Age, period, and cohort effects. Influences on differences between cross-sectional and longitudinal pulmonary function results.

Abstract

It is well known that lung function changes with age, increasing during childhood through most of adolescence and decreasing during most of adulthood; changes at different ages vary by population, place, and time. It is postulated, and fairly wellaccepted, that there are cohort effects in most populations, related mostly to period of birth (i.e., birth cohorts). This will change the relationship of lung function with age in the same population during different periods of time. (It is likely that there is a cohort effect on body/thoracic size that also is reflected in the age-cohort lung function relationships.) Age, cohort, and period effects have been evaluated previously (1-21). The primary purpose of this paper is to evaluate the influences of cohort (and period) on the age-lung function relationship. It is recognized in advance that period effects are the most difficult to evaluate because they occur in specific populations and subpopulations. The rationale for using longitudinal data for this purpose is that longitudinal data provide one with the opportunity to look at age and other effects on lung function in the same individuals over time, while cross-sectional age-lung function relationships are period-specific and are derived from multiple birth cohorts. Various models have been used (2, 4, 5, 8, 10-12, 15-20). Comparisons of single and multiple cross-sectional with longitudinal age-lung function relationships provide the primary basis for determining possible effects of cohorts and time periods in the same population, and also the basis for determining differential velocities/accelerations of growth and decline. Longitudinal data also provide the basis for determining the factors, often also related to cohorts (and possibly periods), that affect the age-lung function relationship. For example, smoking behavior/habits will vary by birth cohort, age, and period and will most likely affect lung function growth and decline for the different cohorts and during different periods. It is conceivable that interactions between age, body habitus, and/or risk factors also vary by cohort and period; these can be determined only by use of longitudinal data. Some of the risk factors are very period-specific. Thus, evaluations of longitudinal data will provide the basis for determining the cohort/time changes, their effects, and their implications. Wehave assumed that cohort effects are likelydue to the different life experiences and exposures in different cohorts. Cohort effects have been reflected clearly in age-specific morbidity rates of both tuberculosis and lung cancer. These age-cohort differences have been ascribed to different living conditions and life styles (e.g., nutrition, sanitation and socioeconomic conditions [SES], smoking) (22-24). Other risk factors that often affect lung function known to be different in the birth cohorts and periods include age-specific prevalence of airway obstructive diseases (ADD) and exposures (occupational and environmental). The