Mammography Revisited

"Here we go again!" is what I thought when I saw the publication of the 25-year follow-up of the Canadian National Breast Screening Study in the British Medical Journal and its corresponding coverage in the New York Times.  If you missed it, the bottom line message from this study and its coverage was:  mammography is worthless.

This is a very dangerous conclusion.

Sources like the New York Times are what many of us turn to to get the facts about local and worldwide events, analyses of current cultural trends and explanations of complex topics.  It has been the 'paper of record' for over a century because its reporting has largely been fair, balanced and believable.  In this case, however, they really blew it.

The Canadian trial, which was completed in 1985 was plagued from the start and likely shouldn't have been published.  Randomly Controlled Trials are the 'gold standard' for testing a new drug, a new treatment, a new screen or any other intervention different from 'usual care.'  In order for the results of any RCT to be accepted, it is crucial that the study itself is carried out with impeccable accuracy. 

First and foremost is the randomization process.  It is paramount that this process is conducted cleanly.  This means that the two groups must be randomly assigned to the intervention 'arm' (in this case mammography) or the usual care arm (the way things were before mammography).  If this process is carried out correctly, both groups should be nearly exactly alike in make up (age, socioeconomic status, etc.).  As such, if neither arm got mammos, you would expect the incidence and death rate from breast cancer in both arms to be the same.  If the intervention (mammography) was of benefit, we would expect to see the death rate from breast cancer to be decreased in the mammo arm.   The death rate in this study was the same for both arms; therefore the conclusion of this trial was that mammography screening was no better at preventing deaths than finding a breast cancer on physical exam.  

Why was this?

In a prior blog, I explained that the three things which determine the prognosis of a breast cancer were size of the tumor, the grade (or aggressiveness) of the tumor and the presence or absence of spread to lymph nodes under the arm.  In this trial, there was a huge excess of patients with advanced cancers in the screening arm compared to the control arm.  These patients had large, palpable breast cancers with at least 4 palpable involved lymph nodes under the arm.  In short, these patients had very poor prognoses and most died during the 5 years of the study, many in the very first year.  There were 17 such women 'randomized' into the mammo arm with only 5 placed in the control arm.  This proved contamination of the randomization process from the start and ensured a worse prognosis of women in the mammo arm.  

Despite worldwide condemnation and criticism, the authors of this trial continue to do follow up analyses of the study and outlets like the NYTs continue to publish these analyses.  This is unfortunate, because there are many, many studies which prove conclusively that mammography saves lives, including this 29-year follow-up of the largest RCT ever conducted which showed a 29% decline in breast cancer death in populations merely invited to be screened and this study which showed an over 60% decline in the death rate among populations which actually got regular screening mammograms.  

There were at least 8 randomly controlled trials, all of which showed a clear benefit in decreasing the death rate from breast cancer, except for this Canadian Trial.  It should be ignored and women 40 and over should get annual screening mammograms-it's the best way to prevent death from this all-too-common disease. 

Check out this interview with the author of the Canadian Trial.  His arguments are uncertain and defensive.  

Until next time...onward and upward!