Korean mammography halved breast cancer deaths. The same women were also half as likely to die of everything else

The headline finding of Korea's mammography evaluation is a mortality rate ratio of 0.43 — screened women dying of breast cancer at less than half the rate of unscreened women. The study is unusual and worth reading closely because its authors also published the number that stops anybody quoting 0.43 as the effect of screening.

The cohort and the headline#

Published in npj Breast Cancer, the study followed 8,300,682 Korean women invited to the National Cancer Screening Program in 2002 and 2003, linking them to the Korea Central Cancer Registry through 2014 and to death certificates through 2015.

The incidence figures are the first thing worth noticing. Screened and unscreened women developed breast cancer at almost the same rate, which argues against large-scale overdiagnosis in this programme — the pattern that dominated Korea's thyroid experience, where screening drove incidence up 6.4-fold without moving mortality.

The number the authors published against themselves#

Women who attend screening are not a random sample of women. They are, on average, healthier, better off and more engaged with medical care, and those attributes reduce mortality from causes that have nothing to do with mammography. The standard way to expose that is to measure deaths screening could not possibly have prevented.

Screened women were about half as likely to die of everything that is not breast cancer. Mammography does not prevent stroke, road traffic injury or liver disease. That 0.52 is therefore a measure of how different the two groups were before any screening happened — and it sits very close to the 0.43 the study reports for breast cancer itself.

The age pattern supports a real effect existing underneath. The largest mortality reduction was in women aged 45 to 54, and there was none observable after 70 — a gradient that healthy-attender bias alone would not naturally produce, since the healthiest attenders are not concentrated in one decade.

What the programme has been doing since#

A separate analysis in the Korean Journal of Radiology examined 42,165,405 mammography screenings performed through the national programme between 2009 and 2020, in women aged 40 and over.

A recall rate of 17.2 percent means that in 2009 roughly one Korean woman in six was called back after a screening mammogram, against a positive predictive value under 1 percent — more than 99 of every 100 recalls led to no cancer. Both figures improved substantially over the decade. The authors nonetheless describe positive predictive value and sensitivity as remaining suboptimal, and regional differences across Korea's 237 districts persisted even as they narrowed.

The rising interval cancer rate — cancers appearing between scheduled screens — moved in the same direction as the detection rate, which is not a contradiction but is a reminder that a screening interval is a compromise rather than a guarantee.

What a woman reading a checkup package should take from this#

Mammography in Korea sits in the same category as upper endoscopy and the faecal immunochemical test: a component of the bundle with real evidence behind it, delivered through a national programme whose performance is measured and published. That is a genuinely different object from an add-on scan sold by dose and price.

It is also not the halving of risk the headline number suggests, the benefit was concentrated in women aged 45 to 54 and absent after 70, and roughly one recall in nine now leads to a cancer. All of that is in the published record, and none of it is in the brochure.