Geographic variations of breast carcinoma incidence in Sweden. Are the differences real?

The validity of the reported geographic variations of breast carcinoma incidence in Sweden was assessed by examination of two possible sources of bias: non-notification to the Cancer Registry of diagnosed carcinoma cases and 'biologically benign' breast carcinoma, i.e. with a low disease-specific lethality, e.g. detected accidentally at autopsy. No significant geographic differences in registration deficit were found even though non-notification tended to be slightly higher for old patients in low-incidence areas. Autopsy cases were estimated to account for less than one per cent of all cases and tended to be more frequent in high-incidence areas but the regional differences were generally small and not significant. An analysis of the relationship between 10-year relative survival and age-standardized incidence in 27 different regions revealed no significant correlation, whereas there was a significant positive correlation between age-standardized incidence and mortality. These findings indicate that non-lethal breast carcinoma cases do not explain the variations of incidence. In conclusion, no evidence was found suggesting that the geographic differences were artifactual. Registration deficit and autopsy cases, however, may have slightly increased the variations among elderly women.