I worked in a lab doing cancer detection via image AI (mammograms, mri, etc) back in 2008 or so, and even then we knew that mammography didn't save lives.
"This is an important, though uncommonly discussed, issue in the translation of evidence from cancer screening trials.1 It is known that overdiagnosis (treatment of cancers that would have been no threat), and high false positive rates (misdiagnosis) lead to medical harms and unnecessary surgeries, chemotherapy, and radiation...
...margin of benefit suggested by the analysis above it seems likely that if there is a benefit to screening mammography it is balanced out by mortal harms from overdiagnosis and false-positives"
Cancer treatment isn't a complete positive. If you subject large numbers of people to it some of them will die from the treatment. If you save 1 person from cancer but kill 5 more that didn't have it, is that a net positive?
I don't know why you're downvoted. The commenters below are misunderstanding things, the NNT page is old and even then it's being mis-cited. I won't go into it but read the caveats section for the author's own explanation.
Screening mammography unequivocally improves cancer-specific mortality. Making the leap to overall mortality is hard, and even a meta analysis is likely too underpowered given the very low overall mortality to begin with. Recall that the smaller the absolute difference is the larger the study will have to be to detect the difference.
For breast cancer we would probably be talking about something like 10 million patients to be adequately powered to draw any conclusion. The older trials also aren't useful/can't be used because the diagnosis and treatment of breast cancer is dramatically different than it was 10 years ago when BI-RADS was in its nascent stage. Accordingly it's not even possible to conduct such a study as it would be unethical to randomize millions of patients to no screening when we know it has proven benefits.
This also raises the question of which outcome measure matters more? All-cause mortality is a good one but it has both pros and cons. Pros being it captures hidden and misattributed deaths and is the least susceptible to bias. Cons include that it underestimates the impact of diseases that aren't high causes of death (i.e. if the patient population is more likely to die of something else the all-cause mortality won't change).
All of this leads to what are we trying to solve with breast cancer screening? It's unequivocal that a screen detected breast cancer is less advanced (i.e. no systemic therapy required) and is associated with cancer-related mortality benefit. The harms of overdiagnosis have also significantly lessened with modern radiology/histology classifications, biopsy techniques and treatment algorithms. Is cancer-specific mortality good enough? I would argue yes given the significant morbidity with systemic therapy and metastatic disease.
To summarize:
1. Screening mammography has been proven to reduce cancer specific mortality in many studies.
2. The only accurate statement about all-cause mortality is "we don't know" rather than yes/no. None of the studies are powered or controlled enough to draw any conclusions.
3. All-cause mortality may not be the best outcome measure to determine whether an intervention is "saving lives" and certainly is not the only measure to consider when deciding on a screening program.
> 1. Screening mammography has been proven to reduce cancer specific mortality in many studies.
So let's assume that an annual X-ray caused another cancer in women who would never develop breast cancer (i.e. 87% of them). You are saying "we don't know", but the authors of that paper are trying to answer exactly that. We may have saved lives in 13% group (that would be < 2.5% of those dying from breast cancer), but may have lost some lives in 87% group. According to the paper the net outcome is around 0.
It's unequivocal that screening reduces breast-cancer specific mortality. None of the studies are powered enough to draw a positive or negative conclusion for overall mortality. The second last link is an explanation of why this is the case, where we're at and what the implications are.
The last link explains the NCCN rationale and decision making process.