effect on expectation of life or on subsequent risk of developing cancer or other fatal diseases." Although participants had significantly higher leukemia rates than controls, the authors attribute that to an abnormally low rate among the controls rather than to a radiation-associated high rate among the participants. That both the control and the participant groups had fewer cancers than expected based on general U.K. population rates (even after controlling for social class) is presented to support that interpretation. The standardized mortality ratio (standardized to the U.K. population) for leukemias in the control groups was 0.56 for the entire follow-up period and 0.34 for the time period 2 to 25 years postexposure.
The results of this study of CROSSROADS participants and the earlier mentioned Five Series Study simultaneously under way should add more information about, and therefore a more stable understanding of, the association between nuclear test participant exposure and mortality. These studies are constructed carefully to include appropriate comparison groups and to avoid known biases (operational as well as conceptual) in the data collection and analysis, and, finally, in their interpretation.