Age, Period, and Cohort Effects to Predict the Future of Despair

The recent discovery that mortality rates were increasing among less-educated middle-aged White men and women^1–3 is an important counterpart to studies about morbidity rates in younger sociodemographic subgroups,^4–6 which provide data on some of the presumable precursors to such “deaths of despair” for which there are interventions. Trends seem to indicate that increases in premature mortality from accidents (e.g., drug overdose) and suicide are increasing across adult age groups of Whites, but not among Blacks and Hispanics.⁷

INCONSISTENT OBSERVATIONS

However, measuring excess mortality by age and race cohorts tells only part of the story and informs predictions for the future only so much. In this issue of AJPH, Gaydosh et al. (p. 774) operationalize the concept of despair by measuring some of its proximal expressions among participants from the Add Health study (born 1974–1983), who were in their 30s and early 40s at the most recent wave of data collection in 2016–2017. Importantly, they stratify sociodemographic subgroups by race/ethnicity and gender, as well as by educational attainment and rurality, to provide more detailed subgroup trends than do other studies.^6,7 In so doing, Gaydosh et al. reveal findings that are sometimes inconsistent with analyses stratifying only by race and gender and inconsistent depending on the particular indicator of despair.

Specifically, even if there were some differences in prevalence between groups, age trends in depression symptoms, suicidal ideation, and to some extent heavy drinking and marijuana use were similar across sociodemographic subgroups. Gaydosh et al. measured despair with numerous indicators of mental health (depression, anxiety, suicidal ideation, and suicide) and substance use (heavy drinking, marijuana and illegal drug use, prescription opioid and other drug abuse). Using a wide range of indicators and comparisons, Gaydosh et al. provide detailed insights into various possible expressions of despair in this longitudinal cohort, when no direct measure of self-reported despair was available. These results run counter to those observed among an older cohort in other studies,^3,7 so any interpretations and future planning must consider the role of age, period, and cohort effects in these results. Table 1 provides a heuristic to illustrate the differences between age, period, and cohort effects.

TABLE 1—

Age, Period, and Cohort Effects

Year	Aged 0 Years	Aged 10 Years	Aged 20 Years	Aged 30 Years	Aged 40 Years	Aged 50 Years
1970	A	A	A	A	A	A
	P C
1980	P	C
1990	P		C
2000	P			C
2010	P				C
2020	P					C

Note. A = age effects; C = cohort effects; P = period effects.

AGE, PERIOD, AND COHORT

With the analysis of a single cohort by Gaydosh et al., it is not possible to disentangle age, period, and cohort effects. However, we can consider their role in these subgroup trends, as the authors themselves do. Conceptually, an age effect would argue that any observed changes in psychiatric symptoms and substance use over time would be from biological and social processes common to some groups of people. Therefore, the time trends observed in this Add Health Gen X cohort would be assumed to be the same for older and younger generations. As noted by the authors, the lack of consistently significantly higher observed despair for less-educated and rural Whites may be explained by an age effect—that is, perhaps the 30s are too young, and the threshold-level of despair does not really emerge until around age 50 years in this subgroup.

Losing one’s job in middle age, for example, can be much more stressful than at younger ages, because it is more difficult to be hired, to move, to gain new training, and to change career paths for those at that age than for people in their 30s. So the despair associated with a factory closing in a small town would hypothetically be more severe among those in their 50s than those in their 30s. With this explanation, it is possible that trends of increased morbidity and mortality-specific to low socioeconomic position (SEP) Whites have yet to emerge in the Add Health cohort.

Alternately, a period effect explanation would argue that historical context has different outcomes for individuals of the same age in different periods or, conversely, that a period may affect multiple birth cohorts similarly at a particular point in time. For example, higher morbidity and mortality among less-educated White men and women in midlife^2,7 may also be manifesting among their 30- and 40-year old counterparts, because of heightened social divisions, income inequality, decreased social mobility, and the availability of opioids and marijuana in recent years. With period effects, we cannot predict what will happen to Gen Xers as they age, as we do not know what the state of the period will be 10 to 20 years from now. However, the high prevalence of indicators of despair across sociodemographic subgroups shown by Gaydosh et al. indicates that early intervention may be important for addressing problems that could be compounded with aging among diverse sociodemographic groups.

AGE X PERIOD EFFECTS

Finally, in epidemiology, cohort effects are conceptualized as an interaction between age and period effects,⁶ meaning that development over the life course occurs differently for cohorts born and maturing in different periods. Consequently, less-educated Whites from Generation X may not follow the same patterns as their baby boomer counterparts. Perhaps the historical, political, economic, and racial histories that defined baby boomers will not apply to Gen Xers from the Add Health study when they reach the age of 50 to 60 years.

For example, baby boomers grew up under Jim Crow and during the Civil Rights and women’s rights movements. They reached adulthood during a period when macroeconomic shifts in labor markets and manufacturing and steep rises in income inequality were beginning and when increasing demand for a more well-educated workforce in the United States was established. Comparatively, the Gen X participants in the Add Health study grew up with more racial and gender equality and integration, more informed expectations about the professional opportunities that exist for those with less than a high school education, and higher average education levels. This could affect the levels of despair-related symptoms and behaviors among sociodemographic subgroups at different stages of their life-spans.

Such a cohort effect could explain the evidence provided by Gaydosh et al. that racial and SEP differences in despair are not observed consistently among Gen X participants in their 30s. Perhaps the “despair” that Gaydosh et al. describe as being ascribed to everything from the opioid epidemic to the 2016 presidential election is peculiar to less-educated Whites from the baby boomer generation and the context of future generations will lead to patterns defined not so starkly by race and SEP.

This study by Gaydosh et al. sets an important benchmark necessary to keep tracking the Add Health cohort over time, to determine the interplay between age, period, and cohort effects in this cohort and its sociodemographic subgroups. Addressing these multiple indicators of despair in diverse sociodemographic subgroups requires a public health approach merged with other macroeconomic and sociological policy changes.

REVERTING TO STIGMA AND CRIMINALIZATION

When researchers interpret research highlighting the need for mental health and substance use prevention and treatment among diverse groups beyond low SEP rural Whites, they must caution against a reversion to stigma and criminalization of drug use and mental illness. By expanding the net of despair to people of color, as public health professionals we must be careful in our messaging—we must ensure that we maintain a public health approach by focusing on prevention, while also leading treatment efforts and equitable provision of services to those in need. As public health professionals, it is our obligation to fight against the historical model of discriminatory legal practices toward drug use and mental illness that threaten to reemerge when these symptoms are associated with people of color. The hope is that we can take advantage of the current public empathetic feelings toward the plight of those affected by the opioid epidemic, even after research is published noting that the despair of people of color also needs to be addressed.