Age–Period–Cohort Distinctions

Age–period–cohort (APC) analysis has played a critical role in studying time-specific phenomena in demography, sociology, and epidemiology for the past 80 years. Broadly defined, APC analysis distinguishes three types of time-related variation in the phenomena of interest: age effects (variation associated with different age groups), period effects (variation over time periods that affect all age groups simultaneously), and cohort effects (changes across groups of individuals who experience an initial event such as birth during the same year or years).

The distinctions of age, period, and cohort effects are especially important in the context of health and aging. Age is associated with the biological process of aging internal to individuals and plays a role in the etiology of most diseases. In general, the incidence and prevalence of disease as well as mortality rates increase systematically with age. The considerable regularity of age variations across time and place reflects the developmental nature of true age changes. In contrast, period and cohort effects reflect the influences of external forces that operate in different ways. The distinction between these two components of temporal changes is often the focus of APC analysis.

Period effects may be due to major historical events such as world wars, economic crises, famine, and pandemics of infectious diseases that elevate the death rates across all ages. They may also arise with public health efforts and diffusion of medical technology innovations that reduce mortality rates for all ages. In addition, a unique period event (e.g., 1991 Kuwait oilfield fires, 2004 Christmas tidal wave, 1986 Chernobyl nuclear disaster) may induce a similar change in disease risk for all individuals alive at a point in time regardless of age. Cohort effects have been conceived as the essence of social change and are evident in many chronic diseases, cancer cites, and human mortality. A birth cohort moves through life together and encounters the same historical and social events at the same ages. Cohorts that experience different historical and social conditions differ in their exposure to socioeconomic, behavioral, and environmental risk factors. Cohort effects represent the effects of these factors that embark at the moment of exposure early in life and act persistently over time to produce health and mortality risk differences in specific cohorts.

In spite of its theoretical merits and conceptual relevance, APC analysis suffers from technical problems that restrict its widespread application in research. Limited analytic approaches produce ambiguous results, and researchers do not agree on methodological solutions to these problems. The fundamental question of determining whether the process under study is some combination of age, period, and cohort phenomena points to the necessity of statistically estimating and delineating the age, period, and cohort effects. Most current statistical APC analyses focus on aggregate population-level data. The data structure is usually in the form of rectangular age by period tables of [Page 21]occurrence/exposure rates of diseases or deaths. Table 1 illustrates this structure using mortality data for older U.S. males. In this table, age–year-specific death rates are tabulated in Age × Period arrays, with eight 5-year age groupings defining the rows and four 5-year periods defining the columns. There are 11 successive 10-year birth cohorts whose death rates fall in diagonal cells of the matrix. The oldest cohort, individuals age 95 years and older during the 1980–1984 period (born before 1885), is at the bottom left cell, the youngest cohort is at the top right cell, and cohorts born in between form the diagonals. The death rates of one such birth cohort are shown by the shaded cells.

...