Fig. 1

Density plots for the distribution of the predicted mean out-of-pocket expenses in 2011 and 2015. Density plot shows the underlying probability distribution of the two-part model predicted mean out-of-pocket expenses in 2011 and 2015 for all cancers, breast and prostate cancer and melanoma, using a kernel density estimation. The height of the curve is scaled such that the area under each curve equals one; the y-axis depends on the maximum out-of-pocket expenses values in the original input file (different for all plots, hence inconsistency in y-axis); Values in the y axis do not have any interpretative value, other than serving as a reference to compare the height of different curves in the same plot. . For all cancers, in 2011 and 2015 most of the out-pocket spending was less than AU$2500; however, the distribution peak is lower in 2015 compared to 2011, indicating fewer patients with expenses less than AU$2500. On the contrary, patients incurred higher expenses ($AU 2500 - $7000) more frequently during 2015 than 2011 (larger area under the curve in 2015 than 2011). For breast cancer, the 2015 curve is less pronounced indicating a somewhat even distribution of costs among most patients ranging from 0 to $AU7,500. Out-of-pocket expenses in prostate cancer, were higher in 2014, with no significant difference to 2015, compared to 2011. 2011 and 2014 curves have the same height but, in the latest year, the curve has moved towards the right, indicating higher prices during that year. Costs ranging from $AU 2500 to $AU8,000 were more common in 2014 compared to 2011. For melanoma, a peak around $AU 500 in 2011 expenses, twice as high as the peak in the 2015 distribution, shows how fewer individuals paid lower out-of-pocket expenses in 2015 compared to 2011. The density of 2011 costs, rapidly flattened from AU$ 2500 to zero, while those in 2015, showed a slow reduction until they flattened in $AU 5000 indicating higher proportion of individuals paying higher prices in 2015. The asymmetry in all plots reflects the typically skewed distribution of cost data, in this case, the fitted gamma distribution. Distributions are adjusted by: a) All cancers: health insurance, sex, age, level of education, cancer type and Year b) Breast cancer: Health insurance, age, year c) Prostate cancer: Health insurance and year d) Melanoma: health insurance, year, level of education