The over-diagnosis, in a graph and a table

December, 3, 2016

Here you will find a representation on a poster (home-made) of the results of meta-analyses from the independent French medical magazine Prescrire, the Cochrane Collaboration (a collective of independent Nordic researchers) and the US Preventive Task Force, the US public health program evaluation agency.

Despite some variations in the data, there are two standout points : on the one hand, the estimates are reasonably close, on a proportional basis, and on the other hand, for each of these three studies, there is a much higher proportion of overdiagnoses and false alarms compared to 'avoided' deaths, pointing out the disadvantages of screening. The benefit/risk balance is therefore far from being as optimistic as it is presented to women…
The Cochrane presents its results on 2000 women screened from age 40 onwards over 10 years,
Prescrire on 1000 women screened from 50 years of age over 20 years,
US Task Force on 1000 women also, screened from 50 to 74 years old, which corresponds most closely to the French situation (click to enlarge).

The brown dots represent the number of women screened and then monitored. The red dots correspond to over-diagnoses, the yellow dots to the number of cancers presumably prevented. The light blue dots are false alarms, and the dark blue ones are unnecessary biopsies.
Here we reproduce a very meaningful graph showing how over-diagnosis occurs, based on the idea of an American researcher, Gilbert Welch, adapted here by Dr. Jean-Baptiste Blanc and with his kind permission to reproduce this image.
Click on the link to read Dr Blanc's article "deconstruction of a manipulation".

Rapidly evolving cancer progresses fast between two mammography examinations and will be "missed" by screening. Often this cancer will have already spread to distant nodes and organs, even though it is not visible.

  • Slowly evolving cancer will certainly be anticipated by screening, but even without screening it will be detected a little later by the appearance of clinical symptoms that will lead the patient to consult in time, as the metastatic time is very long.
  • For the other three forms of cancer, the very slow, not evolving and regressing cancer, screening will detect them but this is a needless detection, the women in whom they have been diagnosed would have died with their cancer but not because of it.