|
|  |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: syntax for MMRM ANCOVA
Dear Stata users,
I have the kind of dataset that statisticians usually don't like to
see . . .small. . . no, tiny!
I would like to do either ANOVA or, preferably, ANCOVA (if the
dataset allows).
The data is gathered from a very small clinical trial that I had to
prematurely discontinue due to poor recruitment.
I have 2 drug groups and 2 sites. Total N was 7, with 4 at one site
and 3 at the other. There was one baseline visit, 4 follow-up visits,
for a total of 5 visits. I planned to use baseline corrected scores,
so the baseline scores wouldn't really count as df.
Thus the total number of datapoints could have been 7 x 4 = 28.
However, a couple of patients dropped out before the end of the study.
If I impute data with the last-observation-carried-forward method,
that is pretty simple. But then, as I understand it, the analysis
only consists of one datapoint per patient, ie. a measly 7 datapoints
total. A simple t-test wouldn't be appropriate because there really
appears to be a site-by-treatment interaction.
I would like to use the mixed model repeated measures (MMRM) method.
I understand that it uses all the available datapoints. In addition,
it includes an algorithm for imputing missing a given patient's
missing data based on the data available for that patient. (I have
heard that it extrapolates a trajectory.)
Mallinckrodt has written extensively on it. I contacted him, but he
does not use Stata and so could not tell me how to implement it in
Stata. I'm hoping that someone out there can tell me how to do this,
assuming it can be done with this dataset.
So . . .
Does anyone know how to run this analysis in Stata?
If not, what would you suggest as an alternate analysis?
Either way, syntax would be appreciated.
Thanks very much.
Erick
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
