Genital warts are not spread evenly across Korea. A national analysis mapped where they cluster, and what tracks with it

Sexually transmitted infections are usually discussed as a matter of individual behaviour. This analysis asks a different question: whether the risk in Korea is distributed across the map in a pattern, and if so what features of a place travel with it.

The pattern is not random#

Researchers took 2019 municipal-level data from the National Health Insurance Service, covering all 250 municipalities, and tested whether condyloma acuminata incidence was spatially autocorrelated — whether high-incidence areas sit next to other high-incidence areas more often than chance would produce.

Moran's I runs from −1 to +1, where values near +1 indicate that similar values cluster together. Both results are positive and significant, tested against 999 Monte Carlo permutations. Local cluster detection using the Getis-Ord Gi* statistic then placed the high-incidence clusters in metropolitan and south-eastern regions.

What travels with the clusters#

Two Bayesian spatial regression models — an intrinsic conditional autoregressive model and a Besag-York-Mollié model — tested municipal-level variables covering health behaviours, socioeconomic indicators, health care access, adult entertainment venue density and risk of sexual violence.

The effect sizes also deserve reading carefully. An adjusted relative risk of 1.005 for welfare spending means a one-point rise in that share is associated with half a percent more incidence — statistically distinguishable from no effect, and small. Only the single-person household figure is large enough to matter much at plausible differences between municipalities.

The welfare-budget association is the one most likely to be standing in for something else. Municipalities that spend a larger share of their budget on social welfare differ from those that do not in density, age structure and deprivation, all of which have their own relationships with sexually transmitted infection. The model adjusts for the variables it includes and cannot adjust for the ones it does not.

The scale behind the map#

This is one year of a condition whose treatment has been rising for over a decade. National claims data shows the number of men treated for anogenital warts rising 3.5-fold between 2007 and 2018, from 32,709 to 114,734, with related costs rising roughly 9.8-fold. An analysis of treatment methods over 2010 to 2019 found most care delivered in primary clinics, with Seoul and Gyeonggi-do carrying the largest patient numbers regardless of treatment type.

That last observation and this one point at the same thing from different data: the burden is concentrated in particular places, and treatment capacity is concentrated there too.

What this does not establish#

1. The design is ecological. Associations hold between municipalities and cannot be transferred to individuals living in them.
2. The data counts people who received insured care coded as condyloma acuminata in one year. Areas differ in how readily people seek treatment and where they seek it, and a hotspot may partly reflect access rather than incidence.
3. Cross-sectional data from a single year cannot establish direction or sequence. Nothing here shows that a municipal characteristic preceded the incidence it correlates with.
4. Adult entertainment venue density is a municipal count, not a measure of anyone's behaviour, and reads as a proxy whose meaning the study does not test.
5. Two of the three significant associations have effect sizes close to 1, and the paper reports them as significant rather than as substantial.
6. The study does not name the specific municipalities in its hotspots beyond describing them as metropolitan and south-eastern.

The finding worth carrying is the first one: this is not evenly spread. A condition usually framed as a private matter turns out to have a geography, and the authors argue that spatial surveillance and locally adapted prevention follow from that — which is an argument about where public health effort goes, not about who is at fault.