a speedy connection with MSN Broadband. Join now!
http://resourcecenter.msn.com/access/plans/freeactivation.asp
|