Monday, July 7, 2014

LIES, DAMN LIES, AND CANCER STATISTICS

Linda with Whiskers, Patches with me
1989
 
Somehow I got on the email list of Lancet, which is a British publisher of medical articles.  Rarely a week goes by that I do not receive notice of a new issue, containing many articles, some of which sound interesting.  Unfortunately, Lancet is not in this game for love; they want $31.50 to read each article, or about $200 for a yearly subscription.  As the odds are strongly against me being able to actually understand any given article, I do not bite.  However, I have discovered a way to out-fox them.  It appears that most important articles generate “Comments”, which I can read in their entirety, for free!  Thus, with a little imagination I can reconstruct the article in question – or at least get a handle on its more controversial parts.  This I did for two Comments on a paper dealing with the labeling of early pre-cancerous lesions.
As cancer doctors apply more and more early detection techniques they find more and more things they call “lesions”, which may or may not develop into cancer.  The argument seems to concern what to call them.  If you call them pre-cancerous lesions you scare the hell out of the patient, who immediately insists on treatment - which costs money and entails greater or lesser discomfort.  If you call them something else the patient doesn’t worry and unnecessary expense is avoided.  However, if one of these things develops into cancer and the patient dies, all hell – in the form of malpractice suits – may break loose.  What to do?
Well, they are arguing about that.  In the meantime, notice what this dilemma does to cancer statistics.  If you label a bunch of these lesion things “cancer” – when many of them aren’t – then it appears that the survival rate for that particular cancer is high.  And, of course, vice versa.  Another reason to be suspicious of cancer statistics.


No comments:

Post a Comment