Agreement estimate among three Italian cancer registries in the coding of multiple primary cancers

Aims: The aim of the study was to compare agreement on the coding of multiple primary cancers (MPs) between three Italian cancer registries, the Ragusa Cancer Registry (RCR), the Cancer Registry of Romagna (RTRo), and the Tuscany Tumor Registry (RTT), that adhere to different rules for accepting MPs and to study whether coding according to common international rules (IARC-IACR) increased comparability. Methods: One hundred cases were randomly extracted from the archives of each registry from those recorded as having more than one cancer. For each of the 300 patients, the number of independent cancers was attributed independently by one coder from each registry. The coders coded the series twice: once following the local registry rules and once according to the IARC-IACR rules. The agreement was estimated by couples of coders by means of Cohen's kappa statistics. Results: The agreement on MP status between coders using local rules and definitions was good between the RTT and RCR (kappa = 0.77) and very good between the RTRo and RCR (kappa = 0.81) and the RTT and RTRo (kappa = 0.96). Exclusion of 23 expected discordant cases increased the agreement. The agreement reached with the use of the IARC-IACR rules was very good (RTRo vs RCR, 0.95; RTT vs RTR, 0.94; RTT vs RTRo, 0.95). Conclusions: The comparison among the RTT, RTRo and RCR confirmed that the number of tumors considered MPs may be modified depending on the rules adopted. There were minor differences between the RTT and the RTRo since their rules were very similar. Most differences in agreement were with the RCR since its classification was conceptually different from the other two. The result on agreement with IARC-IACR rules is encouraging from the point of view of conducting a cooperative study among different registries on the incidence of MPs.