Dear Mickey 191,

first, my expertise is very restricted. I'm not a doctor, I'm interested (and have some knowledge) in research methodology and in decisions based on statistical data. Both interests lead me sometimes to read about medical decisions just as one of several fields of applications.

Mammography and mastectomy are long known in decision research as having a disappointingly low success rate (by the way, a quite similar discussion goes about prostate screening in men; just to worry the other half of humankind). That is, in comparison with a control group of women not undergoing any surgery or screening, the net benefit of medical procedures is low, measured in survival rates (how many women are alive or, if dead, by some other reason after 5 or 10 or whatever years). That has two main reasons: (1) women who die despite operation and (2) women surviving without any medical interference (cancer confined, for instance).

Mind you, on the average, there is a net benefit of medical intervention on survival rates, so that means at least some women who otherwise had not will survive due to medical intervention. However, the intervention has costs and not only benefits. The costs are unnecessary operations or, at a smaller scale, undue worrying after a screening.

Nobody can tell before you if you are the rare woman to be given some extra decades due to the operation or the woman with the unnecessary surgery. What I mean when saying you might have decided otherwise with more information: being told that there is the possibility of a false positive diagnosis, being told that for some women even with cancer the operation is unnecessary for it is confined and not life threatening at all, and being told that for some women even the operation will not help (and all that with the best of numbers they have) you might have decided otherwise. But that is a very personal decision. The idea with the statistical data as a basis for decision is not to tell any woman what the 'correct' decision is but to allow her to make a very personal decision. For some, the small extra risk reduction by the operation is worth the procedure, for some it isn't. None of these decisions can be called wrong by anyone from outside.

Your 'dumb questions' go into the heart of the matter concerning using statistics at all for decisions. All statistics average across very different persons and types of cancer. You always can try to get more 'personalised' statistics, closer to your indivídual case. Very often, these individualised statistics are not available. If they are available for breast cancer I do not know them (but that doesn't mean much, due to my restricted expertise). So all your specific questions do get a 'not known to me' response from me. Sorry. However, a completely personalised statistic is only the single case and that's why some researchers are against using statistics in that field (medicine) at all. I'm not, but I see the problem.

You want to have the data for women of your age group (for better comparison), they are there (I'd guess, but I don't know). You want to have the data for a certain type of cancer, they are there (guess). You want to have the data for a certain level of estrogen, perhaps you can even get those. But when you tell your doctor you want to have the data for your age, your estrogen level and your type of cancer diagnosis, they tell you that there are no such data. For even in a very large sample, the number of women sharing all three (or four, or five) aspects will be very low. And you wouldn't trust very much a statistic based on, say, five women sharing all attributes and 80% have survived for at least five years.

Would there not be a greater percentage of "successful"
masectomies if the C. had been confined? No, if I understand you correctly. The comparison is with an untreated control. If the cancer is confined, then the untreated controls would also survive. The success rate is only measured in more survivals in the treatment than in the control group.

Just to tell you in this example why percentages are sometimes so misleading and absolute numbers are less so. Imagine a group of 200 women with a confined cancer (and assume for the moment we could be sure about the diagnosis). Half of them, 100, are treated with mastectomy, half of them, 100, are untreated. Five years later, 100 of the treated are alive and 99 of the untreated. You can sell that as a 100% success for all (that is, 1) of those who might have died otherwise are saved by the procedure. You can look at the relative risk reduction and see that the risk to die was 1% without the operation and zero with. Then you have a relative risk reduction of just 1 %. You can look at how many women have to undergo operation for 1 to benefit from it. That is 100 have to undergo operation for 1 to profit (in my example).

I have nothing at all against modern medicine in principle, I just don't think they can help much in many instances and they should give better information under which circumstances they do help much and under which they don't. Unlike others here, I don't think that in those instance where modern medicine doesn't help much, alternative methods of treatment do. But that is a very different theme.

Wolfgang