Mortality analyses in a cohort of 18 235 ethylene oxide exposed workers: follow up extended from 1987 to 1998.

To extend mortality follow up from 1987 to 1998 for cohort of 18 235 men and women exposed to ethylene oxide.Standard mortality follow up, life table and Cox regression analysis.There were 2852 deaths, compared with 1177 in the earlier 1987 follow up. There was no overall excess of haematopoietic cancers combined or of non-Hodgkin's lymphoma. However, internal exposure-response analyses found positive trends for haematopoietic cancers which were limited to males (15 year lag). The trend in haematopoietic cancer was driven by lymphoid tumours (non-Hodgkin's lymphoma, myeloma, lymphocytic leukaemia), which also have a positive trend with cumulative exposure for males with a 15 year lag. Haematopoietic cancer trends were somewhat weaker in this analysis than trends in the earlier follow up, and analyses restricted to the post-1987 data did not show any significant positive trends (exposure levels dropped sharply in the early 1980s). Breast cancer did not show any overall excess, although there was an excess in the highest cumulative exposure quartile using a 20 year lag. Internal exposure-response analyses found positive trend for breast cancer using the log of cumulative exposure with a 20 year lag.There was little evidence of any excess cancer mortality for the cohort as a whole, with the exception of bone cancer based on small numbers. Positive exposure-response trends for lymphoid tumours were found for males only. Reasons for the sex specificity of this effect are not known. There was also some evidence of a positive exposure-response for breast cancer mortality.