Methods: The design of the prostate cancer (PC) policy model includes estimation of healthcare costs for all men diagnosed with PC in Ontario in 1992-2002 (n=41,803). Total direct costs are determined by linking Ontario databases describing hospitalization, physician services, same day surgery, physician visits, drugs, radiation, long-term care, and home care. Net and attributable costs are estimated by matching PC cases with population-derived controls using a combination of propensity methods and hard-matching on key covariates. Allocation of observation time for each patient is determined by a period-allocation hierarchy.

Results: Observation time is available for: I) 90 days before diagnosis (100%); II) 180 days post diagnosis (97%); III) 180 days prior to death (20%); IV) 365 days prior to period III (18%); and V) continuing care (between II and IV, 89%). Propensity matching revealed very close distributions of covariates. Total direct costs, excluding radiation therapy, home care and long term care costs, per 100 patient days are: I) ($CAD 2004) $337; II) $1833; III) $950; IV) $684; V) $405. The lifetime net attributable direct cost of prostate cancer was estimated to be $11032. Extremes of age (<60, >80 years), rural residence, and low comorbidity (Charlson score <2), were associated with increased cost of care across most periods. Socioeconomic status did not predict healthcare cost.

Discussion: Attributable Canadian lifetime costs for PC are substantially lower than published estimates based on charges or total costs. Covariates may strongly affect costs. Obtaining data from primary sources may improve the evidentiary basis of the policies supported by decision analytic policy models.