Sun Exposure Risks and Skin Types

We described the differences between skin type groups with regards to personal characteristics, knowledge, attitudes and behaviors (Supplementary Table S1). However, as the balance of the risks and benefits of sun exposure differs according to skin type, we restricted further analyses to people who reported having fair or medium skin who are at markedly higher risk for skin cancer. Analyses of those with olive/dark skin will be presented elsewhere. A priori we considered state/territory of residence, age, sex, skin color and educational attainment to be covariates in multivariable analysis, so participants with missing data for any of these variables were excluded from the analysis.
We described participant characteristics and knowledge, attitudes and behavior variables using numbers and simple percentages. To estimate associations between: (1) personal characteristics and knowledge/attitudes; (2) personal characteristics and behavior; (3) knowledge/attitudes and behavior; and (4) knowledge/attitudes and being sunburnt, we calculated prevalence ratios (PRs) and 95% confidence intervals (95% CI) using log-binomial regression. In cases where the log-binomial model did not converge or was unable to estimate the covariance matrix, we used Poisson regression.
We used directed acyclic graphs to help identify additional potential confounders of the associations of interest, including propensity to sunburn, occupational status and history of skin cancer treatment. In multivariable analysis, we first applied a change in estimate approach (>10%) to decide whether to include or exclude a covariate in the final model. In addition, we repeated analyses separately for men and women to examine whether the associations of interest differed by sex. All analyses were performed using R version 4.1.2.