Research on potential radiation risks in areas with nuclear power plants in Japan: leukaemia and malignant lymphoma mortality between 1972 and 1997 in 100 selected municipalities.

The results of a geographical correlation study using Poisson regression analysis are reported for leukaemia and malignant lymphoma mortality between 1972 and 1997 in 100 selected Japanese municipalities with or without a nuclear power plant (NPP). The data did not support social concerns of an increased risk of malignant lymphoma in the vicinity of Japanese NPPs. However, some estimates of overall excess relative risk (ERR; relative risk minus one) were statistically significantly positive for leukaemia mortality in 20 NPP municipalities compared with mortality in the remaining 80 control areas, taking into account a minimum two-year latency following the start of commercial operation. One estimate was 0.228 (95% CI: 0.074-0.404) from a simple area adjustment using the mortality in all Japan as the external baseline rate. This superficial increase is not due to leukaemia among young people, aged less than 25 years at death. The ERR estimate for ages at death of 50-74 years was confounded to be positive for leukaemia and distorted to be negative for malignant lymphoma. For leukaemia, a positive ERR estimate was seen, especially for females and during specific periods. Confounding of the ERR estimate for two causes was also seen in some NPP areas including a high adult T-cell leukaemia (ATL) area. Temporal area variations associated with ATL misclassification and a temporal increasing trend of leukaemia mortality in the elderly caused the confounding effects. Our findings do not support the hypothesis of a leukaemogenic impact of NPPs in Japan.