Another MSK Nomogram question - I'm SOOO confused!

Hi there
Calling on all statisticians to help me with this one if you might be kind enough please...
When I plug in GGs stats to the MSKK nomogram: "Risk of Dying of Prostate Cancer in Men With a Rising PSA After Radical Prostatectomy" (using a doubling time of 1.5 months), it tells me he has a 6% chance of dying within 5 years and 30% chance at 15 years. However if I change his theoretical Gleason score to 10; i.e. 5+5, it then tells me he has a 4% chance of dying within 5 years, and 20% chance at 15 years.
Whaaaa????
Sorry to open this ol' Pandora's box again, as I know there was a whole other thread about this a short while ago, but doesn't this just point to the fact that this tool should not only be taken with a pinch of salt, but a large tablespoon?!

none of this stuff is perfect...any way... to recognize perfection...we would have to be perfect...if that were true...well...

Maybe our imperfect concept of what perfect is needs to be changed...but then...

I think it gives people that knows numbers something to do. They can build "something"...that resembles...."something".

Post Edited (island time) : 7/9/2017 7:16:01 PM (GMT-6)

Jaybee, I mentioned in neither thread, that when I met withZelefsky, RO at MSKCC, I asked about the nomograms and he said he doesn't put a lot of faith in them, just not enough people in the database that match any one individuals exact situation.

Ah-ha! Thank you TA. Now that would be really interesting! (to see that sample size for the given data inputs). That makes a whole lot of sense now. Don't suppose you'd happen to know which ones provide that particular data? (just out of interest)...

Chask said...
Where is Redwing when you need him. He is the maths wizard around here.

Thanks, but oh, no, no, no.... will not attempt that one!

I think TA is right though. Any of these math tools are only ways to interpret a database of cases. There are not many of the G9-10 cases, and any time sample sizes get small the results can become quite skewed.

The Partin Tables show the number of cases in each category. That link is to a .pdf file.

One positive thought from this - no one can predict the course of any single individual's case. No one. No database, no nomogram, no way. We can all always have hope. On average, we're mostly average. But there are always exceptions.

Allow me to point everyone to perhaps the brightest lighthouse of hope that I know, our own Todd1963. If anyone had the right to throw in the towel right up front he did. But he didn't. And we shouldn't. No matter what any table or any math analysis may say!