Korea's thyroid cancer epidemic was supposed to be over. The incidence curve turned back up in 2015

South Korea is the standard example, worldwide, of what happens when a population is screened for a cancer that most people who have it will never die of. Its thyroid cancer incidence rose further and faster than any other country's, its mortality rate did not follow, and the episode has been taught for a decade as the clearest demonstration that finding more cancer is not the same as having more cancer.

The usual telling ends with the curve coming down. A new analysis of national registry data says it came down and then went back up.

The shape of the curve#

Researchers analysed cancer incidence data from Statistics Korea for 1999 to 2021 and mortality data for 1985 to 2023, calculating standardised rates against Korean resident registration populations and testing for changes in trend with joinpoint regression. The results were published in Endocrinology and Metabolism, the journal of the Korean Endocrine Society.

A tenfold rise to 2012 and a sharp fall to 2015 is the pattern the overdetection account predicts: a wave of screening, then a correction. What that account does not predict is the third segment.

Why the second rise is the interesting part#

Overdetection has a signature. It finds small tumours, it finds them early, and it therefore inflates the count of localised disease while leaving advanced disease untouched. That signature is exactly what an earlier nationwide study documented for the first Korean wave: between 1999 and 2008 incidence rose 6.4-fold, 94.4 percent of the increase consisted of tumours smaller than 20 mm detected mainly by screening, and 97.1 percent of the total increase was localised or regional disease.

The renewed rise since 2015 does not have that signature. The authors report an increase in both localised and distant-stage cancers, and distant-stage disease is not what a scan of an asymptomatic neck turns up. They also report the rise concentrated among men and younger adults, and a histological shift: papillary carcinoma — the subtype that drove the epidemic — has been falling since 2010, while follicular and medullary carcinoma have recently been rising.

Mortality has kept falling throughout#

The death rate tells a steadier story. The standardised mortality rate rose until the early 2000s and has declined continuously since, with the decline most evident among people aged 55 and over — a pattern that held while incidence tripled, collapsed, and rose again.

One methodological detail is worth carrying: the incidence and mortality rates in this study were standardised against different reference populations, 2000 for incidence and 2005 for mortality. Each series is internally consistent and comparable over time, but the two are not directly comparable to each other, and a ratio built from them would not mean what it appears to mean.

Korea is no longer the outlier it was#

A study of 63 countries published in The Lancet Diabetes & Endocrinology in 2024 estimated that more than 85 percent of thyroid cancer cases in women were overdiagnosed in South Korea, Cyprus, China and Türkiye, and that across all the registries examined 1,736,133 of 2,297,057 cases — 75.6 percent — were attributable to overdiagnosis. It also recorded the turn downward that Korea shares with the United States, Canada, Israel and several Western European countries.

Korea's distinction now is not the size of its epidemic but the length of its record. It has more years of post-peak data than anywhere else, which is why what happens to its curve after the correction is worth watching closely.

What registry data cannot settle#

1. A registry records diagnoses, not disease. It cannot distinguish a genuine increase in cancer from a change in who is being examined and with what.
2. Stage at diagnosis is itself partly a function of imaging. More thorough staging of the same tumour can move it from localised to distant without anything about the patient having changed.
3. The study reports the direction of the stage-specific and histological trends, not their magnitude. A rise from a small base is still a rise, and nothing here establishes how many additional advanced cancers are involved.
4. Mortality trends lag incidence by years, so a decline in deaths now is not evidence about diagnoses made since 2015.

The clinical response so far#

In June 2025 the Korean Thyroid Association published the country's first guideline on active surveillance for low-risk papillary thyroid microcarcinoma — monitoring a confirmed small cancer with ultrasound and thyroid function tests every six months for two years and annually thereafter, and operating only if it progresses. That guideline, and whether following a cancer is a decision Korean patients and clinicians will accept, is the subject of our next piece in this series.