Cohort Effects

Cohort effects are variations in factors such as health status or mortality that are attributed to the unique physical and social environment to which a cohort is exposed during its lifetime. A commonly used synonym is generation effects.

A cohort is defined as an aggregate of individuals within a specified population who experience the same event (e.g., birth) at the same time or within the same time interval, and who are observed over time. A cohort is usually identified by the event itself and by the time period during which the event occurs, and also, at least implicitly, by geographic location, that is, the physical environment. For example, the U.S. birth cohort of 2000 includes all persons born in the United States in the calendar year 2000. The time period that defines a cohort can be very short, or it can extend over several years; a decade is commonly used to identify a cohort.

The importance of taking cohort membership into consideration was first noted by demographers, and subsequently by other social and behavioral sciences, including epidemiologists. Cohort studies (and cohort analysis) are conducted to separate the effects due to cohort membership from those due to age or period. However, there is an inherent problem since cohort, age, and period are all related to time, and the interpretation of findings is subject to confounding. Since each of these time-related variables is dependent on the other two, attributing an observed [Page 208]effect to any one of them usually requires knowledge of biological, historical, or behavioral factors that might produce age, period, or cohort effects. Multiple regression techniques can help disentangle age, period, and cohort effects. However, the question remains whether cohort membership has any effect apart from age and period, since cohorts are defined by the sharing of age and period.

A classic example of the attempt to separate age, period, and cohort effects is Wade Hampton Frost's examination of tuberculosis mortality rates in Massachusetts between 1880 and 1930, which showed that apparent changes in the age distribution of tuberculosis mortality could be better interpreted as a decline in mortality over cohorts, rather than as changes in age-specific mortality over time. In a graph of mortality rates, with age along one axis and year of death along the other, the cohort effect can be seen by tracing the diagonal axis of the graph, which plots the mortality of each cohort as it ages over time.

A cohort effect was discerned in the pattern of deafness in New South Wales, Australia; prevalence was especially high in 1911 among those who were 10 to 14 years old, in 1921 among 20to 24-yearolds, and in 1931 among 30to 34-year-olds. The most likely explanation for this prevalence pattern was the 1899 measles epidemic in New South Wales, which resulted in congenital deafness among a large proportion of those born to women exposed to measles during pregnancy.