I read a book about fixing medicare.

In it the writer brought up something I hadn't thought of before.

She said that we should always question 'best treatment' if we have a serious illness because it's tricky interpreting the studies.

E.g. earlier diagnosis skews the survival rate figures:

- You get a cancer
- You can have radiation therapy or surgery or do nothing
- You look at the studies and see that the average person who has treatment x survives 5 years after diagnosis; 10 years ago it would have been 2 years after diagnosis
- Then you find that as a result of aggressive encouragement of early diagnosis, these cancers are being diagnosed earlier in their course of development

Problem: is the survival actually longer due to the treatment, or is it longer because the diagnosis is earlier?

You need to know in order to decide whether to take treatment at all for a cancer that has a poor survival rate.

Where can one get in-depth analysis of this sort of thing without driving one's doctor insane?

quote:

---

Originally posted by babthrower:
Problem: is the survival actually longer due to the treatment, or is it longer because the diagnosis is earlier?

You need to know in order to decide whether to take treatment at all for a cancer that has a poor survival rate.

Where can one get in-depth analysis of this sort of thing without driving one's doctor insane?

---

Excellent question, I've searched for similar information without much success. The treatment success rate does seem to be highly correlated to how early the condition is detected.

I've also discovered that cancers will at times go into remission spontaneously, so how many of those treatment "successes" would have remissed without treatment if not detected? - I guess that's about the same question - I hope someone has an answer!

It depends in part on the type of cancer you're talking about; the issue has been discussed since I was in medical school, which was 30 years ago. And if one uses the old definition of cure as 5-year survival, the question is even more relevant. On the other hand there are plenty of cancers for which early detection isn't really improved, and yet for which survival times are inarguably longer and cure rates higher. The latter include leukemias, lymphomas, testicular and ovarian cancers; the former breast and colon and prostate, in particular. Colon may be a special case in that early detection is possible for pre-malignant stages, making the DEVELOPMENT of the cancer preventable.

For specific treatments, however (as opposed to population statistics for cancer survival in general) the answer is easier: the outcomes of treatments are compared using groups of similar stages, so early detection isn't an issue. In other words, given a particular tumor at a particular stage, comparing one therapy to another determines prospective outcomes of those treatments, and is independent of time of detection. So, for a given stage at discovery, there are plenty of data comparing one treatment to another from which to draw conclusions about which works best. Unfortunately, all such data refer to studies of large numbers of people; there's always a spectrum of responses, and so far not a good way to make specific individualized predictions. In most cases it boils down to making the best decision for a given individual, based on their particular circumstances, preferences, and the judgement of their doctor based on his/her experience over time. One can see, on the distant horizon, the day coming when it will be possible to discover the propensities of a given tumor in a given person and pick the exact best treatment; better, to correct the genetic defect going on within the cancer cell, and revert it to normal.

That's certainly very enlightening. Thank you, Sid.