Ten‐year quality of life outcomes in men with prostate cancer

1. BACKGROUND: Although men with prostate cancer are living longer, they are not necessarily living well, with symptom burden increasing and HRQoL declining over time(1). For many men, the first five years after diagnosis is marked by unmet needs, debilitating disease and treatment-related bowel, urinary and sexual symptoms(2). To date, few studies have examined the trajectories of men over the 10 years after diagnosis(3). To advance survivorship care, it is crucial to understand what factors drive long-term (10 year) health outcomes after a prostate cancer diagnosis. Accordingly, this study extends our previous research (4) to report physical and mental HRQoL, life satisfaction and symptom burden of men over the 10 years after prostate cancer diagnosis. 2. METHOD: 2.1 Study setting and participants This study of newly diagnosed adult men with prostate cancer was conducted in Queensland, Australia. Sampling strategy and methods are previously described(4). Ethical approval was obtained from the Queensland University of Technology Human Research Ethics Committee (Approval No.3629H). In total, 1291 men were approached, 1064 consented, and 598 (56%) completed the final 10-year questionnaire. Self-administered questionnaires and computer-assisted telephone interviews were completed at baseline, and 2,6,12,24,36,48,60,72,84,96,108 and 120 months after the commencement of treatment. 2.2 Measures and statistical analyses: Participants’ demographics and clinical characteristics have been previously described(4). Outcome measures included Disease-specific (Expanded Prostate Cancer Index Composite; EPIC) and HRQoL (Short Form 36; SF-36); Satisfaction with Life (Satisfaction with Life Scale; SWLS) (See Figure 1a,b,c)(5-7). Growth mixture models (GMMs) in Mplus (Muthen and Muthen, 2015, Mplus User’s Guide http://www.statmodel.com/usersguide/chapter8.shtml) were adopted to identify trajectory classes and predictors using 10-year follow-up data, with EPIC longitudinal subscales as time-varying covariates(4). Missing EPIC were computed using multiple-imputation (10 repetitions), but participants with >9 missing values in EPIC or with missing data in other predictors were excluded (sample size reduced to n=928). The number of trajectory classes was determined using the Lo-Mendell-Rubin likelihood ratio test.