Exploring the paradoxical association of sun exposure and melanoma-specific mortality: The Norwegian Women and Health (NOWAC) Study

Background Paradoxically, some studies have found that high sun exposure, a well-known cause of melanoma, is associated with lower mortality among melanoma patients. In the Norwegian Women and Health cohort, we aimed to investigate associations between pre-diagnosis sun exposure and melanoma-specific and overall death, and the potential impact of unmeasured confounding and selection bias (conditioning on melanoma diagnosis) on any observed association between markers of high sun exposure and death. Methods We estimated hazard ratios (HRs) by Cox regression, observed covariate E-values, bias-adjusted HRs, and performed frailty analysis. Results Among 2234 patients with a first melanoma (311 deaths, 168 from melanoma; mean follow-up 8.0 years), we found significantly reduced risk of melanoma-specific (HR=0.41, 95% confidence interval (CI)=0.24–0.68) and overall (HR=0.49, 95%CI=0.33–0.72) death for ever- versus never-sunburn and a negative trend for cumulative number of sunburns (p trend ≤0.001), but not for sunbathing vacations (HR=0.69, 95%CI=0.37–1.27, melanoma-specific death). Neither the E-value approach nor bias-adjusted HRs suggested a large degree of unmeasured confounding for the sunburn-death association. However, frailty analysis suggested that selection may partly explain the association. Conclusion While sunburns appear protective, this association might reflect unobserved heterogeneity in melanoma risk and selection bias, and the associations may not represent a true causal effect.