Overdiagnosis is not misdiagnosis: the pathologist is right, the tumour is real — and the person would have lived and died never knowing it existed. Autopsy studies reveal the reservoir this draws on: papillary thyroid microcarcinomas in up to a third of adults examined systematically, prostate cancer in over a third of men in their seventies, DCIS in a meaningful fraction of women — indolent disease vastly outnumbering the lethal kind. Any technology that looks harder — screening programmes, incidental imaging, sensitive biomarkers — harvests this reservoir, converting healthy people into cancer patients with real treatments, real complications and real fear, for tumours that were never going to act. The concept is counterintuitive precisely because every instinct says finding cancer early must help; overdiagnosis is the name for the cases where it cannot.

The cleanest demonstration ever run was unintentional: South Korea attached cheap thyroid ultrasound to its national screening programme in the early 2000s, and thyroid cancer incidence rose roughly fifteen-fold to become the country's most diagnosed cancer — while thyroid cancer mortality stayed flat. Tens of thousands of thyroidectomies followed, with lifelong hormone replacement and a steady rate of surgical complications, for tumours the mortality curve proves were overwhelmingly harmless. When Korean physicians campaigned against routine thyroid screening, incidence fell by a third within years — the epidemic was a diagnosis epidemic. Melanoma shows the same signature in slow motion across fair-skinned countries: incidence multiplied several-fold over decades of intensifying skin surveillance and falling biopsy thresholds, while mortality moved comparatively little.

Overdiagnosis manufactures fake success through mechanisms every reader of health news should know. Lead-time bias: detecting a cancer three years earlier extends measured survival by three years even if death arrives on the same date. Length bias: screening preferentially catches slow-growing tumours — the ones least needing catching — because they linger longest in the detectable window. And overdiagnosis itself is the ultimate survival-statistic inflator: flood the denominator with harmless cancers and five-year survival soars while the same number of people die. This is why screening claims quoting survival percentages are structurally untrustworthy, and why the only honest metrics are disease-specific and all-cause mortality in randomised trials — the standard the screening hubs on this site apply throughout.

Every overdiagnosed case pays full price for zero benefit: surgery, radiation or drugs with their complication rates intact; the psychological conversion into a cancer patient, with measurable anxiety, insurance and identity effects; and surveillance schedules that generate their own findings. Upstream sits the incidentaloma cascade — the adrenal nodule on a scan done for something else, the thyroid nodule on a carotid ultrasound — where each innocent finding demands workup that can injure. None of this is an argument that screening is bad: it is the cost column in a ledger whose benefit column is real for several cancers. Programmes justify themselves only when the mortality benefit demonstrably outweighs this arithmetic — which is exactly the calculation the PSA, mammography and MCED hubs walk through.

The field's response has matured beyond hand-wringing into working tools: risk-stratified screening that concentrates testing where prior probability is high; diagnostic pathways that interpose better discrimination before biopsy (MRI-first prostate pathways markedly reduce detection of insignificant cancer); active surveillance that diagnoses without automatically treating — now the majority approach for low-risk prostate cancer and under trial for low-risk DCIS and papillary microcarcinoma; nomenclature reform that removes the word carcinoma from indolent lesions; and raised diagnostic thresholds. Each tool concedes the same point: since looking harder always finds more, the discipline must live in what is done with the finding.

Three questions defuse most overdiagnosis risk. Before any screening test: what is the evidence this test reduces mortality for someone with my risk profile — and what is the overdiagnosis rate alongside it? Before any incidental finding cascades: what would happen if we surveilled this rather than biopsied it — many guidelines now bless exactly that. And after any low-risk cancer diagnosis: is active surveillance an option for this tumour type — increasingly, it is the guideline-preferred one. Overdiagnosis is not a reason to refuse screening; it is the reason to choose screening the way evidence-based medicine chooses drugs — by trial-proven net benefit, not by the intuition that looking must always help.