[Cancer incidence estimation in Spain: period 1993-1996].

Objective: To estimate the number of incident cases of cancer in Spain between 1993 and 1996.
Methods: We used data on the incidence of cancer from nine Spanish population-based cancer registries published in the monograph Cancer incidence in Five Continents, vols. VI and VII (period 1983-92). The National Institute of Statistics provided mortality data (period 1983-96). Generalized linear mixed models were used to estimate the number of incident cases. The incidence/mortality ratio was modeled. To account for heterogeneity, the providence of residence was considered as a random effect. Other factors analyzed were sex, age and period. Model parameters were estimated using a Bayesian approach with BUGS software. Estimates were valdated by comparing the observed number of cases with those predicted by the model in the regions with cancer registry data.
Results: The average number of incident cases per annum for all cancer sites except non-melanoma skin cancer was 78,440 for men and 55,480 for women. The most frequent neoplasm in men was lung with 15,480 cases followed by bladder with 9,445 cases and colorectal with 8,876 cases. In women the most frequent cancer was breast with 13,490 cases followed by colorectal with 8,274 cases. These absolute frequencies showed an increasing time trend for all cancers and for the most frequent sites, with the exception of stomach and uterine cervix. Internal validation of the estimates allowed calculation of a relative error smaller than 10%.
Implications: This is the first time that the number of incident cases of cancer in Spain has been estimated with methods that account for heterogeneity. These figures show the importance of cancer as a public health problem in our community.