Estimation of accelerated failure time due to exposure risk using piecewise-exponential model in cohort follow-up study.

Piecewise-exponential excess-relative-risk (ERR) model is used for analysis of large cohort follow-up grouped data of exposure risk. Estimation method for the expected-acceleration-time (EAT) to failure by exposure is proposed using the ERR. Method is applied to all solid cancer mortality atomic-bomb survivor grouped data, and gives EAT to death, 9.9 y (95%CI: 8.4, 11.2 y), at 1 Gy during 50 y follow-up (1950–2000). Longevity (all causes of death) is shorter than the reported before. The exposed younger gives greater EAT. The EAT under carcinogenesis mechanistic model with single-point/chronic exposure is mentioned. Competing causes of cancer death may imply the proneness of double cancer. [ABSTRACT FROM AUTHOR]