Alternatives to the BEIR relative risk model for explaining atomic-bomb survivor cancer mortality.