Dear Statalist,
I am a day or two away from re-submission of a manuscript to a well-reputed
medical journal (after an initial revise and resubmit decision). The paper
looks at a laboratory test, troponin I, as a predictor of in-hospital
mortality in 34,000 patients from two hospitals.
The paper should be relatively easy reading to anybody well-versed in
logistic regression. I believe that the methods and findings are sound (I
have to admit that I re-read a section of Dr. Pagano's text last night
entitled 'Berkson's Fallacy' and had a momentery flash of terror), but as
you all know, that never stopped any reviewer from criticism. In all
seriousness, if there are better ways to analyze or present the data, I'd
rather hear it sooner rather than later. As I have said before on this
list, I am not a formally trained statistician.
Is there anyone out there who is well-versed in the pitfalls of logistic
regression using a highly skewed continuous predictor who would be willing
to take a look within the next couple of days? I'm not necessarily looking
for a meticulous review, just overall thoughts.
It's a lot to ask, and don't expect any takers. But perhaps someone would
be willing to do their part to keep garbage out of the medical literature!
Thanks.
*
* 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/
