Methods: We derived a population-based sample of PC patients from the Ontario Cancer Registry who were residing in selected urban, suburban, and rural areas of Ontario, Canada, and diagnosed in 1993-4, 1997-8 or 2001-2 (n=2749, survivors = 1961). Consenting survivors (n=851) were mailed questionnaires, including the Health Utilities Index (HUI 2/3), Patient-Oriented Prostate Utility Scale (PORPUS-P (psychometric) and PORPUS-U_i (utility)), Functional Assessment of Cancer Therapy-Prostate (FACT-P), the Prostate Cancer Index (PCI), and a consent form for chart review. We constructed univariate and multiple regression models to determine the effects of patient-, disease-, system-, and symptom-related variables on QOL.

Results: 670 patients returned completed questionnaires and 620 charts were reviewed (597 entered for these analyses). Mean patient age was 72.6 years (SD=8.1, range = 43-98). Mean (SD) scores were: PORPUS-P = 71.8 (14.0), PORPUS-U_i = 0.86 (0.11), HUI3 = 0.78 (0.24), and FACT-P = 127.3 (18.4). In univariate analyses with patient-related variables (age, marital status, education, employment, income, and comorbidity), older age, widowhood, lower education, retirement, and comorbidity (Charlson >2) were associated with lower QOL (all p<0.05). In multiple regression analyses, the patient-related variables together explained 18-21% of the variance in scores. With the addition of disease-related variables (Gleason score, treatment with radical prostatectomy, radiation therapy, or hormone therapy, metastases), the model explained 21-25% of the variance. Patients currently on hormone treatment had lower PORPUS-U_i and HUI3 scores than patients treated with hormones in the past or never (p<0.05). System-related variables (diagnosis year and geographic area) contributed little to the explained variance (1-3%). Symptom-related variables (PCI urinary, sexual, and bowel function) were the strongest predictors of QOL. When these variables were added to the model, 62-70% of the variance in PORPUS-P and PORPUS-U_i scores was explained and 37-47% of the variance in FACT-P and HUI3 scores was explained. The effects of the other variables diminished such that no other variable remained significant for the PORPUS-U_i, and comorbidity was the only other significant variable for the other instruments.

Conclusions: Symptoms related to PC and its treatment have a large effect on the QOL of PC survivors. Although many variables are associated with their QOL, only prostate symptoms and comorbidity have independent effects.