HealingWell
Search Close Search
☰

Who has a reliable nomogram?

Support Forums
>
Prostate Cancer
✚ New Topic ✚ Reply
❬ ❬ Previous Thread |Next Thread ❭ ❭
Posted 5/15/2009 8:28 AM (GMT -7)
When I go thru online nomograms I seem to find vast variations in possible outcome probabilities. For example, the Sloan nomogram seems to have the most optimistic outcomes while the nomogram the VA uses seems to have more dire outcomes.
Posted 5/15/2009 8:39 AM (GMT -7)
"A nomogram, graphical calculating device designed to allow the approximate graphical computation of a function.."

You might want to consider just take the "Average" of the tests to get a ballpark idea....

Posted 5/15/2009 9:28 AM (GMT -7)
I would use the Slone K. nomograms. My guess is that the VA nomograms use selective data, ie Veterans that are most likely to be older. Also their treatments may no be as effective as MSKs, hence worse outcomes.
The SK nomograms are highly accurate in that they use a very large base of patients. You must remember that they reflect statistical probabilities and can not predict individual results.
They are valuable in predicting which treatment has the best probability of working given the data that is input, not that the actual treatment will in fact work. There are just too many variables when dealing with PC, not to mention that the data that is input may not be accurate, ie gleason and %of cores.
JT
Posted 5/15/2009 2:06 PM (GMT -7)
Well said John T. I totally agree with your opinion.
Posted 5/16/2009 2:25 AM (GMT -7)
I'll second (third?) the above. Nomograms are an attempt to induce a formula for some clinical outcome (PSA rising, metastasis, probability of dying of PC, etc.) from some set of statistical data. Each one correctly portrays the results a particular research team found, but the results will vary for the reasons stated above (one institution's effectiveness, different age of patients, small sample, etc.)

You can get closer to the underlying data at the American Cancer Society's website
/www.cancer.nexcura.com/Secure/InterfaceSecure.asp?CB=265 so you can sort of pick and choose which study matches your conditions. In some cases they try to do the work for you.

But it all comes down to the problem that your odds of whatever are either zero or one, but you'll never know until after the fact.

Good luck.
Posted 5/16/2009 1:26 PM (GMT -7)
I think nomograms are a good thing and I believe that the statistical manipulation behind the sloan-kettring ones is very solid. On the other hand, if you are going to have 10 year survival rates then you need to be using data from people treated in 1999 and (mostly) before – actually the cut-off is more like 1997 since it takes several years to process the data. Has the detection and treatment of PC changed over the last 12 years? You bet! – and all of us are glad for it! Of course the researchers who build the nomograms know this and have techniques for using five and even three year survival data to “tweak” their models.

Another point in modeling is that the researchers set very high standards for “proof” that certain measures are associated with certain outcomes. You may have noticed that many members here provide greater detail in their sigs than can be entered into the nomogram. Largely this is because when you get too few patients with a certain set of stats you just don’t meet the standards for certainty that the researchers require. So your case may be different in ways that make things better or worse, but just not sufficiently predictable.

Finally, probabilities don’t tell what will happen to you. Some people get a seven on the first roll of the dice and others come up snake eyes. Here is wishing for sevens for each of you.
✚ New Topic ✚ Reply