The cohort mortality perspective: the emperor's new clothes of epidemiology, an illustration using cancer mortality

Cohort analysis of cancer mortality in industrialized countries has led to the generally accepted conclusion that these populations have been exposed to increasing levels of carcinogenic influences. Age-specific cancer mortality rates in the United States from 1962 to 1988 were portrayed in both cross-sectional and cohort manners. Both representations are consistent with the Strehler-Mildvan modification of the Gompertz relationship between aging and mortality. These observations suggest that environmental cohort effects are not responsible for rising cancer mortality. The cohort mortality perspective in epidemiology is inherently biased due to the effects of competing mortality. Competing mortality, like environmental exposures, is applied in a cross-sectional manner across multiple birth cohorts. Competing mortality produces selective mortality, or differential survival, within a birth cohort. Differential survival, an underappreciated form of selection bias, alters the gene pool of surviving birth cohort members. Since cross-sectional competing mortality risks vary with age, the gene pool of different birth cohorts is not uniformly altered. Consequently, differences between birth cohorts in age-specific mortality rates with respect to a particular disorder, for example, cancer, do not specifically imply differences in environmental exposures.