Posted 4/27/2009 7:28 PM (GMT 0)
I thank you for calling our attention to Stephen J Gould’s cautionary tale about cancer statistics. But we need to consider what lesson we should take from it. If you think that the lesson is something like; “Statistics are worse than ‘darned lies’” then you have missed the point. Here, is what I would suggest we learn from Gould’s essay.
1) Know what the statistics cover and what they don’t.
Gould noted that the mortality statistics for abdominal mesothelioma did not take into account age at diagnosis, tumor stage at diagnosis, and the absence of a significant causal factor, workplace exposure to asbestos. Thus he reasoned that his situation; young, early detection, no workplace exposure, might place him in a different group from most of those diagnosed.
2) Recognize that outcomes depend on treatments.
Given that the outcomes of standard treatment were so disappointing, Gould opted for what were then experimental treatments. This is still a good option – the worse your particular odds of cure by standard treatment the more strongly you should consider participating in clinical trials of new treatments.
Finally let me make one more point which was not in Gould’s article.
3) Statistics always lag behind available treatment
If you look at the Sloane-Kettering nomograms you will see only three treatment options: radical prostatectomy, brachytherapy, and external beam radiation therapy. Is this because no other treatments are worth mentioning? No, it is because there is insufficient data on other treatments to establish rigorous probabilities. By the way, among these “unproven” therapies is robotic prostatectomy. Why? Because, while the initial results look good (Egawa, S. Laparoscopic Radical Prostatectomy as Our Bridge to the Future?, Eur Urol (2008),doi:10.1016/j.eururo.2008.11.044) there is not yet enough long term data, such as ten-year progression rate, to be statistically certain.