AbstractA number of patient-reported outcomes (PROs) predict increased mortality after primary cancer treatment. Studies, though, are sometimes affected by methodological limitations. They often use control variables that poorly predict life expectancy, examine only one or two PROs thus not controlling potential confounding by unmeasured PROs, and observe PROs at only a single point in time. To predict all-cause mortality, this study used control variables affording good estimates of life expectancy, conducted multivariate analyses of multiple PROs to identify independent predictors, and monitored PROs two years after diagnosis. We recruited a consecutive sample of 824 patients with uveal melanoma between April 2008 and December 2014. PROs were variables shown to predict mortality in previous studies; anxiety, depression, visual and ocular symptoms, visual function impairment, worry about cancer recurrence, and physical, emotional, social and functional quality of life (QoL), measured 6, 12 and 24 months after diagnosis. We conducted Cox regression analyses with a census date of December 2018. Covariates were age, gender, marital and employment status, self-reported co-morbidities, tumor diameter and thickness, treatment modality and chromosome 3 mutation status, the latter a genetic mutation strongly associated with mortality. Single predictor analyses (with covariates), showed 6-month depression and poorer functional QoL predicting mortality, as did 6–12 month increases in anxiety and 6–12 month decreases in physical and functional QoL. Multivariate analyses using all PROs showed independent prediction by 6-month depression and decreasing QoL over 6–12 months and 12–24 months. Elevated depression scores six months post-diagnosis constituted an increased mortality risk. Early intervention for depressive symptoms may reduce mortality.
Prediction of all-cause mortality from 24 month trajectories in patient-reported psychological, clinical and quality of life outcomes in uveal melanoma patients