The researchers, Weibin Liao and Yaowen Xie, constructed population-weighted measures of ambient ultraviolet availability for multiple lag periods, ranging from lag10 to lag50, meaning the average ultraviolet availability in the years ten to fifty years before cancer diagnosis. Surface ultraviolet data came from the ERA5 reanalysis of monthly downward UV radiation, while population distribution was derived from WorldPop gridded data. This allowed the team to estimate how ultraviolet availability decades earlier related to cancer incidence in each registry area.

To estimate relative risks, the authors used Poisson generalized nonlinear models with fixed effects for registry area and year, so that each area effectively served as its own control over time. The models adjusted for a wide set of potential confounders, including temperature, cloud cover, per-capita GDP, hospital beds, the proportion of residents aged 65 or older, PM2.5 and ozone concentrations, urban-rural status, and population size. Relative risks were expressed per 1 W/m2 increase in ultraviolet availability.

At a lag of 30 years, higher ambient ultraviolet availability was inversely associated with total cancer incidence, with a relative risk of 0.97 (95% CI 0.95-0.99). The total-cancer estimate was also inverse at lag10 (0.94, 95% CI 0.93-0.96), but turned positive at longer lags, reaching 1.05 (95% CI 1.03-1.08) at lag45 and 1.03 (95% CI 1.01-1.05) at lag50. The authors note this lag-dependent reversal as a central feature of the results.

Site-specific patterns were strikingly heterogeneous. After false discovery rate correction, positive lag30 associations were observed for liver cancer (relative risk 1.19, 95% CI 1.12-1.26), thyroid cancer (1.21, 95% CI 1.07-1.37), brain cancer (1.78, 95% CI 1.56-2.02), stomach cancer (1.27, 95% CI 1.19-1.36), and testicular cancer (2.31, 95% CI 1.13-4.71). The testicular estimate carried the widest uncertainty of the significant positive findings.

Inverse lag30 associations appeared for several major cancers: lymphoma (0.81, 95% CI 0.73-0.92), lung cancer (0.87, 95% CI 0.84-0.90), female breast cancer (0.74, 95% CI 0.69-0.78), and prostate cancer (0.73, 95% CI 0.65-0.84). Melanoma, the cancer most classically tied to ultraviolet exposure in individual-level research, was positively associated only at lag50, with a relative risk of 1.89 (95% CI 1.11-3.21), and not at the shorter lags examined.

Sex-stratified analyses showed that estimates differed mainly in magnitude rather than direction, with the principal site-specific patterns broadly preserved across men and women. In sensitivity analyses restricted to registry areas with stable data quality, the results broadly supported the main site-specific findings, and exposure-response analyses were conducted alongside FDR correction, warm-season ultraviolet metrics, and stable-registry checks to probe the robustness of the associations.

The authors emphasize that this is an ecological study: exposure was measured at the area level, and incidence was aggregated across populations, so no individual’s ultraviolet exposure or cancer risk was directly observed. Confounding by factors that vary across regions and over time cannot be fully excluded, even with the extensive adjustment set, and the lag structure means the exposure windows span decades of social and environmental change in China.

Writing in the paper, the researchers conclude that long-term ambient ultraviolet availability showed heterogeneous lag-, cancer-site-, and sex-specific ecological associations with cancer incidence, and they call for individual-level studies to clarify the underlying mechanisms and the causal relevance of the observed patterns. Until such studies are available, the findings stand as population-level associations rather than evidence of benefit or harm for any person.

Subject of Research: Ecological study of ambient ultraviolet radiation and site-specific cancer incidence in China, 2006-2016

Article Title: Associations between long-term exposure to ultraviolet radiation and site-specific cancer incidence: an ecological study in China between 2006 and 2016

Article References: Liao, W., & Xie, Y. (2026). Associations between long-term exposure to ultraviolet radiation and site-specific cancer incidence: an ecological study in China between 2006 and 2016. BMC Public Health. https://doi.org/10.1186/s12889-026-29975-6

Image Credits: AI Generated

DOI: 10.1186/s12889-026-29975-6

Keywords: ultraviolet radiation, cancer incidence, ecological study, China, BMC Public Health, epidemiology, melanoma, cancer registry, lag effects, site-specific cancer, Associations, between