Survival from cancer--up-to-date predictions using period analysis.

OBJECTIVES This period analysis provides Canadian predictions of the short- and long-term relative survival of people recently diagnosed with cancer. Long-term period and cohort-based estimates are also compared. DATA SOURCES Data are from the Canadian Cancer Registry, the Canadian Mortality Data Base, and Statistics Canada life tables. ANALYTICAL TECHNIQUES Relative survival analyses were conducted using the life-table method; expected survival proportions were derived using the Ederer II approach. Period analysis estimates were based on the survival experience of cancer cases followed up in 2002. The cohort analyses involved people diagnosed in 1997 (5-year survival) or 1992 (10-year survival). National estimates exclude Quebec. MAIN RESULTS Relative survival ratios were highest for thyroid (5-year, 97.7%) and prostate (95.2%) cancer and lowest for pancreatic cancer. Survival for many forms of cancer is higher than previously estimated by cohort-based analysis. The largest increases in 10-year relative survival were predicted for cancers of the prostate (13.0%) and rectum (9.7%). The largest predicted increases for 5-year survival were for cancers of the cervix uteri (5.4%) and rectum (4.5%), and for leukemia (3.7%).