Megan Skiendel: Miles, genuinely — did you get a physical this year?
Miles Ashworth: I did not, no. Nor do I intend to, quite.
Megan Skiendel: I'm not going to tell you you're wrong. That's a genuinely uncomfortable place for me to be.
Miles Ashworth: Now that I wasn't expecting. What shifted?
Megan Skiendel: A Swedish study. 750,000 women in a breast cancer screening program. 45% of the cancers they detected — 45 — would never have caused those women harm. And virtually all of them were treated. Surgery, radiation, the whole course.
Miles Ashworth: For something that was going to kill — nothing.
Megan Skiendel: For something that was going to do nothing. And this isn't the system failing — overdiagnosis isn't misdiagnosis, it isn't a false positive. It's its own recognized category. It's what happens when you screen people who have no symptoms. You find things that are real, technically, but clinically inert. The screening made them patients.
Miles Ashworth: Which means mammography — the most debated screening program in existence, estimates of 10% to 30% overdiagnosis of screen-detected cases — those aren't outliers. That's the floor.
Megan Skiendel: And PSA screening is the ceiling nobody wants to talk about. Twelve percent overdiagnosis on the low end. Sixty-three on the high end. One test, one condition, methodology determines whether you've harmed one in eight men or nearly two in three.
Miles Ashworth: Good lord, that is — that range is not a disagreement between scientists. That's a system that cannot evaluate itself.
Megan Skiendel: Which is actually the structural point. Screening is permanently stuck between two harms — missing the dangerous thing and treating the harmless thing — and there is no version of the technology that escapes that. So the question, the one I think nobody running these programs is answering honestly, is: what does 'net benefit' mean when the trap is built into the design?
Miles Ashworth: Well. It means something very convenient for the people selling the screening.
Megan Skiendel: But that's the part that doesn't fit the clean story — because the system isn't broken. Think about it like searching everyone's attic for fire hazards. You will find old wiring that never would have sparked, candles that would never have been lit. Alongside the one frayed wire that genuinely threatened the house. The problem is you cannot tell which is which until after you've rewired everything.
Miles Ashworth: And the wiring is real. That's the thing. You're not imagining the old wiring.
Megan Skiendel: The finding is real. The cancer is real. It just would never have hurt you.
Miles Ashworth: Which is quite different from a false positive. People conflate these. A false positive — the biopsy comes back negative, you go home relieved, it's over. Overdiagnosis means the pathology report confirms cancer. There is a tumour. It's just — well, it was never going to kill you. And now nobody can tell you that, because nobody can identify which tumour it is.
Megan Skiendel: Thyroid cancer. That's the number that stopped me when I first saw it.
Megan Skiendel: Incidence of thyroid cancer rose sharply — not because more people developed dangerous thyroid disease, but because better imaging started finding lesions incidentally. Same with renal cancer. The disease didn't change. The scanner got sharper. And suddenly we had an epidemic that wasn't actually an epidemic.
Miles Ashworth: Which is the structural cruelty of it, frankly — because improved detection is supposed to be progress. Better imaging, earlier catch, better outcome. But what it actually does is worsen overdiagnosis without improving our ability to distinguish the harmless from the harmful. You've found more attics with old wiring. You haven't learned anything about which wiring will spark.
Megan Skiendel: And then the pipeline takes over. Once you're in — surgery, radiation, chemotherapy, long-term medication — those risks are real regardless of whether the original finding was ever a threat. Sepsis from treatment complications. That's a documented cause of death in overdiagnosed patients. Death from the treatment of something that was never going to kill them.
Miles Ashworth: Good lord.
Megan Skiendel: That's not a hypothetical complication. That's the outcome of treating the harmless wiring as though it's the frayed wire.
Miles Ashworth: Now here's the question that actually matters — if we cannot tell harmless from harmful at the individual level, can we at least measure how often we get it wrong across populations? Because that feels like the only lever anyone could actually pull.
Megan Skiendel: And that's the part that breaks the whole frame — because no, we cannot agree on how often we get it wrong. Mammography: somewhere between 10% and 30% of screen-detected cases. Low-dose CT for lung cancer: 13% to 67%. Same disease, same test, and the methodology determines your answer.
Miles Ashworth: The PSA range is the one that actually stops me. Twelve to sixty-three percent. That is — look, that's not a confidence interval. That's a confession.
Megan Skiendel: Cancer Rose did a systematic review — meta-analysis — and what they found was that randomized trial analyses, the ones guideline bodies weight most heavily, actually underestimate overdiagnosis on average. Their pooled number for mammography lands around 27%. Low-dose CT lung screening, around 30%.
Miles Ashworth: Hold on. The method that USPSTF trusts most is the one most flattering to screening programs?
Megan Skiendel: That's the argument, yes. Trial-based analyses are accused of underestimating it. Ecological designs — population-level incidence data — are accused of overstating it. And there is no agreed gold standard between them. The field is just — sitting in that gap.
Miles Ashworth: Which means USPSTF is weighing net benefit using the rosiest available methodology and calling it evidence-based.
Megan Skiendel: To be fair — and I say this having watched this play out — USPSTF did revise. They walked back routine PSA screening. They shifted mammography start ages. The net benefit calculation does include overdiagnosis harms, formally. But here's where it actually gets worse.
Miles Ashworth: The implementation gap.
Megan Skiendel: A 54-year-old man. His GP orders a PSA test in the year after USPSTF walked back routine PSA screening. Not negligence — the old protocol was still the default in that practice. He gets a positive. Enters the biopsy pipeline. The guideline revision never reached him.
Miles Ashworth: The evidence shifted. The machine didn't.
Megan Skiendel: The screening machinery runs on inertia. Institutional inertia, not fraud — that's actually what makes it harder to fix. Nobody made a bad-faith call. The protocol just outlived the evidence that justified it.
Miles Ashworth: Well, that's quite a thing to consent to, isn't it — a procedure whose harm rate is unknown, under a guideline the issuing body has already revised, administered by a practice that hasn't caught up yet.
Megan Skiendel: Which is the consent question, honestly — if we cannot measure the harm consistently, and institutions lag the evidence anyway, what exactly is a patient agreeing to when they say yes to screening? And that gets even stranger once you bring in what better imaging is actually doing to this number — which is where this goes next.
Miles Ashworth: Quite. The technology improves, and somehow the problem gets larger. That's not a paradox — that's the point.
Megan Skiendel: And low-dose CT is where that actually breaks open. Because this was supposed to be the fix — newer technology, more sensitive, finally a lung cancer screening tool that works. The overdiagnosis estimates: 13% to 67%.
Miles Ashworth: Same range as the older methods it replaced.
Megan Skiendel: Exactly the same spread. You improved the resolution. You did not improve the discrimination. And that's — actually, that's the mechanism nobody states plainly enough. Imaging reveals anatomy. It cannot reveal behavior.
Miles Ashworth: Wait, say that again — because that's doing a lot of work.
Megan Skiendel: A scan shows you a nodule. What it cannot tell you is whether that nodule will grow aggressively, grow slowly, or just — sit there forever. Higher resolution finds more nodules. It doesn't answer the biological question. Which abnormality will actually threaten this person's life? The test has never answered that. It just answers: is something there.
Miles Ashworth: And that question — will it threaten the patient — remains unanswerable by imaging alone, regardless of how precise the image becomes. Which is, frankly, a rather extraordinary thing for the imaging industry to have not led with.
Megan Skiendel: Well, liquid biopsies were supposed to solve it. AI-assisted detection. Same problem. More sensitivity, still no discrimination. You find more indolent things. That's all.
Miles Ashworth: Now here is where I want to land the actual weight of this — because the individual gets lost in all the methodology. A woman gets a mammography result. She's told she has cancer. It is real, it exists on the slide. The label lands. And no reassurance about base rates changes what that does.
Megan Skiendel: The psychological harm is partly irreversible once it's communicated. You can't un-tell someone they have cancer.
Miles Ashworth: And the supposedly rational response — active surveillance, watchful waiting, we'll monitor it — requires her to live indefinitely knowing she has a cancer and is choosing not to treat it. Research shows many patients find that more distressing than treatment itself. So the sensible clinical option is practically impossible to sustain.
Megan Skiendel: I mean — the anxiety, the altered self-image, the ongoing surveillance burden. Those are real harms regardless of whether the thing ever progresses.
Miles Ashworth: And this is where shared decision-making — the institutional answer to all of this — becomes genuinely absurd. It asks the patient to hold probabilistic uncertainty that clinicians cannot resolve among themselves. The PSA overdiagnosis range is 12% to 63%. You cannot hand that to a person in a consultation room and call it informed consent.
Megan Skiendel: You're offloading an unresolved scientific debate onto an individual who just heard the word cancer.
Miles Ashworth: Quite. The technology advances. The uncertainty does not shrink. And the person who bears the cost of that gap is the patient, in the room, alone with it.
Megan Skiendel: And that's the thing I keep sitting with, honestly. Because overdiagnosis isn't a bug they can patch. It's not a calibration problem. It is inherent — structurally, permanently — to detecting abnormalities before they announce themselves as symptoms. You screen asymptomatic people, you will find things that would never have harmed them. There is no version of screening that escapes that. Which means the question isn't how to fix it. It's whether medicine can look someone in the eye and actually say: we want to find disease before it hurts you, but we will also find disease that never would have hurt you, and we cannot reliably tell which is which. And if we do find it — the harm of knowing may already be irreversible.
Miles Ashworth: Screening programs require a simpler narrative to survive. That's not cynicism — it's a real tension. Early detection does save lives. That part isn't invented. But the simplification that makes the campaign work, that makes a woman schedule her mammogram, that keeps the programme funded — that same simplification is concealing a permanent, structural harm being distributed quietly across millions of asymptomatic people who came in for reassurance.
Megan Skiendel: Right — but the part I genuinely cannot resolve is whether medicine has the language for that conversation. Not the will, even. The language. USPSTF revised. PSA screening walked back. Mammography start ages shifted. The institutional acknowledgment exists. But I have never — not once — seen awareness campaign materials that say: early detection may find something that would never have harmed you, and we cannot tell which finding that is.
Miles Ashworth: No. They say early detection saves lives. Which is true. And they leave the rest of it on the floor.
Megan Skiendel: That's the thing I'm still turning over. Not whether screening is good or bad — I don't think that's the right frame anymore. But whether medicine has the will to offer genuine uncertainty to people who walked in hoping for reassurance. That's a different ask. That's asking an institution built around decisive intervention to say: we might harm you in ways we cannot measure, at a rate we cannot agree on, and the test that caught it cannot tell us whether you needed catching.
Miles Ashworth: Well. I don't think they're prepared to say it. And I'm not sure the patients asking to be screened would survive hearing it said plainly. Which is quite a place to end up — not good, not bad, just genuinely unresolved. I'll give you that much.
